INHALTSVERZEICHNIS

Dirnbck, T., Dullinger, S. & Grabherr, G.:

A regional impact assessment of climate and land use change
on alpine vegetation.

Dirnbck, T., Dullinger, S., Kck, R. & Grabherr, G.:

Organic matter accumulation following Pinus mugo Turra invasion
into subalpine non-forest vegetation...

32

Dirnbck, T. & Dullinger, S.:

Disturbance and the predictability of plant species distribution
in alpine environments

50

Dirnbck, T., Dullinger, S., Greimler, J. & Grabherr, G.:

A resampling approach for evaluating effects of pasture abandonment
on subalpine plant species diversity...

70

Dullinger, S., Dirnbck, T. & Grabherr, G.:

Patterns of shrub invasion into high mountain grasslands of the
Northern Calcareous Alps (Austria)..

94

Dullinger, S., Dirnbck, T. & Grabherr, G.:

Tree-line shifts prone to climate change: Evaluating
Relative effects of temperature increase, dispersal and
invisibility by means of a plant spread model...

113

Kck, R., Hrtel, E., Holtermann, C., Hochbichler, E. & Hager, H.:
Soil moisture dynamics related to vegetation cover in the subalpine
zone of the Northeastern Calcareous Alps in Austria. Results of case
Studies in the Rax area..

137

Kck, R., Hrtel, E., Hochbichler, E., Hager, H.& Schnthaler, K.E.:
Monitoring hydrological processes in montane and subalpine karst
regions: comparison between different types of vegetation. Experimental
design, techniques and first results.

147

Beitrge bei Fachtagungen.

156

1
Dirnbck, T., Dullinger, S. & Grabherr, G.: A regional impact assessment
of climate and land use change on alpine vegetation.
Manuskript, zur Publikation akzeptiert in der Zeitschrift
Journal of Biogeography

A REGIONAL IMPACT ASSESSMENT OF CLIMATE AND LAND USE

CHANGE ON ALPINE VEGETATION

Thomas Dirnbck1, 2, *, Stefan Dullinger1, and Georg Grabherr1

1

Institute of Ecology and Conservation Biology, University of Vienna,

Althanstrae 14, A-1090 Vienna, Austria
2
Current address: Federal Environment Agency; Spittelauer Lnde 5, A-1090 Vienna, Austria; Tel.:
+43-1-31304-3442; Fax +43-1-31304-3700; E-mail: dirnboeck@ubavie.gv.at
* Corresponding author

ABSTRACT
Aim Assessing potential response of alpine plant species distribution to different future climatic and
land use scenarios.
Location Four mountain ranges totalling 150 km in the North-eastern Calcareous Alps of Austria.
Methods Ordinal regression models of 85 alpine plant species based on environmental constraints and
land use determining their abundance. Site conditions are simulated spatially using a GIS, a Digital
Terrain Model, meteorological station data and existing maps. Additionally, historical records were
investigated to derive data on time spans since pastures were abandoned. This was then used to assess
land use impacts on vegetation patterns in combination with climatic changes.
Results A regionalised GCM scenario for 2050 (+0.65C, -30 mm August precipitation) will only lead
to local loss of potential habitat for alpine plant species. More profound changes (+2C, -30 mm
August precipitation; +2C, -60 mm August precipitation) however, will bring about a severe
contraction of the alpine, non-forest zone, due to range expansion of the treeline conifer Pinus mugo
Turra and many alpine species will loose major parts of their habitat. Precipitation change
significantly influences predicted future habitat patterns, mostly by enhancing the general trend.
Maintenance of summer pastures facilitates the persistence of alpine plant species by providing
refuges, but existing pastures are too small in the area to effectively prevent the regional extinction
risk of alpine plant species.
Main conclusions The results support earlier hypotheses that alpine plant species on mountain ranges
with restricted habitat availability above the treeline will experience severe fragmentation and habitat
loss, but only if the mean annual temperature increases by 2C or more. Even in temperate alpine
regions it is important to consider precipitation in addition to temperature when climate impacts are to
be assessed. The maintenance of large summer farms may contribute to preventing the expected loss
of non-forest habitats for alpine plant species. Conceptual and technical shortcomings of static
equilibrium modeling limit the mechanistic understanding of the processes involved.
Keywords
Alpine plants, European Alps, habitat distribution model, ordinal regression, generalised linear model,
vegetation modeling.
Abbreviations
DD, temperature degree days; DEM, digital elevation model; G, geological unit; NCA, North-eastern
Calcareous Alps; RJ, solar radiation for July; RM, solar radiation for May; RS, solar radiation for
September; S, slope inclination; SC, soil cover type; SCD, snow cover duration; SWB, site water
balance; TPA, time since pasture abandonment; WET, wetness index
3

INTRODUCTION
High mountain ecosystems are considered particularly vulnerable to climate change (Beniston, 1994;
Grabherr et al., 1995; Beniston et al., 1996; Theurillat & Guisan, 2001). The European Alps
experienced a 2C increase in annual minimum temperatures during the 20th century, with a marked
rise since the early 1980ties (Beniston et al., 1997). Upward moving of alpine plants has occurred
(Grabherr et al.1994; Pauli et al., 2001), community composition has changed at high alpine sites
(Keller et al., 2000), and treeline species have responded to climate warming by invasion of the alpine
zone or increased growth rates during the last decades (Gindl, 1999; Paulsen et al., 2000; Motta &
Nola, 2001). Predictions suggest that Austria will experience a further annual mean temperature
increase of 0.8C and 7% annual precipitation anomalies by 2050 compared to the average of the
period 1961-1995 (Lexer et al., 2002).
We present here a study from the North-eastern Calcareous Alps (NCA), which are part of the
catchments for high quality water of the Austrian capital, Vienna, supporting 1.7 million inhabitants.
The mountain ranges are karstic, exhibiting fast drainage. Considerations of plant cover and soils are
therefore crucial for water protection (Dirnbck & Grabherr, 2000). Apart from the potential of
climate changes for direct impact on the hydrological cycle (Loiciga et al., 2000), they will very
probably also change vegetation properties. Nevertheless, in which way and to what extent such
changes will occur remains unclear as yet. An assessment of impacts of climate and land use changes
on the vegetation was therefore of high priority as a prerequisite for future strategies of water
management. From a management perspective it is particularly important to evaluate changes which
might occur at the local scale. In addition to climatic influences, other potential contributing factors
also have to be taken into account of which land use may be most important. Historical human impacts
on the Alp's vegetation have limited most of the approaches attempting to detect the signal solely
attributable to recent climatic changes (Tappeiner et al., 1998; Bolliger et al., 2000; Carcaillet & Brun,
2000; Didier, 2001; Motta & Nola, 2001). In the study area, traditional summer pasturing reaches back
to at least the 16th century shaping subalpine and alpine vegetation (Dullinger et al. 2002). Shrublands
and forests were, like in many other parts of the Alps, clear-cut or burned and the resulting pastures
were grazed in summer. Since pasture abandonment, a long term trend during the last 150 years, is
omnipresent, secondary succession is commonplace and interacts with climate induced perturbation of
plant communities (Dullinger et al., 2002).
To assess vegetation response to different climatic and land use scenarios, distribution models of the
85 most frequent alpine plant species of the study area are derived based on environmental constraints
controlling their abundance. Site conditions are spatially simulated using GIS and represent resource
gradients (site water balance, solar radiation), gradients of direct physiological importance
(temperature, snow cover duration), and indirect variables which reflect soil properties (topographical
indices; geological unit) (see also Austin & Smith, 1989; Guisan & Zimmermann, 2000). Historical
records were investigated to derive data on time spans since pastures became abandoned (Dullinger et
al. 2002). This was then used to assess land use impacts on vegetation patterns in combination with
climatic changes. Static equilibrium models, as used here, were applied in different parts of the world
and various biomes to assess potential impacts of a changing climate (Austin, 1992; Box et al., 1993;
1999; Brzeziecki et al., 1995; Huntley et al., 1995; Gottfried et al., 1998, 1999; Guisan et al., 1998;
Stersdal et al., 1998; Bolliger et al., 2000; Duckworth et al., 2000; Guisan & Theurillat, 2000;
Richardson et al., 2000; Aber et al., 2001; Hansen et al., 2001; Lexer et al., 2002). The present study
provides several improvements over most of these models: (1) Precipitation decrease is incorporated
in addition to temperature increase. (2) Climate scenarios are based on regionalised GCM scenarios.
(3) Land use impacts are investigated in combination with anticipated climatic changes. (4) Species
abundance was used instead of pure presence- absence data. (5) In contrast climate envelop
4

approaches, we aimed to incorporate a full as possible suite of abiotic environmental controls limiting
plant species' distributions.
Climate and land use change scenarios
Climate change scenarios are derived from simulation outputs of the global circulation model
ECHAM4 (Roecker et al., 1996) which were recently downscaled for Austria for the purpose of risk
assessment of forests prone to climatic changes (Lexer et al., 2001, 2002). The simulation is based on
the greenhouse emission scenario IS92a (Houghton et al., 1990). Statistical downscaling techniques
were applied to the GCM scenario which, compared to the average of 1961 1995, resulted in 0.8C
mean annual temperature increase and 7% annual precipitation anomalies for Austria in the year
2050 (average of 2035 2065). The temperature increase was highest during the warm season and in
the northern and westernmost part of Austria, which is in agreement with 20th century trends (Beniston
et al., 1997; Weber et al. 1997). Most precipitation changes are expected in the winter season and the
northern front range of the Alps, where precipitation is predicted to decrease (Lexer et al., 2001). The
NCA should therefore experience major climatic changes during the next decades. For the study area
0.65C increase of the annual temperature and 200 mm reduction in annual precipitation might be
expected by the year 2050 in comparison with the average of 1961-1995 (Lexer et al., 2001). In
addition to the so-called 2050 scenario we assessed two further ones in order to uncover potential
impacts of more pronounced changes. The 2K scenario is based on a 2C increase of the annual mean
temperature, and the 2Kplus scenario assumes a further doubling of the precipitation decrease (Table
1).
Long term trends of snow cover duration of higher altitudes did not show the same decreasing values
as was observed for lower and middle altitudes (Beniston, 1997). Obviously, redistribution patterns
override simple altitudinal controls on snow fall and depletion (Friedel, 1961; Krner, 1999).
Therefore, no snow cover duration change was included in the scenarios.
Land use, in particular logging and grazing in the high mountain areas, influences a plant's distribution
as climate does. Human modified landscapes may pose severe constraints on the re-adjustment of
vegetation patterns subject to altered climatic conditions, in particular due to fragmentation of native
vegetation and altered disturbance regimes (Huntley, 1990; Pitelka & Plant Migration Working Group,
1997; Box et al., 1999; Duckworth et al., 2000; Hobbs, 2000; Hansen et al., 2001; McCarty, 2001). If
climatic conditions change in the NCA , it can be anticipated that land use regimes will interact with
these changes, by either facilitating species resistant to grazing, or by inhibiting the same species
during succession after abandonment (Dullinger et al., 2002). This process can be regarded as
relatively short term (a few decades to several hundreds of years (Wildi & Schtz, 2000)). Our study,
however, intrinsically applies equilibrium models, which are capable of assessing potential long-term
changes, assuming stable habitat conditions which lead to steady state communities. We therefore
restrict the combined scenarios exclusively to two specific land use situations: (1) where all abandoned
sites have reached a stable state and all presently used remain used (LU3), (2) where the whole study
area has become abandoned and all sites have reached a stable state (LU4). Climate change scenarios
are assessed for both land use regimes (Table 1). Comparison is accomplished by using a baseline
scenario with present climate conditions and the LU3 land use status.

METHODS
Study area
The study area covers the subalpine and alpine zone of four neighbouring mountain ranges of the
North-eastern Calcareous Alps in Austria comprising about 150 km area (Mt. Hochschwab, Mt.
5

Schneealpe, Mt. Rax and Mt. Schneeberg, 15 E to 16 E and 4730 N to 4750 N, see Fig. 1).
Summits vary between 1900 and 2300 m a.s.l. The mountain system consists of mesozoic limestone
and dolomites and is characterised by displaced plateaus of different altitudes (Fig. 1(c)). Climatical
conditions are temperate humid. Mean annual temperature approximates 6-8 C in the valleys
decreasing to about 0-2 C in the summit region. Annual precipitation averages 700 mm (valleys) and
1500-2500 mm (summits). Summer pasturing (June to September) in the area dates back at least to the
16th century. Except for rock faces, debris cones and very steep slopes most of the study area has been
historically influenced by livestock grazing. Since the middle of the 19th century grazing intensity has
decreased and much former pastureland has become abandoned (Dullinger et al. 2002). Approximately
30% of the study area is still pastured by free-ranging cattle at an intensity of about 0.5 cattle per
hectar (S. Dullinger, unpublished).
Norway spruce (Picea abies (L.) H. Karsten), sometimes with European larch (Larix decidua Mill.),
dominates from around 1400 m a.s.l. up to a belt of prostrate pine (Pinus mugo Turra). The upper limit
of Pinus mugo is between 1800 and 1900 m a.s.l. (Zukrigl, 1973; Fig. 1(c)). In fact, the subalpine belt
is a mosaic of woody and non-woody vegetation today with spruce and spruce-larch forests rarely
occurring above 1600 m a.s.l. Non-woody vegetation below treeline mainly consists of different kinds
of pastures and natural grasslands with the latter covering disturbed sites like avalanche paths and
exposed ridges. Less frequent, tall herb communities (Adenostyles alliariae (Gouan) Kern., Aconitum
napellus L.) and fens (Carex nigra (L.) Reichard, C. rostrata Stokes) occur. Above treeline, natural
grasslands are dominating with a gradual switch from prevailing Carex sempervirens Vill. to Carex
firma Host grasslands with increasing altitude. Additionally, rock faces, scree and snowbeds are
considerably widespread.
Modeling framework
Three major assumptions cohere with the applied modeling approach: (1) The abiotic environment has
to be the major agent controlling species distribution and abundance together with post-grazing
succession. (2) The models are calibrated using field data, and thus comprise any competitive
constraint a species may force upon or experience from its neighbour. (3) The speed of plant migration
is consistent with that of climate changes so that plant communities are in a permanent equilibrium
with their environment. Admittedly, the third assumption is unrealistic in the mid-term. We know
from observations that migration of alpine plants lag behind climate warming (Grabherr et al., 1994;
Pauli et al., 1996) and that dispersal mechanisms play a crucial role (Theurillat & Guisan, 2001;
Kullman, 2002). As a consequence of the modeling framework applied, i.e. equilibrium models of the
realised niche of plants, we solely depict the spatial distribution of potential habitats in the long term
without severe community deterioration (see Guisan & Zimmermann, 2000).
Vegetation data
The calibration data set is based on a stratified sampling approach, using all combinations of
categorised environmental variables and time of pasture abandonment (see below) as strata. Each
stratum had to be sampled by at least one plot, depending on total area of the stratum. We included a
large set of existing vegetation records (totally 780) and did supplementary records for strata not
covered by these (totally 179). For the latter the position within each stratum was randomly selected
and localised in the field by means of a GPS (Garmin E-Trex, mean root mean square error: 7.5 m).
The existing records had been marked on an infrared or black and white orthophoto (scale 1:5000
1:10000) and were digitised thereafter (estimated geographical accuracy is 10 m). The existing data
was originally collected in the course of mapping the vegetation of the area (Greimler & Dirnbck,
1996; Dirnbck & Greimler, 1997; Dirnbck et al., 1998, 1999; Dullinger et al., 2001). Additionally,
6

57 records stem from several transects at the treeline which cover the lower boundary of Pinus mugo.
The final data comprises records between 1200 m a.s.l. and the summits. All records were collected
between 1994 and 2001 by the authors. Data were collected according to standard phytosociological
techniques: All plant species present on the plot were recorded together with a cover-abundance value
estimated on a 7-level ordinal scale (Braun-Blanquet, 1964). Plot size was approx. 30 m for
grasslands, rock, scree and snowbeds, 150 m for Pinus mugo-krummholz and 300 m for forests.
From these data 85 most frequent (>100 times present) and some further species were selected for
modeling. These are mostly typical alpine species occurring above treeline but also subalpine species
of pastures and avalanche tracks.
Abiotic habitat variables
Several environmental variables were spatially derived using map data, a digital elevation model
(DEM, 20 m resolution, Austrian National Mapping Agency), and meteorological station data. The
DEM was converted into a grid using the triangulated irregular network model, with addition of
breaklines, and bivariate quintic interpolation with the standard methods incorporated in Arc/Info 7.1.
Daily net radiation for 15th May (RM), 15th July (RJ) and 15th September (RS) was derived applying
SOLARFLUX (Rich et al., 1995) under clear sky conditions with the DEM. The program accounts for
relief shading and topographic position. For the whole area atmospheric transmissivity was set to 0.8.
Radiation was calculated at hourly intervals which were then summed for the whole day.
Temperature degree days (DD; daily mean temperature > 0C) were approximated using three
meteorological stations from Mt. Schneealpe (770 m, 820 m and 1740 m a.s.l.; Steinkellner
unpublished data). Half hourly measures were averaged for each day and summed up for the year from
1995 to 1999 (other years were either not measured or were not measured consistently enough for the
whole year). DD show a strictly linear relationship with altitude and decreases with 5.6 days per 100
m (R = 0.9, p-value < 0.001). DD were offset with an increasing base temperature from west to east.
The base temperature was linearly regressed on data from 17 standard meteorological stations located
within and close to the study area (R=0.6, p < 0.05).
Site water balance (SWB) was calculated for August since we assumed that water deficiency does not
occur before the late summer. SWB has been shown to be significant for plant distribution in the area
(Dirnbck et al., 2001). The SWB was approximated by the precipitation less the potential
evapotranspiration (Turc, 1961) and was calculated as an average of the 1st and 15th August, and the 1st
September. Daily net radiation was calculated applying SOLARFLUX (Rich et al., 1995) under the
same settings as described above. Monthly mean temperature was input in the equation for potential
evapotranspiration and calculated applying linear regression on altitude (0.8C per 100m altitudinal
range; R = 0.54; p < 0.001) with data from the above three meteorological stations. Mean temperature
in August does not show a significant longitudinal change within the study area. Mean daily
precipitation for August was calculated using geographic position in addition to altitude because
eastward precipitation decrease is known for the area (Wakonigg, 1978). However, systematic
measurement error, topographical effects, and localised storms probably override general patterns and
may be responsible for the considerably low fit of the model (R = 0.1, p = 0.06). The regression
slopes for monthly precipitation are 1.5 mm per 100 m elevation, and 6 mm per 10 longitudinal
kilometres.
We used 11 SPOT scenes, available for a third of the area, classified into snow cover classes (acquired
1998, 1999, 2000 from February to June, each at about 10:45 am; 20 m resolution; Jansa et al., 2000),
the above mentioned radiation model (Rich et al., 1995), a wind field model (Ross et al., 1988;
Bachmann, 1998), elevation (as a compound surrogate for temperature and precipitation; from the
DEM), relief indices, and plant cover type (krummholz versus grasslands, screes and rocks) from
7

existing vegetation maps to calculate snow cover duration (SCD). The dates when the satellite images
were acquired were used to capture seasonal accumulation and ablation processes. Radiation melting
influence is approximated by calculating the cumulative sum of incoming shortwave radiation
(SOLARFLUX; Rich et al., 1995). Fallen snow is subject to redistribution by wind, avalanches and
sloughing (Liston & Sturm, 1998). These topographical effects were simulated with state variables. To
account for downhill snow transport we used the slope of the surface and a topographic index (steady
state wetness index applying a multiple flow algorithm using Tapes-G; Gallant & Wilson, 1996). This
index was successfully used to depict accumulation areas in relief depressions versus depletion areas
at ridges (Hartmann et al., 1999). Wind redistribution was supposed to be crucial for overall snow
patterns (Jansa et al., 2000; Tappeiner et al., 2001). The study area is exposed to north-westerly winds
and snow is blown from the plateau ridges in nearby leeward slopes and gullies. We used the
diagnostic wind field model NUATMOS (Version 5N, 07/31/91; Ross et al., 1988, integrated in a GIS
by Bachmann, 1998) to derive near surface wind velocities. Snow distribution is inherently
discontinuous, exhibiting non-linear relationships with climatic gradients (Tappeiner et al., 2001). In
order to best account for such a pattern we used a classification tree (Breiman, 1993; Balk & Elder,
2000). A tree is grown by recursively partitioning a training sample, such as to minimise model
deviance in the dependent variable. A binary classification tree was built using random samples
(0.02% of all pixels of each snow cover class, totally 9439 samples) of the 11 SPOT scenes (only full
snow and no snow classes were used for the analysis) and its respective radiation, wind, and
topographic values. The final tree had 20 terminal nodes and a misclassification error rate of 8%. All
variables show a significant influence on snow patterns; julian day is the most important variable,
followed by the topographic accumulation index, elevation and windspeed. In order to facilitate the
extrapolation of snow patterns to the whole study area we calculated all spatial descriptor variables for
the entire region. Following the tree from its origin to a particular terminal node all node probabilities
determine if a cell is covered by snow or not. Therefore, we calculated mean node probability (all
probabilities divided by node number) as the snow cover measure.
A detailed geological map (Geological Survey of Austria, unpublished) was categorised into 5
geological units (limestone, dolomite, clayey weathering carbonates, relict loam, carbonate debris) and
wetland (G).
Soils are important determinants for alpine plant growth as they control water supply and nutrient
status (Krner, 1999). The gross soil distribution was derived from interpretation of aerial
orthophotographs. We delineated bare or almost bare areas (rocks, screes, eroded sites), areas covered
by a closed non-forest vegetation with shallow mineral soils, and forested (krummholz and forests)
areas where mineral soils are covered by organic layers (SC). It has to be mentioned that these three
categories also reflect plant cover and may therefore be interpreted as a gradient of competitive
strength. In particular the forested sites pose significant limitation for alpine and other shade intolerant
non-forest species. Topographically driven erosion and accumulation, mediated by water runoff, was
supposed to be the most important factor controlling soil properties at finer scales. In the study area
relict (but also aeolian), siliceous sediments are widespread and distributed according to topography
(Solar, 1963). Thus, we applied a surface runoff model to account for these processes (TAPES-G,
Gallant & Wilson, 1996) The wetness index (WET) describes the spatial distribution of zones of
saturation as well as runoff generation and proved to have potential use in predicting soil properties
that control plant distribution (Moore et al., 1993; Gessler et al., 1995; McKenzie & Ryan, 1999).
Additionally to WET the slope inclination (S) was used.

Pasture abandonment (TPA)

Spatial land use information was derived from catastral maps and a suite of documents referring to the
land use status of individual parcels at different times (Dullinger et al., 2002). As delineation of
parcels in high mountain areas is usually very coarse we refined their boundaries by adjusting them to
landscape topography in that we excluded all inaccessible rock faces and scree areas from actually or
historically pastured areas. From these historical documents we derived dates of abandonment as
exactly as possible. In some cases personal communication with land owners provided additional
information. Nevertheless, exact dates of abandonment were not available for a considerable part of
historical summer farms. To achieve uniform precision of the land use data for the whole study area
we thus categorised time since abandonment using the six main historical documents as reference
points and assigning each parcel the date of its last documented use as a pasture. In statistical analyses
we used the number of years between the dates of documented pasture uses and the year 2000. Areas
which have never been used were assigned a value of 1000 assuming this period to be sufficient to
completely eliminate all remnants of former pasturing.
Statistical analysis
We derived ordinal logistic regression models (Proportional odds models) using cover-abundance
values of plant species as the response and environmental variables and successional time as the
predictors. Cover-abundance values were categorised into five classes (0: absent, 1: present-1%, 2: 110%, 3: 10-50%, 4: 50-100%). Proportional odds models are based on the cumulative probabilities of
the classes, stemming from successive logistic regression models (McCullagh & Nelder, 1991; Guisan
& Harrell, 2000; Harrell, 2001). Linear functions as well as restricted cubic splines (piecewise third
order polynomials with linear restrictions in the lower and upper tails) with not more than 4 knots
were fitted to the data using the Design library of Splus2000 (Azola & Harrell, 2001). To circumvent
predictor co-linearity we added the three radiation variables separately to the model and used only the
most significant one in the final model. Model and predictor significance were obtained from Wald
test statistic assuming a Chi-square distribution with 1 d.f. (Harrell, 2001). Full models were reduced
by backward elimination, knot reduction where cubic splines were used, and linearisation respectively
(p-value < 0.05). Double linear interactions of DD with all other continuous variables were taken into
account.
Model validation was obtained by resampling from the calibration data (1000 resamples with
replacement) using a bootstrap method (validate function of Azola & Harrell, 2001) to get bias
corrected values of the generalised R (Nagelkerke, 1991; quoted as RN thereafter), judging the
predictive strength of the model, and Somers concordance index Dxy (Harrell, 2001) which is based
on the Wilcoxon-Mann-Whitney two-sample rank test. Dxy takes values between 0 and 1 depicting
totally random versus perfectly discriminating models, respectively. As an additional validation
measure, the maximum absolute error in prediction probability (Emax) was calculated according to
Efron (1983) (see also Harrell, 2001). In order to represent all ordinal response values in each
bootstrap resample the data were stratified by the species abundance. Maps of abundance of each
species were drawn within the GIS summing up probabilities of the inverse logistic link function of
the linear predictors of each ordinal class (Guisan & Harrell, 2000).
Pinus mugo distribution which served also as an indicator of forest soils and therefore as a predictor
for all other species, was treated differently. The land use maps do not reflect the abundance of the
Pinus, which is often due to selective logging or scattered re-invasion in former pastures, leading to a
patchy distribution pattern on a much finer scale (Dullinger, et al. 2002). Consequently, for Pinus
mugo presence or absence is more meaningful for habitat preferences than abundance. Probability of
occurrence was modelled using a binary logistic model (McCullagh & Nelder, 1991; Harrell, 2001).
9

The calibration data used was restricted to those samples which were never used for cattle grazing but
Pinus mugo is absent and those where Pinus mugo is present (n = 477). This procedure guaranteed an
appropriate fit of environmental determinants. The distribution on presently used farmland was kept
constant in the baseline and all LU3 scenarios. In order to circumvent the likely case that Pinus mugo
will be out-competed by subalpine trees (particularly Picea abies) at its lower limit, all comparisons
were restricted to above 1600 m a.s.l.. The model was validated using the same concordance and
discrimination values as for all other species models but by predicting on independent data (totally 809
samples), randomly derived from distribution maps of Pinus mugo comprising the entire area. A
threshold probability, where presence and absence separate was calibrated using the highest Chisquare Test result applied on binary contingency tables from successive prediction thresholds of 0.1,
0.2, 0.3, ... 0.9, resulting to 0.4.
Climate and land use scenarios
An increasing annual temperature of 0.65C (Lexer et al., 2001) corresponds to 8 DD. This is
particularly conservative since minimum temperatures increased above-average during the last
decades in the Alps (Beniston et al., 1997; Weber et al., 1997), and mean temperature may therefore
underestimate DD. According to Lexer et al. (2001) monthly summer temperature will increase by
approximately 0.9 C, and monthly summer precipitation will decrease by 50-30 mm. From the
1980ties onward precipitation amounts in the Alps showed only a modest decrease (Beniston et al.,
1997). Therefore, the lower estimate (30 mm) was used to re-calculate SWB in August for the 2050
scenario. This amounts to 15% (higher elevations and western part) to 30% (lower elevation and
eastern part) less precipitation in August.
The soil cover map was re-adjusted using the predicted Pinus mugo distribution according to the
scenarios. All sites once occupied by either Pinus mugo or subalpine forests (Picea abies, Larix
decidua), from the present distribution to all subsequent scenarios, fell in the forest soil class in the
respective scenario. The two other soil categories were kept constant, simply because no consistent
procedure was at hand. All calculations were restricted to above 1600 m a.s.l., i.e. the subalpine and
alpine area.

RESULTS
Pinus mugo distribution
The environmental factors included in the logistic regression model for the distribution of the
subalpine coniferous shrub Pinus mugo explain 55% of the deviance (Table 2). The model achieves
high predictive accuracy on an independent dataset (RN = 0.68; Dxy = 0.86). Probability of occurrence
is under- and overestimated, but mostly in the lower as well as the upper ranges of probability
estimates, and thus ensures good discrimination between presence and absence (Fig. 2). DD is most
important in explaining occurrence followed by S, G, RS and SCD (Table 2). Pinus mugo shows an
unimodal distribution along the DD gradient with a sharp decline at about 220 DD, and it prefers
intermediate snow cover duration.
Models of alpine plant species
Predictive discrimination (Dxy), as obtained by bootstrap resampling from the calibration data set, is
0.57 on average and RN is 0.25. Poa alpina L. shows the lowest Dxy (0.23) and goodness of fit (RN =
0.09), Helictotrichon parlatorei (Woods) Pilg. the best (Dxy = 0.87; RN = 0.4). Maximal prediction
error of probabilities lies between 0.003 and 0.09 with an average of 0.03 (Fig. 3). All environmental
variables are significant predictors for plant species distribution and abundance, nevertheless, different
10

species comprise different suites of variables. DD is significant (p < 0.05) in almost all models, mostly
with highest predictive power. S is the next most consistent and significant contributor, but shows
relatively low Chi-square values. In contrast, RS is only significant in 45% of the models but its
contribution to variance explanation is comparatively high. SWB comes closely after SC, followed by
G. The latter shows relatively low Chi-square values. Significant in less than 30% of the models is the
TPA, SCD, WET and RJ and RM (Fig. 4)
Potential habitat changes
Pinus mugo expansion due to climate and land use change leads to a 25% loss of non-forest habitat
assuming the 2050 scenario and 48% assuming the K2 and K2plus scenario, if actual pastures remain.
If pastures are eliminated, 42% and 64% will potentially be lost. Those populations of alpine species
of mountains with particularly low summits may experience severe fragmentation (Figs. 5 and 6).
Additionally, land use may significantly reduce the decreasing trend stemming from climate change if
only non-forest area is considered. For lower mountain ranges most of the alpine habitat has already
been lost assuming the 2050 climate scenario, whereas at the higher M. Hochschwab a further decline
is elicited by the model (Fig. 5).
Fig. 7 shows the distribution of relative habitat change (increase and decrease) of all investigated plant
species (except for Pinus mugo). Decrease and increase of suitable habitats occur for all climate
scenarios and with and without maintenance of the summer pastures. Overall, habitat loss
predominates. Under moderate climate change (2050 scenario) 58 species loose habitat, 27 gain
habitat; taking pasture abandonment into account 63 lose and only 22 gain habitat. Severe habitat loss
(i.e. > 50% of its former LU3 area) experience 31 versus 42 species when all pastures become
abandoned. In both of these scenarios losers and winners are quite evenly distributed over the
frequency classes (Fig. 7). This changes severely for the K2 and the K2plus scenario, where losers
clearly predominate. The K2 and K2plus scenario show an almost total habitat loss for more than 40%
of the species when land use systems remain constant and about 50% when all pastures become
abandoned. Comparing K2 and K2plus reveals that further precipitation decrease shifts those species
which already lost habitat in K2 to the extreme side and fewer species gain. However, some species
show reverse trends when further precipitation is considered (see Fig. 8: Deschampsia cespitosa (L.)
P. B. and Silene acaulis (L.) Jacq.). The most extreme deterioration occur when climate changes
strengthen in combination with pasture abandonment. Then, typical pasture species (e.g. Crepis aurea
(L.) Cass., Trifolium pratensis L.), as well as alpine species (e.g. Carex firma, Carex sempervirens,
Silene acaulis) decline. Consequently, the K2plus scenario with total abandonment cause 65 species to
decrease, of which almost all (61 species) lose more than 50% of their actual habitat, and only 20
species will potentially gain additional habitat. Climate change impacts on potential plant distribution
are the more important compared to possible effects of pasture abandonment the more extreme
changes are.
Fig. 8 illustrates that change in abundance is consistent with habitat change, i.e., for those species
which win habitat also abundance increases (Helictotrichon parlatorei (Woods) Pilg.) and vice versa
(Carex firma, Silene acaulis (L.) Jacqu.).
Overall sensitivity
We calculated overall sensitivity of a species as the mean relative area change of all abundance classes
(Table 3). Winners and losers exist in all sensitivity classes. The bulk of species are categorised as
moderately sensitive, however, the species involved may lose almost all of their potential habitat in
singly scenarios. Overall sensitivity of species to climatic and land use changes have to be interpreted
cautiously because sensitivity and model fit are correlated to some extent (e.g., Pearson moment
11

product correlation for Somer's Dxy and Sensitivity shows p = 0.008). Plant species assembling present
communities do not at all respond equally, some may be very sensible to climate and land use change,
others are not. Table 3 also shows that decreasing or increasing trend is not consistent in all
consecutive scenarios. Some exhibit increase in the 2050 scenarios and a decline thereafter (e.g.,
Crepis aurea, Carex capillaris L., Salix alpina Scop.), whereas others decrease under moderate
climate change and increase when pronounced changes are assumed (e.g., Carex ferruginea Scop.,
Scabiosa lucida Vill.) (see also Figs. 7 and 8)

DISCUSSION
Although temperature-limited environments like alpine regions are thought to be particularly
vulnerable to climate change (Beniston, 1994; Grabherr et al., 1995; Beniston et al., 1996; Theurillat
& Guisan, 2001) different results and conclusions were drawn from recent studies on climate change
impacts on alpine plant's distributions. On one side of the extreme are re-visitations of historical sites,
which clearly show that alpine and nival species have responded to climate warming by migrating to
higher positions (Grabherr et al., 1994; Pauli et al., 2001), as did treeline species (Meshinev et al.,
2000; Motta & Nola, 2001; Kullman, 2002). Comparable results, i.e. major re-adjustment of alpine
vegetation patterns, were derived from modeling approaches (Gottfried et al., 1998, 1999; Guisan &
Theurillat, 2000). However, several studies found stasis or insignificant responses to observed
temperature increases. These include some treeline observations on the one hand (Lavoie & Payette,
1994; Paulsen et al., 2000; Cullen et al., 2001) and population studies, either experimental or long
term observations, on the other hand (Molau, 1997; Arft et al., 1999; Totland, 1999; Suzuki & Kudo,
2000; Diemer, 2002). At least for the European Alps Theurillat & Guisan (2001) concluded that,
although alpine vegetation may tolerate an increase of 1-2C of mean air temperature, in the case of
more pronounced increases profound changes may be expected.
Climate and land use change scenarios
In our study, the severe habitat loss of many species is linked to the peculiar plateau shaped
topography of the NCA. The plateau slopes are covered mostly by forests and the treeline is positioned
right where the plateau flattens. The alpine vegetation belt ranges over only a few hundred altitudinal
metres and the alpine flora of each of these mountains is isolated (Fig. 1(c)). Severe loss of alpine
diversity and fragmentation of plant populations due to climate warming were expected for
comparable high mountain systems worldwide (Grabherr et al., 1995; Stersdal et al., 1998; Theurillat
& Guisan, 2001). The majority of habitat decline of alpine plants may be expected to be caused by the
expansion of Pinus mugo into the alpine zone (Figs. 5 and 6). Particularly threatened are moderate
carbonatic slopes with intermediate snow cover duration, and these are the sites which comprise the
major habitats of alpine grasslands today. Table 2 and Fig. 2 illustrate the strong environmental
constraints on the distribution of Pinus mugo. Further evidence exists that growth and fecundity of
Pinus mugo is controlled mostly by temperature (S. Dullinger, unpublished). Nevertheless, recruitment
rates in grasslands strongly depend on the density and canopy height of the community to be invaded.
In particular some pasture grasslands inhibit seedling establishment. Reconstructed postglacial tree
species fluctuations (Kral, 1970) at a nearby mountain range revealed that the treeline was situated
about 400 m higher in the Subboreal (ca. 2950 BP) than today, which is in the range of Holocene
fluctuations found in other parts of the Alps (e.g., Tinner et al., 1996; Carcaillet & Brun, 2000). Pinus
mugo was then much less dominating in the upper subalpine zone, experiencing harsh suppression by
other conifers (e.g., Pinus cembra L.) (Kral, 1970). As a consequence of the uneven invasion process
and underpinned by the paleoecological evidence available, spatio-temporally diverse patterns will
likely evolve which are not exclusively determined by re-adjusting to environmental changes. In the
12

long run, however, and focusing on the gross trend which is elicited by the environmental change
model of Pinus mugo, alpine, non-forest areas of the NCA may shrink considerably if temperature
changes up to 2C and traditional land use declines (Figs. 5 and 6).
When considering the overall impacts, climate change and land use decline may have on alpine plant
species, we see that effects of moderate temperature increases and precipitation decreases are rather
weak, with species gaining and others losing potential habitat (Fig. 7, Table 3). Climate changes
according to the 2050 scenario (+0.65C; -30 mm in August) will cause local loss of potential habitats
for alpine species but will, according to our results, not deteriorate major vegetation patterns and
diversity of the NCA. The cessation of summer farming has been shown to cause a significant long
term decline of plant species diversity at the landscape scale (Dullinger et al., 2002). In combination
with climate warming and drying, the maintenance of pasture systems is important insofar as they
provide refuges for alpine species (illustrated in Figs. 5 and 7, and Table 3; see also Theurillat &
Guisan, 2001). If climate changes more drastically (K2 and K2plus scenarios), actual pastured area is
too small to significantly withhold the severe expected habitat loss of alpine plant species.
Water availability can co-determine spatial distribution patterns leading to species specific responses
under an assumed precipitation decrease. Such constraints are surprising since water availability is
generally thought to have little impact on the vegetation in the humid high elevation environments of
the European Alps (Krner, 1999; but see Guisan et al., 1998). However, water availability can be
significant for alpine plant growth (Walker et al., 1994) or soil formation (Bryant et al., 1997). In the
study area E-W moisture gradients were successfully applied to explain local scale species distribution
pattern (Dirnbck et al., 2001). Thus, water may well matter, particularly in the drier parts of the Alps,
like the easternmost NCA and expected future precipitation decreases may not be restricted to lowland
forests (see e.g. Lexer et al., 2002 for Austrian forests) but also to alpine vegetation of temperate
regions. Overall, a precipitation decrease enhances warming impacts, but some species exhibit reverse
trends (Fig. 8).
Unexplained distribution patterns
Considerable variation of present distribution patterns of alpine plants remained unexplained using
environmental variables as predictors (Fig. 3). Besides technical difficulties to correctly simulate
climate and other environmental factors across landscapes, microtopographic heterogeneity is assumed
to be responsible for such a weak predictability (Guisan & Theurillat, 2000). The suitability of the
Pinus mugo-model, the only taller shrub investigated (illustrated in Table 2 and Fig. 2), points to the
appropriateness of the resolution to the scale where this species acquires resources and is exposed to
environmental stressors. In contrast, the deviation of an alpine plants climate from that measured by
common meteorological stations is well known (Geiger, 1965; Krner, 1999). Micro-climates
significantly affect regeneration, species distribution and richness (Moir et al., 1999; Grytnes, 2000;
Carrer & Carlo, 2001; Erschbamer et al., 2001) and have been shown to differently influence plant
community response to regional climate change (Gavin & Brubakker, 1999). Consequently, microrelief niches enable plant species to occur where they potentially would be missing from the viewpoint
of mesoscale environment, and, in turn, some are missing at fine scales where they potentially occur at
broader scales. Micro niches are captured to such an extent in the models that environmental
conditions at the mesoscale are probably linked to some extent to conditions at finer scales.
Nevertheless, patchiness on a broad range of scales is fundamental for population dynamics,
community organisation and function (e.g., Levin, 1992), thus, predictions about future species
distributions remain uncertain so far.
Interestingly, a study of Gottfried et al. (1998, 1999) showed, that a considerable refinement of the
DEM to even one metre resolution, does not substantially enhance model accuracy. Although they
13

used different statistical techniques and indirect topographic gradients, it could be asked whether the
unexplained variance can be attributed to reasons other than the spatial resolution of the input data?
Composition and diversity of alpine plant communities is strongly determined by specific disturbance
regimes (Chambers, 1995; Bhmer, 1999). The consequence for environmental response models is
that disturbance driven patterns are correlated solely with static environmental gradients in an assumed
equilibrium setting (Dirnbck et al., 2002). Part of these patterns will be explained by the comprised
environmental variables (e.g. slope inclination will partly explain frequency of avalanches), but more
complex ones will remain unexplained (see e.g White et al., 2001).
For all of these considerations we may conclude, that the predicted patterns will be characterised by
significant within-cell and between-cell variation, and hence that they represent a response trend rather
than real expected distributions. Although static habitat distribution modeling does not allow much
conclusions on mechanisms underlying a plants adaptation to environmental changes, such an
analysis can elicit landscape scale patterns which we might observe in the farther future, taking certain
assumptions on plant responses to climate and land use change into account.
Acknowledgements
We are grateful to M. Steinkellner, G. Mandl, G. Bryda, J., Jansa, K. Kraus, R. Tscheliesnig for
meteorological, geological, land use, and snow cover data, and J. Greimler, I. Schmidsberger, N.
Sauberer and T. Englisch for field data collection. Further thanks go to N. Zimmermann, A.
Bachmann, J. Gallant, J. Wilson for several GIS software. H. Pauli, W. Willner, M. Abensperg-Traun,
and two anonymous referees helped improving the manuscript. The study was funded by the Austrian
Federal Ministry for Education, Science and Culture and the Viennese Water Works.

REFERENCES
Aber, J., Neilson, R.P., McNulty, S., Lenihan, J.M., Bachelet, D. & Drapek, R.J. (2001) Forest processes and
global environmental change: predicting the effects of individual and multiple stressors. Bioscience. 51, 735751.
Adler, W., Oswald, K. & Fischer, R. (1994) Exkursionsflora von sterreich. Verlag Eugen Ulmer, Stuttgart,
Wien.
Arft, A.M., Walker, M.D., Gurevitch, J., Alatalo, J.M., Bret-Harte, M.S., Dale, M., Diemer, M., Gugerli, F.,
Henry, G.H.R., Jones, M.H., Hollister, R., Jnsdttir, I.S., Laine, K., Lvesque, E., Marion, G.M., Molau, U.,
Mlgaard, P., Nordenhll, U., Rasvhizin, V., Robinson, C.H., Starr, G., Stenstrm, A., Stenstrm, M.,
Totland, ., Turner, L., Walker, L., Webber, P., Welker, J.M. & Wookey, P.A. (1999) Response patterns of
tundra plant species to experimental warming: A meta-analysis of the international tundra experiment.
Ecological Monographs. 69, 491-511.
Austin, M.P. & T.M. Smith (1989) A new model for the continuum concept. Vegetatio. 83, 35-47.
Austin, M.P. (1992) Modelling the environmental niche of plants: implications for plant community response to
elevated CO2 levels. Australian Journal of Botany. 40, 615-630.
Azola, C. & Harrell, F.E. (2001) An introduction to S-Plus and the Hmisc and Design libraries. University of
Virginia School of Medicine, http://hesweb1.med.virginia.edu/biostat/s/splus.html.
Bachmann, A. (1998) Coupling NUATMOS with the GIS ARC/INFO. Final report for MINERVA 2. Department
of Geography, University Zurich, CH.
Balk, B. & Elder, K. (2000) Combining binary decision tree and geostatistical methods to estimate snow
distribution in a mountain watershed. Water Resources Research. 36, 13-26.
Beniston, M. (1994) Mountain Environments in Changing Climates. Routledge Publishing Co., London, New
York.

14

Beniston, M. (1997) Variations of snow depth and duration in the Swiss Alps over the last 50 years: links to
changes in large-scale climatic forces. Climatic Change. 36, 281-300.
Beniston, M., Fox, D.G., Adhikary, S., Andresson, R., Guisan, A., Holten, J.I., Innes, J., Maitima, J., Price, M. &
Tessier, L. (1996) The impacts of climate change on mountain regions. Second assessment report of the
intergovernmental panel on climate change (IPCC), Chapter 5, Cambridge University Press, UK.
Beniston, M., Diaz, H.F. & Bradley, R.S. (1997) Climate change at high elevation sites: An overview. Climatic
Change. 36, 233-251.
Bhmer, H.J. (1999) Vegetationsdynamik im Hochgebirge unter dem Einflu natrlicher Strungen.
Dissertationes Botanicae. 311, 1-180.
Bolliger, J., Kienast, F. & Bugmann, H. (2000) Comparing models for tree distributions: concept, structures and
behaviour. Ecological Modelling. 134, 89-102.
Box, E.,O., Crumpacker, D.W. & Hardin, E.D. (1993) A climate-based model of plant species locations in
Florida, U.S.A. Journal of Biogeography. 20, 629-644.
Box, E.,O., Crumpacker, D.W. & Hardin, E.D. (1999) Predicted effects of climatic change on distribution of
ecologically important native tree and shrub species in Florida. Climatic Change. 41, 213-248.
Braun-Blanquet, J. (1964) Pflanzensoziologie, Grundzge der Vegetationskunde. 3rd edition. Springer Verlag,
Vienna.
Breiman, L. (1993) Classification and regression trees. Chapman & Hall, New York.
Bryant, D.M., Holland, E.A., Seastedt, T.R. & Walker, M.D. (1997) Analysis of litter decomposition in an alpine
tundra. Canadian Journal of Botany. 76, 1295-1304.
Brzeziecki, B., Kienast, F. & Wildi, O. (1995) Modelling potential impacts of climate change on the spatial
distribution of zonal forest communities in Switzerland. Journal of Vegetation Science. 6, 257-268.
Carcaillet, C. & Brun, J.-J. (2000) Changes in landscape structure in the northwestern Alps over the last 7000
years: lessons from soil charcoal. Journal of Vegetation Science 11: 705-714.
Carrer, M. & Carlo, U. (2001) Spatial analysis of structural and tree-ring related parameters in a timberline forest
in the Italian Alps. Journal of Vegetation Science. 12, 643-652.
Chambers, J.C. (1995) Disturbance, life history strategies, and seed fates in alpine herbfield communities.
American Journal of Botany 82: 421-433.
Cullen, L., Stewart, G.H., Duncan, R.P. & Palmer, G. (2001) Disturbance and climate warming influences on
New Zealand Nothofagus tree-line population dynamics. Journal of Ecology. 89, 1061-1071.
Didier, L. (2001) Invasion patterns of European larch and Swiss stone pine in subalpine pastures in the French
Alps. Forest Ecology and Management. 145, 67-77.
Diemer, M. (2002) Population stasis in a high-elevation herbaceous plant under moderate climate warming.
Basic and Applied Ecology. 3, 77-83.
Dirnbck, T. & Grabherr, G. (2000) GIS assessment of vegetation and hydrological change in a high mountain
catchment of the Northern Limestone Alps. Mountain Research and Development. 20, 172-179.
Dirnbck, T. & Greimler, J. (1997) Subalpin-alpine Vegetationskartierung der Raxalpe, nordstliche Kalkalpen,
Vegetationskarte 1:12500. Linzer biologische Beitrge. 29, 299-339 and 623-640.
Dirnbck, T., Greimler, J. & Grabherr, G. (1998) Die Vegetation des Zeller-Staritzen-Plateaus
(Hochschwab,Steiermark) und ihre Bedeutung fr den Quellschutz. Mitteilungen des Naturwissenschaftlichen
Vereins der Steiermark. 128, 123-183.
Dirnbck, T., Dullinger S., Gottfried, M. & Grabherr, G. (1999) Die Vegetation des Hochschwab (Steiermark),
alpine und subalpine Stufe. Mitteilungen des Naturwissenschaftlichen Vereins der Steiermark. 129, 111-251.
Dirnbck, T., Dullinger, S. & Grabherr, G. (2001) A new grassland community in the Eastern Alps (Austria):
Evidence of environmental distribution limits of endemic plant communities. Phytocoenologia. 31, 521-536.
Dirnbck, T., Hobbs, R.J., Lambeck, R.J. & Caccetta, P.A. (2002) Vegetation distribution in relation to
topographically driven processes in south-western Australia. Applied Vegetation Science. 5, 147-158.

15

Duckworth, J.C., Bunce, R.G.H. & Malloch, A.J.C. (2000) Modelling the potential effects of climate change on
calcareous grasslands in Atlantic Europe. Journal of Biogeography. 27, 347-358.
Dullinger, S., Dirnbck, T. & Grabherr, G. (2000) Reconsidering endemism in the North-eastern Limestone
Alps. Acta Botanica Croatica. 59, 55-82.
Dullinger, S., Dirnbck, T. & Grabherr, G., (2001) Die subalpine und alpine Vegetation der Schneealpe,
Steiermark. Mitteilungen des Naturwissenschaftlichen Vereins der Steiermark. 131, 83-127.
Dullinger, S., Dirnbck, T. & Grabherr, G. (2002) Effects of summer farming on subalpine plant species
diversity. Proceedings 32nd Annual conference of the Ecological Society of Germany, Austria and
Switzerland. Cottbus, Germany, 2002, 16. - 20 September.
Efron, B. (1983) Estimating the error rate of a prediction rule: Improvement on cross-validation. Journal of the
American Statistical Association. 78, 316-331.
Erschbamer, B., Kneringer, E. & Niederfringer-Schlag, R. (2001) Seed rain, soil seed bank, seedling recruitment,
and survival of seedlings on a glacier foreland in the Central Alps. Flora. 196, 304-312.
Friedel, H. (1961) Schneedeckendauer und Vegetationsverteilung im Gelnde. Mitteilungen Forstliche
Bundesversuchsanstalt Mariabrunn. 59, 317-370.
Gallant, J.C. & Wilson, J.P. (1996) Tapes-G: a grid-based terrain analysis program for the environmental
science. Computers & Geoscience. 22, 713-722.
Gavin, D.G. & Brubaker, L.B. (1999) A 6000-year soil pollen record of subalpine meadow vegetation in the
Olympic Mountains, Washington, USA. Journal of Ecology. 87, 106-122.
Geiger, R. (1965) The climate near the ground. Harvard University Press, Cambridge, MA.
Gessler, P.E., Moore, I.D. & McKenzie, N.J. (1995) Soil-landscape modelling and spatial prediction of soil
attributes. International Journal of GIS. 9, 421-432.
Gindl, W. (1999) Climatic significance of light rings in timberline spruce, Picea abies, Austrian Alps. Arctic,
Antarctic, and Alpine Research. 31, 242-246.
Gottfried, M., Pauli, H. & Grabherr, G. (1998) Prediction of vegetation patterns at the limits of plant life: A new
view of the alpine-nival ecotone. Arctic and Alpine Research. 30, 207-221.
Gottfried, M., Pauli, H., Reiter, K. & Grabherr, G. (1999) A fine-scaled predictive model for changes in species
distribution patterns of high mountain plants induced by climate warming. Diversity and Distributions. 5, 241251.
Grabherr, G., Gottfried, M. & Pauli, H. (1994) Climate effects on mountain plants. Nature. 369, p. 448.
Grabherr, G., Gottfried, M., Gruber, A. & Pauli, H. (1995) Patterns and Current Changes in alpine Plant
Diversity. Arctic and alpine biodiversity (ed. by F.S. Chapin and C. Krner), pp. 167-181. Ecological Studies,
Vol. 113, Springer, New York.
Greimler, J. & Dirnbck, T. (1996) Die subalpine und alpine Vegetation des Schneebergs,Niedersterreich.
Vegetationskarte im Mastab 1:10000 und Beschreibung der Vegetation. Linzer biologische Beitrge. 28,
437-482.
Grytnes, J.A. (2000) Fine-scale vascular plant species richness in different alpine vegetation types: relationships
with biomass and cover. Journal of Vegetation Science. 11, 87-92.
Guisan, A. & Harrell, F.E. (2000) Ordinal response regression models in ecology. Journal of Vegetation Science.
11, 617-626.
Guisan, A. & Theurillat, J.-P. (2000) Equilibrium modeling of alpine plant distribution: how far can we go?.
Phytocoenologia. 30, 353-384.
Guisan, A. & Zimmermann, N.E. (2000) Predictive habitat distribution models in ecology. Ecological
Modelling. 135, 147-186.
Guisan, A., Theurillat, J.-P. & Kienast, F. (1998) Predicting the potential distribution of plant species in an
alpine environment. Journal of Vegetation Science. 9, 65-74.

16

Hansen, A.J., Neilson, R.P., Dale, V.H., Flather, C.H., Iverson, L.R., Currie, D.J., Shafer, S., Cook, R. &
Bartlein, P.J. (2001) Global change in forests: response of species, communities, and biomes. Bioscience. 51,
765-779.
Harrell, F.E. (2001) Regression modeling strategies with application to linear models, logistic regression, and
survival analysis. Springer Series in Statistics, Springer, New York.
Hartman, M.D., Baron, J.S., Lammers, R.B., Cline, D.W., Band, L.E., Liston, G.E. & Tague, C. (1999)
Simulation of snow distribution and hydrology in a mountain basin. Water Resources Research. 35, 15871603.
Hobbs, R.J. (2000) Land-use changes and invasions. Invasive species in a changing world (ed. by H.A. Mooney
and R.J. Hobbs), pp. 55-64. Island press Washington, USA..
Houghton, J.T., Jenkins, G.J. & Ephraums, J.J. (1990) Climate change The IPCC climatic assessment. Report
prepared for IPCC by Working Group 1. Cambridge University Press, 365 pp.
Huntley, B. (1990) European post-glacial forests: compositional changes in response to climatic change. Journal
of Vegetation Science. 1, 507-518.
Huntley, B., Berry, P.M., Cramer, W. & McDonald, A.P. (1995) Modelling present and potential future ranges
of some European higher plants using climate response surfaces. Journal of Biogeography. 22, 967-1001.
Jansa, J., Kraus, K., Blschl, G., Kirnbauer, R. & Kuschnig, G. (2000) Modelling snow melt processes in alpine
areas. International Archives of Photogrammetry and Remote Sensing. 33, pp. 6.
Keller, F., Kienast, F. & Beniston, M. (2000) Evidence of response of vegetation to environmental change on
high-elevation sites in the Swiss Alps. Regional Environmental Change. 1, 70-77.
Krner, C. (1999) Alpine plant life: functional plant ecology of high mountain ecosystems. Springer, Berlin,
Heidelberg, New York.
Kral, F. (1970) Pollenanalytische Untersuchungen zur Waldgeschichte des Dachsteinmassivs. University for
Agriculture and Forestry, Vienna, Austria.
Kullman, L. (2002) Rapid recent range-margin rise of tree and shrub species in the Swedish Scandes. Journal of
Ecology. 90, 68-77.
Lavoie, C. & Payette, S. (1994) Recent fluctuations of the lichen-spruce forest limit in subarctic Quebec. Journal
of Ecology. 82, 725-734.
Levin, S.A. (1992) The problem of pattern and scale in ecology. Ecology. 73, 1943-1967.
Lexer, M.J., Hnninger, K., Scheifinger, H., Matulla, Ch., Groll, N., Kromp-Kolb, H., Schadauer, K., Starlinger,
F. & Englisch, M. (2001) The sensitivity of the Austrian forests to scenarios of climatic change.
Monographien, Umweltbundesamt Wien. 132, 1-132.
Lexer, M.J., Hnninger, K., Scheifinger, H., Matulla, Ch., Groll, N., Kromp-Kolb, H., Schadauer, K., Starlinger,
F. & Englisch, M. (2002) The sensitivity of Austrian forest to scenarios of climatic change: a large-scale risk
assessment based on a modified gap model and forest inventory data. Forest Ecology and Management. 162,
53-72.
Liston, G.E. & Sturm, M. (1998) A snow-transport model for complex terrain. Journal of Glaciology. 44, 498516.
Loiciga, H.A., Maidment, D.R. & Valdes, J.B. (2000) Climate-change impacts in a regional karst aquifer,
Texas, USA. Journal of Hydrology. 227, 173-194.
McCarty, J.P. (2001) Ecological consequences of recent climate change. Conservation Biology. 15, 320-331.
McCullagh, P. & Nelder, J.A. (1991) Generalized linear models. 2nd edition, Chapman & Hall, London, UK
McKenzie, N.J. & Ryan, P.J. (1999) Spatial prediction of soil properties using environmental correlation.
Geoderma. 89, 67-94.
Meshinev, T., Apostolova, I. & Koleva, E.S. (2000) Influence of warming on timberline rising: a case study on
Pinus peuce Griseb. in Bulgaria. Phytocoenologia. 30, 431-438.

17

Moir, W.H., Rochelle, S.G. & Schoettle, A.W. (1999) Microscale patterns of tree establishment near upper
treeline, Snowy Range, Wyoming, U.S.A.. Arctic, Antarctic, and Alpine Research. 31, 379-388.
Molau, U. (1997) Responses to natural climatic variation and experimental warming in two tundra plant species
with contrasting life forms: Cassiope tetragona and Ranunculus nivalis. Global Change Biology. 3, 97-107.
Moore, I.D., Gessler, P.E., Nielsen, G.A. & Peterson, G.A. (1993) Soil attribute prediction using terrain analysis.
Soil Science Society of American Journal. 57, 443-452.
Motta, R. & Nola, P. (2001) Growth trends and dynamics in sub-alpine forest stands in the Varaita Valley
(Piedmont, Italy) and their relationships with human activities and global change. Journal of Vegetation
Science. 12, 219-230.
Nagelkerke, N.J.D. (1991) A note on a general definition of the coefficient of determination of determination.
Biometrika. 78, 691-692.
Pauli, H., Gottfried, M. & Grabherr, G. (1996) Effects of climate change on mountain ecosystems- upward
shifting of alpine plants. World Resource Review. 8, 382-390.
Pauli, H., Gottfried, M. & Grabherr, G. (2001) High summits of the Alps in a changing climate. "Fingerprints"
of Climate Change, adapted behaviour and shifting species ranges (ed. by Walther G.-R., A. Burga and P.J.
Edwards), pp. 139-149. Kluwer, New York.
Paulsen, J., Weber, U.M. & Krner, C. (2000) Tree growth near treeline: abrupt or gradual reduction with
altitude. Arctic, Antarctic, and Alpine Research. 32, 14-20.
Pitelka, L.F. & Plant Migration Workshop Group (1997) Plant migration and climate change. American Scientist.
85, 464-473.
Rich, P.M., Hetrick, W.A. & Saving, S.C. (1995) Modeling topographic influence on Solar Radiation: a manual
for the SOLARFLUX Model. Departmentment of Systematics & Ecology, Los Alamos, New Mexico.
Richardson, D.M., Bond, W.J., Dean, W.R.J., Higgins, S.I., Midgley, G.F., Milton, S.J., Powrie, L.W.,
Rutherford, M.C., Samways, M.J. & Schulze, R.E. (2000) Invasive alien species and global change: a South
African perspective. Invasive species in a changing world (ed. by H.A. Mooney and R.J. Hobbs), pp. 303-350.
Washington, Island Press.
Roecker, E. Oberhuber, J.M., Bacher, A., Christopch, M. & Kirchner, I. (1996) ENSO variability and
atmospheric response in a global coupled atmosphere-ocean GCM. Climate Dynamics 12, 737-754.
Ross, D.G., Smith, I.N., Manins, P.C., & Fox, D.G., (1988) Diagnostic wind field modeling for complex terrain:
Model development and testing. Journal of Applied Meteorology. 27, 785-796.
Stersdal, M., Birks, H.J.B. & Peglar, S.M. (1998) Predicting changes in Fennoscandian vascular-plant species
richness as a result of future climatic change. Journal of Biogeography. 25, 111-122.
Solar, F. (1963) Zur Kenntnis der Bden auf dem Raxplateau. Mitteilungen der sterreichischen
bodenkundlichen Gesellschaft. 8, 3-72.
Suzuki, S. & Kudo, G. (2000) Response of alpine shrubs to simulated environmental change during three years
in the mid-latitude mountain, northern Japan. Ecography. 23, 553-564.
Tappeiner, U., Tappeiner, G., Aschenwald, J., Tasser, E. & Ostendorf, B. (2001) GIS-based modelling of spatial
pattern of snow cover duration in an alpine area. Ecological Modelling. 138, 265-275.
Tappeiner, U., Tasser, E. & Tappeiner, G. (1998) Modelling vegetation patterns using natural and anthropogenic
influence factors: preliminary experience with a GIS based model applied to an alpine area. Ecological
modelling. 113, 225-237.
Theurillat, J.-P. & Guisan, A. (2001) Potential impact of climate change on vegetation in the European Alps: a
review. Climatic Change. 50, 77-109.
Tinner, W., Ammann, B. & Germann, P. (1996) Treeline fluctuations recorded for 12,500 years by soil profiles,
pollen, and plant macrofossiles in the Central Swiss Alps. Arctic and Alpine Research 28: 131-147.
Totland, . (1999) Effects of temperature on performance and phentypic selection on plant traits in alpine
Ranunculus acris. Oecologia. 120, 242-251.

18

Turc, L. (1961) valuation des besoins en eau dirrigation, vapotranspiration potentielle. Annales
Agronomiques, Paris. 12, 13-49
Wakonigg, H. (1978) Witterung und Klima in der Steiermark. University of Technology, Graz, Austria.
Walker, M.D., Webber, P.J., Arnold, E.H. & Ebert-May, D. (1994) Effects of interannual climate variation on
aboveground phytomass in alpine vegetation. Ecology. 75, 393-408.
Weber, R.O., Talkner, P., Auer, I., Bhm, R., Gali-Capka, M.,Zaninovi, K., Brzdil, R. & Fasko, P. (1997)
20th-century changes of temperature in the mountain regions of Central Europe. Climatic Change. 36, 327344.
White, P.S., Wilds, S.P. & Stratton, D.A. (2001) The distribution of heath balds in the Great Smoky Mountains,
North Carolina and Tennessee. Journal of Vegetation Science. 12, 453-466.
Wildi, O. & Schtz, M. (2000) Reconstruction of a long-term recovery process from pasture to forest.
Community Ecology. 1, 25-32.
Zukrigl, K. (1973) Montane und subalpine Waldgesellschaften am Alpenostrand. Mitteilungen der Forstlichen
Bundesversuchsanstalt Mariabrunn. 101, 1387.

19

TABLES
Table 1: Scenarios of climate and land use change. a annual mean compared to the average of 1961
1995 (according to Lexer et al. 2001); b monthly precipitation sum in August compared to the average
of 1961 1995 (according to Lexer et al. 2001).
land use change
scenario
present land use
maintained
all pastures become

climate change scenario

as present

+0.65Ca ; -30 mmb

+2Ca; -30 mmb

+2Ca, -60 mmb

LU3

LU3-2050

LU3-K2

LU3-K2plus

LU4-2050

LU4-K2

LU4-K2plus

baseline
LU4

abandoned

20

Table 2: Significant predictor variables of the logistic regression model for Pinus mugo Turra
distribution. Deviance = change in deviance when dropping a term from the full model. d.f. = change
in degrees of freedom. Significance tested using Wald-Chi Test. a restricted cubic spline with 3 knots;
b
6 categories.
Variable
Deviance
df
p (Wald Chi2)
degree days a

250.9

<0.001

slope inclination a

26.3

<0.001

snow cover duration a

8.8

0.02

geological unit b

22.6

0.001

September radiation a

21.3

<0.001

total

330

13

<0.001

595.5

476

null model

21

sensitivity

LU3-2050

LU3-K2

LU3-K2plus

LU4-2050

LU4-K2

LU4-K2plus

low

Table 3: Overall sensitivity of 85 plant species as the mean relative area change of all existing
abundance classes (compared to the baseline scenario LU3). Relative change of area of potential
habitats was compared to the baseline scenario with actual pasture use (LU3). + indicates general
increasing trend, - decrease. Sensitivity categories (categories represent means of all scenarios; before
calculation no change was set to 0, 100% decrease or increase to 1, > 100% increase was set to 1): low
= 0-0.2; moderate = 0.2-0.4; high = 0.4-0.6; very high = 0.6-0.8; extreme = 0.8-1. Nomenclature of
species follows Adler et al. (1994).

+
+
-

+
+
-

+
-

+
+
-

+
+
+
+
-

+
+
-

Adenostyles alliariae, Betonica alopecuros, Chaerophyllum hirsutum, Trisetum

alpestre, Vaccinium myrtillus

Veratrum album

Campanula alpina, Loiseleuria procumbens, Pedicularis rostratocapitata, Salix alpina

Acinos alpinus, Carex ferruginea

+
+
-

+
-

+
+
-

+
-

+
+
+
+

+
+
-

+
+
-

+
+
-

+
-

+
-

+
+
-

+
+
-

+
+
+

+
-

+
-

+
-

+
+
+

+
-

+
+
+

+
+
-

+
-

+
+
-

plant species

Helictotrichon parlatorei, Homogyne alpina, Vaccinium vitis-idaea

Galium anisophyllon, Sesleria albicans
Rhododendron hirsutum, Scabiosa lucida
Viola biflora

Leucanthemum atratum
Carex capillaris
Agrostis alpina, Androsace chamaejasme, Anthoxanthum odoratum agg., Anthyllis
vulnerari ssp. alpestris, Armeria alpina, Aster bellidiastrum, Campanula pulla, Carex
firma, Carex sempervirens, Dryas octopetala, Festuca pumila, Festuca versicolor ssp.
brachystachys, Helianthemum alpestre, Helianthemum glabrum, Juncus monanthos,
Luzula glabrata, Minuartia sedoides, Poa alpina, Persicaria vivipara, Potentilla aurea,
Primula clusiana, Ranunculus alpestris, Ranunculus montanus agg., Salix reticulata,
Salix retusa, Silene acaulis

moderate

Deschampsia cespitosa, Nardus stricta

Trollius europaeus
Leontodon hispidus

high

Gentiana pumila
Dianthus alpinus
Achillea clavenae, Alchemilla anisiaca, Phyteuma orbiculare, Thymus praecox ssp.
polytrichus
Trifolium pratense
Tofieldia calyculata
Bartsia alpina, Campanula scheuchzeri, Euphrasia salisburgensis, Galium noricum,
Geum montanum, Ligusticum mutellina, Minuartia gerardii, Myosotis alpestris,
Parnassia palustris

very high

Calamagrostis varia

Anemone narcissiflora
Crepis aurea
Euphrasia picta, Festuca rupicaprina, Gentiana clusii, Homogyne discolor, Lotus
corniculatus, Luzula multiflora, Pedicularis verticillata, Soldanella alpina

22

extreme

Heracleum austriacum, Rumex alpestris, Silene alpestris

FIGURE CAPTIONS

Figure 1: A) Location of the study area in the European Alps (topographical shading). A
Austria, CH Switzerland, D Germany, F France, FL Liechtenstein, I Italy, MC Monaco,
SLO Slovenia; B) the four mountain ranges considered; C) View of the easternmost
mountain range (Mount Schneeberg; 2075 m a.s.l) exemplifying the specific relief.
Alpine non-forest sites (snow covered by a light autumn snowfall event) predominantly
occur at the plateau, slopes are covered by Pinus mugo Turra (upper dark area) and Picea
abies (L.) H. Karsten (lower dark area).

24

Figure 2: Validation of the logistic model of occurrence of Pinus mugo against an

independent data set derived randomly from map data (n = 809). Calibrated risk
distribution (grouped proportions versus mean predicted probability in group) is shown
by means of a logit and a non-parametric calibration curve additionally to the histogram
of logistic-calibration probabilities. Triangles represent grouped proportions. Somer's Dxy
rank correlation and Nagelkerke's RN between predicted probabilities and
presence/absence outcome of the independent data set.

non-forest area (ha)

M. Hochschwab
4000
3000
2000
1000
0
baseline

2050

K2

non-forest area (ha)

M. Raxalpe
600
500
400
300
200
100
0
baseline

2050

K2

non-forest area (ha)

M. Schneealpe
600
500
400
300
200
100
0
baseline

2050

K2

non-forest area (ha)

M. Schneealpe
600
500
400
300
200
100
0
baseline

2050

25

K2

Figure 3: Frequency distribution of RN, Somers Dxy, and Emax in proportional odds
models of 85 alpine and subalpine plant species of the area. The bars show corrected
values derived from bootstrap evaluation (n = 1000 resamples with replacement); at least
one sample per resample from each ordinal cover-abundance classes of the species were
conditioned. Due to singularity and divergence respectively for 13 species, n is below
1000 (smallest n = 868).

26

40

80

30

60

20

40

10

20

(S
)

ty
pe
so
il

in
cl
in
at
io

(D
da
ys

si
te

de
gr
ee

sl
op
e

27

rel. frequency (%)

100

at
(S
er
C
ba
)
la
nc
e
(S
W
ge
B)
ol
og
ic
Se
a
lu
pt
em
ni
t(
be
G
rr
)
pa
ad
st
ia
ur
t
io
e
n
ab
(R
an
S)
do
nm
sn
ow
en
t(
co
TP
ve
A)
rd
ur
at
io
n
w
(
et
SC
ne
D
ss
)
in
de
x
(W
Ju
ET
ly
)
ra
di
at
io
n
M
(R
ay
J)
ra
di
at
io
n
(R
M
)

50

Chi-square - df

Figure 4: Relative frequency and importance of environmental variables. Solid line:

Percentage of PO-models of alpine plant species distribution (n = 85) the different
environmental and land use variables significantly contribute to (Wald test statistic;
p<0.05); bars: Mean Chi -df value (= explanatory power) of the respective variables in
these models.

Figure 5: Total area (> 1600 m a.s.l.) of non-forest habitats of each isolated mountain
range considered which remain after Pinus mugo reaches its potential distribution which
can be expected from the 2050 and the K2 and climate and land use scenarios. K2plus
scenario is not shown since SWB is not a significant explanatory variable for Pinus mugo
distribution.

28

Figure 6: Present distribution of Pinus mugo above 1600 m a.s.l. (presence with black
shade) and probability of occurrence (see legend) of Pinus mugo prone to climate and
land use change; exemplified by a part of the mountain range of M. Hochschwab, the
westernmost of the study area (see Fig. 1(b)). Only LU4-2050 and LU4-K2 scenarios are
shown (Table 1). Pinus mugo is successively covering alpine areas leaving only scattered
non-forest habitats. Disappearance at the lower limit reflects competitive exclusion by
subalpine forest trees.

29

Figure 7: Frequency distribution of expected potential habitat change of all 85 alpine

plant species (in percent area of the baseline scenario LU3 = 100%) due to climate and
land use change. Area increase > 210% was pooled. Total number of species increasing
(+) versus decreasing (-) due to the assumed climate and land use scenario is shown.

30

Figure 8: Some plant species examples (Carex firma Host, Silene acaulis acaulis (L.)
Jacq., Helictotrichon parlatorei (Woods) Pilg., Deschampsia cespitosa (L.) P. B.) of area
change of abundance classes due to climate change and pasture abandonment (according
to Table 1) compared to the baseline scenario with actual pasture use (LU3). The
darkness gradient of the shades represents increasing abundance.

31

2
Dirnbck, T., Dullinger, S., Kck, R. & Grabherr, G.: Organic
matter accumulation following Pinus mugo Turra invasion into
subalpine non-forest vegetation.

Manuskript, zur Publikation eingereicht in der Zeitschrift

Basic and Applied Ecology

32

Organic matter accumulation following Pinus mugo TURRA

invasion of subalpine non-forest vegetation

Dirnbck, T.1,2, Dullinger, S.1, Kck, R.3, Grabherr, G.1

Institute of Ecology and Conservation Biology

University of Vienna, Althanstrasse 14, A-1090

Current address: Federal Environment Agency

Spittelauer Lnde 5, 1090 Vienna, Austria

Institute of Silviculture
Universitt fr Bodenkultur Vienna, Peter Jordanstrasse 70, A-1190

Corresponding author:
Dr. Thomas Dirnbck
Federal Environment Agency
Spittelauer Lnde 5, 1090 Vienna, Austria
Tel: +43-1-31304-3042
Fax +43-1-31304-3700
E-mail dirnboeck@ubavie.gv.at.

NUMBER OF WORDS (EXCLUDING TABLE AND FIGURE CAPTION):

5066

33

ABSTRACT
Pinus mugo invasion into subalpine pastures, mediated by abandonment and livestock
reduction since the mid 19th century, is a common process in the Northern Calcareous
Alps in Austria and Germany. Pinus mugo invasion has several ecological consequences,
which are important in the study area since it is part of the karstic catchments of Viennas
water resources. In this study we focus on the organic matter accumulation associated
with Pinus mugo invasion of non-forest plant communities. Age determination of Pinus
mugo is used in combination with soil and climatic data to investigate the mechanisms
involved. Humus horizons accumulate with a mean increase of 12.5 cm in the first 100
years and considerable variation dependent on site conditions. Apart from the age of
Pinus mugo, temperature, site water balance, and the existence of a clayey soil is
significantly related with humus thickness. The harsh alpine climate, in particular low
temperature, and the adverse effect of the Pinus mugo canopy on temperature, limits litter
decay rates and facilitates deep Moderhumus development. Clay content of the soils
enhances decomposition leading to significantly less humus accumulation under
otherwise similar climatic conditions.
Durch den Rckgang der Almwirtschaft in den Nrdlichen Kalkalpen (sterreichs und
Deutschlands) seit der Mitte des 19. Jhd., ist die natrliche Wiederbesiedlung mit Pinus
mugo (Latsche, Legfhre oder Krummholz) eine auffllige Erscheinung. kologische
Konsequenzen dieser Vernderungen sind im Untersuchungsgebiet, dem KarstwasserEinzugsgebiet der Wiener Wasserversorgung, von grosser Bedeutung. In dem
vorliegenden Artikel wird die Humusakkumulation nach Etablierung von Pinus mugo
untersucht. Durch Altersbestimmung von Pinus mugo-Individuen, Bodenuntersuchungen
und klimatischen Daten wurden die relevanten Prozesse studiert. Die
Gesamthumusmchtigkeit nahm im Mittel um 12,5 cm in den ersten 100 Jahren nach
Aufkommen von Pinus mugo zu, wobei erhebliche standortsbedingte Abweichungen
beobachtet wurden. Abgesehen vom Alter spielen Temperatur, Wasserbilanz und der
Tongehalt des Bodens eine Rolle. Extreme Klimabedingungen, insbesondere die
niedrigen Temperaturen und die zustzliche Verschlechterung im Bestand verringern den
Humusabbau, wodurch tiefgrndiger Moderhumus entsteht. Tongehalte der Bden
verbessern bei gleicher klimatischer Situation den Humusabbau und weniger Humus
akkumuliert.
Keywords: Alpine - European Alps - humus - shrub invasion - treeline - pasture - land
use change

34

INTRODUCTION
Shrub invasion into subalpine pastures mediated by abandonment and livestock reduction
since the mid 19th century is a common process in the Northern Calcareaous Alps in
Austria and Germany. Especially Pinus mugo Turra is able to effectively invade
subalpine and alpine non-forest sites. Pinus mugo invasion into pastures and natural
grasslands has several ecological consequences. Shrub invasion generally triggers
considerable carbon and nitrogen accumulation in soils, primarily by litterfall but also by
root development (van Breemen & Finzi 1998, De Kovel et al. 2000, Post & Kwon 2000).
Tree cohorts are effective traps for snow and aeolean dust, thereby additionally enhancing
water and nutrient input (Holtmeier & Broll 1992, R. Kck unpublished). Pasture soils
frequently are prone to severe humus loss and compaction by livestock grazing (Bochter
et al. 1981, Liss 1990). Plant roots are known to increase pore volume in soils and thereby
proliferating infiltration (Angers & Caron 1998, and literature therein). Canopy
interception increases to about 10% of the annual precipitation in mature stands,
interception in humus layers is significant in subalpine calcareous forest soils, and the
dense shrub canopy keeps back surface runoff during storm events and probably
decreased drainage due to enhanced evapotranspiration (Gattermayr 1976, Hafenscherer
& Mayer 1986, Fuxjger 1998). The large areas of the Northern Calcareous Alps covered
by Pinus mugo therefore contribute significantly to the catchments water flows
(Dirnbck & Grabherr 2000). Functional consequences of Pinus mugo invasion are of
particular interest because it is supposed that this may affect drinking water production
for Vienna, the capital of Austria. .
The most salient soil change accompanying Pinus mugo invasion is the accumulation of
organic matter (see particularly Zttl 1965). However, how much organic matter
accumulates in a given time is largely unknown as are the environmental constraints of
litter decomposition. In cold climates litter decay is triggered by temperature and
humidity as well as litter chemical composition (O'Lear & Seastedt 1994, Bryant et al.
1997, Berg 2000, Seastedt & Adams 2001). Additionally, parent material influences
organic matter accumulation under Pinus mugo. Dolomite commonly limits litter decay
by decreasing water holding capacity (Kilian 1959, Margl 1973); clayey contents of
carbonate brown loams (Terra fusca, Chromic Cambisol, for nomenclature see Nestroy
et al. 2000) facilitate the binding of organic substances (Solar 1963, Zttl 1965). In this
study we especially focus on the organic matter accumulation associated with Pinus mugo
invasion into non-forest communities. Thereby we ask the following questions: (1) How
much organic matter accumulates over time in terms of horizon thickness? (2) Is the
process of organic matter accumulation determined by climate and other habitat
conditions?

METHOD AND MATERIALS

Study area
The study area is situated in the Northeastern Calcareous Alps, Austria, from the
Hochschwab mountain range to Mount Schneeberg (15 E to 16 E and 4730 N to
4750 N) and covers ca. 150 km. Plateau-mountains, with steep slopes and precipices at
the plateau edges, are built up foremost by different kinds of carbonate bedrock, mostly
Wettersteinkalk and Wettersteindolomit. Meso- and micro-topography is characterised by
35

a variety of karst landforms like poljes, dolines and karren. Flat terrain and the fact that
glaciation was restricted to comparatively small areas during the Pleistocene caused
tertiary sediments to persist on the plateaus (see e.g., Dullinger et al. 2000). These loams,
and constant aeolian deposition from nearby siliceous mountains, influenced the
carbonate soils of the area. Typically a catena appears with Rendzina (Lithic, Calcaric
and Rendzic Leptosol, for nomenclature see Nestroy et al. 2000) soils at positions prone
to erosion (peaks, ridges, steep slopes) overlying pure carbonate followed by a gradual
increase of clayey subsoils (carbonate brown loam, Terra fusca, Chromic Cambisol, for
nomenclature see Nestroy et al. 2000). Loam accumulated in karst depressions to extreme
depths of more than 1.5 m lacking any fraction of coarse carbonate. A mixture of loams
and debris of carbonate occurs at intermediate slope positions (for more detail see Solar
1963). Organic C, N, Carbonats, and pH typically decreases from the Ah to the A and the
B horizon, whereas clay content increases. Ah show particularly high silt content; organic
matter content is high in the A and Ah horizon (Table 1). Soil texture ranges from Tu4,
Tu3, Tu2 to Tt, (nomenclature according to AG Boden 1994; T. Dirnbck unpublished)
The area mirrors a typical alpine climate with steep altitudinal gradients: mean annual
temperature approximates 6-8 C in the valleys (400 700 m a.s.l.) decreasing to about 02 C at the summits (2000 2300 m a.s.l.); annual precipitation amounts to 700 mm
(valleys) and 1500-2500 mm (summits) with a distinct summer peak and snow
precipitation in winter (Hydrographischer Dienst in sterreich 1951-94).
At the actual treeline (around 1500 m a.s.l.) spruce forests (Picea abies (L.) H. Karsten)
and larch forests (Larix decidua Mill) dominate. The adjacent krummholz-zone is, in fact,
a mosaic of prostrate pine shrub (Pinus mugo) and different types of alpine pastures.
Between 1700 and 1800 m a.s.l. there is a gradual transition to the alpine belt in the strict
sense. Natural alpine grasslands are formed mainly by Carex sempervirens Vill., Sesleria
albicans Kit. ex Schult. and Carex firma Host. Additionally, rock faces, scree and
snowbeds are considerably widespread from the valley bottoms up to the summit area.
For more details on the land use status, vegetation and soils see Dullinger et al. (2001)
and references therein.
Sampling design, age determination and soil data
The position of field records were selected by stratified randomised sampling applying
site variables (temperature, potential evapotranspiration, geology, topography, land use)
provided by a GIS. Environmental stratification provides a way of identifying areas of the
landscape that are similar with respect to the environmental variables of interest. Soils
and vegetation of the selected plots are subject to further analyses within a broader survey
to assess climate and land use changes on alpine karstic catchments. Areas actually used
for summer farming were excluded as well as those which have never been used at all. A
land use map was derived from catastral maps combined with historical and actual farm
census data. Inaccuracies of the catastral maps (e.g., pastures extending into steep rock
faces or debris cones) were corrected using infrared and b/w aerial orthophotos. Totally
140 sampling plots were selected and geo-coded in order to be found in field using a GPS
(Garmin Etrex).
All living and dead individuals of Pinus mugo in each study plot (20 * 20 m) were
sampled. For each individual we recorded origin (seed or layer),vitality (dead or alive),
height, canopy-diameter (mean of largest and smallest diameter), and other data not
considered here. Three of the individuals per plot (largest, median and smallest with
36

regard to canopy-diameter) were sectioned as close to the root collar as possible and rings
were counted under a stereoscope (14x magnification) after fine-grinding. To eliminate
underestimation of the age due to weakly developed rings, four radii were counted and
the maximum value was used. The age of all individuals whose root collar was no more
identifiable (55% of all original individuals) was estimated as a linear function of their
diameter, height and the average growth rate(linear regression, R = 0.69) on the
respective plot. This growth rate was calculated by dividing the length of the sectioned
branch by its ring count. For the present study we used only generative individuals with
canopy-diameter > 0.5 m, therefore focussing on only 33 from the 140 plots, totalling to
89 individuals.
Three soil profiles per plot were recorded, from the Pinus mugo individuals with assigned
age, and the non-forest matrix. The matrix sample was positioned in the predominant
plant community within the plot area. The soil below Pinus mugo was sampled at its
germination origin. If the root collar couldnt be identified any more the origin was
estimated taking branch thickness, patch form, and relief into account. Soil horizons were
described and soil types were classified, both according to Englisch & Kilian (1998).
Since soil types with clayey subsoils exist apart from pure Rendzinas we distinguished
three types, dependent on the coarse carbonate fraction (Englisch & Kilian, 1998). We
define these as no loam (all Rendzinas) loam (< 40% coarse fraction per volume soil,
typical carbonate brown loam) and intermediate (> 40% coarse fraction, mixed soil).
Loam content (no loam, intermediate, loam) is assumed to account for most of the
differences in soil water holding capacity, nutrient status, and the potential to form
organo-clay complexes.
Humus and mineral soil horizons determination were done according to Englisch &
Kilian 1998 (see also Green et al. 1993). The only mineral horizon which was included in
these analyses is the Ah. In order to facilitate differentiation of the Ah from the organic
matter horizons (particularly the Oh) in the field, visible mineral sliver was used
additionally to its common features (matter, aggregation, etc.). It is worthwhile to
mention, that the field methods differentiating soil horizons most probably impose
considerable data noise. For each soil horizon, organic as well as mineral soil, depth (in
0.5 cm), aggregation, delimitation, root abundance < 2 mm thickness, fungal mycelia and
fabric was recorded.
In addition to these data 40 soil profiles below Pinus mugo individuals were described
during a series of altitudinal transects throughout the study area. Data on Pinus mugo
individuals were equally recorded and georeferenced using aerial orthophotographs. Soil
data was restricted to horizon depth, soil type and humus form. All field data were
acquired in the summer months of 2000 and 2001 and pooled for this study.
Climatic data
Temperature degree days (daily mean temperature > 0C) were approximated using
meteorological data from three stations of the area at 770 m, 820 m and 1740 m a.s.l.
(Steinkellner unpublished, 30 minute interval). We used a single lapse rate over altitude
and an increasing base temperature from west to east. The base temperature (annual
mean) was linearly regressed on data from 17 standard meteorological stations located
within and close to the study area; Central Institute of Meteorology and Geodynamics,
Austria). Annual mean temperature shows a 0.82C westward decrease per 100 km when
altitude is kept constant (R=0.6, p < 0.05). High altitudinal stations (Mount Raxalpe and
37

Schneealpe) were excluded from this analysis. The calculated degree days show a strictly
linear relationship with altitude (5.6 days decrease per 100 m) and 12 degree days per 1C
change of annual mean temperature.
Potential for water deficiency was calculated as the site water balance for August since
we assumed that water deficiency does not occur before the late summer. We used the
common Turc-equation (Turc 1961) to calculate potential evapotranspiration (PET)
besides using exact measurements of net radiation. Net radiation was calculated applying
SOLARFLUX (Rich et al. 1995) under clear sky conditions with a Digital Elevation
Model (DEM, 20 m resolution, Austrian National Mapping Agency ). The program
accounts for relief shading and topographic position. Temperature for August was
approximated by multiple linear regression on altitude (R = 0.5; p < 0.001) with data
from the meteorological stations mentioned above. Total precipitation for August (R =
0.1; p = 0.06) was calculated using geographic position in addition. Precipitation of the
area is strongly mediated by local storms, which is probably the main reason for the low
fit of the model. The regression slope for monthly precipitation is 1.5 mm per 100 m
elevation, and 6 mm per 10 longitudinal km.
At three sites (ALM: 1820 m. a.s.l., HO: 1855 m a.s.l., PO: 1880 m. a.s.l.) we measured
detailed temperature and humidity data (Rotronic Hygromer MP 100A, 2 m offset),
incident solar radiation (Delta-T Devices Ltd. Energy Sensor ES2) and precipitation
(Delta-T Devices Ltd. Raingauge, Type RG1). An automatic data logger (Minicube,
Kucera Environmental Measuring Systems) recorded all parameters in a 10 minute
interval. Data for the snow free period of 2000 was used for the analysis. Daily PET was
calculated with the Turc-equation (Turc 1961) using temperature and solar radiation. Six
Pinus mugo and three matrix samples were recorded from these plots according to the
methods described above.
Statistical Analyses
We analysed the relationship between organic matter accumulation, age and habitat
conditions within the framework of Generalised Linear Models (GLM, McCullagh &
Nelder 1991). Age, temperature degree days, site water balance, and loam content were
used to estimate humus horizon thickness. The response was assumed to follow a
Gaussian distribution and an identity link function was applied. After visual exploration
of scatterplots and with one exception, only linear response curves were considered; the
age-Ol relationship was fitted using a third order polynomial. The change in deviance
when dropping terms from the full model was used to test for significant explanatory
variables (F-Test for nested models) and variable importance (deviance change).
Temperature and PET differences between the meteorological stations (ALM, HO, PO)
were analysed by a paired T-Test after testing for normality using a Quantile-quantile
plot. Accordingly, PET was log-transformed before applying the T-Test. In order to test if
climatic differences lead to different organic matter accumulation we predicted the
overall depth of organic matter at PO, HO and ALM using GLMs as above with age and
loam content as explanatory variables and calculated residual depths.
The significance of soil morphology parameters between different age classes were
analysed with One-way ANOVA and Kruskal-Wallis rank sum test. SPlus2000 was used
for all statistical analyses.

38

RESULTS
Changes of soil morphology
While humus forms of natural alpine soils are mostly Moders with a considerable organic
matter content, pasture soils, being situated at lower altitudes, tend more to Mullmoders,
characterised by a shallower organic matter horizon (Solar 1963). Only the Oh has a
noteworthy depth (3 cm on average, with extremes of 30 cm for some alpine soils) and
the Ah reaches even 9 cm on average with extremes of 35 cm (Table 2). In the first stage
of Pinus mugo invasion organic matter of the former grassland soils decays, whereas
needle litter accumulates forming a shallow Ol. The depth of Ol rises until about 50 years
after invasion, however, thereafter decreases implicating enhanced litter decomposition.
The Of, Oh, and Ah, instead, increase steadily (Table1, Fig. 1). Consequently, the humus
form is changing from Mull or Mullmoder to Moder or Tangelmoder, the last is
characterised by deep and loose Of-horizons but comparably shallow Ah (Nestroy et al.
2000). The change of the humus form is accompanied by decreasing root abundance in
the organic matter horizons since Pinus mugo has a far less dense fine-root system than
the preceding grasses (Table 2). Coarse roots of Pinus mugo deeply penetrate the soil
profile (approx. 40 60 cm depth).
Age dependent humus accumulation
Total accumulated humus increases linearly with age of Pinus mugo (Fig. 1). Mean
increase is about 12.5 cm in the first 100 years but variation is considerable (95%
confidence limits: 10 cm, 15 cm). In particular litter thickness is not satisfactorily
explained by the age of Pinus mugo pointing to considerable site specific variation (Fig.
1). Whereas all horizons show a linear relationship, undecomposed litter (Ol) thickness
increases only until about 30 to 50 years, then decreases until about 70 to 100 years and
thereafter remains rather constant (Fig. 1). Although significant, age alone is of limited
explanatory value [R between 0.17 (Ol) and 0.47(Of)].
Climatic and habitat constraints
Besides the dominant effect of age, humus thickness is co-determined by climatic
conditions (especially temperature degree days) and loam content of the mineral soil
(Table 3). The litter horizon is additionally controlled by PET and precipitation during the
summer (see site water balance in Table 3). At colder and drier conditions litter
accumulation is enhanced, hence decomposition seems to be suppressed. Soil organic
matter of the mineral soil is mostly controlled by loam content, overriding climatic
constraints. Litter decomposition increases from rendzic soils to clayey carbonate brown
loams soils. Age, climate and loam content together do not explain more than 50% of the
depth-variance of different horizons (Table 3).
Temperature is significantly different at the three climatic stations (HO, PO, ALM)
(Table 4). ALM had about 0.4C higher daily mean temperature (June to October 2000)
than HO and PO (p<0.001). HO showed only an 0.07C increase in daily mean
temperature compared to PO (p<0.05). Daily potential evapotranspiration of ALM is
significantly (p<0.001) higher in comparison to PO (0.9 mm) and HO (1 mm). PO and
HO showed no significant difference (p = 0.36). Precipitation is quite similar for all
stations besides a weak trend to higher daily precipitation in PO (p<0.05, 1,5 mm more
than HO and 0.6 mm more than ALM). In sum, ALM is warmer and drier than the other
39

station sites. Although HO is positioned on a south-east facing slope PET does not
deviate much from PO, both stations being located at the upper limit of Pinus mugo
growth, whereas ALM reflects the situation at the predominant altitudinal distribution of
the krummholz belt. The snow regime was not measured, since one or two years give
little allusion on the snow regime. Nevertheless, especially PO is extremely wind-prone
and the total snow free period may be longer, in particular in comparison with ALM.
In order to test if these climatic differences lead to deviating organic matter
decomposition rates we predicted the overall depth of organic matter at PO, HO and
ALM using age and loam content as explanatory variables and calculated residual depths.
Organic matter horizon depth of ALM is overestimated, HO and PO are underestimated
(Table 5). That means that apart from the influence of age and loam, increasingly colder
conditions from ALM to HO and PO coincide with accumulation of organic horizons.
Although HO is characterised by lower humus accumulation than PO, PET does not
differ significantly between these plots (Table 4). Therefore, water shortage due to
enhanced PET may play a minor role, however, values show considerable variation and
significance could not be tested due to the limited number of replications.

DISCUSSION
Alpine soil formation is essentially driven by biological processes (Krner 1999). Woody
plant invasion into non-forest vegetation thus pose severe changes on soil properties. The
most salient change is increasing humus thickness. On average, we found an
accumulation rate of 12.5 cm during the first 100 years. In general, humus build-up in
cold environments can be a long-term process (thousands of years, Berg 2000). Zttl
(1965) and Kral (1970) estimated much lower rates than we found, however, these studies
give estimates for the accumulation since the Pleistocene in which disturbance may have
been significant and climate was fluctuating. After Pinus mugo invades in non-forest
sites, organic matter increases linearly during the investigated time span of 150 years.
From our results we can not infer when the accumulation may saturate or if disturbance
events are repulsing humus build-up to earlier stages.
Humus accumulation is a synergistic effect of increased litter input and decreased litter
decay rates due to cooler conditions inside the shrub canopy. We did not consider litter
quality since exclusively Pinus mugo was investigated (see Nardi et al. 2000 on Pinus
mugo litter quality). Zttl (1965) did not found much differences of the chemical
composition between Pinus mugo litter and that from other nearby plant species, and
concluded that humus accumulation must therefore be triggered by increased productivity
of Pinus mugo in combination with the harsh alpine climate. Seastedt & Adams (2001)
reported on limited litter decay inside tree-islands (Picea engelmannia and Abies
lasiocarpa) due to cooler conditions, particularly during the snow free period. Lowered
soil temperature during the summer months was also found under Pinus mugo compared
to non-forest vegetation (R. Kck submitted). On the other hand, decay rates increase
under snow pack, since deeply covered sites remain warmer (O'Lear & Seastedt 1994,
Bryant 1997). Snow depths and depletion dates are significantly different between Pinus
mugo shrubs and nearby low-stature vegetation in the study area. Snow packs are deeper,
occur earlier in winter (snow interception in the dense branches) and last shorter in the
spring (R. Kck submitted).
We found a very subtle signal of the significance of PET and precipitation for humus
accumulation. The importance of water deficiency for litter decay in alpine environments
40

was, however, consistently reported for non-forest vegetation by Bryant (1997) and
O'Lear & Seastedt (1994), although not verified for tree-island soils (Seastedt & Adams
2001). PET and precipitation alone do not account for water holding capacity of the soils
itself, which increases with the existence of clayey soils. In addition, clay contents of
loamy horizons facilitate enhanced humification and formation of stable organo-clay
complexes (Solar 1963, Zttl 1965, Rehfuess 1990). Consequently, loam contents play an
important role for the organic matter accumulation after Pinus mugo invasion. Total
organic matter thickness rarely exceeds 10 cm after 150 years on carbonate brown loams,
whereas on pure limestone up to 30 cm humus was accumulated.
Decomposition of fresh litter is retarded the first 30 to 50 years after Pinus mugo
invasion. Undecomposed needles build a loose layer up to 15 cm, often without any Of
and a very shallow Oh below. This is in line with the expected decrease of litter
decomposition due to climatic deterioration inside a growing shrub mentioned above.
However, thereafter the Ol even decreases, pointing to enhanced decay rates. Why do
plant residues decompose faster under bigger shrub individuals? From our data we can
not give an unambiguous explanation. The growth rate of Pinus mugo is low (approx. 3 to
10 cm shoot growth per year, own unpublished data); 50 or less years old individuals
have less than 3 m canopy-diameter and 1 m height. Snow pack must therefore still be
shallow and low temperatures might limit this early phase of humus development.
Another explanation may be found in the soil fauna. Soil and litter invertebrates, which
comminute plant residues before they can be decomposed by microbes, were reported to
recolonise slowly in early years of succession (Haslam et al. 1998). There, post-fire soils
accumulated litter for the first 20 to 40 years without substantial decomposition.
However, Schinner (1982) found no difference in litter decomposition including or
excluding soil fauna of alpine soils. That organic matter horizons of soils preceding Pinus
mugo invasion become decomposed within the first few decades might (illustrated in
Table 2) possibly point to necessary adaptations of the decomposers to the novel plant
residues.
We conclude from our data that the process of humus accumulation when Pinus mugo
invades non-forest subalpine communities is dependant upon litter input, low temperature
and the occurrence of a clayey subsoil. Organic matter accumulates steadily over time
and no saturation of this process can be seen within the first 150 years. The harsh alpine
climate, in particular low temperature, and the adverse effect of the Pinus mugo canopy
on temperature, limits litter decay rates and facilitates deep Moderhumus development.
Clay contents of the soils, derived from relict sediments and aeolian deposition,
considerably enhances decomposition. Decomposition of shed Pinus mugo needles shows
a lag phase of 30 to 50 years which we can not explain unambiguously from our data.

ACKNOWLEDGEMENTS
We are grateful to Christian Holtermann, Max Steinkellner, Karin Wriessnig for their
technical support, data and soil analyses. Michael Englisch and Eduard Hochbichler
revised an early draft of the manuscript and gave valuable comments. The study was
funded by the Austrian Federal Ministry for Education, Science and Culture and the
Viennese Water Works.

41

REFERENCES
AG Boden (1996) Bodenkundliche Kartieranleitung. 4. Aufl., Schweizerbartsche
Verlagsbuchhandlung, Stuttgart.
Angers DA, Caron J (1998) Plant-induced changes in soil structure: Processes and feedbacks.
Biogeochemistry 42: 55-72.
Berg B (2000) Litter decomposition and organic matter turnover in northern forest soils. Forest
Ecology and Management 133: 13-22.
Bochter R, Neuerburg W, Zech W (1981) Humus und Humusschwund im Gebirge. Nationalpark
Berchtesgaden Forschungsberichte 2: 1-110.
Bryant DM, Holland EA, Seastedt TR, Walker MD (1997) Analysis of litter decomposition in an
alpine tundra. Canadian Journal of Botany 76: 1295-1304.
De Kovel CGF, Mierlo EM, Wilms YJO, Berendse F (2000) Carbon and nitrogen in soil and
vegetation at sites differing in successional age. Plant Ecology 149: 43-50.
Dirnbck T, Grabherr G (2000) GIS assessment of vegetation and hydrological change in a high
mountain catchment of the Northern Limestone Alps. Mountain Research and Development 20:
172-179.
Dullinger S, Dirnbck T, Grabherr G (2000) Reconsidering endemism in the North-eastern
Limestone Alps. Acta Botanica Croatica 59: 55-82.
Dullinger S, Dirnbck T, Grabherr G (2001) Die subalpine und alpine Vegetation der Schneealpe.
Mitteilungen Naturwissenschaftlicher Verein Steiermark 131: 83-127.
Englisch M, Kilian W (1998) Anleitung zur Forstlichen Standortskartierung in sterreich. FBVABerichte, Forstlichen Bundesversuchsanstalt Wien 104: 1-108.
Fuxjger C (1998) Niederschlagsinterzeption von Vegetation und Auflagehumus in fnf
Waldbestnden de Nationalparks Kalkalpen, Obersterreich. Diplomarbeit Universitt fr
Bodenkultur Wien.
Gattermayr W (1976) Vergleichende Messung und Berechnung der Verdunstung, der
Evapotranspiration und Interzeption zur Abschtzung des Wasserhaushalts der Karsthochflche
Dachstein-Oberfeld. Dissertation Universitt Innsbruck.
Green RN, Trowbridge RL, Klinka K (1993) Towards a taxonomic classification of humus forms.
Forest Science Monograph, 29: 1-49.
Hafenscherer J, Mayer H (1986) Standort, Aufbau, Entwicklungsdynamik und Verjngung von
Latschenbestnden im Karwendeltal/Tirol. Schweizerische Zeitschrift Forstwesen 137: 177203.
Haslam SFI, Chudek JA, Goldspink CR, Hopkins DW (1998) Organic matter accumulation
following fires in a moorland soil chronosequence. Global Change Biology 4: 305-313.
Holtmeier F-K, Broll G (1992) The influence of tree islands and microtopography on
pedoecological conditions in the forest-alpine tundra ecotone on Niwot Ridge, Colorado Front
Range, U.S.A.. Arctic and Alpine Research 24: 216-228.
Hydrographischer Dienst in sterreich (1951-94) Beitrge zur Hydrographie sterreichs 23, 26,
38, 46, 52.
Kilian W 1959: Vorschlag zu einer Karststandsaufnahme sterreichs dargestellt am Beispiel des
stlichen Dachsteingebietes. Dissertation Universitt fr Bodenkultur, Wien.
Kck R, Hrtel E, Holtermann C, Hochbichler E, Hager H, Schnthaler KE (submitted)
Monitoring hydrological processes in montane and subalpine karst regions: Comparison
between different vegetation types - Experimental design, techniques and first results. UNESCO
Technical Documents in Hydrology.

42

Krner C (1999) Alpine plant life: Functional plant ecology of high mountain ecosystems.
Springer.
Kral F (1970) Pollenanalytische Untersuchungen zur Waldgeschichte des Dachsteinmassivs.
Habilitationsschrift an der Universitt fr Bodenkultur, Wien.
Liss B-M (1990) Beweidungseffekte im Bergwald, Ergebnisse aus fnfjhriger Untersuchungen
zur Waldweide unter besonderer Bercksichtigunhg des Wildverbisses. In: Schuster E. (ed)
Zustand und Gefhrdung des Bergwaldes, Forstwissenschaftliche Forschungen, Paul Parey
Verlag, Hamburg, pp 50-65.
Margl H (1973) Waldgesellschaften und Krummholz auf Dolomit. Angewandte
Pflanzensoziologie 21: 1-50.
McCullagh P, Nelder JA (1991) Generalized linear models. 2nd edition, Chapman & Hall, London.
Nardi S, Pizzeghello D, Reniero F, Rascio N (2000) Chemical and biochemical properities of
humic substances isolated from forest soils and plant growth. Soil Science Society America
Journal 64: 639-645.
Nestroy O, Danneberg OH, Englisch M, Gel A, Hager H, Herzberger E, Kilian W, Nelhiebel P,
Pecina E, Pehamberger A, Schneider W, Wagner J (2000) Systematische Gliederung der Bden
sterreichs, sterreichische Bodensystematik 2000. Mitteilungen der sterreischischen
Bodenkundlichen Gesellschaft 60: 1-99.
O'Lear HA, Seastedt TR (1994) Landscape patterns of litter decomposition in alpine tundra.
Oecologia 99: 95-101.
Post WM, Kwon KC (2000) Soil carbon sequestraion and land-use change: processes and
potential. Global Change Biology 6: 317-327.
Rehfuess KE (1990) Waldbden, Entwicklung, Eigenschaften und Nutzung. 2nd edition, Pareys
Studientexte 29, Parey, Hamburg.
Rich PM, Hetrick WA, Saving SC (1995) Modeling topographic influence on solar radiation: a
manual for the SOLARFLUX model. Departmentment of Systematics & Ecology, Los Alamos,
New Mexico.
Schinner F (1982) Soil microbial activities and litter decomposition related to altitude. Plant Soil
65: 87-94.
Schlichting E, Blume H-P, Stahr K (1995) Bodenkundliches Praktikum. 2. Aufl., Blackwell
Wissenschafts-Verlag, Berlin, Wien.
Seastedt TR, Adams GA (2001) Effects of mobile tree islands on alpine tundra soils. Ecology 82:
8-17.
Solar F (1963) Zur Kenntnis der Bden auf dem Raxplateau. Mitteilungen der sterreichischen
bodenkundlichen Gesellschaft 8: 3-75.
Turc L (1961) valuation des besoins en eau dirrigation, vapotranspiration potentielle. Annales
Agronomiques, Paris. 12, 13-49
van Breemen N, Finzi AC (1998) Plant-soil interactions: ecological aspects and evolutionary
implications. Biogeochemistry 42: 1-19.
Zttl H (1965) Zur Entwicklung der Rendzinen in der subalpinen Stufe. I. Profilmorphologie, II.
Chemisch-biologische Dynamik. Zeitschrift fr Pflanzenernhrung, Dngung u. Bodenkunde
110: 109-126.

43

Table 1: Mean of chemical and physical soil horizon characteristics derived from 23 soil samples of
the study area.

mg.g-1
soil
horizon

pH [CaCl2]

CaCo3

Ah
A
B

6.1
6
5.2

45.7
60.3
2.4

grain size [%.g-1] d

N
53.4
16.7
12.5

Corg

2-0.063
mm

0.063-0.002
mm

< 0.002
mm

319.1
198.9
123.2

9
24.7
1.9

52.6
40.1
41.4

38.4
35.2
56.7

11
7
5

44

Table 2: Soil organic matter horizons of different age classes. Mean horizon depth and standard deviation [cm]; sample size (n); median of root abundance with
sample size in brackets (number of roots < 2mm dm-). Most frequent humus aggregations and humus forms with rare ones in brackets. Significance tested using
One-way ANOVA for horizon depth, and Kruskal-Wallis rank sum test for root abundance (*** p < 0.001; n.s. not significant)

Age class

Horizon depth [cm]

Ol***

Of***

Oh***

Root abundance [n.dm-]

Ah n.s.

Ol n.s.

Aggregation

Humus form

Of ***

Oh ***

Ah ***

Ol

Of

Oh

>50 (21)

>50 (18)

21-50 (26)

loose

loose

loose,
compact

Mull,
Mullmoder

Moder

0.6 0.5 0.9 0.9 3.2 5.5

9.0 7.1

30 1-5 (3)

1 - 20

2.1 1.7 0.5 0.8 1.9 3.4

8.4 3.9

18

21 - 50

4.3 2.5 1.2 1.2 0.7 1.3

7.6 7.6

30 1-5 (2) 6-10 (12)

11-20 (11)

21-50 (18)

loose

loose,
agglutinate

loose,
(compact)

Moder

51 - 100

4.1 2.4 2.8 2.1 3.2 3.7

8.7 7.4

27 1-5 (3) 6-10 (25)

11-20 (23)

6-10 (25)

loose,
(agglutinate)

loose,
agglutinate

loose
(compact)

Moder

101 - 160

3.7 1.6 4.7 1.6 7.9 8.4

12.9 12.8

10 1-5 (2) 11-20 (10)

11-20 (10)

6-10 (8)

loose,
agglutinate

loose,
agglutinate

loose

Tangelmoder

45

Table 3: Deviance reduction when dropping terms from the Generalised Linear Models of organic
matter horizon depths, using age of the respective Pinus mugo individual, number of days per year
with > 0C mean temperature, site water balance (precipitation potential evapotranspiration), and
loam content in the mineral soil (loam, intermediate, no loam) as explanatory variables. n.s. not
significant; * p<0.05; ** p<0.01; *** p<0.001. a Litter-age relationship was fitted using a third order
polynomial, all others as linear terms.

Horizon depth

Agea

Degree
daysa

Site water
balancea

Loam
content

Null
deviance

Explained
Deviance (%)

Litter (Ol horizon)

55**

27**

78***

n.s.

394

41

Humus layer(Ol, Of, and

Oh horizon)
Humus layer and Ah
horizon

1153***

94*

n.s.

122*

2739

50

3285***

n.s.

n.s.

834*

11869

35

46

Table 4: Mean and standard deviation of daily mean temperature (daily means from 10 minute interval
measures) and daily potential evapotranspiration (PET, using daily mean temperature and solar
radiation) between June and September 2000 from three stations in the study area. ALM (1820 m
m.s.l.), HO (1855 m m.s.l.), PO (1880 m m.s.l.). Significance of differences in means was tested using
paired T-test (** p-value < 0.05, *** p-value < 0.001, n.s. not significant).

Meteorological Stations

Mean of Differences

ALM

HO

PO

ALM-HO

ALM-PO

HO-PO

Daily mean
Temperature (C)

7.3 4.2

6.2 3.6

6.0 3.7

0.4 ***

0.4 ***

0.1 **

Daily PET (mm)

6.7 5.7

5.1 4.7

4.9 4.5

1 ***

0.9 ***

-0.1 n.s.

47

Table 5: Mean residual of humus depths (Ol, Of and Oh) and standard deviation from 6 soil samples of
three meteorological stations (ALM 1820 m m.s.l., HO 1855 m m.s.l., PO 1880 m m.s.l.). GLMs
(gaussian distribution with identity link; linear responses) were fitted using age (years after Pinus
mugo invasion) and loam content of the mineral soil (loam, intermediate, no loam) from 89 Pinus
mugo individuals as explanatory variables. Negative values indicate overestimation positive
underestimation.

Meteorological Station

residual depth [cm]

ALM

HO

PO

-1.6 3.2

+1.2 6.2

+7.2 9.5

48

Figure 1: Relationship between humus horizon depths [cm] and age [a] of the invaded Pinus mugo
individual. Solid lines give linear regressions, besides for the Ol-horizon, which is a cubic power
function, 95% confidence limits are shown by dotted lines; n = 87.

11
10
9
8
7
6
5
4
3
2
1
0

10
9
8
7
6
5
4
3
2
1
0

R = 0.17

p < 0.001

Ol

Of

p = 0.001

R = 0.47

10

30

50

70

90

110

130

150

10

30

50

30
27
24
21
18
15
12
9
6
3
0

R = 0.22
p < 0.001

10

30

50

70

90

30
27
24
21
18
15
12
9
6
3
0

110

130

150

p < 0.001

10

110 130 150

30

50

70

90

110 130 150

Alter

70

R = 0.16

R = 0.28

60

p = 0.002

50

Humus + Ah

Ah

90

R = 0.39

Alter

30
27
24
21
18
15
12
9
6
3
0

70

Age

Humus total

Oh

Age

p < 0.001

40
30
20
10
0

10

30

50

70

90

110

130

10

150

30

50

70

90

Alter

Alter

49

110

130

150

3
Dirnbck, T. & Dullinger, S.: Disturbance and the predictability of plant
species distribution in alpine environments.
Manuskript, zur Publikation eingereicht in der Zeitschrift
Journal of Vegetation Science

50

Disturbance and the predictability of plant species distribution in alpine

environments
Dirnbck, T.1, 2, * & Dullinger, S.1
1

Institute of Ecology and Conservation Biology, Univ. Vienna, Althanstrae 14, 1090 Vienna, Austria;
2
Current address: Federal Environment Agency; Spittelauer Lnde 5, 1090 Vienna, Austria;
*
Corresponding author; Fax +431313043700; E-mail dirnboeck@ubavie.gv.at.

ABSTRACT
As a basis for assessing causes of diversity we evaluated the influence of disturbance on the
predictability of alpine plant species distribution using static environmental response models. Firstly,
abundance data of 71 plant species were correlated with a comprehensive set of environmental
variables using Proportional Odds Models. Thereafter, the residual spatial autocorrelation in the data
was explored. The additional amount of variance explained by variables accounting for spatial
structuring was compared with a set of functional traits assumed to be positively or negatively
associated with the existence of disturbance. We found significant residual spatial autocorrelation in
the habitat models of most of the species that were analysed. The amount of this autocorrelation is
positively related with the dispersal capacity of the species, levelling off with increasing spatial scale.
Both trends indicate that dispersal and colonisation processes, caused by disturbance, influence the
distribution of many alpine plant species. Since environmental response models commonly ignore
such processes they are of limited reliability when predicting local to landscape scale plant distribution
or community diversity. The additional existence of residual spatial autocorrelation for many of the
more competitive species indicate, that, apart from disturbance, there are additional factors that
confound species-environment relationships
Keywords alpine plants, spatial autocorrelation, functional traits, habitat distribution model, ordinal
response model
Nomenclature Adler et al. (1994)
Number of words: 5590

51

INTRODUCTION
Disturbance, is a fundamental agent structuring plant communities (Hobbs & Huenneke 1992; Huston
1994; Pickett & White 1985; White & Jentsch 2001). Particularly its control over biotic diversity has
attained much interest since Connell (1978) has formulated the Intermediate Disturbance Hypothesis.
Disturbance regimes interfere with environmental gradients (Pausas & Austin 2001) as well as with
biotic interactions, particularly with the process of competitive displacement (Connell 1978; Huston
1994). Moreover, disturbance filters the regional species pool which promotes the occurrence of
certain life history strategies (Belsky 1992; Dupr & Diekman 2001; McIntyre et al. 1995) and
functional traits (Grime 2001; Lavorel et al. 1997; McIntyre et al. 1999) which either allow for
persistence in or for rapid re-invasion of disturbed habitats. Attempts to predict community attributes
(e.g., species composition, species richness or abundance) exclusively from habitat variables, like
climate and soil properties, may thus have limited success in disturbance prone landscapes (e.g.,
DeAngelis & Waterhouse 1987; Dirnbck et al. 2002; White et al. 2001). Nevertheless, many
modeling studies used the conceptual framework of equilibrium of species with their environment to
predict future plant distribution and/or diversity under conditions of environmental change (see for a
review Guisan & Zimmermann 2000). To validate the use of equilibrium-based modeling approaches
a method that allows for disentangling environmental and disturbance effects on species distribution
would be a useful tool. Given data on the joint distribution of alpine plant species and environmental
conditions we propose a method comprising two essential steps. First, we demonstrate the existence of
spatial pattern in the distribution of biota that cannot be explained by habitat conditions, and second,
we relate this pattern to disturbance effects using general theory on functional adaptation of plant
species to disturbance. Conceptually, the first part of the approach is based on Legendre (1993).
Accordingly, the distribution of species is spatially autocorrelated partly because environmental
variables are, most commonly, represented as patches or gradients in real landscapes or due to
spatially structured processes like dispersal and colonisation, competition, historical events,
conspecific attraction and disturbance. Disturbance is likely to produce autocorrelated patterns because
of its patchy nature and on account of the fact that most disturbance events initiate processes of recolonisation which are dependent upon a propagule source (e.g. Greene & Johnson 1986, Jongejans &
Schippers 1999). Species, characterised by functional traits related to disturbance (see particularly
McIntyre et al. 1999), should therefore show significant spatial patterns in the residuals of habitat
models at the scale of the disturbance patch size and the dispersal abilities. On the other hand,
competitors are usually assumed to predominate in undisturbed habitats under equilibrium conditions
(Grime 2001). Therefore, we expect species with traits that point to competitive advantages not to
show residual spatial patterns independent from the environment. In this paper we formalise two
specific questions: (1) Do, and if so, to what extent, exist spatial distribution patterns of alpine plants
which are independent of environmental variables? (2) Do functional traits, positively or negatively,
related to disturbance explain those residual patterns.

STUDY AREA
The study area covers the non-forest alpine and subalpine zone of four neighbouring mountain ranges
of the North-eastern Calcareous Alps in Austria comprising about 150 km area (Mt. Hochschwab, Mt.
Schneealpe, Mt. Rax and Mt. Schneeberg, 15 E to 16 E and 4730 N to 4750 N). Summits vary
between 1900 and 2300 m a.s.l. The mountain system consists of mesozoic limestone and dolomites
and is characterised by displaced plateaus of different altitudes. Meso- and microrelief are shaped by a
variety of karst landforms like dolines and karren. Climatic conditions are temperate humid. Mean
annual temperature approximates 6-8 C in the valleys decreasing to about 0-2 C in the summit
52

region. Annual precipitation averages 700 mm (valleys) and 1500-2500 mm (summits) with a decisive
peak during the summer season. Summer pasturing (June to September) in the area dates back at least
to the 16th century. Except for rock faces, debris cones and very steep slopes, most of the study area
has been historically influenced by livestock grazing. Since the middle of the 19th century grazing
intensity has decreased and much former pastureland has been abandoned. Approximately 30% of the
study area is still pastured by free-ranging cattle at an intensity of about 0.5 cattle per hectar. Other
forms of land use (e.g., tourism) are of minor importance.
Norway spruce (Picea abies) and European larch (Larix decidua ), dominate from around 1400 m
a.s.l. up to a belt of prostrate pine (Pinus mugo). The upper limit of Pinus mugo is between 1800 and
1900 m a.s.l.. Non-woody vegetation below treeline mainly consists of different kinds of pastures and
natural grasslands with the latter covering disturbed sites like avalanche paths and exposed ridges.
Less frequent, tall herb communities (Adenostyles alliariae, Aconitum napellus) and fens (Carex
nigra, C. rostrata) occur. Above treeline, natural grasslands (mostly with Carex sempervirens and
Carex firma) dominate. Additionally, rock faces, scree and snowbeds are considerably widespread
(totalling 25% of the study area) from the valley bottoms up to the summits.

METHODS
Firstly, plant distribution and abundance data were correlated with a comprehensive set of
environmental variables, derived using a GIS (we will refer to environmental variables exclusively
as abiotic, relatively stable physico-chemical, climatological or geomorphological factors; natural
disturbance is considered separately). Then the residual spatial structure in the data was explored by
spatial vectors of presence or absence of a species at different distances. And lastly, we correlated the
additional variance explained by these spatial vectors with a set of functional traits assumed to be
positively or negatively associated with the existence of disturbance.
Data
Vegetation data
The calibration data set is based on a stratified sampling approach, using all combinations of
categorised environmental variables and time of pasture abandonment (see below) as strata. Each
stratum had to be sampled by at least one plot, depending on total area of the stratum. We included a
large set of existing vegetation records (totally 780. Dullinger et al. 2001 and references therein) and
did supplementary records for strata not covered by these (totalling 236). The positions were either
measured by means of a GPS or marked on an infrared or black and white orthophoto. Data were
collected according to standard phytosociological techniques. All plant species present on the plot
were recorded together with a cover-abundance value estimated on a 7-level ordinal scale (BraunBlanquet 1964). From the total of 650 species occurring in the sample set, 71 (>100 times present)
were selected for modeling. This subset comprises typical alpine species occurring above the treeline,
species of pastures and of avalanche tracks. Understorey species of Pinus mugo shrubland and forests
were excluded, as well as species which comprise ecologically different subspecies or variants not
separated during field work (e.g., Anthoxanthum odoratum and A. alpinum, Biscutella laevigata).
Moreover, we excluded endemic species with restricted geographical ranges within the study area
(e.g., Alchemilla anisiaca, D. calcareum)
Environmental variables
Several environmental variables were spatially derived using map data, a digital elevation model
(DEM, 20 m resolution, Austrian National Mapping Agency), and meteorological station data. A
53

detailed description of the various procedures is given in Dirnbck et al. (submitted). Daily net
radiation for 15th May, 15th July and 15th September was derived applying SOLARFLUX (Rich et al.
1995) under clear sky conditions with the DEM. Temperature degree days (daily mean temperature >
0C) were calculated using three meteorological stations from Mt. Schneealpe (770 m, 820 m and
1740 m a.s.l.; half hourly measures; Steinkellner unpublished data) and data from 17 standard
meteorological stations located within and close to the study area (annual means; Central Institute of
Meteorology and Geodynamics, Austria). Site water balance was calculated for August since we
assumed that water deficiency does not occur before the late summer. It was approximated by the
potential evapotranspiration (Turc 1961), using net radiation from SOLARFLUX and mean
temperature from regression over altitude (data as above), subtract from the monthly precipitation.
Latter was calculated using geographic position in addition to altitude because eastward precipitation
decrease is known for the area.
We used 11 SPOT scenes (acquired 1998, 1999, 2000 from February to June, each at about 10.45 am;
20 m resolution), available for a third of the area and classified into snow cover classes, the abovementioned radiation model (Rich et al. 1995), a wind field model (Bachmann 1998 unpubl.), elevation
(as a compound surrogate for temperature and precipitation; from the DEM), relief indices (wetness
index, see below), and plant cover type (krummholz versus grasslands, screes and rocks) from existing
vegetation maps to parameterise a snow distribution model applying a Classification Tree procedure
(Splus 2000). The final model had a misclassification error rate of 8% using the calibration data. This
model was used to extrapolate a snow cover duration index for the whole study area.
A detailed geological map (Geological Survey of Austria, unpubl.) was categorised into 5 geological
units (limestone, dolomite, clayey weathering carbonates, relict loam, carbonate debris) and wetland.
The gross soil distribution was derived from interpretation of aerial orthophotographs (bare or almost
bare areas, areas covered with soils). Topographically driven erosion and accumulation, mediated by
water runoff, was supposed to be the most important factor controlling soil properties at finer scales.
We calculated the topographical wetness index according to Gallant & Wilson (1996). In addition to
the wetness index, the slope inclination was used.
Land use
Since we attempted to exclusively address effects of natural disturbance, land use information was
included in the environment-only models. Logging and summer grazing significantly influenced the
vegetation of the area. To account for this human induced disturbance regime, spatial land use
information was derived from catastral maps and a suite of documents referring to the land use status
of individual parcels at different times. From historical documents and personal communication we
derived dates of abandonment as exactly as possible. In statistical analyses we used the number of
years between the last dates of documented pasture use and the year 2000. Areas which have never
been used were assigned a value of 1000 assuming this period to be sufficient to completely eliminate
all remnants of former pasturing. It must be assumed that these data will only partly explain land use
driven distribution patterns of plants due to differential grazing and spatially as well as temporally
heterogeneous logging.
Spatial vectors
We adopt the method proposed by Borcard & Legendre (1994) and applied by Leathwick (1998). To
account for spatial clustering of the data and the residuals which remain after environmental variation
was filtered, spatial vectors of plant species occurrence are calculated. For each record and species we
calculated if the same species occurs in records at distances form the focal record or not. A new
predictor variable was derived for each species reflecting the spatial clustering or autocorrelation.
54

Vectors were calculated using moving circles of 40, 80, 160, and 320 m radius. These vectors are used
to explore the residual spatial structure which remains after fitting the environment-only models. Not
all records have neighbouring records in the respective windows, particularly the 40 m radius. To
retain all records while testing for the influence of the spatial vectors, we imputed those values by zero
where no records were detectable, thus circumventing the problem of different calibration data with
respect to different vectors and species. Some very isolated records were excluded from analysis.
Functional traits
Both theoretical concepts and modeling studies (e.g., Grime 2001; Noble & Slayter 1980; Oksanen &
Ranta 1992; Schippers et al. 2001) as well as recent field research (e.g. Belsky 1992; Chambers et al.
1993, 1995; Dupr & Diekman 2001; McIntyre et al. 1995) have pointed to correlations of
morphological and regenerative plant traits with establishment and survival in disturbed habitats.
Drawing from this accumulated knowledge, McIntyre et al. (1999) have published a checklist of
functional traits that are assumed to be either positively or negatively associated with disturbance.
From the multitude of traits screened in those and other papers, we have selected a subset for which
data were readily available from the literature on the Middle European flora (Adler et al. 1994; Conert
et al. 1931-1984; Jger, et al. 2001; Jger & Werner 2002; Mller-Schneider 1986; Oberdorfer 2001;
Schrter 1926) (Table 1). Concerning regenerative traits, we chose two attributes associated with
dispersal capacity: diaspore size (as the length) and dispersal mode. With respect to dispersal mode
diaspores were categorised into two classes according to the occurrence or lack of morphological
adaptations to wind dispersal (plume, pappus, wings). Concerning morphological traits we used mean
plant height, maximal lateral spread and above ground cover density (Table 1). All of these traits are
assumed to be positively correlated with competitive ability and negatively with disturbance (Table 1).
Statistical analysis
We derived ordinal logistic regression models (Proportional odds models) using cover-abundance
values of plant species as the response and environmental variables and time after pasture
abandonment as the predictors. Cover-abundance values were categorised into five classes (absent,
present-1%, 1-10%, 10-50%, 50-100%). Proportional odds models are based on the cumulative
probabilities of the classes, stemming from successive logistic regression models (Guisan & Harrell
2000; Harrell 2001). Linear functions as well as restricted cubic splines (piecewise third order
polynomials with linear restrictions in the lower and upper tails) with no more than four knots were
fitted to the data using the Design library of Splus2000. Model and predictor significance were
obtained from Wald test statistic assuming a Chi distribution with 1 d.f. (Harrell 2001). Full models
were reduced by backward elimination, knot reduction where cubic splines were used, and
linearisation respectively (p-value < 0.05). Double linear interactions of DD with all other continuous
variables were taken into account. Validation of environmental models was obtained by resampling
from the calibration data (1000 resamples with replacement) using bootstrap (validate function of
Splus 2000) to get bias corrected values of Nagelkerke's R (quoted as RN thereafter), judging the
predictive strength of the model, and Somers concordance index Dxy (Harrell 2001) which is based on
the Wilcoxon-Mann-Whitney two-sample rank test. Dxy takes values between 0 and 1 depicting totally
random versus perfectly discriminating models, respectively.
The four spatial vectors were added to the environment-only models separately and percent change of
the null deviance minus the residual deviance was used to account for additional variation explained.
Deviance measures are calculated from the logarithm of a ratio of likelihoods and are commonly used
to account for the discrepancy between model and data (McCullagh & Nelder 1991).

55

Correlations of residual spatial patterns and functional traits were analysed by two-way ANOVA and
covariates with the deviance additionally explained by spatial vectors as the response. Main effects
and first order interactions, as well as spatial vector radius were included. Explained deviance was logtransformed in order to fulfil statistical assumptions for ANOVA. Homogeneity of variances and
normal distribution of errors were verified by examining plots of the distribution of residuals and of
residuals against fitted values.

RESULTS
Predictive discrimination (Somers Dxy), as obtained by bootstrap resampling from the calibration data
set, is 0.56 on average and RN is 0.25. Poa alpina shows the lowest Dxy (0.23) and goodness of fit
(RN = 0.09), Helictotrichon parlatorei the best (Dxy = 0.87; RN = 0.4) (Table 2). Dxy is an appropriate
measure for discrimination of ordinal responses and should be around or above 0.6 to be useful to
predict the response of individual subjects (Harrell 2001). Most, but not all, species show Dxy > 0.6.
All environmental variables are significant predictors for plant species distribution and abundance,
nevertheless, different species comprise different suites of significant descriptors. Temperature degree
days are significant (p < 0.05) in almost all models, mostly with highest predictive power. Slope
inclination is the next most consistent and significant contributor, but shows relatively low Chivalues. In contrast, September radiation is only significant in 45% of the models but its contribution to
variance explanation is comparatively high. Site water balance comes closely after soil cover, followed
by geological unit. The latter shows relatively low Chi-values. Significant in less than 30% of the
models is the time since pasture abandonment, snow cover duration, wetness index and July and May
radiation.
Although for most investigated species residual spatial pattern remains after fitting the environmentonly model, the additional deviance reduction in the Proportional odds models is low. However, only
eight out of 71 species do not show any significant deviance reduction when adding spatial vectors (p
< 0.05; Table 2). The relationship between environment-only model fit (deviance explained) and the
degree of spatial autocorrelation (proportional deviance reduction by spatial vectors) is weak (R for
all radii independently = 0.02 0.17). Thus, the residual spatial pattern of the environment-only
models is only to a low degree caused by insufficient or inadequate environmental variables.
Higher and more frequent deviance reduction occurs at larger scales than at smaller ones. Although
spatial autocorrelation generally increases with larger scales some species show highest (Dryas
octopetala, Euphrasia salisburgensis, Festuca rupicaprina, Nardus stricta, Salix alpina) or lowest
(Acinos alpinus, Bartia alpina, Minuartia sedoides,) values at intermediate scales. Those plant species
with a marked residual pattern at smaller scales (40 m radius) are Minuartia gerardii, Poa alpina,
Primula clusiana, and Thlaspi alpestre (Table 2).
Traits associated with dispersal capacity are significantly correlated with the degree of deviance
reduction caused by the incorporation of spatial vectors. Overall, wind dispersed species have higher
residual spatial autocorrelation, especially, but not significantly so, at broader scales (Table 3, Fig. 2).
Diaspore length is negatively related with spatial residuals at smaller scales, but this correlation
interacts with the radius of the spatial vector (Table 3, Fig. 1). Additionally, a significant interaction
exists between diaspore length and dispersal mode with wind dispersed species showing a more
pronounced negative correlation of diaspore length and spatial residuals than those which do not have
specific adaptations (Table 3, Fig. 2). Concerning morphological traits, only lateral spread and above
ground cover density are significantly associated with deviance reduction (Table 4), particularly at
broader scales (Fig. 3). Plant height does not explain the degree or the scale of spatial patterns
significantly (Table 4).

56

DISCUSSION
In our analysis we have detected residual spatial autocorrelation in the habitat models of most of the
species analysed (Table 2). This result points to a significant impact of spatially structured processes
for alpine plant distribution which are independent from the environment. However, the mere
existence of residual patterns does not allow for any straightforward identification of the driving
processes. We thus collected evidence for the role of disturbance in creating these patterns by
examining the correlation of residual patterns with plant functional traits assumed to be positively or
negatively associated with disturbance (Table 1). The results of these analyses are equivocal. On the
one hand, there is a significant correlation of the amount of residual spatial autocorrelation with the
dispersal capacity of the respective species (Table 3, Figs. 1 and 2). Plants with small diaspores and
with adaptations for wind dispersal show higher residual spatial autocorrelation. This correlation is
levelling off with increasing spatial scale (Fig. 1). Both trends indicate that dispersal and colonisation
processes cause species to occur in habitats classified as unsuitable in environment-only models and
vice versa. Numerous studies (e.g. Beerling 1995; Grabherr 1989; Sousa 1985; Richardson & Bond
1991) have shown that invasion of additional species into established communities is strongly
facilitated by disturbance events which open up closed canopies and thus create recruitment
opportunities. Disturbances like substrate movement, recurrent avalanches, and soil frost action shape
the structure, composition, and diversity of alpine communities (Chambers 1995; Bhmer 1999;
Ellenberg 1988; Erschbamer 1989; Fox 1981; Grabherr 1989; Johnson & Billings 1962; Malanson &
Butler 1986; Schrter 1926). The individual biology of many of the investigated species showing
significant residual patterns at small to intermediate spatial scales additionally points to the importance
of disturbance in driving these patterns. Among them are well known alpine ruderals like Poa alpina,
annual species (Euphrasia salisburgensis), pioneers in open habitats (Dryas octopetala, Salix spp.) or
small-statured, gap filling species with low competitive ability (Androsace chamaejasme, Minuartia
gerardii).
We had expected plant traits that confer competitive advantages to be negatively correlated with the
degree of residual spatial autocorrelation. However, the results do not confirm this expectation. Plant
height is uncorrelated and lateral spread as well as above ground cover density are even positively
correlated with residual patterns (Table 4, Fig. 3). This may either indicate that the assumed
relationship between disturbance and competitiveness does not hold or that there are additional
processes creating autocorrelated residual patterns. There is indeed some support for both hypotheses.
A closer look at the individual species being both autocorrelated and competitive, reveals that many of
them are either confined to or occur frequently in avalanche paths (e.g. Helictotrichon parlatorei,
Calamagrostis varia, Deschampsia cespitosa, Trollius europaeus, Carex ferruginea). The effect of
avalanches is mainly to remove or to permanently prevent the establishment of late successional
species in subalpine habitats, i.e. forest trees and shrubs. Therefore, we have to consider that the
gradient representing competitive ability is truncated by the exclusion of trees and shrubs, the late
successional dominants of subalpine habitats. Also the apparent correlation between lateral spread and
residual spatial pattern might be spurious. Lateral spread is particularly characteristic for dwarf shrubs
(like Dryas octopetala, Salix spp., Thymus praecox ssp. polytrichus) which occupy disturbed sites by
vegetative spread and predominate mid-seral stages in disturbed grasslands and snowbeds (Ellenberg
1988; Pachernegg 1973). Many of these species show wind dispersal, too.
In addition to the above rationale, we hypothesise that historic perturbation (e.g., post-glacial climate
fluctuations, pasturing effects not included in our variable used) and ongoing climatic changes may
cause the apparent broad scale pattern independent from environmental variables used. It was shown
that many alpine plants were unable to keep pace with ongoing rapid climate changes (Grabherr 1994).
Retarded successional processes are assumed to be caused by low growth rates and the extraordinary
57

persistence of alpine key-species (Steininger et al. 1996) and whole communities (Blinnikov 1994).
Over time scales of decades or centuries gradual climatic changes thus produce non-equilibrium
patterns which are not confined to the scale of disturbance patches. Rather they will produce largerscale autocorrelated patterns which are due to the gradual invasion of migrating species into new
habitats starting from some dispersal foci. This may well explain the frequent occurrence of residual
spatial pattern at scales which are beyond the usual area of most high mountain disturbances as well as
the seed shadow of herbaceous or graminoid plants (e.g. Jongejans & Schippers 1999). Given long
enough time scales, these autocorrelated non-equilibrium patterns caused by migratory adaptations to
environmental change should be detectable for species of both high and low competitive ability.
Even if the resulting correlations of functional traits and residual patterns are thus reasonably
interpretable, there remains a considerable amount of unexplained variance (see Table 2). Certainly,
spatial inaccuracies of the GIS-derived environmental variables contribute to leaving some spatial
patterns unexplained (Dullinger et al. 2003; Keitt et al. 2002). Moreover, only a subset of plant traits
assumed to be correlated with existence in disturbed habitats has been analysed due to limited data
availability. As an example, dormant seed banks are assumed to be one of the most successful tools to
deal with recurrent disturbances even in alpine habitats (Archibold 1984; Diemer & Prock 1993,
Bernhardt 1996; Chambers 1993 1995). It is remarkable in this context that the few species for which
persistent seed banks are known (Deschampsia cespitosa, Gentiana clusii, Luzula multiflora, Nardus
stricta, and Thymus praecox ssp. polytrichus; see Bernhardt 1996; Grime 2001; Onipchenko et al.
1998) show comparatively large spatial autocorrelation.
The low additional variance explained in our study by spatial patterns independent from the
environment (see Table 2) points to the strong dependence of disturbance regimes and environmental
gradients, at least in alpine environments. However, the results also demonstrate the existence of
spatial patterns which are independent from environmental variables and attributable to disturbance.
Additional broad scale non-equilibrium of plant species with environmental variables exists, to which
long-term climatic changes may contribute. We conclude from our study that environment-only
models ignore significant local to landscape scale community level processes, and that any inference
derived from such models should be treated with caution.
Acknowledgements
We thank M. Steinkellner, G. Mandl, G. Bryda, J., Jansa, K. Kraus for meteorological, geological,
land use, and snow cover data, and J. Greimler, I. Schmidsberger, N. Sauberer and T. Englisch for
field data collection. K. Hlber, H. Pauli, S. Ertl and G. Grabherr helped to conceptualise the
functional traits database. We also gratefully acknowledge the Austrian Federal Ministry for
Education, Science and Culture and the Vienna Water Works for financial support.

REFERENCES
Adler, W., Oswald, K. & Fischer, R. 1994. Exkursionsflora von sterreich. Verlag Eugen Ulmer, Stuttgart,
Wien.
Archibold, O.W. 1984. A comparison of seed reserves in arctic, subarctic and alpine soils. Canadian Field
Naturalist 98: 337-344.
Beerling, D.J. 1995. General aspects of plant invasions. an overview. In. Pysek, P., Prach, K., Rejmnek, M., &
Wade, W. (eds.) Plant invasions, general aspects and special problems, pp. 3-13. SPB Academic Publishing,
Amsterdam, The Netherlands.
Belsky, A.J. 1992. Effects of grazing, competition, disturbance and fire on species composition and diversity in
grassland communities. J. Veg. Sci. 3: 187-200.
Bernhardt, K.-G. 1996. Die Diasporenbank einer alpinen Pflanzengesellschaft. Diss. Bot. 258: 295-304.

58

Blinnikov, M.S. 1994. Phytolith analysis and holocene dynamics of alpine vegetation. Verff. Geobot. Inst. Eidg.
Tech. Hochschule, Stiftung Rbel 115: 23-40.
Bhmer, H.J. 1999. Vegetationsdynamik im Hochgebirge unter dem Einflu natrlicher Strungen. Diss. Bot.
311: 1-180.
Borcard, D. & Legendre, P. 1994. Environmental control and spatial structure in ecological communities. an
example using orbatid mites (Acari, Oribatei). Environmental and Ecological Statistics 1: 37-61.
Braun-Blanquet, J. 1964. Pflanzensoziologie, Grundzge der Vegeationskunde. 3rd ed. Springer Verlag, Vienna.
Chambers, J.C. 1993. Seed and vegetation dynamics in an alpine herbfield. effects of disturbance type. Can. J.
Bot. 71: 471-485.
Chambers, J.C. 1995. Disturbance, life history strategies, and seed fates in alpine herbfield communities. Am. J.
Bot. 82: 421-433.
Conert, H.J., Hamann, U., Schultze-Motel, W. & Wagenitz, G. (eds.) 1931-1984. Illustrierte Flora von
Mitteleuropa. Verlag Paul Parey, Berlin, Hamburg.
Connell, J.H. 1978. Diversity in tropical rain forests and coral reefs. Science 199: 1302-1310.
DeAngelis, D.L. & Waterhouse, J.C. 1987. Equilibrium and nonequilibrium concepts in ecological models.
Ecological Monographs 57: 1-21.
Diemer, M. & Prock, S. 1993. Estimates of alpine seed bank size in two Central European and one Scandinavian
subarctic plant communities. Arct. Alp. Res. 25: 194-200.
Dirnbck, T., Hobbs, R.J., Lambeck, R.J. & Caccetta, P.A. 2002. Vegetation distribution in relation to
topographically driven processes in southwestern Australia. Appl. Veg. Sci. 5: 147-158.
Dullinger, S., Dirnbck, T. & Grabherr, G. 2001 Die subalpine und alpine Vegetation der Schneealpe,
Steiermark. Mitt. Naturwiss. Vereins Steiermark. 131: 83-127.
Dullinger, S., Dirnbck, T., Greimler, J. & Grabherr, G. (2003) A resampling approach for evaluating effects of
pasture abandonment on subalpine plant species diversity. J. Veg. Sci., in press.
Dupr, C. & Diekman, M. 2001. Differences in species richness and life-history traits between grazed and
abandoned grasslands in southern Sweden. Ecography 24: 275-286.
Ellenberg, H. 1988. Vegetation Ecology of Central Europe. Cambridge University Press, Cambridge, UK.
Erschbamer, B. 1989. Vegetation on avalanche paths in the Alps. Vegetatio 80. 139-146.
Fox, J.F. 1981. Intermediate levels of soil disturbance maximize alpine plant diversity. Nature 293: 564-565.
Gallant, J.C. & Wilson, J.P. 1996. Tapes-G. a grid-based terrain analysis program for the environmental science.
Computers & Geoscience 22: 713-722.
Grabherr, G., Gottfried, M. & Pauli, H. 1994. Climate effects on mountain plants. Nature 369: 448.
Grabherr, G. 1989. On community structure in high alpine grasslands. Vegetatio 83: 223-227.
Greene, D.F. & Johnson, E.A. 1986. Dispersal capacity and seed production in anemochorous plants. Oecologia
68: 629-631.
Grime, J.P. 2001. Plant strategies, vegetation processes, and ecosystem properties. 2nd ed. John Wiley & Sons,
Chichester, UK.
Guisan, A. & Harrell, F.E. 2000. Ordinal response regression models in ecology. J. Veg. Sci. 11: 617-626.
Guisan, A. & Zimmermann, N.E. 2000. Predictive habitat distribution models in ecology. Ecol. Mod. 135: 147186.
Harrell, F.E. 2001. Regression modeling strategies. with application to linear models, logistic regression, and
survival analysis. Springer Series in Statistics, Springer, New York.
Hobbs, R. & Huenneke, L.F. 1992. Disturbance, diversity, and invasion. Implications for conservation. Cons.
Biol. 6: 324-337.
Huston, M.A. 1994. Biological diversity. The coexistence of species on changing landscapes. Cambridge
University Press, Cambridge.

59

Jger, E., Schubert, R. & Werner, K. 1991. Exkursionsflora von Deutschland, Atlas der Gefpflanzen, 3. 8th ed.
Volk und Wissen Verlag, Berlin.
Jger, J. & Werner, K. 2002. Exkursionsflora von Deutschland, Gefpflanzen, 4. Spektrum Akademischer
Verlag, Heidelberg, Berlin.
Johnson, P.L. & Billings, W.D. 1962. The alpine vegetation of the Beartooth Plateau in relation to
cryopedogenic processes and patterns. Ecological Monographs 32: 105-135.
Jongejans, E. & Schippers, P. 1999. Modeling seed dispersal by wind in herbaceous species. Oikos 87: 362-372.
Keitt, T.H., Bjrnstad, O.N., Dixon, P.M. & Citron-Pousty, S. 2002. Accounting for spatial pattern when
modeling organism-environment interaction. Ecography 25: 616-625.
Lavorel, S., McIntyre, S., Landsberg, J. & Forbes, T.D.A. 1997. Plant functional classifications. from general
groups to specific groups based on response to disturbance. Trends in Ecology and Evolution 12: 474-478.
Leathwick, J.R. 1998. Are New Zealand's Notofagus species in equilibrium with their environment. J. Veg. Sci.
9. 719-732.
Legendre, P. 1993. Spatial autocorrelation. trouble or new paradigm. Ecology 74: 1659-1673.
Malanson, G.P. & Butler, D.R., 1986. Floristic patterns on avalanche paths in the northern Rocky Mountains,
USA. Physical Geographer 7: 231-238.
McCullagh, P. & Nelder, J.A. 1991. Generalized linear models. Chapman & Hall, 2nd ed. London, UK.
McIntyre, S., Lavorel, S. & Tremont, R.M. 1995. Plant life-history attributes. their relationship to disturbance
response in herbaceous vegetation. J. Ecol. 83: 31-44.
McIntyre, S., Lavorel, S., Landsberg, J. & Forbes, T.D.A. 1999. Disturbance response in vegetation - towards a
global perspective on functional traits. J. Veg. Sci. 10: 621-630.
Mller-Schneider, P. 1986. Verbreitungsbiologie der Bltenpflanzen Graubndens. Verffentlichung des
Geobotanischen Institut ETH, Stiftung Rbel 85: 1-263.
Noble, I.R. & Slatyer, R.O. 1980. The use of vital attributes to predict successional changes in plant
communities subject to recurrent disturbance. Vegetatio 43: 5-21.
Oberdorfer, E. 2001. Pflanzensoziologische Exkursionsflora. Fr Deutschland und angrenzende Gebiete. Ulmer,
Stuttgart.
Oksanen, L. & Ranta, E. 1992. Plant strategies along mountain vegetation gradients. a test of two theories. J.
Veg. Sci. 3: 175-185.
Onipchenko, V.G., Semenova, G.V. & van der Maarel, E. 1998. Population strategies in severe environments:
alpine plants in the northwestern Caucasus. J. Veg. Sci. 9: 27-40.
Pachernegg, G. 1973. Struktur und Dynamik der alpinen Vegetation auf dem Hochschwab (NO-Kalkalpen).
Diss. Bot. 22.
Pausas, J.G. & Austin, M.P. 2001. Patterns of plant species richness in relation to different environments: an
appraisal. J. Veg. Sci. 12: 153-166.
Pickett, S.T.A. & White, P.S. 1985. The ecology of natural disturbance and patch dynamics. Academic press,
Inc., San Diego, US.
Rich, P.M., Hetrick, W.A. & Saving, S.C. 1995. Modeling topographic influence on Solar Radiation. a manual
for the SOLARFLUX Model. Dep. Syst. & Ecol., Los Alamos, New Mexico.
Richardson, D.M. & Bond, W.J. 1991. Determinants of plant distribution. Evidence from pine invasions.
American Naturalist 137: 639-668.
Schippers, P., van Groenendal, J.M., Vleeshouwers, L.M. & Hunt, R. 2001. Herbaceous plant strategies in
disturbed habitats. Oikos 95: 198-210.
Schrter, C. 1926. Das Pflanzenleben der Alpen. 2nd ed., Verlag von Albert Raustein, Zrich.
Sousa, W.P. 1985. Disturbance and patch dynamics on rocky intertidal shores. In: Picket, S.T.A. & White P.S.
(eds.) The ecology of natural disturbance and patch dynamics. pp. 101-124. Academic press, Inc., San Diego,
US.

60

Steininger, T., Krner, C., & Schmid, B. 1996. Long-term persistence in a changing climate. DNA analysis
suggests very old ages of clones of alpine Carex curvula. Oecologia 105: 94-99.
Turc, L. 1961. valuation des besoins en eau dirrigation, vapotranspiration potentielle. Annales Agronomiques,
Paris 12: 13-49
White, P.S. & Jentsch, A. 2001. The search for generality in studies of disturbance and ecosystem dynamics.
Progress in Botany 62: 401-450.
White, P.S., Wilds, S.P., & Stratton, D.A. 2001. The distribution of heath balds in the Great Smoky Mountains,
North Carolina and Tennessee. J. Veg. Sci., 12: 453-466.

61

Table 1. Functional traits, measurement units and sources that were used. Assumed positive (+) or
negative (-) response to disturbance. a adaptations to wind dispersal (plume, pappus, or wings). b
estimated as the projected shoot cover density.
Functional trait

unit

dispersal modea

yes/no

diaspore length

source

response to
disturbance

Conert et al., 1931-1984; Mller-Schneider, 1986;

Oberdorfer, 2001

mm

Mller-Schneider, 1986

mean plant height

cm

Adler et al., 1994

lateral spread

cm

Adler et al., 1994; Conert et al., 1931-1984; Jger, et

al. 2001; Jger & Werner, 2002; Schrter, 1926

above ground cover

densityb

Jger, et al. 2001; Jger & Werner, 2002

62

Table 2. Statistics of the Proportional Odds Models exclusively with environmental variables [n,
observations; d.f., degrees of freedom used, Somer's Dxy; Nagelkerke's RN; (the two latter represent
corrected values derived from bootstrap validation with 1000 resample with replacement); null
deviance; percentage deviance explained] and proportional deviance decrease (in percentage of the
Null Model deviance; 1% indicated as bold letters) when adding spatial vectors of species
occurrence in different distances to the environment-only models. *** p < 0.001, ** p < 0.01, * p <
0.05, otherwise not significant.
plant species

Environment only model

Dxy

Proportional deviance decrease

RN null dev exp dev

d.f.

40

80

160

320

Achillea clavenae

1016

0,58 0,20

816

15,7

0.4

0.7*

0.4

1.8***

Acinos alpinus

1008

0,64 0,27

1223

18,4

0.4*

0.1

1.1***

Agrostis alpina

1008

0,58 0,29

1911

16,0

0.3*

0.5**

0.8***

1***

Androsace chamaejasme

1016

0,61 0,29

1399

18,3

2.2***

2.8***

5***

6.4***

Anemone narcissiflora

1016

0,50 0,15

866

11,6

0.1

0.4*

1.7***

Anthyllis vulneraria ssp. alpestris

1008 19

0,63 0,31

1542

20,4

0.1

0.9***

0.9***

0.6**

Armeria alpina

1016 10

0,74 0,40

1127

29,7

1.4***

2.1***

2.1***

3.6***

Aster bellidiastrum

1016

0,38 0,13

1634

7,7

0.2

0.3*

0.4*

Bartsia alpina

1016

0,61 0,27

1260

18,3

0.7**

0.3

0.1

1.6***

Betonica alopecuros

1016 19

0,77 0,40

1288

29,3

0.3

0.2

2.3***

2.5***

Calamagrostis varia

1016 13

0,85 0,41

712

39,3

0.9**

0.5*

1.7***

Campanula alpina

1016

0,73 0,35

884

27,4

0.1

0.1

0.4

0.3

Campanula pulla

1016 10

0,60 0,31

1242

21,4

0.1

0.3*

0.5*

0.1

Campanula scheuchzeri

1008 11

0,24 0,08

2062

4,4

0.2*

0.2*

0.4**

0.2

Carex capillaris

1016

0,56 0,21

1317

13,7

0.2

0.1

0.1

Carex ferruginea

1008

0,61 0,25

1435

15,3

0.4*

0.1

1.1***

0.4*

Carex firma

1016 10

0,58 0,32

2300

16,0

0.7***

0.8***

0.6***

1.4***

Carex sempervirens

1008 20

0,52 0,28

2351

13,5

0.1

Chaerophyllum hirsutum

1016

0,61 0,20

932

15,5

0.2

0.5*

0.3

1**

Crepis aurea

1008

0,37 0,09

1273

6,6

0.3

0.3*

Deschampsia cespitosa

1016 15

0,59 0,31

1962

17,8

1.1**

3.2***

7.7***

14***

Dianthus alpinus

1016

0,59 0,26

1234

17,9

0.3

0.1

0.7**

0.1

Dryas octopetala

1016

0,65 0,36

1844

20,5

0.6***

1.2***

1.3***

0.8***

Euphrasia picta

1016 12

0,45 0,14

1300

10,0

0.2

0.6**

0.1

0.8**

Euphrasia salisburgensis

1008

0,38 0,10

1250

6,5

0.6**

1.5***

0.9***

0.2

Festuca versicolor ssp. brachystachys

1008 11

0,57 0,23

1427

14,9

0.6**

1.4***

1.6***

1.4***

Festuca pumila

1008 15

0,63 0,41

2167

22,5

0.3*

0.1

0.3*

0.5**

Festuca rupicaprina

1016 13

0,32 0,07

1478

6,0

0.4*

1.3***

1.2***

0.3*

Galium anisophyllum

1016 13

0,40 0,19

2106

9,5

0.2*

0.3**

0.1

Galium noricum

1016 11

0,55 0,24

1308

16,8

0.4*

0.7**

1.1***

0.8**

Gentiana clusii

1016

0,47 0,11

786

9,3

0.1

0.3

0.8*

2.8***

Gentiana pumila

1008

0,66 0,26

697

21,7

0.5*

0.8*

0.9*

1.7***

Geum montanum

1016 11

0,58 0,22

1107

16,3

0.5*

1.2***

2.4***

2.6***

Helianthemum alpestris

1016 11

0,67 0,35

1473

23,3

0.4*

0.7**

0.4*

Helictotrichon parlatorei

1016 16

0,87 0,40

636

36,2

2.7***

2.9***

4.9***
1.8***

Heracleum austriacum

1016

0,66 0,28

1132

19,7

0.3

0.1

1.3***

Homogyne alpina

1016 15

0,67 0,31

1326

21,3

0.4

1.1

1.4

1.9

Homogyne discolor

1008

0,44 0,18

1745

10,2

0.3*

0.9***

1.5***

Juncus monanthos

1016 10

0,52 0,16

1049

12,6

0.6**

0.4*

1.3***

1.7***

Leontodon hispidus

1008 12

0,38 0,11

1646

7,0

0.2*

0.3*

0.9***

1.6***

Leucanthemum atratum

1016 10

0,57 0,23

1350

14,8

0.2

0.6**

0.9***

1.4***

Ligusticum mutellina

1016 15

0,56 0,24

1509

15,9

0.9**

0.7**

1.7***

1.9***

Loiseleuria procumbens

1016

0,85 0,35

438

31,8

0.6

0.2

2.2**

1.5*

Table 2: continued

63

plant species

Environment only model

Dxy

Proportional deviance decrease

RN null dev exp dev

d.f.

40

80

160

Lotus corniculatus

1008

0,60 0,30

1625

17,7

0.1

0.2

320
0.2

Luzula glabrata

1016 15

0,51 0,17

1146

13,4

0.8**

1***

1.5***

1.2***

0.6**

2.3***

2.7***
1.5***
1.4***

Luzula multiflora

1016 15

0,67 0,28

981

23,4

Minuartia gerardii

1008 10

0,58 0,23

981

17,7

1.8***

1**

0.7**

Minuartia sedoides

1016 11

0,68 0,36

1554

22,5

1***

0.8***

1.3***

Myosotis alpestris

1016 10

0,53 0,18

799

15,6

0.7*

0.6*

0.5*

0.9**

Nardus stricta

1008 13

0,81 0,38

980

30,4

0.1

1.4***

0.9***

Pedicularis rostratocapitata

1016 13

0,61 0,27

1096

20,1

1.3***

1.1***

2.2***

1.1***

Pedicularis verticillata

1016 10

0,56 0,21

991

16,3

0.4*

0.5*

0.3

0.5*

Persicaria vivipara

1016

0,55 0,33

2171

16,2

0.1

0.1

Phyteuma orbiculare

1016 14

0,51 0,24

1641

14,8

0.1

0.1

0.1
0.2*

Poa alpina

1008

0,23 0,09

2363

3,7

1***

0.7***

0.3*

Potentilla aurea

1008 17

0,49 0,25

2004

14,3

0.2*

0.8***

0.3*

Primula clusiana

1016

0,57 0,28

1822

15,5

0.6***

0.6***

0.2*

0.2*

Ranunculus alpestris

1016 15

0,73 0,43

1530

29,5

0.1

0.1

Ranunculus montanus

1016 16

0,26 0,07

2109

4,2

0.1

0.3**

0.3*

0.3*

Salix alpina

1016 15

0,57 0,20

1072

16,0

0.1

1.9***

1.9***

0.9**

Salix reticulata

1016 13

0,69 0,31

1045

25,0

0.9**

1.7***

3.3***

3.4***

Salix retusa

1016

0,66 0,33

1591

20,2

0.1

0.2

0.9***

Sesleria albicans

1008 14

0,48 0,23

2106

12,3

0.1

0.3*

1***

Silene acaulis

1016

0,65 0,40

2158

21,4

0.1

0.2

Silene alpestris

1016 14

0,49 0,18

1200

13,1

0.2

0.1

0.3*

Soldanella alpina

1008

0,41 0,15

1660

8,7

0.1

0.3*

0.6**

0.4*

Thlaspi alpestre

1016

0,27 0,06

1328

4,3

1.6***

0.7**

1.1***

1.4***

Thymus praecox ssp. polytrichus

1008 16

0,54 0,20

1408

13,3

0.1

0.5**

1.4***

4***

Tofieldia calyculata

1008 16

0,56 0,17

732

17,4

0.2

0.3

0.7*

2.1***

Trollius europaeus

1008 13

0,41 0,13

1153

10,6

0.3

0.6**

1.4***

0.9**

Veratrum album

1016 17

0,75 0,41

1197

29,9

0.2**

1.1***

0.8***

Viola biflora

1016 12

0,38 0,17

2092

9,1

0.3**

0.6***

0.9***

1.1***

64

Table 3. ANOVA results of the proportional deviance reduction when adding spatial vectors (at four
different radii) to the environment-only models explained by colonisation traits found in 71 alpine
plant species.
Sum of Sq
d.f.
F
p-value
radius of spatial vector

23.1

15.8

< 0.001

dispersal mode

1.9

3.9

0.048

8.1

0.005

length of diaspore * dispersal mode

2.9

5.9

0.016

radius * dispersal mode

1.6

1.1

0.364

radius * length of diaspore

7.8

5.3

0.001

radius * dispersal mode * length of diaspore

2.1

1.4

0.236

Residuals

126

258

length of diaspore

65

Table 4. ANOVA results of the proportional deviance reduction when adding spatial vectors (at four
different radii) to the environmental only models explained by competition traits found in 71 alpine
plant species. No interactions are significant (not shown, p> 0.5).
Sum of Sq
d.f.
F
P
radius of spatial vector

15.8

12.7

< 0.001

lateral spread

3.2

7.7

0.006

mean height

0.4

0.325

canopy closure

3.7

8.9

< 0.003

103.9

250

residuals

66

Figure 1. Relationship between residual spatial deviance reduction (% decrease when adding spatial
vectors to the environment-only models) of all plant species and their dispersal length (mm) at four
different radii and separated by dispersal mode (see table 1 for details on functional attributes).

0.0

1.5

1.0

0.5

0.0

0
2

10

160 m radius

wind disperser

0.5

no wind disperser

15

10

320 m radius

10

4
5
2
0
15

0
8
6

no wind disperser

proportional deviance reduction %

1.0

80 m radius

wind disperser

40 m radius

1.5

10

4
5
2
0

0
0

10

diaspore length mm

67

10

Figure 2. Residual spatial deviance reduction (% decrease when adding spatial vectors to the
environment-only models) of plant species with and without wind dispersal at four different radii (see
table 1 for details on functional attributes). Box plots indicate median (white line), interquartile
distance (IQD; gray boxes), extreme values (whiskers at 1.5 x IQD) and outliers (dots beyond 1.5 x
IQD).

80 m radius

40 m radius

proportional deviance reduction %

0
0
8

320 m radius

160 m radius
13

5
8

2
3

no

yes

no

wind disperser

68

yes

Figure 3. Relationship of residual spatial deviance reduction (% decrease when adding spatial vectors
to the environment-only models) of plant species with different potential for lateral spread at four
different radii (see table 1 for details on functional attributes).

40 m radius

80 m radius

proportional deviance reduction %

0
0
8
160 m radius

320 m radius
13

5
8

2
3

20

40

60

80

100

lateral spread mm

69

20

40

60

80

100

4
Dullinger, S., Dirnbck, T., Greimler, J. & Grabherr, G.: A resampling
approach for evaluating effects of pasture abandonment on subalpine plant
species diversity.

Manuskript, zur Publikation akzeptiert in der Zeitschrift

Journal of Vegetation Science

70

A resampling approach for evaluating effects of pasture

abandonment on subalpine plant species diversity

Dullinger, Stefan1*; Dirnbck, Thomas2; Greimler, Josef3 & Grabherr, Georg1

1

Institute of Ecology and Conservation Biology, University of Vienna, Althanstrae 14, A-1090
Vienna;
2
Federal Environmental Agency; Spittelauer Lnde 5, A-1090 Vienna, Austria;
3
Department of Higher Plant Systematics and Evolution, Institute of Botany, University of Vienna,
Rennweg 14, A 1030 Vienna;
*Corresponding author; FAX +43 1 4277 9542; E-mail: dull@pflaphy.pph.univie.ac.at

ABSTRACT
The decline of species-rich semi-natural calcareous grasslands is a major conservation problem
throughout Europe. Maintenance of traditional animal husbandry is often recommended as an
important management strategy. However, results that underpin such management recommendations
were derived predominantly from lowland studies and may not be easily applicable to high mountain
areas. In this study we analyse the importance of traditional low-intensity summer farming (cattle
grazing) for vascular plant species diversity of a subalpine region in the Northern Calcareous
Alps/Austria by resampling from an existing dataset on its vegetation. Results indicate a significant
long term decline of plant species diversity following abandonment at the landscape scale. In contrast,
within-community effects of pasture abandonment on plant species diversity are equivocal and
strongly depend on the plant community. We suppose these differences to be due to diet preferences of
cattle as well as to the differential importance of competition for structuring the respective
communities. From our results we infer that the main mechanism by which pasture abandonment
affects vascular plant species diversity, at least during the first about 100 years documented here, are
not local-scale competitive exclusion processes within persisting communities. Instead, postabandonment successional community displacements that cause a landscape scale homogenisation of
the vegetation cover seem to be primarily responsible for a decline of species diversity. We conclude,
that successful management of vascular plant species diversity in subalpine regions of the
Northeastern Calcareous Alps will depend on the maintenance of large scale pasture systems with a
spatially variable disturbance regime.
Keywords: Biodiversity; Conservation; Geographical Information System; Grazing; High mountain
summer farming; Land use history; Northern Calcareous Alps; Scale; Succession
Nomenclature: Adler et al. (1994)
Abrreviations: DEM = Digital Elevation Model; GIS = Geographical Information System; WET =
Topographic Wetness Index; EROS = Topographic Erosion Index; GLM = Generalized Linear Model;

71

S = Species Richness; H = Shannon Index of diversity; E = Evenness; CC = Number of plant

communities; dC = Berger-Parker-Index of community diversity
Number of words: 6247

72

INTRODUCTION
In temperate and mediterranean Europe many secondary, semi-natural grasslands are getting lost due
to economically motivated cessation of traditional pasture systems (Plachter 1991, Luick 1998, Zervas
1998). These grasslands, and especially the calcareous ones, are often particularly species-rich (e.g.
Kull & Zobel 1991, Gigon & Leutert 1996, Niemel & Baur 1998, Austrheim et al. 1999). Thus, the
maintenance or re-establishment of low-intensity grazing systems has become an important strategy of
biological conservation throughout Europe (Bobbink & Willems 1993, Zobel et al. 1996, Zervas 1998,
Ostermann 1998, Austrheim et al. 1999, Pykl 2000, Barbaro et al. 2001).
However, although vascular plant species richness and diversity of various natural and semi-natural
grasslands have been demonstrated to increase under moderate grazing (Rusch & Oesterheld 1997,
Collins et al. 1998, Zhang 1998, Sternberg et al. 2000), effects of grazing on biodiversity vary
considerably among ecosystems (Huntly 1991, Milchunas & Lauenroth 1993) and among different
taxa (Chaneton & Facelli 1991, Milchunas et al. 1998, Olff & Ritchie 1998, Prt & Sderstrm 1999,
Rambo & Fateh 1999, Balmer & Erhardt 2000). Moreover, patterns of biodiversity and their driving
processes obviously vary with spatial and temporal scale (Auerbach & Shmida 1987, Palmer & White
1994, Huston 1994, 1999, Cornell & Karlson 1997, Weiher 1999, Rahbek & Graves 2000, Crawley &
Harrel 2001, Whittaker et al. 2001). The maintenance or re-introduction of traditional pasture systems
should, therefore, not be uncritically promoted as a prime strategy of biodiversity conservation without
verification (Pykl 2000).
Seasonal farming in high mountain areas has a long tradition in many parts of the world (Messerli &
Ives 1997). Subalpine and alpine areas of the Northeastern Calcareous Alps have been used for
summer pasturing at least since medieval times (Btzing 1991) with profound effects on subalpine
vegetation (Mayer 1976, Ellenberg 1988). Indigenous coniferous forests have been fragmented to a
mosaic of forest, shrub and grassland patches and the treeline has been lowered substantially (Kral
1970). However, as a consequence of declining summer farming during the last 150 years (Zwittkovits
1974, Groier 1993) subalpine forest and shrubland species are successively re-invading abandoned
grasslands.
It is inevitable that the decline or even the end of traditional summer farming will cause the loss of
valuable cultural landscapes throughout the Alps (Stone 1992, Krner 1999), but the effects of pasture
abandonment on biodiversity are poorly studied in high mountain areas (Austrheim & Eriksson 2001).
In this paper we investigate the consequences of pasture abandonment on the vascular plant species
diversity of four mountain ranges of the Northeastern Calcareous Alps in Austria. The area
investigated is part of Viennas drinking water catchments and subject to an interdisciplinary research
project on the ecology of karst ecosystems and different aspects of karst water management (see e.g.
Dirnbck & Grabherr 2000). Analyses are done by resampling from an extensive dataset collected in
the coarse of mapping the vegetation of the area and additional research projects. We address the
following questions:
(1) Does the cessation of low-intensity subalpine summer farming significantly alter plant species
diversity at the landscape scale and, if so, does this effect vary with the use of different plot sizes?
(2) Are local scale within-community effects of pasture abandonment on species diversity similar to
landscape scale effects?

73

STUDY AREA

The study area covers the subalpine and alpine zone of four neighbouring mountain ranges of the
Northeastern Calcareous Alps (Mt. Hochschwab, Mt. Schneealpe, Mt. Rax and Mt. Schneeberg, 15 E
to 16 E and 4730 N to 4750 N, approximately 150 km, see Fig. 1). Summits vary between 1900
and 2300 m asl. The mountain system is formed by Mesozoic limestone and dolomites. Climatic
conditions are temperate humid.
Summer pasturing (June to September) in the area dates back at least to the 16th century (Hafner 1979,
Zwittkowits 1974). Most of the area has been historically influenced by livestock grazing. Since the
middle of the 19th century grazing intensity has decreased and much former pastureland has become
abandoned. Approximately 30% of the area is still pastured by free-ranging cattle at an intensity of
about 0.5 cattle/ha (data from official sources, Appendix: No. 22), other forms of land use are of minor
importance.
Under natural conditions Norway spruce (Picea abies) or Norway spruce and European larch (Larix
decidua) forests together with prostrate pine (Pinus mugo) shrubland dominate from 1400 m asl. up to
the treeline (1800 - 1900 m asl.) (Zukrigl 1973). Today, the subalpine belt is a mosaic of woody and
non-woody vegetation with spruce and spruce-larch forests rarely occurring above 1600 m asl. Nonwoody vegetation below the treeline mainly consists of different kinds of pastures and natural
grasslands (see Table 1) with the latter covering disturbed sites like avalanche paths and exposed
ridges or substituting abandoned pastures as a first step of secondary succession. Less frequent, tall
herb communities (Adenostyles alliaria, Aconitum napellus) and fens (Carex nigra, C. rostrata) occur.
Above the treeline, natural grasslands dominate with a gradual switch from prevailing Carex
sempervirens to Carex firma grasslands with increasing altitude. Additionally, rock faces, scree and
snowbeds are widespread from the valley bottoms up to the summits, covering about 25% of the area.

METHODS

Data sets
The analyses in this paper were done by resampling data originally collected for mapping the
vegetation of the area (Greimler & Dirnbck 1996, Dirnbck & Greimler 1997, Dirnbck et al. 1998,
1999, Dullinger et al. 2001) and research on climate and land use change driven vegetation dynamics
(Dirnbck et al., in press). These data consist of
(1) 960 vegetation samples (= relevs) distributed over the whole study area. These samples represent
a subset of a larger database selected with respect to the below mentioned plot size criteria. Relevs
were done according to standard phytosociological techniques: All plant species present in the relev
were recorded using a 7-level ordinal cover-abundance scale (Braun-Blanquet 1964). Plot size was
chosen to represent the minimum area of the respective vegetation types: 30 m for grasslands, rock,
74

scree and snowbeds, 150 m for Pinus mugo-shrubland and 300 m for forests (see e.g. Westhoff &
van der Maarel 1973). Position of the relevs was determined subjectively regarding homogeneity of
the stand and representation of all habitats in the area. The position was either marked on an Infrared
(IR) or Black and White (BW) orthophoto (scale 1:5000 1.10000) or registered by a GPS (GARMIN
E-TREXc, mean Root Mean Square Error: 7.5 m). All relevs were collected between 1994 and 2001
by the authors and syntaxonomically classified according to Mucina et al. (1993a), Grabherr &
Mucina (1993) and Mucina et al. (1993b).
(2) A vegetation map (1:10.000) of about 75% of the study area. Mapping was done in the field by
delineating plant communities on IR- and BW-orthophotos (scale: 1:5.000 1:10.000). We
distinguished patches to a minimal size of about 10 x 10 m. Field maps were digitised and captured in
a Geographical Information System (GIS).
Land use information about the study area was derived from cadastral maps with individual parcels
and a suite of documents referring to their land use status at different times (see Appendix for a list of
the documents). From these documents we derived dates of abandonment for each parcel of the
cadastre as exactly as possible, usually as the period between the last record of active pasturing and the
first one reporting the parcel to be abandoned. As delineation of parcels in high mountain areas is
usually very coarse we refined the boundaries by adjusting them to landscape topography in that we
excluded all inaccessible rock faces and scree areas from actual- or historical-pastured areas. For this
refinement we used the same orthophotos as for vegetation mapping.
For information on abiotic habitat conditions we used geological maps (scale 1:50000, Geological
Survey of Austria, unpublished), and a digital elevation model (DEM, 20 m resolution, Austrian
National Mapping Agency). Based on the DEM we calculated altitude and slope (inclination in ) as
well as solar radiation at the beginning (15th of May), in the midst (15th of July) and at the end (15th of
September) of the growth period. Solar radiation was calculated using SOLARFLUX (Dubayah &
Rich 1996) assuming clear sky conditions. Additionally, we derived two topographical indices from
the DEM addressing topographically driven habitat properties. The topographic wetness index (WET)
provides relative estimates of soil moisture conditions based on slope inclination and drainage area
(Moore et al. 1993). It was also used to model snow distribution patterns (Hartmann et al. 1999).
EROS is an erosion index aimed at characterising the spatial distribution of soil loss and deposition
potential (Moore et al. 1993). Both indices were calculated by means of the program package Tapes-G
(Gallant & Wilson 1996).

Resampling design Landscape scale effects

Resampling strategies varied for landscape scale and within-community analyses. For landscape scale
analyses we used nested grid cells of five different sizes (termed plots hereafter). The basic plot size
was set to 500 x 500 m as 25 ha is about the area used for cattle pasture on the smallest of the summer
farms in the study area (data from official sources, Appendix Nr. 22). The study area was partitioned
into grid cells of that size and 16 such cells were selected for each treatment (actually pastured vs.
abandoned). The selection was based on the following constraints:
As succession in high mountain habitats is supposed to proceed very slowly (e.g. Wildi & Schtz
2000) we restricted comparisons on areas abandoned before 1927, most of them between 1850 and
1900. This date was chosen in order to maximise the time span since abandonment and retain an area
sufficiently large for analysis.
To exclude abiotic habitat effects on species diversity, we derived homogeneous site clusters. The
mean and standard deviation of all DEM-based habitat variables as well as the percentage of
limestone, dolomite and Quartenary sediments (mainly moraine materials) for each 500 x 500 m grid
75

cell were used as independent cluster variables. The standard deviation of the respective variable is
supposed to be an indicator of intra-plot environmental variability. We used Partitioning Around
Medoids (Kaufmann & Rousseeuw 1990, implemented in S-Plus 2000 for Windows) for clustering the
pooled set of pastured and abandoned grid cells (178 grid cells, partitioned into 40 clusters). Sampling
plots were selected by drawing pairs of abandoned and pastured grid cells from clusters containing
cells of both land use types with the restriction that there may not remain any significant differences (p
< 0.05) in none of the tested variables (see Table 2).
For each of the 32 selected 500 x 500 m plots we used nested grid cells of 500, 500/3 (= 166.667 m),
500/5 (100 m), 500/7 (71.528) and 500/9 (55.55 m) metres cell size. According to conventional
definitions (e.g. Whittaker et al. 2001) all these plot sizes may be termed landscape scale plots, except
for the smallest being of transitional size between landscape and local scale.
Within these plots we defined the 9 intersection points of a regular 16 cell-subgrid as sampling points.
Any such intersection point was assigned a relev (from the overall pool of 960) with respect to its
vegetation type (derived from the vegetation map), its land use status (derived from the cadastral
maps) and its abiotic habitat conditions (derived from the DEM-based habitat variables). The
assignment procedure selected all relevs from areas of the same land use type (pastured or abandoned
before 1927) representing the same plant community as mapped at the respective point and than
picked out the one most similar relev in terms of abiotic habitat conditions (Euclidean Distance on
standardised habitat variables).
Resampling design Within-community effects
To investigate within-community effects of abandonment we focused on the seven most important
grassland types in the study area. For this analysis we used all relevs of the respective grasslands
provided that they had been collected either on parcels of the cadastral map belonging to still active
summer farms or to farms which were abandoned before 1927.

Data analysis Landscape scale effects

At the landscape scale species richness (S) of each plot was calculated as the cumulative sum of
different species in all selected 9 relevs. We also calculated the mean Shannon diversity index (H)
and Evenness (E) (Margurran 1988) for each plot after numerical transformation of ordinal coverabundance values according to Dierschke (1994). Community diversity was measured as the number
of different plant communities (community count = CC) per plot and by means of the Berger-ParkerIndex (dC) (Berger & Parker 1970). The importance of land use for explaining existing differences in
these biodiversity measures was evaluated using Generalized Linear Models (GLM). We used altitude,
slope, solar radiation in July, WET, EROS, and land use status as independent variables and tested for
the increase in Residual Deviance when excluding land use information from the model. We
performed an ANOVA model comparison for formally testing these differences (see e.g. Selvin 1998).
GLMs were formulated as Poisson models with a log-link for count data (S, CC) and as Gaussian
models with an identity-link for all indices (H, E, dC). Poisson models were compared by Chi-Square
tests, Gaussian models by F-tests, respectively.
Within the landscape scale we tested for scale dependency of abandonment effects by plotting the
results for S, H, E, CC and dC against the logarithm of plot size (500, 166, 100, 72, 55 m). Linear
regression lines were then calculated and their slopes were compared between pastured and abandoned
areas. The rationale of this test was that if land use affected species diversity differently at different
scales the slopes of these curves should differ significantly between pastured and abandoned

76

conditions (Olff & Ritchie 1998). Formally, we tested differences in means of the slope coefficient
for each nested set of plots.
Data analysis Within-community effects
Within-community responses to abandonment were evaluated by comparing S, H and E of relevs of
each community form either pastured or abandoned areas. The significance of land use status for the
explanation of within-community diversity patterns was evaluated by the same type of GLM-analysis
as described above.
We used the Geographical Information System ARC-Info (7.1) for handling and analysing spatial data
and S-Plus for Windows 2000 for all statistical analyses. The test statistics for sample comparisons
were a standard two sided t-test for continuous variables if normally distributed (examined by onesample Kolmogorov-Smirnoff test), and a Wilcoxon rank sum test otherwise as well as for count data
(S, CC).

RESULTS
Landscape scale effects
Differences in mean plant species richness S between pastured and abandoned conditions are highly
significant no matter which plot size is used (Table 3). A comparison of effective sampling areas
(Table 3) demonstrates that the decrease in S is not an artefact due to systematically smaller relevs of
abandoned plots. On the contrary, sampling areas are considerably larger there, a fact mainly due to
the higher proportion of Pinus mugo-stands (cf. Table 1 and Fig. 3). We also tested the results against
possible bias due to systematically assigning ecologically more similar relevs to either pastured or
abandoned sampling points, but means of Euclidean Distances (per plot) do not differ significantly
(Table 3).
Results for the diversity indices H and E are similar to those for S (Table 3). Thus, abandonment
causes not only a decline in species numbers but also a less even distribution of abundances. The
trends for community diversity are the same as those for species diversity, though differences in means
are significant only for the two larger plot sizes. The main successional pathway after pasture
abandonment, visible from Fig. 3, is a considerable increase in Pinus mugo-shrubland and a rapid
decline of Leontodon hispidus-Crepis aurea pastures.
The GLMs, using both abiotic variables and land use information, explain between 30 and 70% of the
variance in S, H and E (Table 3). In any case land use status significantly improves model performance
and in most cases accounts for more than 50% of the explained variance.
There is no indication for any scale effects on these results. The slopes of the regression lines for
species diversity measures are nearly identical among pastured and abandoned plots, those for
community diversity measures differ slightly but not significantly (Fig. 2).

Within-community effects
Considering within-community effects the unequivocal results of the landscape scale patterns are not
confirmed. Diversity effects of abandonment are obviously dependent on the community under
analysis. Just one of six grassland types shows a significant decrease in S (Table 4). Comparisons of H
77

and E show some more significancies. Moreover, the decrease in E may be underestimated due to the
coarse resolution of the Braun-Blanquet cover-abundance values.
The GLMs, using abiotic habitat and land use information, explain up to 65% of within-community
diversity patterns (Table 4). However, model performance is quite variable and a few models are not
even significant (e.g. S for Carex sempervirens-grasslands, H for Agrostis alpina-Festuca pumilagrasslands). Neither S nor H or E show consistent trends across the different grasslands analysed and
most of the models are not significantly improved when incorporating land use information. Thus,
some of the existing differences (e.g. H for Nardus stricta-pastures, E for Carex sempervirensgrasslands) should not be attributed to abandonment. On the other hand, some effects of abandonment
seem to be hidden by differences in abiotic habitat conditions (e.g. S for Deschampsia cespitosagrasslands, E for Leontodon hispidus-Crepis aurea-pastures). The only communities with a rather
unequivocal negative diversity response to abandonment when abiotic habitat conditions are
statistically kept constant are Carex ferruginea- and Deschampsia cespitosa-grasslands.

DISCUSSION
Before interpreting the results of this study some problems associated with the methodological
approach are in need of discussion. First, the data we resample from were collected for other research
and not for testing the hypotheses we focus on in these analyses. Instead, they should primarily
provide an inventory of phytosociologically defined plant communities of the area. The use of
phytosociological data as a basis for diversity assessments has recently been criticised with respect to
a lack of objectivity in localising and determining the size of a relev plot (Chytr 2001). This
criticism is valid especially with regard to analyses of large databases which many authors have
contributed to. However, since all relevs we resample from in this paper were done by the same small
group of investigators using constant plot sizes for certain vegetation types we suppose the data to be
appropriate to biodiversity analyses.
A second point is that at the landscape scale we used an assignment procedure that shifted relevs
taken elsewhere to certain points in our sampling design. Though the filters used for selection of these
relevs should guarantee the assignment of the most probable one out of quite a large number of such
relevs from the same study area, it is beyond question that such a method will never exactly meet the
real species composition of the respective sampling points. Neither species composition nor species
numbers of the landscape scale plots must therefore be taken as real values but as approximations.
Hence, we may not be able to derive from these data exact rates of species loss due to abandonment on
the plots analysed. Nevertheless, as these constraints uniformly apply to both pastured and abandoned
plots, we do not assume that they affect comparisons between them seriously, at least if they are
restricted to the issues raised here.
At last, we emphasise that our approach is strictly focused on comparisons of present day vegetation.
No information on the vegetation of individual relev-plots at the time of abandonment was available.
Thus, the post-abandonment successional history of plots covered by the same plant community today
may differ. Taking account for a possible impact of such historical variability would require long-term
monitoring data which are not available for time frames spanning several to many decades. However,
Figure 3 illustrates that intra-grassland succession has likely played a minor role as almost all
grassland types have lost area and only Pinus mugo shrublands have expanded considerably. Most of
the grasslands still present in abandoned areas have probably persisted in situ.

78

Our results clearly demonstrate an unequivocal decrease of vascular plant species diversity as a long
term trend of pasture abandonment at the landscape scale. This effect does not vary with plot sizes
over two orders of magnitude. Simultaneously a homogenisation of the vegetation cover, that is a
reduction of plant community diversity, occurs. Hence, the main mechanism by which summer
farming promotes subalpine species diversity at the landscape scale seems to be making the vegetation
more patchy. Grazing, shrub clearing and shrub encroachment shape subalpine pastures usually
towards a mosaic of grasslands and fragments of Pinus mugo-shrubland and, at lower altitudes, Picea
abies-Larix decidua forests. Moreover, differential grazing intensity and behavioural characteristics of
cattle may promote heterogeneity of the grassland matrix itself by reallocation and selective
concentration of nutrients, soil disturbance and compaction (Hobbs & Huenneke 1992, Krner 1999,
Austrheim & Eriksson 2001). Making the vegetation more patchy is thus a result of increasing habitat
heterogeneity, primarily in terms of light availability (opening up closed forest and shrubland
canopies), but also in terms of other properties of the physical environment (e.g. loss of organic
matter, snow regime, wind exposure). Pasture abandonment reverses these trends and homogenises the
vegetation cover towards large closed Pinus mugo-stands. Our results may therefore well be
interpreted in agreement with the theory of habitat diversity as an essential driver of species diversity
at landscape scales (Auerbach & Shmida 1987, Huston 1994, Rosenzweig 1995).
At the local scale within-community effects of grazing on plant species diversity are commonly
interpreted in the context of the Intermediate Disturbance Hypothesis (Conell 1978). It states that
diversity is promoted by disturbance levels that effectively reduce the vitality of competitive
dominants but are not strong enough to completely exclude species from the community (e.g. Huntly
1991, Milchunas & Lauenroth 1993). However, both empirical studies and mathematical model
analyses have shown that, in general, predator effects on prey diversity will depend on both exact
competitive relationships among prey species and on diet preferences of the predators (e.g. Lubchenko
1978, Pacala & Crawley 1992). Our findings which indicate considerable variation in the effect of
herbivores on within-community species diversity across different grassland types confirm these
predictions, even though we have focused on a group of closely related communities, both in terms of
floristic composition and geographic distribution. A negative correlation between competitive ability
and palatability prevents positive effects of herbivores on plant diversity because they will then not
preferably feed on the competitive dominant species (Pacala & Crawley 1992). This is likely to be the
case for grassland types dominated by sclerophyllous sedges and grasses like Carex firma, Nardus
stricta or, to a lesser degree, Carex sempervirens (see e.g. Ellenberg 1974 for species information on
leaf sclerophylly). Furthermore, models of herbivore effects on plant diversity usually imply distinct
hierarchies in competitive rank of plant species (e.g. Shurin & Allen 2001). Such effects will
presumably decrease with decreasing differences in competitive ability between the species of a
community and this may explain why communities lacking any pronounced dominants like Leontodon
hipidus-Crepis aurea pastures or Agrostis alpina-Festuca pumila grasslands do not respond to pasture
abandonment in terms of plant species diversity or even show an increase of diversity with
abandonment (see increasing E for Leontodon hipidus-Crepis aurea pastures in Table 4). Disturbance
is even less likely to promote diversity, if biotic interactions among plants are primarily facilitative
instead of competitive. In fact, recent research has provided indication for a prevalence of facilitation
in stressful alpine environments (Callaway et al. 2002). In the study area, Carex firma- and, to a lesser
degree, Agrostis alpina-Festuca pumila-grasslands predominate in the harshest habitats. Although not
significant, both communities show a trend towards increasing species richness with pasture
abandonment.
In contrast, we may expect positive effects of herbivores on species diversity of productive habitats
that are dominated by high growing and thus highly competitive graminoids (Grime 2001, Weiher et
79

al. 1999). Indeed, the two communities that proved most sensitive to pasture abandonment in terms of
species diversity, Carex ferruginea and Deschampsia cespitosa grasslands, are among the most
productive grassland communities of the Calcareous Alps (Ellenberg 1988, Ellmauer & Mucina 1993,
Mucina & Karner 1993). Above-ground net primary production was also the factor most closely
correlated to a species richness response to grazing in a world-wide comparison of grassland studies
performed by Milchunas & Lauenroth (1993).
Our findings suggest that the decline of vascular plant species diversity resulting from the cessation of
traditional low-intensity summer farming in subalpine regions of the Northeastern Calcareous Alps is
not primarily due to the influence of cattle grazing on the structure of individual communities. The
main effect on species diversity comes from successional community displacements. The local scale
effect of these displacements is ambiguous as they do not show an unequivocal trend from more
diverse to less diverse communities: Both species poor (e.g. Pinus mugo-shrubland) and species rich
(e.g. Carex sempervirens-grasslands) communities may profit from abandonment, at least during the
first around one hundred years documented in our study (see Fig. 3). Instead, the decrease of vascular
plant species diversity is rather a landscape scale process. It is driven by the balance of successional
pathways being strongly biased towards an increase of comparatively species poor Pinus mugo
shrubland and by a general reduction of community diversity. This homogenisation of the vegetation
cover may cause a reduction of landscape-scale diversity even at increasing frequencies of species-rich
communities like Carex sempervirens-grasslands, trading a higher alpha- against a lower betadiversity. From these findings we conclude that to function as a successful tool of plant species
diversity management summer farming should best be applied to large areas coupling a heterogeneous
abiotic environment with a spatially varied disturbance regime.

ACKNOWLEDGEMENTS
We would like to thank Andreas Exner, Johannes Peterseil and Norbert Sauberer for fruitful
discussions, Max Abensperg-Traun for linguistic improvement and Martin Zobel, Andreas Stampfli,
Gunilla Olsson and an anonymous referee for valuable criticism on earlier versions of the manuscript.
The collection of data analysed in this study was financed by the Water Department of the City of
Vienna.

REFERENCES
Adler, W., Oswald, K. & Fischer, R. 1994. Exkursionsflora fr sterreich. Eugen Ulmer, Stuttgart.
Auerbach, M. & Shmida, A. 1987. Spatial scale and the determinants of plant species richness. Trends Ecol.
Evol. 2: 238-242.
Austrheim, G., Gunilla, E., Olsson, A. & Grontvedt, E. 1998. Land-use impact on plant communities in seminatural subalpine grasslands of Budalen, central Norway. Biol. Conserv. 87: 369-379.
Austrheim, G. & Eriksson, O. 2001. Plant species diversity and grazing in the Scandinavian mountains patterns
and processes and different scales. Ecography 24: 683-695.
Balmer, O. & Erhardt, A. 2000. Consequences of succession on extensively grazed grasslands for central
european butterfly communities: rethinking conservation practices. Conserv. Biol. 14: 746-757.

80

Barbaro, L., Dutoit, T. & Cozic, P. 2001. A six year experimental restoration of biodiversity by shrub-clearing
and grazing in calcareous grasslands of the French Prealps. Biodivers. and Conserv. 10: 119-135.
Btzing, W. 1991. Die Alpen. Entstehung und Gefhrdung einer europischen Kulturlandschaft. C.H. Beck,
Mnchen.
Berger, W.H. & Parker, F.L. 1970. Diversity of planktonic Foraminifera in deep sea sediments. Science 168:
1345-1347.
Bobbink, R. & Willems, J.H. 1993. Restoration management of abandoned chalk grassland in the Netherlands.
Biodivers. and Conserv. 2: 616-626.
Braun-Blanquet, J. 1964. Pflanzensoziologie. 3rd ed. Springer, Wien.
Callaway, R.M., Brooker, R.W., Choler, P., Kikvidze, Z., Lortie, C.J., Michalet, R., Paolino, L., Pugnaire, F.I.,
Newingham, B., Aschehoug, E.T., Armas, C., Kikodze, D. & Cook, B.J. 2002. Positive interactions among
alpine plants increase with stress. Nature 417: 844-848.
Connell, J.H. 1978. Diversity in tropical rainforests and coral reefs. Science 199: 1302-1310.
Cornell, H.V. & Karlson, R.H. 1997. Local and regional processes as controls of species richness. In: Tilman, D.
& Kareiva, P. (eds.) Spatial Ecology, pp. 250-270. Princeton University Press, Princeton.
Chaneton, E.J. & Facelli, J.M. 1991. Disturbance effects on plant community diversiy: spatial scales and
dominance hierarchies. Vegetatio 93: 143-155.
Collins, S.L., Knapp, A.K., Briggs, J.M., Blair, J.M. & Steinauer, E.M. 1998. Modulation of diversity by grazing
and mowing in native tallgrass prairie. Science 280: 745-747.
Crawley, M.J. & Harral, J.E. 2001. Scale dependence in plant biodiversity. Science 291: 864-868.
Dierschke, H. 1994. Pflanzensoziologie. Eugen Ulmer, Stuttgart.
Dirnbck, T., Dullinger, S., Gottfried, M. & Grabherr, G. 1999. Die Vegetation des Hochschwab (Steiermark).
Alpine und subalpine Stufe. Mitteilungen des Naturwissenschaftlichen Vereins der Steiermark 129: 111-251.
Dirnbck, T., Dullinger, S. & Grabherr, G. 2003. A regional impact assessment of climate and land use change
on alpine vegetation. J. Biogeography, in press.
Dirnbck, T. & Grabherr, G. 2000. GIS assessment of vegetation and hydrological change in a high mountain
catchment of the Northern Limestone Alps. Mountain Research and Development 20: 172-179.
Dirnbck, T. & Greimler, J. 1997. Subalpin-alpine Vegetationskartierung der Raxalpe, Nordstliche Kalkalpen.
Linzer Biologische Beitrge 29: 299-339.
Dirnbck, T., Greimler, G. & Grabherr, G. 1998. Die Vegetation des Zeller Staritzen Plateaus (Hochschwab,
Steiermark) und ihre Bedeutung fr den Quellschutz. Mitteilungen des Naturwissenschaftlichen Vereins der
Steiermark 128: 123 183.
Dubayah, R. & Rich, P.M. 1996. GIS-based solar radiation modeling. In: Goodchild, M.F., Steyaert, L.T. &
Parks, B.O. (eds.) GIS and Environmental Modeling, pp. 129-134. John Wiley & Sons, New York.
Dullinger, S., Dirnbck, T. & Grabherr, G. 2001. Die subalpine und alpine Vegetation der Schneealpe,
Steiermark. Mitteilungen des Naturwissenschaftlichen Vereins der Steiermark 131: 83-128.
Ellenberg, H. 1988. Vegetation Ecology of Central Europe. Cambridge University Press, Cambridge.
Ellenberg, H. 1974. Zeigerwerte der Gefpflanzen Mitteleuropas. Scripta Geobotanica XI: 1-122. 2nd edition.
Goltze, Gttingen.
Ellmauer, T. & Mucina, L. 1993. Molinio-Arrhenatheretea. In: Mucina, L., Grabherr, G. & Ellmauer, T. 1993
(eds.) Die Pflanzengesellschaften sterreichs. Teil I, pp. 297-401. Gustav Fischer Verlag, Jena.
Gallant, J.C. & Wilson, J.P. 1996. Tapes-G: a grid based terrain analysis program for the environmental science.
Computers and Geoscience 22: 713-722.
Gaston, K.J. 2000. Global patterns in biodiversity. Nature 405: 220-227.
Gigon, A. & Leutert, A. 1996. The dynamic keyhole-key model of coexistence to explain diversity of plants in
limestone and other grasslands. J. Veg. Sci. 7: 29-40.

81

Grabherr, G. 1997. The high mountain ecosystems of the Alps. In: Wielgolaski, F.E. (ed.) Ecosystems of the
World, Vol. 3, Polar and Alpine Tundra, pp. 97-122. Elsevier, Amsterdam
Grabherr, G. & Mucina, L. 1993. Die Pflanzengesellschaften sterreichs. Teil II. Gustav Fischer Verlag, Jena.
Greimler, J. & Dirnbck, T. 1996. Die subalpine und alpine Vegetation des Schneebergs, Niedersterreich.
Linzer Biologische Beitrge 28: 437-482.
Grime, J.P. 2001. Plant Strategies, Vegetation Processes, and Ecosystem Properties. 2nd ed. John Wiley & Sons,
New York.
Hafner, F. 1979. Steiermarks Wald in Geschichte und Gegenwart. sterreichischer Agrarverlag, Wien.
Hartman, M.D., Baron, J.S., Lammers, R.B., Cline, D.W., Band, L.E., Liston, G.E. & Tague, C. 1999.
Simulation of snow distribution and hydrology in a mountain basin. Water Resources Research 35: 15871603.
Huntly, N. 1991. Herbivores and the dynamics of communities and ecosystems. Ann. Rev. Ecol. Syst. 22: 477503.
Huston, M.A. 1994. Biological diversity. Cambridge University Press, Cambridge.
Huston, M.A. 1999. Local processes and regional patterns: Appropriate scales for understanding variation in the
diversity of plants and animals. Oikos 93: 393-401.
Kaufmann, L. & Rousseeuw, P.J. 1990. Finding Groups in Data: An Introduction to Cluster Analysis. John
Wiley & Sons, New York.
Krner, C. 1999. Alpine Plant Life. Springer, Berlin.
Kral, F. 1970. Pollenanalytische Untersuchungen zur Waldgeschichte des Dachsteinmassivs. University for
Agricultural Sciences, Vienna.
Kull, K. & Zobel, M. 1991. High species richness in an Estonian wooded meadow. J. Veg. Sci. 2: 715-718.
Lubchenco, J. 1978. Plant species diversity in a marine intertidal community: importance of herbivore food
preferences and algal competitive abilities. Am. Nat. 112: 23-39.
Luick, R. 1998. Ecological and socio-economic implications of livestock-keeping systems on extensive
grasslands in south-western Germany. J. Appl. Ecol. 35: 979-982.
Margurran, A.E. 1988. Ecological diversity and its measurement. Chapman & Hall, London.
Mayer, H. 1976. Gebirgswaldbau Schutzwaldpflege. Gustav Fischer, Stuttgart.
Messerli, B. & Ives, J.D. 1997. Mountains of the World. The Parthenon Publishing Group, New York.
Messerli, P. 1989. Mensch und Natur im alpinen Lebensraum. Risiken, Chancen, Perpektiven. Paul Haupt, Bern.
Milchunas, D.G. & Lauenroth, W.K. 1993. Quantitative effects of grazing on vegetation and soils over a global
range of environments. Ecol. Monogr. 63: 327-366.
Milchunas, D.G., Lauenroth, W.K. & Burke, I.C. 1998. Livestock grazing: animal and plant biodiversity of
shortgrass steppe and the relationship to ecosystem function. Oikos 83: 65-74.
Moore, I.D., Gessler, P.E., Nielsen, G.A. & Peterson, G.A. 1993. Soil attribute prediction using terrain analysis.
Soil Science of America 57: 443-452.
Mucina, L., Grabherr, G. & Ellmauer, T. 1993. Die Pflanzengesellschaften sterreichs. Teil I. Gustav Fischer
Verlag, Jena.
Mucina, L., Grabherr, G. & Wallnfer, S. 1993. Die Pflanzengesellschaften sterreichs. Teil III. Gustav Fischer
Verlag, Jena.
Mucina, L. & Karner, P. 1993. Mulgedio-Aconitetea. In: Grabherr, G. & Mucina, L. (eds.) Die
Pflanzengesellschaften sterreichs. Teil II, pp. 468-505. Gustav Fischer Verlag, Jena.
Niemel, J. & Baur, B. 1998. Threatened species in a vanishing habitat: plants and invertebrates in calcareous
grasslands in the Swiss Jura mountains. Biodivers. and Conserv. 7: 1407-1416.
Olff, H. & Ritchie, M.E. 1998. Effects of herbivores on grassland plant species diversity. Trends Ecol. Evol. 13:
261-265.
Pacala, S.W. & Crawley, M.J. 1992. Herbivores and plant diversity. Am. Nat. 140: 243-260.

82

Palmer, M.W. & White, P.S. 1994. Scale dependence and the species-area relationship. Am. Nat. 144: 717-740.
Prt, T. & Sderstrm, B. 1999. Conservation value of semi-natural pastures in Sweden: contrasting botanical
and avian measures. Conserv. Biol. 13: 755-765.
Plachter, H. 1991. Naturschutz. Gustav Fischer Verlag, Jena.
Pykl, J. 2000. Mitigating human effects on European biodiversity through traditional animal husbandry.
Conserv. Biol. 14: 705-712.
Rahbek, C. & Graves, G.R. 2000. Detection of macro-ecological patterns in South American hummingbirds is
affected by spatial scale. Proc. R. Soc. Lond. B 267: 2259-2265.
Rambo, J.L. & Fateh, S.H. 1999. Effect of vertebrate grazing on plant and insect community structure. Conserv.
Biol. 13: 1047-1054.
Ricklefs, R.E. 1987. Community diversity: relative roles of local and regional processes. Science 235: 165-171.
Rosenzweig, M.L. 1995. Species diversity in space and time. Cambridge University Press, Cambridge.
Rusch, G.M. & Oesterheld, M. 1997. Relationship between producitivity, and species and functional group
diversity in grazed and non-grazed Pampas grassland. Oikos 78: 519-526.
Selvin, S. 1998. Modern Applied Biostatistical Methods Using S-Plus. Oxford University Press, Oxford.
Shurin, J.B. & Allen, E.G. 2001. Effects of competition, predation, and dispersal on species richness at local and
regional scales. Am. Nat. 158: 624-637.
Sternberg, M., Gutman, M., Perevolotsky, A., Ungar, D. & Kigel, J. 2000. Vegetation response to grazing
management in a mediterranean herbaceous community: a functional group approach. J. Appl. Ecol. 37: 224237.
Stone, P.B. 1992. The State of the Worlds Mountains. Zed Books, London.
Weiher, E. 1999. The combined effect of scale and productivity on species richness. J. Ecol. 87: 1005-1011.
Weiher, E., van der Werf, A., Thompson, K., Roderick, M., Garnier, E. & Erikson, O. 1999. Challenging
Theophrastus: A common core list of plant traits for functional ecology. J. Veg. Sci. 10: 609-620.
Westhoff, V. & van der Maarel, E. 1973. The Braun-Blanquet approach. In: Witthaker, R.H. (ed.) Ordination
and classification of communities, pp. 617-627. Junk, The Hague.
Whittaker, R., Willis, K. & Field, R. 2001. Scale and species richness: towards a general, hierarchical theory of
species diversity. J. Biogeography: 453-470.
Wildi, O. & Schtz, M. 2000. Reconstruction of a long-term recovery process from pasture to forest. Community
Ecology 1: 25-32.
Zervas, G. 1998. Quantifying and optimizing grazing regimes in Greek mountain systems. J. Appl. Ecol. 35:
983-986.
Zhang, W. 1998. Changes in species diversity and canopy cover in steppe vegetation in Inner Mongolia under
protection from grazing. Biodivers. Conserv. 7: 1365-1381.
Zukrigl, K. 1973. Montane und subalpine Waldgesellschaften am Alpenostrand. Mitteilungen der Forstlichen
Bundesversuchsanstalt Wien 101: 1-387.
Zwittkovits, F. 1974. Die Almen sterreichs. Selbstverlag, Zillingdorf.

83

Figure 1: Geographical position of the study area. Only the Alps are represented by topographical
shadings (A Austria, CH Switzerland, D Germany, F France, FL Liechtenstein, I Italy, MC
Monaco, SLO Slovenia).

84

Species Richness

130

p = 0.73

100

70

4.0

4.5

5.0

5.5

6.0

4.5

5.0

5.5

6.0

5.5

6.0

p = 0.35

Shannon-Index

3.3

2.7

2.1
4.0

0.74

Evenness

p = 0.58

0.66

0.58

Number of plant communities

4.0

4.5

5.0

p = 0.78
4

Berger-Parker-Index of community diversity

4.0

4.5

5.0

5.5

6.0

4.5

5.0

5.5

6.0

p = 0.13

2.3

1.7

1.1

4.0

Ln(Plot Size)

85

Figure 2: Comparison of linear regression lines for mean species richness, mean Shannon-Index, mean
Evenness, mean number of different plant communities and mean Berger-Parker-Index of plant
community diversity on the natural logarithm of plot size (55 m, 71 m, 100 m, 167 m, 500 m) for
pastured and abandoned plots. P-values come from comparison of mean slope coefficient between
16 sets of nested plots on actually pastured and 16 sets of nested plots on abandoned areas. Solid
symbols: Plots on actually pastured areas; Empty symbols: Plots on abandoned areas.

86

55 x 55 m

71 x 71 m

80
70

70
60

60
50

Count

Count

50
40

40
30

30
20

20

10

10

CFi

AF

CS

LC

NS

DS

PM

PL

CFi

AF

CS

Community

CFe

LC

NS

DS

PM

PL

DS

PM

PL

Community

Pastured
Abandoned

Pastured
Abandoned

100 x 100 m

167 x 167 m

80
70
70
60
60
50

Count

Count

50
40

40
30

30
20

20

10

10

0
CFi

AF

CS

CFe

LC

NS

DS

PM

PL

CFi

AF

Community

CS

CFe

LC

NS

Community

Pastured
Abandoned

Pastured
Abandoned

500 x 500 m

70

60

Count

50

40

30

20

10

0
CFi

AF

CS

LC

NS

DS

PM

PL

Community
Pastured
Abandoned

Figure 3: Frequencies of the most important plant communities of the study area on grazed vs.
abandoned plots (16 plots each) for all 5 plot sizes analysed. CFi Carex firma-grasslands, AF
Agrostis alpina-Festuca pumila-grasslands, CS Carex sempervirens-grasslands, CFe Carex
ferruginea-grasslands, LC - Leontodon hispidus-Crepis aurea pastures, NS Nardus stricta-pastures,
DS Deschampsia cespitosa-pastures, PM - Pinus mugo-shrubland, PL - Picea abies-Larix deciduaforests.

87

Community group

plant community

rock and scree-communities

rock communities

Natural alpine grasslands

Pastures

Woody vegetation

plot size (m)

area (%)

relevs

species

30

33

20

scree communities

30

48

20

snowbed communities

30

62

26

Carex firma grassland

30

118

28

Agrostis alpina-Festuca pumila grassland

30

67

37

Carex sempervirens grassland

30

140

39

Carex ferruginea grassland

30

46

41

Leontodon hispidus-Crepis aurea pasture

30

44

38

Nardus stricta pasture

30

40

27

Deschampsia cespitosa pasture

30

41

23

Pinus mugo shrubland

150

20

45

25

Picea abies (-Larix decidua) forests

300

25

20

42

Table 1: Plant communities covering more than 1% of the study area, percentage cover of the study
area, size of respective sampling plots (relevs) and mean number of vascular plant species per relev.

88

Variable
Altitude Mean (m)
Altitude StD (m)
Slope Mean ()
Slope StD ()
EROS Mean
EROS StD
WET Mean
WET StD
Radiation May Mean (MJ/m)
Radiation May Std (MJ/m)
Radiation July Mean (MJ/m)
Radiation July Std (MJ/m)
Radiation September Mean (MJ/m)
Radiation September Std (MJ/m)
Area Limestone (ha)
Area Dolomite (ha)
Area non-calcareous substrates
Area Quartenary sediments (ha)

abandoned

pastured

1704
47
22
10
-0.09
3.35
5.79
1.36
24.84
2.64
26.07
2.50
15.44
3.70
15.87
3.59
0
2.57

1712
41
20
10
-0.05
2.82
5.91
1.31
24.82
3.22
26.15
3.01
14.88
4.21
13.22
7.68
0
1.59

0.84
0.37
0.3
0.87
0.29
0.33
0.35
0.37
0.97
0.17
0.87
0.24
0.49
0.28
0.79
0.22
0.07

Table 2: Abiotic habitat conditions on 32 selected 500 x 500 m plots. Values are means for 16 plots on
abandoned vs. 16 plots on actually used summer farms. Mean and standard deviation of numerical
variables (Altitude, Slope, EROS, WET, Radiation) as well as the total area covered by geological
units were calculated for each plot based on a Digital Elevation Model with 20 m cell size. WET
Topographic Wetness Index, EROS Erosion Index. For explanation of indices see text.

89

55 x 55 m 71 x 71 m 100 x 100 m 167 x 167 m 500 x 500 m

91
96
106
108
128
63
65
71
79
100
0.0074
0.0042
0.0021
0.0029
0.0018
373
372
366
290
179
272
250
199
167
87
334
309
250
228
149
< 0.0001 < 0.0001
< 0.0001
< 0.0001
< 0.0001

Mean-p
Mean-a
p
ND
RD+
RDp

Mean-p
Mean-a
p
ND
RD+
RDp

3.14
2.36
0.0002
11.8
6.7
10.2
0.001

3.17
2.38
< 0.0001
10.5
5.1
8.4
0.0004

3.27
2.44
0.0001
12.1
5.6
8.8
0.0008

3.28
2.53
< 0.0001
9.2
3.6
7.1
0.0004

3.53
2.85
< 0.0001
7.3
2.6
6.2
< 0.0001

Mean-p
Mean-a
p
ND
RD+
RDp

0.70
0.58
0.0003
0.28
0.16
0.25
0.0008

0.70
0.58
< 0.0001
0.22
0.10
0.18
0.0001

0.70
0.58
0.0001
0.26
0.12
0.20
0.0004

0.70
0.59
< 0.0001
0.22
0.09
0.17
0.0009

0.73
0.62
< 0.0001
0.19
0.07
0.24
< 0.0001

CC

Mean-p
Mean-a
p
ND
RD+
RDp

2.3
1.9
0.37
16.2
14.9
15.2
0.61

2.6
2.1
0.17
15.2
12.2
13.1
0.57

3.1
2.2
0.056
15.6
11
11.6
0.45

3.4
2.4
0.037
17.1
10.2
12
0.175

4.5
3.4
0.017
14.2
7.3
10.2
0.09

dC

Mean-p
Mean-a
p
ND
RD+
RDp

1.55
1.34
0.15
5.2
4.4
4.6
0.27

1.61
1.41
0.22
6.7
5.5
5.6
0.48

1.74
1.46
0.11
7.7
6.1
6.3
0.54

2.07
1.50
0.009
10.8
6.4
8.4
0.009

2.74
1.89
0.005
22.9
13.3
18.6
0.004

Mean-ED-p
Mean-ED-a
p

1.10
0.91
0.445

1.05
0.89
0.359

0.96
0.91
0.736

0.98
1.03
0.723

1.03
1.02
0.966

ESA-p
ESA-a

8910
15780

9630
15180

9570
15240

10980
15060

10710
15120

Table 3: Comparison of vascular plant species richness (S), Shannon-Index (H), Evenness (E), number
of different plant communities (CC), and Berger Parker-Index of community diversity (dC) between
16 plots on abandoned and 16 plots on actually used summer farms using five different plot sizes.
Mean-p, Mean-a: Mean values under pastured or abandoned conditions. ND, RD+, RD-: Null
deviance, Residual deviance of a Generalized Linear Model (GLM) including, and Residual Deviance
of a GLM not including land use information. For abiotic habitat variables in the model see Text.
Mean-ED: Mean Euclidean Distance between abiotic habitat conditions of the 9 samples (relevs)
assigned to a plot and the abiotic habitat conditions of the sampling points the relevs were assigned to
(p: actually used; a: abandoned summer farms). ESA-p, ESA-a: Effective sampling area, i.e.
cumulative sum of area of all 9 relevs assigned to a plot (p: actually used; a: abandoned summer
farms).

90

Carex ferruginea grasslands

Deschampsia cespitosa pastures
Nardus stricta pastures
Agrostis alpina-Festuca pumila grasslands
Leontodon hispidus-Crepis aurea pastures
Carex firma -grasslands
Carex sempervirens -grasslands

n
29
23
29
32
26
38
61

Mean-p
48
24
30
36
40
29
40

Carex ferruginea grasslands

Deschampsia cespitosa pastures
Nardus stricta pastures
Agrostis alpina-Festuca pumila grasslands
Leontodon hispidus-Crepis aurea pastures
Carex firma -grasslands
Carex sempervirens -grasslands

29
23
29
32
26
38
61

Carex ferruginea grasslands

Deschampsia cespitosa pastures
Nardus stricta pastures
Agrostis alpina-Festuca pumila grasslands
Leontodon hispidus-Crepis aurea pastures
Carex firma -grasslands
Carex sempervirens -grasslands

29
23
29
32
26
38
61

Mean-a
34
21
22
37
36
32
39

p
0.0002
0.38
0.09
0.89
0.46
0.29
0.79

ND RD+ RDp
88
36
57 < 0.0001
91
65
72
0.009
0.170
118
42
44
0.175
50
47
49
0.220
85
51
52
0.378
79
66
67
0.824
46
42
43

Mean-p Mean-a
2.73
2.13
1.96
1.56
2.15
1.47
2.36
2.43
2.59
2.70
2.14
2.06
2.53
2.36

p
0.0002
0.1086
0.027
0.5151
0.5926
0.5792
0.0839

ND
5.88
7.96
8.49
2.41
2.79
5.25
4.29

RD+
3.01
5.53
4.11
2.20
1.71
4.10
3.47

RD4.60
7.08
4.79
2.25
2.06
4.10
3.51

p
0.003
0.050
0.070
0.459
0.064
0.993
0.401

Mean-p Mean-a
0.71
0.61
0.62
0.52
0.64
0.47
0.66
0.68
0.71
0.76
0.64
0.60
0.69
0.65

p
0.0057
0.093
0.0438
0.5484
0.2382
0.1086
0.0412

ND
0.23
0.46
0.43
0.11
0.07
0.25
0.22

RD+
0.15
0.35
0.24
0.09
0.04
0.17
0.18

RD0.19
0.44
0.29
0.09
0.05
0.17
0.18

p
0.015
0.058
0.035
0.754
0.021
0.671
0.350

Table 4: Comparison of vascular plant species richness (S), Shannon-Index (H) and Evenness (E) of
the seven most important grassland communities of the study area on actually used and on abandoned
summer farms. n number of samples (relevs). Mean-p, Mean-a: Mean values under pastured or
abandoned conditions. ND, RD+, RD-: Null deviance, Residual deviance of a Generalized Linear
Model (GLM) including, and Residual Deviance of a GLM not including land use information. For
abiotic habitat variables in the model see Text.

91

APPENDIX DATA SOURCES ON HISTORICAL AND ACTUAL LAND

USE
published data sources
Anonymus, no year. General-Waldt-Bereit-Berain- und Schtzungs-Commissions-Beschreibung im
Erbherzogthum Steyer de anno 1755. 28 volumes. Steyr.
Schmidl, A. 1831. Der Schneeberg in Unter Oesterreich. Doll, Wien.
Rabl, S. 1877. Die Raxalpe. Hlder, Wien.
Biedermann, C. 1882. Fhrer auf die Hohe Veitsch und die Schneealpe. Lechner, Wien.
Ronninger, K. 1893. Frsters Touristenfhrer in Wiens Umgebungen. 8th ed. Hlder, Wien.
Benesch, F. 1894. Special-Fhrer auf die Raxalpe. Artaria & Co, Wien.
Benesch, F. 1897. Special-Fhrer auf den Schneeberg. Artaria & Co, Wien.
Wittschieben, O. 1910. Die Alpen im Bezirke Aflenz in der Steiermark. Statistische Monatsschrift 36: 526-646.
Schneiter, F. 1930. Statistik und Hebung der Steirischen Almwirtschaft. Selbstverlag des Steiermrkischen
Landes-Alminspektorats, Graz.
Leeder, F. 1932. Rax und Schneeberg. Oesterreichische Vierteljahreschrift fr Forstwesen 82: 1-16.
Ballik, K. 1973. Hochlagenaufforstung Rax. Magistrat der Stadt Wien, Wien.
Zwittkovits, F. 1974. Die Almen sterreichs. Selbstverlag, Zillingdorf.
Hafner, F. 1979. Steiermarks Wald in Geschichte und Gegenwart. sterreichischer Agrarverlag, Wien.

Unpublished data sources

Franziszeischer Kataster. Kartenbltter fr die Katastralgemeinden Gutenstein, Hirschwang, Klein- und Groau,
Puchberg, Schwarzau, Altenberg, Aschbach, Flz, Kapellen, Mrzsteg, Seewiesen, Schattenberg, St. Ilgen,
Weichselboden, Wildalpen. Katastralarchiv des sterreichischen Bundesamtes fr Eich- und
Vermsessungswesen (Cadastral Archive of the Austrian Mapping Agency)
Protocolle und Operate zum Franziszeischen Kataster. Katastralgemeinden Gutenstein, Hirschwang, Klein- und
Groau, Puchberg, Schwarzau. Niedersterreichisches Landesarchiv, St. Plten (Historical Archive of Lower
Austria).
Protocolle und Operate zum Franziszeischen Kataster. Katastralgemeinden Altenberg, Aschbach, Flz, Kapellen,
Mrzsteg, Seewiesen, Schattenberg, St. Ilgen, Weichselboden, Wildalpen. Steirisches Landesarchiv, Graz
(Historical Archive of Styria).
Auszug aus dem Vermessungs- und Schtzungs-Anschlage zum Zwecke der Einbringung von Reclamationen
aufgrund des 37 des Gesetzes vom 6. April 1879. Steirisches Landesarchiv, Graz (Historical Archive of
Styria).
Regulierungsbltter zum Franziszeischen Kataster. Katastralgemeinden Gutenstein, Hirschwang, Klein- und
Groau, Puchberg, Schwarzau. Katastralarchiv des sterreichischen Bundesamtes fr Eich- und
Vermsessungswesen (Cadastral Archive of the Austrian Mapping Agency).
Almbcher der Almen Steiermarks von 1927. Agrarbezirksbehrde Leoben, Archive (Historical Archive of the
Agricultural Agency Leoben, Styria)
Alm-Erhebungsbcheln der Almen sterreichs von 1950/52. Agrarbezirksbehrde Leoben, Archive (Historical
Archive of the Agricultural Agency Leoben, Styria).
Ergebnisse der Alpkatasterbltter von 1950/52. Agrarbezirksbehrde Leoben, Archive (Historical Archive of the
Agricultural Agency Leoben, Styria).

92

Erhebung der Almen 1986: Alpkatasterbltter. Agrarbezirksbehrde Leoben, Archive (Historical Archive of the
Agricultural Agency Leoben, Styria).

93

5
Dullinger, S., Dirnbck, T. & Grabherr, G.: Patterns of shrub invasion into
high mountain grasslands of the Northern Calcareous Alps (Austria).
Manuskript, zur Publikation eingereicht in der Zeitschrift
Arctic, Antarctic and Alpine Research

94

Patterns of shrub invasion into high mountain grasslands

of the Northern Calcareous Alps (Austria)

S. Dullinger, T. Dirnbck, and G. Grabherr

Institute of Ecology and Conservation Biology, University of Vienna,
Althanstrae 14,
A-1090 Vienna, Austria.
dull@pflaphy.pph.univie.ac.at

ABSTRACT
Both land use and expected climate change will probably cause range shifts of tree and shrub species
in the European Alps. Attempts to predict the magnitude and direction of these processes will produce
reliable results only if they consider both abiotic habitat conditions and biotic interactions. In this
study we analyse recruitment patterns of Pinus mugo Turra in different grassland communities of the
Northern Calcareous Alps (Austria). Pinus mugo is the most important invader of abandoned
subalpine pastures in the area and the predominant woody plant at the current timberline. Results
indicate strong dependence of colonisation success on propagule pressure and differential invasibility
of grassland types but only a marginal impact of local scale site conditions, at least within the species
current altitudinal distribution limits. As the grassland matrix at and above the current treeline is
dominated by a particularly invasible grassland type a possible climate change driven upward
movement of Pinus mugo shrublands may take place quite rapidly. In contrast, encroachment of
abandoned subalpine pastures is frequently delayed by competition with vigorous grassland canopies.

INTRODUCTION
Global climate change is expected to alter vegetation patterns of high mountain ecosystems, especially
in temperate and boreal mountains (Guisan et al. 1995, Grabherr et al. 1994, 1995, Harte and Shaw
1995, Theurillat and Guisan 2001). Possible range expansions of subalpine forests and shrublands will
probably be one of the most important successional processes in terms of ecosystem functioning and
landscape structure. In fact, recent invasions of trees into high mountain grasslands have been reported
from many parts of the world (Magee and Antos 1992, Rochefort and Peterson 1996, Hessel and
Baker 1997, Moir et al. 1999, Meshinev et al. 2000, Wearne and Morgan 2001), but the impact of
long-term climatic changes on these colonisation processes is often difficult if not impossible to
disentangle from short- or mid-term climatic fluctuations (Kullmann 1993, Szeicz and MacDonald
1995), aperiodic disturbance events (Cullen et al. 2001), or land use changes (Motta and Nola 2001,
95

Didier 2001). The latter is especially relevant in the European Alps where many high mountain
grasslands, traditionally used for summer farming, have been abandoned during the last 150 years
(Zwittkovits 1974, Btzing 1991).
Recent advances in invasion ecology have demonstrated that for an organism to expand its range, it
must effectively disperse propagules, it must cope with habitat conditions, and it must successfully
compete within the existing biota (Richardson et al. 2001, Shea and Chesson 2002). As for the
colonisation of grasslands by tree and shrub species, various studies have demonstrated that this
process is strongly affected by competition with vigorous grass and herb layers (Magee and Antos
1992, Richardson and Bond 1992, Richardson and Higgins 1998). Predictions of shifts in species
ranges that exclusively focus on abiotic habitat conditions and that disregard biotic interactions may
thus produce unreliable results (Davis et al. 1998).
In the Northern Calcareous Alps of Austria abandoned subalpine pastures are subject to encroachment
mainly by prostrate pine (Pinus mugo). The same species dominates at the current timberline and will
therefore probably be the most important agent of supposed future treeline dynamics induced by
climate warming. The purpose of this study is to analyse patterns of Pinus mugo-invasion into
different high mountain grasslands of the area. We thereby focus particularly on the following
questions:
1. Do local scale abiotic habitat conditions control recruitment success of Pinus mugo in high
mountain grasslands?
2. Are there any differences in invasibility among the grassland communities of the area, and if so,
does their current spatial distribution cause any emergent landscape scale patterns of invasibility?

MATERIAL AND METHODS

Study area
The study area covers the subalpine and alpine zone of four neighbouring mountain ranges of the
Northeastern Calcareous Alps in Austria (Mt. Hochschwab, Mt. Schneealpe, Mt. Rax and Mt.
Schneeberg, 15 E to 16 E and 4730 N to 4750 N, approximately 150 km, see Fig. 1). Summits
vary between 1900 and 2300 m asl. The mountain system is formed by mesozoic limestone and
dolomite and is characterised by displaced plateaus of different altitudes. Meso- and microrelief are
shaped by a variety of karst landforms like dolines and karren. Climatical conditions are temperate
humid. Mean annual temperature approximates 6-8 C in the valleys, decreasing to about 0-2 C in the
summit region. Annual precipitation averages 700 mm (valleys) and 1500-2500 mm (summits), with a
distinct peak during the summer season.
Summer pasturing (June to September) in the area dates back at least to the 16th century (Zwittkowits
1974). Most of the study area has been historically influenced by livestock grazing, involving patchy
forest and shrub clearing. Since the middle of the 19th century grazing intensity has decreased and
much former pastureland has become abandoned. Approximately 30% of the study area is still
pastured by free-ranging cattle (Farm census data 1986, Statistic Austria, unpublished). Other forms of
land use are of minor importance.
Under natural conditions Norway spruce (Picea abies) or mixed spruce and European larch (Larix
decidua) forests predominate in the lower parts of the subalpine belt and prostrate pine (Pinus mugo)
shrubland at the current treeline ecotone between around 1750 m and 1900 m asl. In fact, the subalpine
belt is a mosaic of woody and non-woody vegetation today with spruce and spruce-larch forests rarely
occurring above 1600 m asl. Non-woody vegetation below the treeline mainly consists of different
96

kinds of pastures and natural grasslands with the latter covering disturbed sites like avalanche paths
and exposed ridges or substituting abandoned pastures as a first step of secondary succession. Less
frequently, tall herb communities (Adenostyles alliaria, Aconitum napellus) and fens (Carex nigra, C.
rostrata) occur. Above the treeline, natural grasslands are dominating. Additionally, rock faces, scree
and snowbeds are widespread, ranging from the valley bottoms up to the summits.
Species
Pinus mugo Turra has an obligatory prostrate growth form. Maximum canopy heights of 5 metres have
been reported (Richardson and Rundel 1998), but in the study area plant height usually varies between
0.3 and 2.5 metres. The geographical range of the species spans the mountains of eastern Europe
(Willis et al. 1998). Within its natural range it is most frequent at treeline ecotones where it establishes
monodominant, dense and extensive shrublands. For convenience, we use the term treeline for the
upper altitudinal range margin of Pinus mugo despite its shrubby growth form. Seedling establishment
seems to be inhibited by low light availability (Hafenscherer and Mayer 1986) and deep litter layers
(Michiels 1993). Thus, within-stand regeneration is entirely dependent on clonal propagation by
means of layering. However, recruitment of seedlings in grasslands and other open habitats is
widespread. The species is well known for being a successful woody pioneer in naturally disturbed
sites or fallow pastures of subalpine regions (Ellenberg 1988, Michiels 1993). Moreover, recent
observations suggest that it is as effective an invader of new habitats outside its natural range as many
other species of the genus (Richardson and Higgins 1998). Seeds of Pinus mugo are primarily wind
dispersed. Secondary distribution of seeds by birds and small mammals has been observed (MllerSchneider 1986).
Environmental data
Environmental information on the study area was provided by means of a raster based Geographical
Information System (GRID-module of ARC-Info 7.1, see Tab. 1). Grid cell size was uniformly 20 x
20 m according to the resolution of the Digital Elevation Model (DEM, Austrian National Mapping
Agency) used. Based on meteorological station data and the DEM we derived envelops for abiotic
habitat variables which are known to influence the performance of high mountain plants (Krner
1999). Temperature was represented by Degree Days (number of days with a mean > 0C). Degree
Days were calculated by means of lapse rates for altitude and geographical longitude according to data
from 20 meteorological stations in or nearby the study area. Solar radiation was calculated for the
early (15th of May), the mid (15th of July) and the late growing period (15th of September) with the
program SOLARFLUX (Dubayah and Rich 1996) assuming clear sky conditions. SOLARFLUX
accounts for relief shading and topographic position. Potential evapotranspiration (PET) was estimated
for August using the formula of Turc (1961). PET-calculations were restricted to August because in
the humid and snow-rich climate of the study area we expected occasional water stress not to occur
before late summer. Topography was represented by slope inclination, a topograhic wetness index
(WET) and a topograhic soil erosion index (EROS). Both indices were calculated by means of the
program-package TAPES-G (Gallant and Wilson 1996). WET computes topographically controlled
average soil moisture conditions based on elevation, flow direction, drainage area and slope fields
(Barling et al. 1994). EROS estimates the spatial distribution of soil erosion and deposition potential
using the same topographic variables (Wilson and Gallant 1996). As we assumed snow-ice abrasion
during winter to influence the performance of Pinus mugo we simulated near surface wind velocities
using the diagnostic wind field model NUATMOS (Version 5N; Ross et al. 1988) which was
integrated in a GIS by Bachmann (1998). NUATMOS was initialised assuming strong winds (22 m/s)
97

from the main wind direction during winter (north-west = 280). Information on snow distribution was
available from 11 classified satellite images of a subarea comprising a third of the study area (Mt.
Schneealpe and western parts of Mt. Rax, acquired between February and June 1998, 1999 and 2000,
resolution: 20 m). From these data we extrapolated a snow cover index applying a binary classification
tree with solar radiation, wind velocity, elevation, and topographical indices as predictor variables
(misclassification error = 8%, Dirnbck et al., submitted).
Besides climatological and topographical descriptors we derived information on bedrock material
from a geological map of the study area (scale: 1:50.000, Geological Survey of Austria, unpublished).
Bedrock mineralogy was compiled into 4 classes: limestone, dolomite, Quartenary sediments and
aeolian deposits (Tertiary loams). Spatial land use information comes from catastral maps combined
with historical and actual farm census data covering the time span between 1770 and the year 2000.
We used time since pasture abandonment as a measure for land use impact setting the concrete values
to the difference in years between the date of the most recent document reporting a parcel to be used
and the year 2000.
For the whole study area a map of Pinus mugo-shrubland was produced by visual interpreation of
infrared- and black and white orthophotos (scale: 1.5000 1:10000). Using this map we calculated the
distance of each sampling plot (see below) to the next Pinus mugo-stand as well as the deviation of
compass direction of this proximate shrubland patch from the main wind direction (north-west = 280)
during seed-release period (Octobre, Mller-Schneider 1986). These two variables should provide an
estimate of seed input into each sampling plot.
Sampling design
Selection of field plots was based on a stratified random sampling design. Our stratification considered
land use, Degree Days, PET, the wetness index and geology (see Table 1). Concerning land use,
sampling was restricted to former pastures not yet overgrown by Pinus mugo shrubland or Picea
abies-Larix decidua forests. Areas actually used for summer farming were not considered, neither
were those which have never been used at all. Continuous variables were categorised into classes and
each possible combination of values was defined a separate stratum. Sampling was restricted to strata
larger than 4 ha resulting in a total of 99 relevant strata. Each stratum had to be sampled by at least one
plot, strata larger than 40 ha by two and some very large ones (> 400 ha) by 3 sample plots. Plot
locations were selected randomly. A total of 140 plots were selected this way, each 20 x 20 m in size.
Plots were localised in the field by means of a GPS (GARMIN E-TREXc, mean Root Mean Square
Error: 7.5 m).
Data collection
All living and dead individuals of Pinus mugo in each study plot were sampled (698 individuals in
total). Besides attributes not considered in this study, we recorded for each individual origin (seed or
layer), vitality (dead or alive), height, diameter (mean of largest and smallest diameter), and the
vegetation type it has germinated in (rock, scree or particular grassland type). Small individuals
(diameter < 0.5 m) were aged by counting bud scares if readily visible (n = 196). If larger individuals
were present on the plot, a maximum of three of them (largest, smallest and median) were sectioned as
close to the root collar as possible and rings were counted under a stereoscope (14x magnification)
after grinding the surface (n = 116). For each individual sectioned, a growth rate was calculated by
dividing the length of the sectioned branch by its ring count. The age of all remaining individuals (n =
386) was estimated as a function of their diameter, height and the average growth rate of the sectioned
individuals on the respective plot (linear regression, R = 0.69).
98

For each plot we further recorded the percentage cover of all tree and shrub species present and the
percentage cover of all non woody vegetation types. For the dominant non woody vegetation type
(mainly grasslands) we recorded some structural attributes (canopy height and canopy cover) and
documented the species composition by means of a 30 m-relev according to Braun-Blanquet (1964).
Relevs were assigned to plant communities according to Mucina et al. (1993a, b) and Grabherr and
Mucina (1993).
As some grassland types hardly predominated on the sampling plots their structural properties were
poorly represented. To extend information on mean structural attributes of all major grassland
communities of the area we used an additional dataset of 1116 relevs done in the course of mapping
the areas subalpine and alpine vegetation during the years 1994 1999 (Greimler and Dirnbck 1996,
Dirnbck and Greimler 1997, Dirnbck et al. 1998, 1999, Dullinger et al. 2001). All relevs come
from the same area and have been done in the same way as on the plots of this study by the authors
themselves. Mean canopy height and canopy cover of the respective communities in this larger dataset
were then used to analyse correlation between vegetation structure and invasibility (see below).
Data analysis
Dependence of recruitment patterns on abiotic habitat conditions was tested by means of multivariate
regression analysis. As response variable we used the recruitment rate (= RR, in recruits per year) at
each plot during the last 50 years , i.e the number of recruits younger than 50 years divided by 50. We
restricted the analysis to this time span as for some older individuals we could not decide
unambiguously if they came from one seedling or from two or more that had merged. All abiotic
habitat variables including distance to and direction of next seed source as well as land use
information were included in the model. Solar radiation was only represented by calculations for the
15th of July, because May, July and September values were closely correlated and the mid-season
gradient proved to have highest predictive power. Distance to seed source was log-transformed prior
to analysis to account for the well known exponential decrease of seed rain density with distance from
the propagule source (Nathan and Muller-Landau 2000). Non-linear effects were tested for by using
restricted cubic spline functions with 4 knots (Harrell 2001). Cubic spline functions were introduced
separately for each variable to avoid seriously overfitting the model. The final model comprised all
non-linear terms that proved significant. Goodness of fit (R2) was validated by bootstrapping (1000
resamples, function validate implemented in the Design-library of S-Plus 2000 for Windows, see
Harrell (2001) for details). The final multivariate model was compared to a univariate one containing
only the single most important variable by means of an F-test for the comparision of nested models.
For evaluating dependence of recruitment patterns on the host-community we established a null
model which is based on the assumption of equal recruitment rate in each of the major grassland
communities of the study area. The expected number of recruits in any one of these grassland types
under the null model was calculated as
(n*(Ai /As )) / As

(1)

with n = the total number of recruits, Ai = the proportional area of the ith community and As = Ai.
Significant deviation of the observed distribution of recruits among grassland communities from the
expected one under the null model was tested for by means of a Chi-square-test.
We used the ratio of observed versus expected number of recruits in any one grassland community as
an index of its invasibility. The dependence of this index on structural properties (mean canopy height,
mean canopy cover) and the position of the different communities on environmental gradients was
99

examined by means of univariate regression models. For each variable original and log-transformed
values were tested separately. Position on environmental gradients was calculated as an area-weighted
mean of the respective habitat variable in our sampling plots, i.e. by calculating for each grassland
community
(pi,j*va,j) / pi,j

(2)

with pi,j = proportion of community i on the area of plot j and va,j = value of variable a on plot j.
Moreover, we tested the dependence of the invasibility index on the mean age of recruits in a
grassland community to control for a possible bias due to an age dependent decrease of mortality: If
mortality should have a peak during the first years or decades following germination, uneven
transition probabilities among age classes will result in higher invasion rates for communities with
younger recruits. This relationship was tested by means of a univariate regression model using the
mean age of recruits in the patches of a community as predictor of its respective index of invasiblity.
Finally we attempted to analyse landscape scale patterns of invasibility using the current distribution
of grassland types across the study area as an indicator. This current spatial distribution was derived
from vegetation maps (scale: 1:10.000, Greimler and Dirnbck 1996, Dirnbck and Greimler 1997,
Dirnbck et al. 1998, 1999, Dullinger et al. 2001).We calculated the proportional area of each
grassland community of the overall grassland matrix in altitudinal steps of 100 metres. The
proportions were multiplied by the respective indices of invasibility and the sum of these values was
used as an estimate of the overall invasibility of the grassland matrix of each altitudinal belt.
Nomenclature follows Adler et al. (1994). ARC-Info (7.1) was used to analyse spatial data and S-Plus
2000 for all statistical calculations.

RESULTS
The multivariate regression model of RR against abiotic habitat variables and land use history is
highly significant but explains only 33% of the overall variance in RR when using a bootstrapvalidation correcting for model overfit (Tab. 2). It has just two significant predictors: Distance to seed
source and deviation of compass direction of the next seed source from the prevailing wind direction
during seed release (Fig. 2). Solar radiation in July (p = 0.05) and wind velocity (p = 0.07) are
marginally non-significant with wind velocity having a significant non-linear effect. However,
compared to the impact of propagule pressure, local scale abiotic habitat conditions seem to have little
influence on recruitment success (Fig. 2). The predominant importance of seed input is confirmed by
the comparison between the Full Model and a univariate one using distance to seed source as the only
predictor of RR (Tab. 2). No significant (p > 0.05) improvement of goodness of fit can be achieved by
adding habitat information provided by the 11 additional predictores in the Full Model.
While testing the influence of the host community on recruitment patterns we accounted for the
influence of seed input by holding distance to seed source constant, i.e. we restricted our analysis to
sampling plots immediately adjacent to established Pinus mugo-shrublands. The -statistic, using only
these 50 immediately adjacent plots, reveals a highly significant deviance of the observed distribution
of recruits among the different grassland communities from an equal distribution expected under the
null model (Tab. 3). Apart from tall herb communities which seem to be completely resistent to Pinus
mugo invasion, the index of invasibility differs by almost 2 orders of magnitude. Carex firma-swards
turn out to be by far the most invasible grassland community of the study area followed by Carex
sempervirens-grasslands and Leontodon hipidus-Crepis aurea-pastures. Besides tall herb
100

communities, Nardus stricta- and Deschampsia cespitosa-pastures, as well as Helictotrichon

parlatorei-grasslands are particularly difficult to establish in for Pinus mugo.
Univariate regression analysis of the invasibility index against structural properties or position of the
respective grasslands on environmental gradients reveals that soil erosion potential is the only
significant environmental predictor of invasibility (Fig. 3). However, the main factor controlling the
differential recruitment success of Pinus mugo seems to be the canopy structure of the respective
grassland community. Both canopy height and canopy cover are significantly correlated with the
invasibility index. For EROS and canopy height logarithmic transformation increases goodness of
model fit indicating a non-linear relationship with recruitment rates. There is no indication for any bias
of the results due to a dependence of the invasibility index on the mean age of recruits of a
community.
Area-statistics derived from the vegetation map demonstrate considerable variability in the
composition of the grassland matrix among different altitudinal steps (Fig. 4). The most obvious trend
is a nearly linear increase of the proportion of Carex firma-grasslands between 1500 and 2200 metres
a.s.l. which in turn proved to be most invasible in the preceding analysis. Plotting the overall
invasibility index of each altitudinal belt against the current distribution of Pinus mugo-shrublands
demonstrates that invasibility does not match the current Pinus mugo distribution but is highest at and
above its present altitudinal range margin.

DISCUSSION
Our results show that patterns of Pinus mugo recruitment in high mountain grasslands of the Northern
Calcareous Alps are mainly a function of propagule pressure and the differential invasibility of the
respective host communities. In contrast, there is little indication for a major impact of abiotic
habitat conditions on recruitment success. This is surprising given the well known influence especially
of temperature on the germination behaviour of alpine plants (Krner 1999). However, meteorological
stations measure air temperature at 2 metres above ground and it is well known that, in particular in
alpine regions, temperature conditions at the soil surface, and especially in dense canopies of
herbaceous or dwarf shrub vegetation, may considerably deviate from 2 metre air temperature (Geiger
1965, Cernusca 1975, 1976). Moreover, habitat conditions of the seed bed should be measured at the
scale of the regeneration niche (Grubb 1977) which may cover a few square centimetres only.
Microtopographic variation as well as neighbouring plants (e.g. Moen 1993) may strongly influence
climatic patterns at this scale. Lack of correlation between recruitment success and the abiotic
environment in our study may therefore be partly due to a deviance of microclimatological patterns
from local scale gradients. Additionally, the presumably crucial influence of temperature (Krner
1999) on the position of the current upper altitudinal range margin of Pinus mugo may be masked by
partial collinearity with propagule pressure, because for sampling plots above the current treeline there
is a close negative correlation between decreasing Degree Days and increasing distance to seed source.
It is thus hardly possible to disentangle the relative roles of dispersal limitation and temperature
thresholds in determining the lack of recruitment success above the current treeline. However, what
we may unambiguously conclude from our results is that within its current distribution limits local
scale environmental gradients have little impact on Pinus mugo-colonisation of high mountain
grasslands.
The predominant impact of propagule pressure on colonisation success is a simple consequence of
basic models of population dynamics (Silvertown and Lovett Doust 1993). Holding mortality constant,
an enhanced seed input must increase recruitment rates. However, as a consequence, application of a
null model of equal recruitment rates among grassland types may be biased by differential seed input.
101

Although we have held distance to seed source constant, homogenous seed input over the whole area
of a 20 x 20 metre sampling plot is quite unlikely to occur. Short distance seed dispersal kernels are
usually exponential in shape (Nathan and Muller-Landau 2000) with the majority of seeds deposited
within a few plant heights of the source (Bullock and Clarke 2000). Besides stochastic events and
differential fecundity of neighbouring Pinus mugo-stands the very size of the plots will thus cause a
gradient of seed rain density from the immediate edge of the propagule source to the opposite border
of the plot. Hence, comparison of recruitment data among grassland types will be biased if some of
them show a systematic trend to occur closer to the edge of Pinus mugo shrublands. In fact, such
habitat preferences are known for Deschampsia cespitosa- and Carex ferruginea-grasslands (e.g.
Greimler and Dirnbck 1996) which probably profit from enhanced water and nutrient supply at
shrubland edges (Holtmeier and Broll 1992). However, contrary to expectation, these communities
have a very low index of invasibility. Thus, non-random differences in seed input are unlikely to be
responsible for the deviance of observed recruitment patterns from the null model.
The relationship between invasibility index and grassland canopy structure suggests that colonisation
success of Pinus mugo is controlled by varying intensities of competition in the respective host
communities. Accordingly, competitive interactions with non woody plants, notably grasses, have
been demonstrated to be a key factor controlling pine invasions in different habitats and parts of the
world (Richardson and Bond 1992, Richardson and Higgins 1998) but also colonisation of subalpine
grasslands by other tree species (Magee and Antos 1992). Experimental tests will be necessary to
determine the limiting resource, but given the comparatively high light requirements of seedlings of
different Pinus species (Keeley and Zedler 1998), and of Pinus mugo in particular (Hafenscherer and
Mayer 1986, Michiels 1993), shading by dense and high grassland canopies may be supposed to be the
most important factor. Additionally, there is a significant relationship between the invasion index of a
community and its position on a gradient of soil erosion potential. Considering the low explanatory
power of EROS in the abiotic habitat model this correlation is possibly coincidental and does not
indicate a mechanistic process. However, topsoil erosion may in fact facilitate Pinus mugo recruitment
as heavy litter seems to inhibit successful seedling establishment (Michiels 1993). A similar inhibitive
effect of thick litter layers was demonstrated for Picea abies in central European montane grasslands
(Prach et al. 1996).
As a consequence of their differential invasibility the current spatial distribution of grassland
communities will variegate Pinus mugo colonisation rates in different parts of the landscape.
Considering possible climate change effects the predominance of highly invasible Carex firma
grasslands at and above the current treeline is particularly remarkable. If climate warming will relax
the environmental constraints that limit its current altitudinal distribution, an effect wich has been
hypothesised to depend on exceeding certain thresholds (Paulsen et al. 2000), a consecutive range
expansion of Pinus mugo into former alpine grasslands may thus take place quite rapidly. Such an
accelerated upward movement of Pinus mugo-shrublands may pose a major threat to populations of
many herbaceous alpine species if they are unable to change their ranges at comparable rates
(Theurillat and Guisan 2001, Thomas et al. 2001). In contrast, even under the favorable climatic
conditions of lower subalpine regions, Pinus mugo will probably colonise the vigorous high growing
grasslands of many abandonend pastures at quite low rates. Taken together, the results of this study
thus suggest that bioic interactions will have a major impact on land use and climate change driven
successional dynamics which may shape the vegetation cover of many temperate and boreal high
mountain landscapes during the next decades and centuries.

102

ACKNOWLEDGEMENTS
We are grateful to Andi Bachmann who gave us access to his GIS-application of NUATMOS, to M.
Steinkellner for meteorological, to G. Mandl and G. Bryda for geological, and to J. Jansa for snow
cover data, to M. Abensperg-Traun for linguistic improvement and to J. Greimler and D. Moser for
valuable criticism. J. Greimler, I. Schmidsberger, N. Sauberer and T. Englisch helped with field work.
Financial support came from the Water Department of the City of Vienna and the Austrian Federal
Ministry for Education, Culture and Science.

REFERENCES CITED
Adler, W., Oswald, K., and Fischer, R., 1994: Exkursionsflora fr sterreich. Stuttgart: Ulmer. 1180 pp.
Bachmann, A., 1998: Coupling NUATMOS with the GIS ARC/Info. Final Report for MINERVE 2. Department
of Geography of the ETH Zrich, Division of Spatial Data Handling. 13 pp.
Barling, R.D., Moore, I.D., and Grayson, R.B., 1994: A quasi-dynamic wetness index for characterizing the
spatial distribution of zones of surface saturation and soil water content. Water Resources Research, 30: 10291044.
Btzing, W., 1991: Die Alpen. Entstehung und Gefhrdung einer europischen Kulturlandschaft. Mnchen: C.H.
Beck. 286pp.
Braun-Blanquet, J., 1964: Pflanzensoziologie. 3rd ed. Wien: Springer. 865 pp.
Bullock, J.M. and Clarke, R.T., 2000: Long distance seed dispersal by wind: measuring and modelling the tail of
the curve. Oecologia, 124: 506-521.
Cernusca, A., 1975: Standrtliche Variabilitt und Engergiehaushalt alpiner Zwergstrauchbestnde.
Verhandlungen der Gesellschaft fr kologie, Wien, 9-21.
Cernusca, A., 1976: Energy Exchange within individual layers of a meadow. Oecologia (Berlin), 23: 141-149.
Cullen, L., Stewart, G.H., Duncan, R.P., and Palmer, G., 2001: Disturbance and climate warming influences on
New Zealand Nothofagus tree-line population dynamics. Journal of Ecology, 89: 1061-1071.
Davis, A.J., Jenkinson, L.S., Lawton, J.L., Shorrocks, B., and Wood, S., 1998: Making mistakes when predicting
shifts in species range in response to global warming. Nature, 391: 783-786.
Didier, L., 2001: Invasion patterns of European larch and Swiss stone pine in subalpine pastures in the French
Alps. Forest Ecology and Management, 145: 67-77.
Dirnbck, T. and Greimler, J., 1997: Subalpin-alpine Vegetationskartierung der Raxalpe, Nordstliche
Kalkalpen. Linzer Biologische Beitrge, 29: 299-339.
Dirnbck, T., Dullinger, S., Gottfried, M., and Grabherr, G., 1999: Die Vegetation des Hochschwab
(Steiermark). Alpine und subalpine Stufe. Mitteilungen des Naturwissenschaftlichen Vereins der Steiermark,
129: 111-251.
Dirnbck, T., Greimler, G., and Grabherr, G., 1998: Die Vegetation des Zeller Staritzen Plateaus (Hochschwab,
Steiermark) und ihre Bedeutung fr den Quellschutz. Mitteilungen des Naturwissenschaftlichen Vereins der
Steiermark, 128: 123 183.
Dubayah, R. and Rich, P.M., 1996: GIS-based solar radiation modeling. In Goodchild, M.F., Steyaert, L.T. and
Parks, B.O. (eds.), GIS and Environmental Modeling. New York: Wiley, 129-134.
Dullinger, S., Dirnbck, T., and Grabherr, G., 2001: Die subalpine und alpine Vegetation der Schneealpe,
Steiermark. Mitteilungen des Naturwissenschaftlichen Vereins der Steiermark, 83-128.
Ellenberg, H., 1988: Vegetation ecology of central Europe. Cambridge: Cambridge University Press. 731 pp.
Gallant, J.C. and Wilson, J.P., 1996: Tapes-G: a grid based terrain analysis program for the environmental
science. Computers and Geoscience, 22: 713-722.
Geiger, R., 1965: The climate near the ground. Cambridge, MA: Harvard University Press. 277 pp.

103

Grabherr, G. and Mucina, L., 1993: Die Pflanzengesellschaften sterreichs. Teil II. Jena: Gustav Fischer. 523
pp.
Grabherr, G., Gottfried, M., and Pauli, H., 1994: Climate effects on mountain plants. Nature, 369: 448.
Grabherr, G., Gottfried, M., Gruber, A., and Pauli, H., 1995: Patterns and Current Changes in alpine Plant
Diversity. In Chapin, F.S. and Krner, C. (eds.), Arctic and alpine biodiversity: pattern causes and ecosystem
consequences. New York: Springer, 167-181.
Greimler, J. and Dirnbck, T., 1996: Die subalpine und alpine Vegetation des Schneebergs, Niedersterreich.
Linzer Biologische Beitrge, 28: 437-482.
Grubb, P.J., 1977: The maintenance of species-richness in plant communities: The importance of the
regeneration niche. Biological Review, 52: 107-145.
Guisan, A., Holten, J.I., Spichiger, R., and Tessier, L. (eds.), 1995: Potential ecological impacts of climate
change in the Alps and Fennoscandian Mountains. Genve: Conservaitoire and Jardin Botaniques de Genves.
194 pp.
Hafenscherer, J. and Mayer, H., 1986: Standort, Aufbau, Entwicklungsdynamik und Verjngung von
Latschenbestnden im Karwendeltal/Tirol. Schweizerische Zeitschrift fr Forstwesen, 137: 177-203.
Harrell, F.E. Jr., 2001: Regression Modeling Strategies. New York: Springer. 568 pp.
Harte, J. and Shaw, R., 1995: Shifting dominance within a montane vegetation community: Results of a climatewarming experiment. Science, 267: 876-880.
Hessl, A.E. and Baker, W.L,. 1997: Spruce and fir regeneration and climate in the forest-tundra ecotone of
Rocky Mountains National Park, Colorado, U.S.A. Arctic and Alpine Research, 29: 173-183.
Holtmeier, F.K. and Broll, G., 1992: The influence of tree islands and microtopography on pedoecological
conditions in the forest-alpine tundra ecotone on Niwot Ridge, Colorado Front Range, U.S.A. Arctic and
Alpine Research, 24: 216-228.
Keeley, J.E. and Zedler, P.H., 1998: Evolution of life histories in Pinus. In Richardson, D.M. (ed.), Ecology and
biogeography of Pinus. Cambridge: Cambridge University Press, 219-250.
Krner, C., 1998: A re-assessment of high elevation treeline positions and their explanation. Oecologia, 115:
445-459.
Krner, C., 1999: Alpine Plant Life. New York: Springer. 338 pp.
Kullman, L., 1993: Pine (Pinus sylvetris L.) tree-limit surveillance during recent decades, Central Sweden.
Arctic and Alpine Research, 25: 24-31.
Michiels, H.G., 1993: Die Stellung einiger Baum- und Straucharten in der Struktur und Dynamik der Vegetation
im Bereich der hochmontanen und subalpinen Waldstufe der Bayrischen Kalkalpen. Forstliche
Forschungsberichte Mnchen, 135. 300 pp.
Moen, J., 1993: Positive versus negative interactions in a high alpine block field: Germination of Oxyria digyna
seeds in a Ranunculus glacialis community. Arctic and Alpine Research, 25: 201-206.
Motta, R. and Nola, P., 2001: Growth trends and dynamics in sub-alpine forest stands in the Varaita Valley
(Piedmont, Italy) and their relationships with human activities and global change. Journal of Vegetation
Science, 12: 219-230.
Mucina, L., Grabherr, G., and Ellmauer, T., 1993a: Die Pflanzengesellschaften sterreichs. Teil I. Jena: Gustav
Fischer. 578 pp.
Mucina, L., Grabherr, G., and Wallnfer, S., 1993b: Die Pflanzengesellschaften sterreichs. Teil III. Jena:
Gustav Fischer. 353 pp.
Mller-Schneider, P., 1986: Verbreitungsbiologie der Bltenpflanzen Graubndens. Verffentlichungen des
Geobotanischen Institutes der Eidg. Techn. Hochschule, Stiftung Rbel, 85. 263 pp.
Nathan, R. and Muller-Landau, C., 2000: Spatial patterns of seed dispersal, their determinants and consequences
for recruitment. Trends in Ecology and Evolution, 15: 278-285.

104

Paulsen, J., Weber, U.M., and Krner, Ch., 2000: Tree growth near treeline: Abrupt or gradual reduction with
altitude? Arctic, Antarctic, and Alpine Research, 32: 14-20.
Prach, K., Leps, J., and Michlek, J., 1996: Establishment of Picea abies-seedlings in a central European
mountain grassland: an experimental study. Journal of Vegetation Science, 7: 681-684.
Price, M.F. and Barry, R.G., 1997: Climate change. In Messerli, B. and Ives, J.D. (eds.), Mountains of the
World. New York: Parthenon, 409-446.
Richardson, D.M. and Bond, W.J., 1991: Determinants of plant distribution: Evidence from pine invasions.
American Naturalist, 137: 639-668.
Richardson, D.M., Bond, W.J., Richard, W., Dean, J., Higgins, S.I., Midgley, G.F., Milton, S.J., Powrie, L.W.,
Rutherford, M.C., Samways, M.J. and Schulze, R.E., 2000: Invasive alien species and global change: A south
African perspective. In Mooney, H.A. and Hobbs, R.J. (eds.), Invasive species in a changing world.
Washington: Island Press, 303-350.
Richardson, D.M. and Higgins, S.I., 1998: Pines as invaders in the southern hemisphere. In Richardson, D.M.
(ed.), Ecology and biogeography of Pinus. Cambridge: Cambridge University Press, 450-474.
Richardson, D.M. and Rundel, P.W., 1998: Ecology and biogeography of Pinus: an introduction. In Richardson,
D.M. (ed.), Ecology and biogeography of Pinus. Cambridge: Cambridge University Press, 3-48.
Rochefort, R.M. and Peterson, D.L., 1996: Temporal and spatial distribution of trees in subalpine meadows of
Mount Rainier National Park, Washington, U.S.A. Arctic and Alpine Research, 28: 52-59.
Ross, D.G., Smith, I.N., Manins, P.C., and Fox, D.G., 1988: Diagnostic wind field modeling for complex terrain:
Model development and testing. Journal of Applied Meteorology, 27: 785-796.
Silvertown, J.W. and Lovett Doust, J., 1993: Introduction to plant population biology. 3rd ed. Oxford, Blackwell.
210 pp.
Szeicz, J.M. and MacDonald, G.M., 1995: Recent white spruce dynamics at the subarctic alpine treeline of
north-western Canada. Journal of Ecology, 83: 873-885.
Theurillat, J.-P. and Guisan, A., 2001: Potential impact of climate change on vegetation in the European Alps: A
review. Climatic Change, 50: 77-109.
Thomas, C.D., Bodsworth, E.J., Wilson, R.J., Simmons, A.D., Davies, Z.G., Musche, M., and Conradt, L., 2001:
Ecological and evolutionary processes at expanding range margins. Nature, 411: 577-581.
Turc, L., 1961: valuation de besoins en eau dirrigation, vapotranspiration potentielle, formule simplifie et
mise en jour. Annales Agronomiques, 12: 13-49
Wearne, L.J. and Morgan, J.W., 2001: Recent forest encroachment into subalpine grasslands near Mount
Hotham, Victoria, Australia. Arctic, Antarctic and Alpine Research, 33: 369-377.
Willis, K.J., Bennet, K.D., and Birks, H.J.B., 1998: The late Quartenary dynamics of pines in Europe. In
Richardson, D.M. (ed.), Ecology and biogeography of Pinus. Cambridge: Cambridge University Press, 107121.
Wilson, J.P. and Gallant, J.C., 1996: EROS: A grid based program for estimating spatially-distributed erosion
indices. Computers and Geosciences, 22: 707-712.
Zwittkovits, F., 1974: Die Almen sterreichs. Zillersdorf: Selbstverlag. 419 pp.

105

FIGURES AND TABLES

Figure 1: Geographical position of the study area. Only the Alps are represented by topographical
shadings (A Austria, CH Switzerland, D Germany, F France, FL Liechtenstein, I Italy, MC
Monaco, SLO Slovenia).

106

Distance***
Direction*
Solar Radiation
Wind speed
Slope Inclination
Geology
Degree Days
Snow Cover Index
PET August
Pasture abandonment
Soil erosion Index
Wetness Index

10

20

30

40

50

60

Akaike Information Criterion

Figure 2: Akaike Information Criterion for predictor variables in a multivariate regression model of
Pinus mugo-recruitment rate in high mountain grasslands of the Northern Calcareous Alps. Distance =
Distance to seed source; Direction = Deviation of compass direction of the next seed source from the
prevailing wind direction during the main seed release period (Octobre); PET = Potential
Evapotranspiration; Pasture abandonment = Time since pasture abandonment. *** = p < 0.0001, * = p
< 0.05.

107

p = 0.0035

Observed/Expected Number of recruits

1.9

1.4

0.9

0.4

-0.1

10

20

30

40

50

60

Mean canopy height (cm)

p = 0.0049

Observed/Expected number of recruits

2.0

1.5

1.0

0.5

0.0

75

80

85

90

95

100

Mean canopy cover (%)

p = 0.0049

Observed/Expected number of recruits

2.0

1.5

1.0

0.5

0.0

0.00

0.25

0.50

0.75

1.00

1.25

1.50

1.75

Soil Erosion Index + 1

Figure 3: Univariate regression models using the index of invasibility of the 10 predominant grassland
types of the study area as response and their canopy structure respectively their position on different
environmental gradients as predictors. Models for Degree Days, solar radiation, potential
evapotranspiration, slope inclination, wetness index, wind velocity, snow cover index, land use,
geology and mean age of recruits in the respective communities are not shown. They all have p-values
> 0.1.

108

Proportional area of grassland type

1.0

0.8

0.6

0.4

0.2

2200 2300

2100 2200

2000 2100

1900 2000

1800 1900

1700 1800

1600 1700

1500 1600

1400 1500

1300 1400

1200 1300

0.0

Altitudinal step

2.0

1.0

2200 2300

2100 2200

2000 2100

1900 2000

1800 1900

1700 1800

1600 1700

1500 1600

0.0

1400 1500

1300 1400

0.5

Area of Pinus mugo shrubland (km)

1.5

1200 1300

Index of Invasibility

Altitudinal step

Figure 4a: Proportional area of the 10 predominant grassland communities of the study area in
altitudinal steps of 100 metres. = Carex firma grassland; = Carex sempervirens grassland; =
Carex ferruginea grassland; = Calamagrostis varia grassland; = Tall herb communities; =
Agrostis alpina-Festuca pumila grassland; = Deschampsia cespitosa pasture; = Helictotrichon
parlatorei grassland;

= Leontodon hispidus-Crepis aurea pasture;

= Nardus stricta pasture.

Figure 4b: Index of invasibility of the overall grassland matrix (= sum of the proportional areas of the
10 predominant grassland types weighted by their respective index of invasibility, = solid line) and
current distribution of Pinus mugo shrublands (= dotted line) in altitudinal steps of 100 metres.

109

Variable
Degree Days
Radiation 15th of May (MJ/m)
Radiation 15th of July (MJ/m)
Radiation 15th of September (MJ/m)
PET August
Slope Inclination ()
Wetness Index
Soil Erosion Index
Wind speed (m/s)
Snow Cover Index
Time since pasture abandonment
(years)
Distance to seed source (m)
Deviation from main wind direction ()
Geology (number of plots)

Min

Max

Mean

StdDev

197
10.95
13.54
3.126
1.07
1
2.83
-8.22
3.74
0.301
40

246
28.92
29.73
23.19
4.65
53
13.02
35.37
45.2
0.502
180

222
24.3
25.56
14.97
2.85
23
6.24
0.92
16.02
0.385
96

12
4.05
3.63
5.84
0.79
11
1.84
5.24
7.06
0.074
48

20
1

937
316

161
156

207
101

Lime
77

Dol
39

QS
20

AD
4

Table 1: Minimum, maximum, mean and standard deviaton of numerical environmental variables at
140 sampling plots and number of plots in each class of bedrock material (Lime = limestone; Dol =
dolomite; QS = Quartenary Sediments; AD = Aeolian Deposits).

110

Full Model
Univariate Model
F15,123

df
16
1

R-original R-corrected
p
0.47
0.33
< 0.0001
0.31
0.3
< 0.0001
> 0.05
0.367

Table 2: Comparision of the full multivariate model of Pinus mugo-recruitment rate (10 linearily fitted
numerical variables, a categorical one with 4 classes and a numerical one fitted by a restricted cubic
spline function with 4 knots, see Figure 2) and a univariate one using distance to seed source as the
only predictor. Corrected R-values come from boostrap-validation (1000 resamples). The F-test was
calculated based on corrected R-values.

111

Grassland Type
Carex firma grassland
Carex sempervirens grassland
Leontodon hispidus-Crepis aurea pasture
Agrostis alpina-Festuca pumila grassland
Calamagrostis varia grassland
Carex ferruginea grassland
Nardus stricta pasture
Deschampsia cespitosa pasture
Helictotrichon parlatorei grassland
Tall herb communities
X
d.f.
p

Area (m) Observed Expected Deviance Obs/Exp

1018
61
28
37
2.18
6188
264
173
48
1.53
583
20
16
1
1.25
1243
29
35
1
0.83
408
7
11
2
0.61
846
9
24
9
0.38
320
2
9
5
0.22
2013
11
56
36
0.19
1433
2
40
36
0.05
420
0
12
12
0.00
14472
405
405
188
188
9
< 0.001

Table 3: Chi-square statistic comparing the observed number of Pinus mugo recruits in 10
predominant grassland types of the study area to an expected number under a neutral model assuming
equal distribution of recruits among these grassland communities. Expected numbers were calculated
with respect to the proportional area of each grassland type in the sampling plots. Obs/Exp = Ratio of
observed and expected number of recruits.

112

6
Dullinger, S., Dirnbck, T. & Grabherr, G.: Tree-line shifts prone to
climate change: Evaluating relative effects of temperature increase,
dispersal and invisibility by means of a plant spread model.
Manuskript, vorbereitet zur Einreichung bei der Zeitschrift
Journal of Ecology

113

TREE-LINE SHIFTS PRONE TO CLIMATE CHANGE: EVALUATING RELATIVE EFFECTS

OF TEMPERATURE INCREASE, DISPERSAL AND INVASIBILITY BY MEANS OF A
PLANT SPREAD MODEL
Stefan Dullinger, Thomas Dirnbck & Georg Grabherr
Abstract
1. Predicted global warming will probably shift tree-lines upslopes and polewards in alpine and arctic
environments, respectively. However, regional tree-lines have responded ambiguously to the last
centurys climatic trends. We hypothesised that these equivocal responses may partly be caused by
dispersal and recruitment limitations specific to potentially expanding tree species.
2. To test this hypothesis we established and parameterised a temporally and spatially explicit plant
spread model and analysed its sensitivity to varied assumptions on forthcoming climatic trends,
functional form of the dispersal kernel and consideration of spatially inhomogenous invasion
resistance of resident non-woody vegetation. As a model system we used a high mountain landscape
of the Northern Calcareous Alps in Austria whose tree-line is built up by a shrubby growing pine
(Pinus mugo TURRA).
3. Results demonstrate low growth rates, long generation times, considerable dispersal and recruitment
limitation and thus an overall slow range expansion of the pine under climate warming.
4. Running the model for the next 1000 years the area covered by pines is predicted to increase from
currently 9.96% to a range between 24.4% and 59.3% of the study landscape.
5.Effects of dispersal capacity and spatial patterns of competitively controlled recruitment suppression
are at least as severe for range size dynamics as those of variations in expected climatic trends.
Moreover, invasibility and dispersal function interact with each other due to peculiarities of the
regional vegetation cover.
6. Ambiguous transient responses of regional tree-lines may thus well be caused by variation in these
species-specific traits.

INTRODUCTION
Predicted global warming will probably affect range sizes and geographical distribution of biota at
various scales (e.g. Grabherr et al. 1994, Parmesan 1996, Sturm et al. 2001, Thomas et al. 2001,
Walther et al. 2002, Parmesan & Yohe 2003). As climatically determined ecotones arctic and alpine
tree-lines are assumed to be especially sensitive to altered temperature regimes and climate warming is
expected to drive tree-lines upslopes and poleward at the expense of alpine and arctic ecosystems,
respectively (Kittel et al. 2000, Theurillat & Guisan 2001, Hansen et al. 2001 ). However, field studies
and remote sensing analyses investigating recent tree-line shifts in response to increasing temperatures
during the last century have provided ambiguous results which span the whole gradient from rapid
dynamics to apparently complete inertia (e.g. Kullmann 1993, Lavoie & Payette 1994, Szeicz &
MacDonald 1995, Hessl & Baker 1997, Meshinev et al. 2000, Paulsen et al. 2000, Cullen et al. 2001,
Didier 2001, Masek 2001, Motta & Nola 2001, Sturm et al. 2001, Klasner & Fagre 2002, Kullmann
2002). Apart from variation in regional climatic trends these equivocal findings likely result from
differences between individual tree-line systems. In analogy to alien plant invasions one may
hypothesise that dispersal capacity of potentially expanding tree species (e.g. Kot et al. 1996, Neubert
& Caswell 2000, Bullock et al. 2002, Shigesada & Kawasaki 2002, Watkinson & Gill 2002) and
competitive interactions with resident alpine vegetation (Richardson & Bond 1991, Magee & Antos
114

1992, Rochefort & Peterson 1996, Moir et al. 1999) will be major sources of variation in tree-line
responses to climate change.
Throughout most of the Northeastern Calcareous Alps the tree-line is currently dominated by a
shrubby growing pine species (Pinus mugo TURRA) known as a successful invader of naturally and
human-disturbed habitats (e.g. Ellenberg 1996). In a prior study (Dirnbck et al. 2003) we have used
an equilibrium-based static modelling approach (Guisan & Zimmermann 2000) to evaluate effects of
forthcoming climate warming on the regional distribution of this species in some mountain ranges at
the northeastern fringe of the Alps. Results of these models suggested a considerable range expansion
of Pinus mugo with only moderate levels of temperature increase. As a consequence habitats of low
competitive alpine plant species would become severely reduced and fragmented eventually driving a
couple of regional endemics to extinction (Grabherr et al. 1995, Dullinger et al. 2000, Pauli et al.
2003). However, additional work on Pinus mugo dynamics suggested that the response of the pinus to
forthcoming climate warming in terms of a possible range shift may be hampered by restricted
dispersal as well as by competitive inhibtion of recruitement in dense grassland layers and predicted
potential distributions may thus not be accomplished in the mid-term (Dullinger et al. submitted).
Unfortunately, most existing simulation models that explore potential range shifts and forest dynamics
in response to climate change either do not represent transient dynamics (in the case of equilibrium
based, bioclimatic habitat distribution models, see e.g. Guisan & Zimmermann 2000) or they disregard
processes that occur beyond the boundaries of the simulation plot (in the case of gap models, e.g.
Loehle & LeBlanc 1996, Schwalm & Ek 2001). Models which integrate both the spatial and temporal
dimensions have until now primarily been applied to problems of alien plant spread (see e.g. Higgins
& Richardson 1996a, 1999, Neubert & Caswell 2000, Shigesada & Kawasaki 2002). However, they
have also been used to reconstruct Holocene vegetation dynamics (e.g. Clark 1998) and their potential
to simulate possible response of plant species to forthcoming climate change has been inferred from
such applications (Pitelka et al. 1997, Malanson & Cairns 1997, Clark et al. 1998a, Clark et al. 1999,
Higgins & Richardson 1999).
In this paper we explicitly address the relative roles of climate warming, dispersal capacity, and
competitive interaction with resident alpine vegetation on the range expansion process of Pinus mugo
by means of a spatially and temporarily explicit plant spread simulator. Focussing on a 1000-year time
frame we particularly ask how predictions of pine shrub distribution are affected by (1) different
degrees of temperature increase, (2) different functional forms of the dispersal kernel, and (3) the
consideration of varied invasibility of resident alpine vegetation types. We moreover examine the
possible interactions among these driving processes.
METHODS
Study system
The study area covers the tree-line ecotone and the alpine belt of Mt. Hochschwab (4734 to 4738
latitude and 1500 to 1518 longitude, uppermost summit: 2277 m a.s.l.) which is part of the
Northeastern Calcareous Alps of Austria. The mountain range is characterised by displaced plateaus of
different altitudes with surfaces shaped by Pleistocene glaciation and karst landform development.
Soils are predominantly lithic, calcaric and rendzic Leptosols as well as chromic Cambisols. Climatic
conditions are temperate humid. Mean annual temperature approximates about 0-2 C in the summit

115

region. Annual precipitation averages to 2000-2500 mm (summits) with a decisive peak during the
growth period. The alpine areas are covered by snow for approximately 6-8 months (October May).
Fine scale snow cover duration is highly variable due to the rugged relief and strong winds.
Summer pasturing (June to September) in the area dates back at least to the 16th century. For 30% of
the modeling landscape at least some historical life stock grazing is documented. Since the middle of
the 19th century grazing intensity has decreased and much former pastureland has become abandoned.
Today, only % of the modeling landscape are still pastured by free-ranging cattle at an intensity of
about 0.5 cattle per hectare (Dullinger et al. 2003).
The predominant woody plant species of the upper subalpine belt is prostrate pine (Pinus mugo Turra).
The upper limit of single Pinus mugo-individuals is currently at about 1950 metres a.s.l. In fact, the
subalpine belt is a mosaic of woody and non-woody vegetation today. Non-woody vegetation below
the tree-line mainly consists of different kinds of pastures and natural grasslands with the latter
covering disturbed sites like avalanche paths and exposed ridges. Above tree-line, natural grasslands
are dominating with a gradual switch from prevailing Carex sempervirens Vill. to Carex firma Host
grasslands with increasing altitude. Additionally, rock faces, scree and snowbeds are widespread from
the valley bottoms up to the summits.

Study species
Pinus mugo Turra is an obligatory prostrate pine with canopy heights of adults varying between ca. 0.3
and 2.5 metres in the study area (for convenience, we use the term tree-line for the upper altitudinal
range margin of Pinus mugo despite its shrubby growth form). Seedling establishment seems to be
inhibited by low light availability (Hafenscherer and Mayer 1986) and deep litter layers (Michiels
1993). Thus, within-stand regeneration is entirely dependent on clonal propagation by means of
layering. Intensive, multidirectional layering makes clones virtually immortal and inhibits gap-phase
regeneration processes in established stands (Hafenscherer and Mayer 1986). Nevertheless,
recruitment of seedlings in grasslands is common. Seeds of Pinus mugo are primarily wind dispersed.
Secondary redistribution of seeds by birds and small mammals has been observed (Mller-Schneider
1986).
Representation of environmental conditions
A digital elevation model (DGM, Austrian Mapping Agency) with a cell size of 20 metres served as
main input for the representation of abiotic habitat conditions. The derivation of environmental
envelops from the DGM is described in detail in Dirnbck et al. (2003). In brief, the following
variables were calculated:

116

(1) Climatic conditions, represented by annual degree days (= days with a mean daily temperature >
0C, DD), solar radiation income at the beginning (15th of May, SRM), in the middle (15th of July,
SRJ), and at the end (15th of September, SRS) of the growth period, and site water balance in August.
(2) Topography, characterised by slope inclination (SLOPE), a topographical wetness index (WET), a
topographical soil erosion index (EROS), and an estimate of topographically modified near-surface
wind velocity during strong, northwesterly winds (WSP).
Additionally, we provided datasets on bedrock mineralogy (GEO), spatial distribution of chromic
Cambisols, and current vegetation cover. Bedrock mineralogy was derived from recently updated
geological maps (scale: 1:50.000, Geological Survey of Austria, unpublished). The distribution of
chromic Cambisols was extrapolated from 573 sample points in the study area and adjacent mountain
ranges using a Binary Classification Tree procedure with topographical variables (altitude, slope,
EROS, WET and their interactions) as predictors (misclassfication rate: 15%). Information on current
vegetation cover including actual pine shrub distribution comes from fine-scale vegetation maps
(1:10000, Dirnbck et al. 1999) and high-resolution IR-orthophotographs (acquired: 1994-07-23, pixel
resolution: 25 cm). All these datasets were resampled to meet the resolution of the DGM (Table 1).
Model concept and assumptions
In accordance with environmental descriptors the modelling landscape is represented by a twodimensional grid with a cell size of 20 metres. Overall, it spans about 53 km (131901 cells). Pine
shrub dynamics across this landscape is tracked as the changing percentage of Pinus mugo-cover per
individual grid cell (= site) over time. These changes result from the the spatial distribution of recruits
originating from a site and from the growth and mortality of individual pine shrubs growing at the
respective site.
Model formulation is based on several simplifying assumptions:
(1) Time passes in discrete steps. Our parameterisation data demonstrates very low recruitment rates
(see below). We thus decided to use a rather long time step of 50 years. Moreover, using a long
time step avoids bias in recruitment rate estimation due to variable seed production (e.g. Herrera et
al. 1998, De Steven & Wright 2002). Though little is known about inter-annual patterns in seed
production for Pinus mugo, masting behaviour is widespread among tree-line species of the
Northern Calcareous Alps and masting frequency is very low at high elevations (e.g. > 10 years
for Picea abies, Mayer 1976).
(2) The canopy-increment of an individual Pinus mugo shrub per time step is adequately represented
by a growing circle. In fact, the canopy shape is irregular but the multi-stem growth-form
(Hafenscherer & Mayer 1986) justifies this geometrical approximation.
(3) Due to intensive clonal propagation by multidirectional layering mortalilty does not have an age
dependent component but is entirely due to catastrophic disturbances. The most important
disturbance regimes are avalanches and extreme weather events. Such events do not cause
mortality of just one individual but usually kill the whole population of a site. We thus assume
mortality events to reset the canopy cover of a site to 0 %.
(4) Recruitment and mortality of pines include a stochastic component which is due to
unpredictability of annual snow fall and melting processes, unusual weather events or spatial
pattern of seed and seedling predation and secondary seed dispersal (e.g. Vander Wall 1992, Keely
and Zedler 1998, Greene and Johnson 1997).
(5) Pine shrub populations of already densely covered cells invade neighbouring cells vegetatively.
For ease of computation, the invading front is assumed to be a straight line with a width of one
cell size. We used a threshold cover of 90% to trigger this process of vegetative invasion into
adjacent sites.
117

(6) Rock faces lack appropriate space and soil substrate to support a dense Pinus mugo-canopy.
According to our field experience we set the maximum value of pine shrub cover in rock habitats
to 10%.
(7) Besides seed availability and climatic constraints mechanical stress mainly controls pine shrub
colonisation of debris cones. Successful invasion is thus only possible if the site is positioned on a
ridge or if there is a neighbouring cell above the focal site which gives shelter by means of an
already existing dense pine-cover. These rules produce the typical colonisation pattern of pine
shrubs in debris cones which is propagating downhill in a conical fashion.
The model starts from the current distribution of pine shrub populations across the landscape. Initially,
all actually occupied sites are assigned a cover of 100% and an age of 100 years. The model first
calculates the fecundity of the population of each site as a function of the age of the individuals and of
environmental conditions. Sites with populations above a certain threshold of fecundity (see below)
are defined as seed sources. Next, it determines the number of newly germinating recruits per
individual site during one time step. Year of germination is chosen randomly, replicates are possible.
The canopy of all individuals of a site increases by a site-specific growth rate independently from age
(in accordance with results of Michiels 1993). The resulting increase in pine shrub cover per site is
augmented by vegetative invasion from neighbouring cells. The model proceeds with simulating
catastrophic mortality events. A random number generator (0 1, uniform distribution) is used and for
each cell the created number is compared to the site specific probability of occurrence of such an
event. If the random number is greater than this probability for a given site its pine shrub cover is reset
to 0%. At last, the model recalculates the age of each Pinus mugo-population at the end of one time
step. Age of a population is represented as a cover-weighted mean of the age of all its individuals.
Parameter estimation
Parameters of recruitment, growth, fecundity and mortality functions were estimated using data
collected on 140 20 x 20 metre plots in the study area and on adjacent mountain ranges. Selection of
plots was based on a stratified random sampling design (Dullinger et al., submitted). Sites already
possessing a dense pine-shrub cover were not considered. We also excluded all sites on currently used
pastures to avoid bias in recruitment rate estimation due to cattle browsing or logging. For growth and
fecundity parameters we additionally used 196 pine shrub individuals selected during random walks in
the same areas.
Recruitment, dispersal functions and invasibility The site-specific 50-year recruitment rate was
determined by counting the number of individuals younger than 50 years on each plot. Age of pine
shrubs was determined for small individuals (diameter < 0.5 m) by counting bud scares if readily
visible (n = 196). If larger individuals were present on the plot, a maximum of three of them (largest,
smallest and median) were sectioned as close to the root collar as possible and rings were counted
under a stereoscope (14x magnification) after grinding the surface (n = 116). For each individual
sectioned, a growth rate was calculated by dividing the length of the sectioned branch by its ring
count. The age of all remaining individuals (n = 386) was estimated as a function of their diameter
(mean of smallest and largest diameter), height and the average growth rate of the sectioned
individuals on the respective plot (R = 0.69, p < 0.0001).
We fitted a recruitment kernel to this data using the distance (two-dimensional Euclidean distance) to
the nearest pine shrub stand (= grid cell with a pine shrub cover > 10% as determined from aerial
photographs) as a predictor and the 50-year recruitment rate as a response. Two alternative functional
forms were applied: a negative exponential and a restricted cubic spline with four knots. Restricted
cubic splines are third-order polynomials within intervals of the predictor forced to be smooth at the
join points (= knots) and constrained to be linear in the tails (Stone & Koo 1985, Harrell 2001).
118

In simulation runs stochastic variation in the number of recruits per site was implemented by drawing
a random number from an exponential distribution for each site with the site-specific recruitment rate
determined from the recruitment kernel as the respective mean. We used an exponential distribution to
mimic the error pattern in fitted recruitment kernels for the plots 0 metres and 20 metres apart from the
nearest seed source.
Finally, the number of recruits per site and time step as calculated by the stochastically varied
recruitment kernels was weighted by two alternative invasibility layers. These layers were derived
from existing fine-scale vegetation maps. The first one assumed equal invasibility (weighting factor =
1) across all types of alpine vegetation (but holding forests and snowbeds uninvasible), the second one
assigned each plant community a specific invasibility value. Derivation of these invasibility values is
described in detail in Dullinger et al. (submitted). In brief, they were calculated as the ratio between
observed recruitment rates in individual plant communities and expected rates under a null model of
equal invasibility using the same parameterisation data set as for demographic variables. These ratios
vary between 0.1 and 2.1 for the plant communities of the study area.
Growth The vegetative growth rate of 231 individuals was measured as the mean length-increment
of a randomly selected major branch in the years 1996 1999. This growth rate was regressed against
abiotic environmental conditions using ordinary least-squares regression. Both linear and non-linear
effects were considered. Non-linear effects were tested for by restricted cubic splines with four knots.
We moreover tested for all two-way interactions among predictors. Model and predictor significance
were obtained from Wald test statistic assuming a Chi-square distribution with one d.f. (Harrell 2001).
The full model was reduced by backward elimination, knot reduction, and linearisation, respectively
(threshold p-value < 0.05). The final coefficient of determination was corrected for possible overfit by
means of bootstrapping (validate-function as implemented in the Design-library of S-PLUS by Alzola
and Harrell 2001; 1000 resamples with replacement).
Fecundity Using the number of cones produced by an individual as an indicator each pine shrub was
assigned an ordinal fecundity value on a four-level scale (0 3). Fecundity was regressed against
environmental conditions, age of the individual, and occurrence of damages (see below) by means of a
proportional odds model (= ordinal logistic regression model, Harrell 2001) using the same procedure
as for the growth function. Instead of least-squares R we report Nagelkerkes R and Somers Dxy as a
measure of concordance between regression model predictions and data.
In the simulation model the population of a site was defined a seed source if its predicted fecundity
value was greater than 1. Age of the population was represented as a cover-weighted mean of the age
of all its individuals.
Mortality We used the distribution of dead individuals across our sampling plots to estimate the site
specific risk of adult mortality by means of logistic regression. The procedure of regression analysis
was the same as for growth and fecundity functions.
In the simulation model the site specific probability of finding a dead individual was set as the site
specific adult mortality per time step. This somewhat arbitrary definition was found to produce
reasonable spatio-temporal mortality patterns in exploratory model runs: Well-known avalanche paths
which are unlikely to ever support closed Pinus mugo cover on account of frequent severe disturbance
do not become overgrown whereas the establishment and persistence of closed populations at less
exposed sites is not affected.
We moreover adapted the concept usually applied to mortality routines in forest gap model
formulations (see Keane et al. 2001 for a review) and defined sites with growth rates lower than the
minimum observed in our parameterisation data set to be not suitable for permanent Pinus mugo
establishment. In the model this was realised by resetting the population of all such sites to 0% cover
after each time step.
119

For all individuals we had also recorded occurrence of damages due to climatic constraints (frost
desiccation, snow-ice abraison) on a four-level ordinal scale. Occurence of damages turned out to be a
significant predictor in both growth and fecundity models. In the simulation model we thus introduced
the site-specific intensity of damages (estimated by means of a proportional odds model, cf. Tab. ) to
the growth- and fecundity functions interpreting it as a weighted interaction term of individual age and
abiotic site conditions.
Simulation runs and evaluation of temperature rise, dispersal, and invasibility effects
on pine shrub expansion
A factorial design was established combining four different climatic scenarios with the two distance
dependent recruitment functions (exponential and restricted cubic spline) and the two alternative
assumptions on spatial invasibility patterns (homogenous and varied invasibility).
As Pinus mugo performance in parameterisation plots did not show any significant effect of water
balance (cf. Tab. ) climatic scenarios only took account for temperature rise. The four different
climatic scenarios were: (1) A baseline scenario assuming current temperature conditions to remain
unchanged. (2) A scenario derived from simulation outputs of the global circulation model ECHAM4
(Roecker et al., 1996) downscaled for Austria by Lexer et al. (2001, 2002). The simulation is based on
the greenhouse emission scenario IS92a (Houghton et al., 1990). According to this scenario, a 0.65C
increase of annual temperature might be expected for the study area by the year 2050 (mean 2035
2065) in comparison with the average of 1961-1995. The temperature regime is assumed to remain
constant thereafter. (3) Like scenario 2, but with a further increase to +1.2C by the year 2100. (4)
Like scenario 3, but with a further increase to +2C by the year 2150. The different scenarios were
implemented into the model by adapting the DD-values of all sites according to the assumed
temperature regime.
Overall, this 2 x 4 x 2 fixed effects-combination yields 16 different scenarios in a fully orthogonal
design. 10 replicates of each combination of treatments were run. For each time step of each replicate
we recorded the total percentage of pine shrub cover, the elevation of the uppermost population and
the maximum distance a new recruit had established from an existing seed source. All scenarios were
run under the assumption that current land use, i.e. cattle grazing and logging, will cease
instantaneously. To exclude edge effects we ran the model for the whole landscape (131901 sites, 52.7
km), but results are reported only for the area above 1700 metres asl. (84827 sites, 33.9 km).
ANOVA was used to test for main effects and interactions of climate scenario, recruitment kernel and
invasibility on the recorded response variables. Assumptions of ANOVA were evaluated using
diagnostic plots (normal probability plots, see e.g. Scheiner & Gurevitch 2001) and calculating
Cochrans C for homogeneity of variances.
Results
Parameterisation of recruitment, growth, fecundity and mortality functions
With an average of about 0.0005 individuals.m-2.year-1 recruitment of Pinus mugo is generally sparse
even where seed sources are available within a radius of 20 metres. This initially low rate declines
sharply with distance form seed sources approaching 7.6x10-6 or 5.9x10-5 individuals m-2.year-1
respectively at 100 metres, depending on the recruitment kernel used. The main differences between
the two recruitment functions are that the restricted cubic spline-model predicts higher recruitment in
the immediate vicinity of a seed source and has a longer tail whereas the exponential model simulates
more recruits at intermediate distances (50 300 m, see Fig. 1). Both functions fit the data

120

significantly though residual variance is considerably lower for the more flexible restricted cubic
spline kernel (see Tab. ).
Mean annual growth is extremely slow. We found pine shrub branches growing only about 5 cm.year-1
(four-year average) with the minimum at 1.8 cm.year-1 and the maximum at 12.cm.year-1. Among
abiotic site conditions, variation in growth rate is most sensible to temperature (see Tab. and Fig. ).
Age of individuals primarily controls fecundity levels. Pine shrubs usually do not start cone
production until they are about 15 - 20 years old. High fecundity rates are hardly achieved by
individuals younger than 50 years. For individuals older than 80 years fecundity becomes independent
of age. The temperature regime significantly affects pine shrub fecundity but has rather low impact
(Tab. , Fig. ).
Topography is the most important factor controlling adult mortality (Tab. ). Mortality is low (2-3%,
see Fig. 2) on average. High values are spatially clustered on exposed ridges, steep slopes with high
erosion potential and risk of avalanches, and on the bottom of slopes where snow is accumulating.
Among climatic variables, temperature is again the most effective predictor. However, in contrast to
growth and fecundity functions its effect is non-linear with lowest mortality at intermediate DDvalues.
Simulation runs and evaluation of temperature rise, dispersal, and invasibility effects on pine
shrub expansion
Pine shrubs are currently covering 9.96% of the study area. Within the next 1000 years the spread
model predicts Pinus mugo to increase its cover to a range between 24.4% and 59.3% depending on
the degree of climatic warming, the assumed recruitment function and the invasibility matrix used in
simulation runs (Fig. 3). Within-scenario variance of predicted values is generally low.
As expected, the rate of pine shrub expansion increases with rising temperatures. The effect of higher
temperature is most pronounced when comparing the baseline scenario with an assumed increase of
0.65C and lowest when switching from a +1.2C- to a +2C-scenario (Fig. ).
Assuming an exponential recruitment kernel considerably accelerates pine shrub spread. After 1000
years simulations run with an exponential kernel predict Pinus mugo to cover an area 20% larger on
average than simulations with a restricted cubic spline-kernel. Similarly, spatially varied invasibility of
the resident vegetation facilitates pine spread. Simulations that take account of varied invasibility
among individual plant communities predict an additional increase of about 15% in the area covered
by Pinus mugo after 1000 years (see Fig. ).
ANOVA results demonstrate that all main effects as well as all interactions significantly affect model
predictions. ANOVA coefficients are highest for the +0.65C-climate scenario and the recruitment
function used (Fig. ). Among interactions, the one between recruitment function and invasibility
matrix is especially pronounced whereas all others are of minor importance.
The uppermost position of Pinus mugo individuals shifts from currently 1935 to a range between 2076
and the highest peak of the modeling area (2273 m asl.) depending on the respective scenario. Main
effects and interactions change model predictions in a similar way as for the area covered by pine
shrubs. Again, the most pronounced interaction is the one between recruitment kernel and invasibility
(Tab. 2).
Tab and Fig. 6 illustrate why this latter interaction is salient in both cases. If an exponential kernel is
used, the maximum distance of a new recruit from the nearest seed source will only marginally be
influenced by the way invasibility is implemented. In contrast, when applying a restricted cubic spline
kernel, this maximum distance is affected considerably by the way invasibility is dealt with.

121

DISCUSSION
Results of this simulation study indicate that pine shrubs will invade and displace current alpine
vegetation under future climate change scenarios rather slowly. The inertia of the system is obviously
due to recruitment and dispersal limitation as well as to long generation times and slow growth rates.
Whereas our estimates of growth rates and generation times agree with values reported in Schroeter
(1926) and Michiels (1993) comparable data on the intensity and spatial pattern of Pinus mugo
recruitment in subalpine and alpine environments are lacking. In general, recruitment limitation seems
to be a widespread phenomenon in temperate forests (Clark et al. 1998) but the number of recruits
detected in our study is particularly low. Ribbens et al. (1994), for example, have defined recruitment
limitation to occur where expected number of recruits drops below 1 individual m-2.year-1 and Mayer
(1976) reports 0.1 0.5 successful recruitment events m-2.year-1 in gaps of subalpine spruce forests
(Picea abies) of the European Alps. Moreover, the initially low rate declines sharply with distance
form seed sources. As seed dispersal distance is closely linked to seed release height (e.g Nathan &
Muller-Landau 2000), we may assume this sharp decline in recruitment to be caused by narrow seed
shadows of the low statured pine shrubs.
Nevertheless, despite such a delayed response pine shrub cover of the modeling landscape is predicted
to increase considerably within the next 1000 years. With climatical conditions unchanged most of this
range expansion process may be due to encroachment of former summer farms. However, even under
a moderate level of temperature rise the area predicted to be covered by pine shrubs in the year 3000
largely exceeds the extent of historically or currently used pastures (see Dullinger et al. 2003 for a list
of historical documents).
The rate of this range expansion process is significantly affected by the assumed climatic scenario as
well as by the recruitment kernel used and the spatial pattern of invasibility. As expected, it is
increasing with rising temperatures, an effect which directly follows from higher growth rates and
fecundity levels and an overall lower risk of mortality. In contrast, the accelerating effect of using an
exponential dispersal kernel seems somewhat counterintuitive as this function approaches to zero more
rapidly than the restricted cubic spline kernel (Fig. 1) and the length of the kernel tail has been
repeatedly demonstrated to control the speed of migration or invasion processes (Kot et al. 1996, Clark
1998, Higgins & Richardson 1999, Neubert & Caswell 2000, Bullock et al. 2002, Shigesada &
Kawasaki 2002). But in fact, probability of recruitment in the tail of the restricted cubic spline-kernel
is too low to create significant differences in the maximum distance of successful dispersal events per
time step (cf. Tab. 2). Thus, higher predicted recruitment at intermediate distances is responsible for
the kernel-effect on pine shrub expansion. In accordance with simulation results of Shigesada &
Kawasaki (2002) more intense short distance dispersal, as predicted by the restricted cubic spline
kernel, has negligible effects on simulation results.
We found a pronounced interaction of dispersal and invasibility (Fig. 6, Tab. 2). Assuming spatially
differential invasibility increases the maximum distance of successful dispersal when using a restricted
cubic spline-kernel but leaves dispersal distances nearly unchanged for the exponential kernel. This
interaction is explained by peculiarities of regional vegetation distribution. The grassland community
most invasible for pine shrubs (swards of Carex firma Mygind) is rare in subalpine regions where the
current vegetation mosaic implies seed sources to be available within rather small distances nearly
everywhere. However, it predominates at the currrent tree-line and in alpine areas. Thus, under
assumptions of differential invasibility a range expansion of pine shrubs into the alpine belt is
facilitated (cf. Dullinger et al., submitted) because of lower levels of competitive inhibition of
germination and establishment. The interaction with the recruitment kernel is caused by an increased
probability of seeds dispersed over long-distances into these alpine areas to reach a habitat with
competitively favourable recruitment conditions. Thus, incorporating differential invasibility into the
122

model strengthens the effect of the thin but longer tail of the restricted cubic spline-kernel due to the
specific regional pattern of habitats suitable to recruitment.
Though invasibility of resident vegetation has long been recognised as a major control of alien plant
spread (Richardson & Bond 1991, Robinson et al. 1995, Wiser et al. 1998, Lonsdale 1999, Shea &
Chesson 2002, Cohen 2002) its possible impact on the rate of climate change driven altitudinal treeline shifts has rather been ignored. In fact, evidence is available that resident vegetation may strongly
affect rates of woody plant encroachment of non-forest subalpine habitats (Magee & Antos 1992,
Rochefort & Peterson 1996, Moir et al. 1999, Dullinger et al., submitted). Competition with
herbaceous vegetation was also invoked as a possible explanation for regionally observed
unexpectedly slow expansion of boreal forests into arctic tundra during the recent decades (Masek
2001). Given the functional similarities of these encroachment processes to e.g the invasion of exotic
pine species into native non-forest vegetation (Richardson & Higgins 1998, Higgins et al. 2001) the
effect of invasibility on rates of altitudinal tree-line shifts is not surprising. Again, the accelerating
effect of assuming spatially varied invasibility on the range expansion process is due to the particular
regional distribution of more and less invasible plant communities. Accounting for differential
invasibility increases rates of pine shrub expansion because more easily invasible communities
dominate most of the newly available terrain above the current tree-line.
In conclusion, results of this modeling study suggest that patterns and rates of climate change driven
transient tree-line dynamics may be rather idiosyncratic. Species will not only vary in their response to
increasing temperatures in terms of their growth-rates or life history schedules. They also possess
different dispersal capacities and competitive abilities during their recruitment phase. Our study
demonstrates that these species-specific traits may affect future range dynamics at least as strong as
variation in regional climatic trends. Moreover, they interact with each other and with peculiarities of
the regional vegetation cover and thus further complicate the task of accurately predicting climate
change driven transient dynamics. Additional modifications of these dynamics may also come from
drivers not considered within this study, e.g. from interactions of temperature increase with other
features of the abiotic environment like water and nutrient availability (Weisberg & Baker 1995,
Chapin et al. 1995) or with disturbance regimes (Cullen et al. 2001), from interference of climate and
land use changes (Didier 2001, Motta & Nola 2001) or from presence of physical migration barriers
(Rupp et al. 2001). Given this complex multi-factorial causation of range dynamics the ambiguous
response of regional tree-lines to the last centurys climatic trends are less surprising.
Acknowledgements
References
Aber, J., Neilson, R.P., McNulty, S., Lenihan, J.M., Bachelet, D., Drapek, R.J. 2001: Forest processes and global
environmental change: predicting the effects of individual and multiple stressors. - BioScience, 51: 735-751.
Bergelson, J., Newman, J.A. & Floresroux, E.M. 1993: Rates of weed spread in spatially heterogenous
environments. - Ecology, 74: 999-1011.
Bolliger, J., Kienast, F. & Bugmann, H. 2000: Comparing models for tree distributions: concepts, structures and
behavior. - Ecological Modelling, 134: 89-102.
Bugmann, H., Pfister, C. 2000: Impacts of interannual climate variability on past and future forest composition. Regional Environmental Change, 1: 112-125.
Bullock, J.B., Moy, I.L., Pywell, R.F., Coulson, S.J., Nolan, A.M. & Caswell, H. 2002: Plant dispersal and
colonization processes at local and landscape scales. In: Bullock, J.M., Kenward, R.E., Hails, R.S. (eds.),
Dispersal Ecology, pp. 279 302. Blackwell.

123

Cain, M.L., Damman, H., Muir, A. 1998: Seed dispersal and the Holocene migration of woodland herbs. Ecological Monographs, 68: 325-347.
Clark, J.S., Beckage, B, Camill, P., Cleveland, B., HilleRisLambers, J., Lichter, J., McLachlan, J., Mohan, J. &
Wyckoff, P. 1999: Interpreting recruitment limitation in forests. - American Journal of Botany, 86: 1-16.
Clark, J.S., Macklin, E. & Wood, L. 1998: Stages and spatial scales of recruitment limitation in southern
Appalachian forests. - Ecological Monographs, 68: 213-235.
Clark, J.S., Silman, M., Kern, R., Macklin, E. & HilleRisLambers, J. 1999: Seed dispersal near and far: Patterns
across temperate and tropical forests. - Ecology, 80: 1475-1494.
Cuevas, J.G. 2002: Episodic regeneration at the Nothofagus pumilio alpine timberline in Tierra del Fuego, Chile.
- Journal of Ecology, 90: 52-60.
Cullen, L., Stewart, G.H., Duncan, R.P. & Palmer, G. 2001: Disturbance and climate warming influences on
New Zealand Nothofagus tree-line population dynamics. - Journal of Ecology, 89: 1061-1071.
De Steven, D. & Wright, J. 2002: Consequences of variable reproduction for seedling recruitment in three
neotropical tree species. - Ecology, 83: 2315-2317.
Didier, L. 2001: Invasion patterns of European larch and Swiss stone pine in subalpine pastures in the French
Alps. - Forest Ecology and Management, 145: 67-77.
Dirnbck, T., Dullinger S., Gottfried, M. & Grabherr, G. 1999: Die Vegetation des Hochschwab (Steiermark) Alpine und Subalpine Stufe. Mitteilungen des Naturwissenschaftlichen Vereins der Steiermark 129: 111-251.
Dirnbck, T., Dullinger, S. & Grabherr, G. 2003. A regional impact assessment of climate and land use change
on alpine vegetation. Journal of Biogeography, in press.
Dubayah, R. & Rich, P.M. 1996: GIS-Based Solar Radiation Modeling. - in: Goodchild, M.F., Steyaert, L.T. &
Parks, B.0.: GIS and Environmental Modeling, John Wiley & Sons.
Gallant, J.C., Wilson, J.P. 1996: Tapes-G: a grid-based terrain analysis program for the environmental science. Computers & Geoscience, 22 (7): 713-722.
Grabherr, G., Gottfried, M., Gruber, A. & Pauli, H. 1995: Patterns and Current Changes in alpine Plant
Diversity. - Ecological Studies, Vol. 113: 167-181.
Greene, D.F. & Johnson, E.A. 1997: Secondary dispersal of tree seeds on snow. - Journal of Ecology, 85: 329340.
Guisan, A., Zimmermann, N.E. 2000: Predictive habitat distribution models in ecology. - Ecological Modelling,
135: 147-186.
Hafenscherer, J. & Mayer, H. 1986: Standort, Aufbau, Entwicklungsdynamik und Verjngung von
Latschenbestnden im Karwendeltal/Tirol. - Schweiz. Zeitschrift f. Forstwesen, 137/3: 177-203.
Hansen, A.J., Neilson, R.P., Dale, V.H., Flather, C.H., Iverson, L.R., Currie, D.J., Shafer, S., Cook, R., Bartlein,
P.J. 2001: Global change in forests: response of species, communities, and Biomes. - BioScience, 51: 765779.
Harrell, F.E. 2001: Regression modeling strategies: with application to linear models, logistic regression, and
survival analysis. - Springer Series in Statisitcs, Springer Verlag, New York, pp. 568.
Herrera, C.M., Jordano, P., Guitin, J. & Traveset, A. 1998: Annual variability in seed production by woody
plants and the masting concept: Reassessment of principles and relationship to pollination and seed dispersal.
- American Naturalist, 152: 576-594.
Hessl, A.E. & Baker, W.L. 1997: Spruce and fir regeneration and climate in the forest-tundra ecotone of Rocky
Mountains National Park, Colorado, U.S.A.. - Arctic and Alpine Research, 29: 173-183.
Higgins, S.I. & Richardson, D.M. 1996: A review of models of alien plant spread. - Ecological Modelling, 87:
249-265.
Higgins, S.I. & Richardson, D.M. 1999: Predicting plant migration rates in a changing world: The role of longdistance dispersal. - American Naturalist, 153: 464-475.

124

Higgins, S.I., Richardson, D.M., Cowling, R.M. 2001: Validation of a spatial simulation model of a spreading
alien plant population. - Journal of Applied Ecology, 38: 571-584.
Keane, R., Austin, M., Field, C., Huth, A., Lexer, M.J., Peters, D., Solomon, A. & Wyckoff, P. 2001: Tree
mortality in gap models: Application to climate change. - Climatic Change, 51: 509-540.
Keeley, J.E. & Zedler, P.H. 1998: Evolution of life histories in Pinus. - Richardson, D.M. (ed.): Ecology and
Biogeography of Pinus, Cambridge University Press, Cambridge, pp. 219-251.
Kittel, T.G.F., Steffen, W.L., Chapin, F.S. 2000: Global and regional modelling of arctic-boreal vegetation
distribution and its sensitivity to altered forces. - Global Change Biology, 6: 1-18.
Klasner, F.L., Fagre, D.B. 2002: A half century of change in alpine treeline patterns at Glacier National Park,
Montana, U.S.A.. - Arctic, Antarctic, and Alpine Research, 34: 49-56.
Krner, C. 1998: A re-assessment of high elevation treeline positions and their explanation. - Oecologia, 115:
445-459.
Krner, Ch. 1999: Alpine plant life: Functional plant ecology of high mountain ecosystems. - Springer, pp 338.
Kot, M., Lewis, M.A. & van den Driessche, P. 1996: Dispersal data and the spread of invading organisms. Ecology, 77: 2027-2042.
Kullman, L. 1993: Pine (Pinus sylvetris L.) tree-limit surveillance during recent decades, Central Sweden. Arctic and Alpine Research, 25: 24-31.
Kullman, L. 2002: Rapid recent range-margin rise of tree and shrub species in the Swedish Scandes. - Journal of
Ecology, 90: 68-77.
Lavoie, C. & Payette, S. 1994: Recent fluctuations of the lichen-spruce forest limit in subarctic Quebec. - Journal
of Ecology, 82: 725-734.
Loehle, C. & LeBlanc, D. 1996: Model-based assessments of climate change effects on forests: a critical review.
- Ecological Modelling, 90: 1-31.
Lonsdale, W.M. 1999: Global patterns of plant invasions and the concept of invasibility. - Ecology, 80: 15221536.
Magee, T.K. & Antos, J.A. 1992: Tree invasion into a mountain-top meadow in the Oregon Coast Range, USA. J. Veg. Sci., 3: 485-494.
Malanson, G.P. & Cairns, D.M. 1997: Effects of dispersal, population delays, and forest fragmentation on tree
migration rates. - Plant Ecology, 131: 67-79.
Masek, J.G. 2001: Stability of boreal forest stands during recent climate change: evidence from Landsat satellite
imagery. - Journal of Biogeography, 28: 967-976.
Mayer, H. 1976: Gebirgswaldbau - Schutzwaldpflege. - G. Fischer Verlag, Stuttgart, pp 436.
Meshinev, T., Apostolova, I., Koleva, E.S. 2000: Influence of warming on timberline rising: a case study on
Pinus peuce Griseb. in Bulgaria. - Phytocoenologia, 30: 431-438.
Michiels, H.G. 1993: Die Stellung einiger Baum- und Straucharten in der Struktur und Dynamik der Vegetation
im Bereich der hochmontanen und subalpinen Waldstufe der Bayrischen Kalkalpen. - Forstliche
Forschungsberichte Mnchen, 135, 300 pp..
Moir, W.H., Rochelle, S.G. & Schoettle, A.W. 1999: Microscale patterns of tree establishment near upper
treeline, Snowy Range, Wyoming, U.S.A.. - Arctic, Antarctic, and Alpine Research, 31: 379-388.
Motta, R., Nola, P. 2001: Growth trends and dynamics in sub-alpine forest stands in the Varaita Valley
(Piedmont, Italy) and their relationships with human activities and global change. - J. Veg. Sci., 12: 219-230.
Mller-Schneider, P. 1986: Verbreitungsbiologie der Bltenpflanzen Graubndens. - Verff. des Geobot. Inst
ETH, Stiftung Rbel, Zrich, 85: 1-263.
Nathan, R. & Muller-Landau, C. 2000: Spatial patterns of seed dispersal, their determinants and consequences
for recruitment. - Trends in Ecology and Evolution, 15: 278-285.

125

Nathan, R., Horn, H.S., Chave, J. & Levin, S.A. 2001: Mechanistic models for tree seed dispersal by wind in
dense forests and open landscapes. - In : Levey, D.J., Silva, W.R. & Galetti, M. (eds.), Seed dispersal and
Frugivory: Ecology, Evolution and Conservation, CAB International Press, Oxfordshire.
Neubert, M.G. & Caswell, H. 2000: Demography and dispersal: Calculation and sensitivity analysis of invasion
speed for structured populations. - Ecology, 81: 1613-1628.
Overpeck, J.T., Bartlein, P.J. & Webb, T. 1991: Potential magnitude of future vegetation change in eastern North
America: Comparisons with the past. - Science, 254: 692-695.
Parmesan, C. 1996: Climate and species range. - Nature, 382: 765-766.
Parmesan, C. & Yohe, G. 2003: A globally coherent fingerprint of climate change impacts across natural
systems. - Nature, 421: 37-42.
Paulsen, J., Weber, U.M. & Krner, C. 2000: Tree Growth near Treeline: Abrupt or Gradual Reduction with
Altitude. - Arctic, Antarctic, and Alpine Research, 32: 14-20.
Payette, S., Filion, L., Delwaide, A. & Bgin, C. 1989: Reconstruction of tree-line vegetation response to longterm climate change. - Nature, 341: 429-432.
Ribbens, E., Silander, J.A. & Pacala, S.W. 1994: Seedling recruitment in forests: Calibrating models to predict
patterns of tree seedling dispersion. - Ecology, 75: 1794-1806.
Richardson, D.M. & Bond, W.J. 1991: Determinants of plant distribution: Evidence from pine invasions. American Naturalist, 137: 639-668.
Richardson, D.M. & Higgins, S.I. 1998: Pines as invaders in the southern hemishpere. - Richardson, D.M. (ed.):
Ecology and Biogeography of Pinus, Cambridge University Press, Cambridge, pp. 450-473.
Robinson, G.R., Quinn, J.F. & Stanton, M.L. 1995: Invasibility of experimental habitat islands in a california
winter annual grassland. - Ecology, 76: 786-794.
Rochefort, R.M. & Peterson, D.L. 1996: Temporal and spatial distribution of trees in subalpine meadows of
Mount Rainier National Park, Washington, U.S.A.. - Arctic and Alpine Research, 28: 52-59.
Rupp, S.T., Chapin, F.S., III & Starfield, A.M. 2001: Modeling the influence of topographic barriers on treeline
advance at the forest-tundra ecotone in northwestern Alaska. - Climatic Change, 48: 399-416.
Rupp, T.S., Starfield, A.M. & Chapin, S. III 2000: A frame-based spatially explicit model of subarctic vegetation
response to climatic change: comparison with a point model. - Landscape ecology, 15: 383-400.
Schwalm, C.R. & Ek, A.R. 2001: Climate change and site: relevant mechanisms and modeling techniques. Forest Ecology and Management, 150: 241-257.
Shea, K. & Chesson, P. 2002: Community ecology theory as a framework for biological invasions. - Trends in
Ecology and Evolution, 17: 170-176.
Shigesada, N. & Kawasaki, K. 2002: Invasion and the range expansion of species: effects of long-distance
dispersal. In: Bullock, J.M., Kenward, R.E. & Hails, R.S. (eds.), Dispersal Ecology, pp. 350-373.
Starfield, A. M. & Chapin, III, F.S. 1996: Model of transient changes in arctic and boreal vegetation in response
to climate and land use change. Ecological Applications 6: 842-864.
Stone, C.J. & Koo, C.Y. 1985: Additive splines in statistics. Proceedings of the Stastical Computing Section
ASA: 45-48. Washington, D.C.
Szeicz, J.M. & MacDonald, G.M. 1995: Recent white spruce dynamics at the subarctic alpine treeline of northwestern Canada. - Journal of Ecology, 83: 873-885.
Theurillat, J.-P. & Guisan, A. 2001: Potential impact of climate change on vegetation in the European Alps: A
review.. - Climatic Change, 50: 77-109.
Thomas, C.D., Bodsworth, E.J., Wilson, R.J., Simmons, A.D., Davies, Z.G., Musche, M. & Conradt, L. 2001:
Ecological and evolutionary processes at expanding range margins. - Nature, 411: 577-581.
Tinner, W., Ammann, B. & Germann, P. 1996: Treeline fluctuations recorded for 12500 years by soil profiles,
pollen, and plant macrofossils in the Central Swiss Alps. - Arctic, Antarctic, and Alpine Research, 28: 131147.

126

Vander Wall, S.B. 1992: The role of animals in dispersing a "wind-dispersed" pine. - Ecology, 73: 614-621.
Walther, G.-R., Post, E., Convey, P., Menzel, A., Parmesan, C., Beebee, T.J.C., Fromentin, J.-M., HoeghGuldberg, O. & Bairlein, F. 2002: Ecological responses to recent climate change. - Nature, 416: 389-395.
Weisberg, P.J. & Baker, W.L. 1995: Spatial variation in tree seedling and krummholz growth in the forest-tundra
ecotone of Rocky Mountain National Park, Colorado, U.S.A.. - Arctic and Alpine Research, 27: 116-129.

Figures

127

Figure1: Recruitment kernel of Pinus mugo fitted to data from 140 field plots. Straight line: Restricted
cubic spline function. Dashed line: Exponential function. Each site is 400 m, the time step is 50 years.

10

Recruits/TimeStep/Site

100

200

300

Distance (m)

128

400

500

600

700

6
5
2

cm.year-1

Figure 2: Effects of temperature (number of days with mean temperature > 0C) on growth, fecundity
and mortality of Pinus mugo. Dashed lines represent 95%-confidence intervals. Additional variables in
multivariate regression models are set to the following values: Growth:Damages = 0, SOIL = 1, SRJ =
2.71 MJ, WET = 5.94; Fecundity: Age = 40, Damages = 0, SRM = 2.58 MJ, GEO = limestone.
Mortality: WET = 5.9, EROS = -0.27, SLOPE = 18.5

210

215

220

225

230

235

225

230

235

225

230

235

0.6
0.5
0.4

Prob(Fecundity > 1)

0.7

Degree Days

210

215

220

0.97
0.96
0.95

Prob(Mortality = 0)

0.98

0.99

Degree Days

210

215

220
Degree Days

129

Figure 3: Predicted increase of the area covered by Pinus mugo during the next 1000 years. Maximum,
minimum and median of 160 model runs. Error bars represent standard deviations

70

60

%Area

50

40

30

20

10

0
1800

2000

2200

2400

2600

Year

130

2800

3000

3200

30

40

50

60

Figure 4: Effects of climate change scenario, type of recruitment kernel and invasibility matrix on the
area predicted to be covered by Pinus mugo 1000 years A.P. exp exponential recruitment kernel; rcs
restricted cubic spline recruitment kernel.

+0.65C

+1.2C

+2C

30

40

50

60

+0C

rcs

homogeneous

variable

30

40

50

60

exp

131

Figure 5: ANOVA-coefficients for factor levels a) response: area predicted to be covered by Pinus
mugo 1000 years A.P. b) Uppermost position of Pinus mugo individuals 1000 years A.P. ClimScen:
Climate change scenario, Invas: Invasibility, RecrKern: Recruitment kernel. Colons symbolize
interaction terms.

RecrKern:Invas
ClimScen3:Invas
ClimScen2:Invas
ClimScen1:Invas
ClimScen3:RecrKern
ClimScen2:RecrKern
ClimScen1:RecrKern
Invas
RecrKern
ClimScen3
ClimScen2
ClimScen1

-6

-2

RecrKern:Invas
ClimScen3:Invas
ClimScen2:Invas
ClimScen1:Invas
ClimScen3:RecrKern
ClimScen2:RecrKern
ClimScen1:RecrKern
Invas
RecrKern
ClimScen3
ClimScen2
ClimScen1

-30

-10

10

132

30

Figure 6: Interacting effects of recruitment kernel and invasibility on the maximum distance from a
seed source a new recruit is predicted to establish during one 50 year time step. hom: Invasibility
assumed to be homogenous across different alpine plant communities; var: Invasiblity assumed to be
varied across different plant communities.

400
380

MaxDistance

360
340
320
300
280
260
hom

var

Invasibility

133

Table 1: Variables representing abiotic habitat conditions and sources they were derived from
Variable

Source

Abbreviation

Degree Days

DGM, climate station data

DD

Solar radiation in May, July and

September

DGM, SOLARFLUX (Dubayah & Rich 1996)

SRM, SRJ, SRS

Water Balance in August

DGM, climate station data, SOLARFLUX (Dubayah

& Rich 1996)

WBA

Wind speed

DGM, climate station data, NUATMOS (Ross et al.

1988, Bachmann 1998)

WSP

Slope Inclination

DGM

SLOPE

Soil erosion potential

DGM, TAPES-G (Gallant & Willson 1996)

EROS

Topographic Wetness Index

DGM, TAPES-G (Gallant & Willson 1996)

WET

Bedrock mineralogy

Geological map

GEO

Distribution of chromic Cambisols

DGM, 573 sample points

SOIL

134

Table 2: Regression functions for parameterising the demographic processes of the pine shrub spread
model. LS Least squares regression, LR logistic regression, PO proportional odds regression. For
each multiple regression function only significant predictors are listed. Distance Distance to nearest
pine shrub stand; Damages Probability of climatic damages; Age Age of pine shrub individual. For
abbreviations of abiotic habitat variables see Table 1. *** - p < 0.001, ** - p < 0.01, ** - p < 0.05. Rand Somers Dxy-values were corrected for possible overfit by bootstrapping (1000 resamples). exp
exponential recruitment kernel, rcs restricted cubic spline recruitment kernel.
Model
Type

Predictors

R-orig

R-corr

Dxy-orig

Dxy-corr

Recruitment_EXP

LS

Distance***

0.62

0.59

< 0.0001

Recruitment_RCS

LS

Distance***

0.88

0.85

< 0.0001

Growth

LS

DD***
Damages***
SOIL*,
WET*
SRJ*

0.49

0.44

< 0.0001

Fecundity

PO

Age***
GEO***
Damages ***
SRM**
DD*

0.57

0.54

0.80

0.79

< 0.0001

Mortality

LR

EROS***
WET***
DD**
SLOPE*

0.27

0.24

0.69

0.66

< 0.0001

Occurence of
damages

PO

DD***
SRJ***
Age***
SLOPE*
WET*

0.4

0.38

0.7

0.68

< 0.0001

135

Table 3: Effects of climate change scenario (ClimScen), type of recruitment kernel (RecrKern),
invasibility matrix (Invas), and their interaction terms on the area covered by Pinus mugo 1000 years
A.P. (= Area), the uppermost position of Pinus mugo-individuals in 3000 (= Altitude), the maximum
distance from a seed source a new recruit is predicted to establish during one time step (=
MaxDistance). Results of fixed-effects factorial ANOVA. Df Degrees of freedom.
Treatment
ClimScen
RecrKern
Invas
ClimScen:RecrKern
ClimScen:Invas
RecrKern:Invas

Area

Altitude

Df

F-value

F-value

3
1
1
3
3
1

26150
32568
8020
900
181
4241

< 0.0001
< 0.0001
< 0.0001
< 0.0001
< 0.0001
< 0.0001

530
1817
929
8
23
347

136

MaxDistance
p

F-value

< 0.0001 5.3

0.001
< 0.0001 0.2
0.65
< 0.0001 123.4 < 0.0001
0.0005 3.2
0.02
< 0.0001 3.5
0.02
< 0.0001 112.4 < 0.0001

7
Kck, R., Hrtel, E., Holtermann, C., Hochbichler, E. & Hager, H.: Soil
Moisture Dynamics Related to Vegetation Cover in the Subalpine Zone of
the Northeastern Calcareous Alps in Austria. Results of Case Studies in the
Rax Area.
Artikel, publiziert in der Zeitschrift
Centralblatt fr das gesamte Forstwesen

137

138

139

140

141

142

143

144

145

146

8
Kck, R., Hrtel, E., Holtermann, C., Hochbichler, E. & Hager, H.:
Monitoring hydrological processes in montane and subalpine karst regions:
comparison between different types of vegetation. Experimental design,
techniques and first results.
Manuskript, zur Publikation akzeptiert in der Zeitschrift
UNESCO Technical Documents of Hydrology

147

Monitoring hydrological processes in montane and subalpine karst

regions: comparison between different types of vegetation.
Experimental design, techniques and first results.
Kck, R.*1, Hrtel, E.**, Holtermann, C.**, Hochbichler,E.*, Hager, H.***, Schnthaler,
K.**
* Institute of Silviculture, University of Agricultural Sciences, Austria.
** Institute of Landscape Architecture and Landscape Management, Univ. of Agricult. Sciences, Austria.
*** Institute of Forest Ecology, University of Agricultural Sciences, Austria.
1
Corresponding author: Institute of Silviculture, Peter Jordanstr. 70, A 1190 Vienna, Austria.

Abstract
On the Rax Alpe, an eastern elevation of the Northeastern Calcareous Alps of Austria, five experimental plots
have been selected within representative types of vegetation cover (forest, krummholz and subalpine grassland)
in order to describe hydrological parameters for these vegetation classes. The research area is located in the
water resources protection zone for the Viennese drinking water supply. Three vegetation types were analysed in
the montane zone (1000 m asl), a mixed spruce-fir-beech stand (B1), a homogeneous spruce plantation (F1) and
a regeneration area (B2). Two vegetation types were analysed in the subalpine zone (1800 m asl), a subalpine
grassland (A1) and a krummholz area (L1 Pinus mugo). A comparable set of measuring instruments was
installed on each experimental plot. Differences between the specific vegetation types, which seem to be
significant for the hydrology of these karstic headwaters, could be outlined by the comparison of various
parameters.
Keywords: Karst, forest, krummholz, subalpine grassland, interception, soil moisture.

Introduction
The City of Vienna receives about 95 % of its drinking water from the Northeastern Calcareous Alps
of Austria. The watershed areas are covered with forests, alpine grasslands or are simply barren rock
areas. In order to estimate hydrological differences between representative types of vegetation, five
experimental plots have been installed on the Rax Alpe, which is a mountain within the water
resources protection zone of the City of Vienna. Within the montane and within the subalpine zone,
the investigated vegetation classes are compared regarding specific hydrological parameters. By the
means of comparison, hydrological differences between the vegetation classes can be emphasized. The
study also contributes to the formulation of management guidelines, which define the most adequate
forest composition, forest structure and land use practices in order to reach the purpose of an
optimisation of water quality and water yield for the drinking water supply of the City of Vienna.
Methods
Comparison between the specific vegetation classes, which are represented by the experimental plots,
has been conducted on three plots within the montane zone and on two plots within the subalpine
zone. In order to represent wide spread forest types of the montane zone, a natural old grown mixed
stand of spruce, fir and beech (B1, natural vegetation about 150 years old), a homogeneous 60 years
old spruce plantation (F1) and a regeneration area (B2, young trees and clearcut vegetation) have been
148

selected as appropriate for the installation of three experimental plots. They are situated on the same
altitude (1040 m asl), they have the same inclination (30), similar aspects (from SW to W) and
relatively similar soil types (regarding to the highly variable karst soils). The soils are calcareous
leptosols and the bedrock is dolomite, which is mixed with limestone particles. On the forested plots
the humus layers are thick and variable and on the regeneration plot the humus layers are shallow and
relatively homogenous. The thickest humus layers can be found on the B1 plot (the natural mixed
stand of spruce, fir and beech). On all three plots the potential natural forest community is HelleboroAbieti-Fagetum typicum (Zukrigl 1973, Kck et al. 2002). On the Rax plateau, two experimental plots
have been installed in 1840 m asl. The first one is situated within a krummholz area (Pinus mugo L1)
and the second one within an open subalpine grassland (A1). Krummholz vegetation and subalpine
and alpine grasslands cover large areas of these altitudinal regions. The area on the Rax plateau is
level and slightly undulating, the soils on both plots are chromic cambisols and the bedrock is
limestone. The humus layer on the krummholz area (L1) is shallow and homogeneous (around 5 cm),
while there is only mull-humus on the grassland-plot (A1).
On all five plots a comparable set of measuring instruments has been installed. In order to guarantee
measurement cycles throughout the year, data loggers have been installed in plastic boxes within the
soils. The data loggers are MiniCube loggers from EMS (Environmental Measuring Systems, Brno).
Air temperature and relative humidity were measured at a height of 2 m on all five plots. Also soil
temperature was measured in 5 cm, 15 cm, 30 cm and 45 cm depth on all five plots. Wind speed was
measured 2 m above the ground on all experimental plots except L1 (krummholz-plot) and 2 m above
the crowns of two dominant trees in B1 (spruce-fir-beech plot) and in F1 (spruce plantation plot).
Solar radiation was measured 3 m above the ground at two places, respectively in B2 (regeneration
plot), which is the open reference base within the montane zone and in A1 (subalpine grassland),
which is the open reference base within the subalpine zone. Soil moisture was measured in a depth of
20 cm and 50 cm in the montane zone, using Frequency Domain (FD) probes (Delta-T Devices Ltd.
1999). This design has been installed with 5 replications in B1 and with three replications in F1 and
B2. Within the subalpine zone, soil moisture probes have been installed at a depth of 20 cm and 35 cm
below the soil surface. Both on the A1 plot and on the L1 plot the measuring design has been installed
with two replications. Because of the high gravel contents of the karstic soils, the theta probes have
been installed horizontally in that way, that the rods are surrounded by homogenized autochthonous
soil material, settled to its natural bulk density, instead of inserting them into the natural layered soil.
This has been executed because of the significant underestimation of soil moisture contents, which is
caused by gaps of soil around the rods, which occurs more than likely, if the sensors are placed into
the undisturbed rocky soils (v. Wilpert et al. 1998). The soil, where the sensor is inserted should also
not be compacted (Robinson et al. 1999). Only within the naturally dense and homogeneous humus
layers, the Theta probes (Delta-T Devices Ltd.) have been installed into the undisturbed soil. Brner et
al. (1996) found a strong linear relationship between the volumetric water content of litter measured
by TDR probes and the litter water content calculated from gravimetric values.
In order to describe the precipitation events, bulk raingauges have been installed for measuring
precipitation in the open area (gross rainfall) in B2, close to B1 and in A1. Within the montane zone,
canopy throughfall collectors have been installed at all three plots, five in B1, three in F1 and two in
B2. In the subalpine zone, six canopy throughfall collectors have been installed in L1 within the
krummholz bushes. At the B1 plot, where beech is growing, stemflow gauges have been installed on
three beech trees in order to characterize interception loss for this mixed stand (spruce, fir, beech,
maple). Stemflow on spruce occurs with quantities around 1 % or less and was therefore considered
negligible (Hager & Holzmann 1997). During the winter season, snow cover (snow depth) and the
149

water equivalent of the snow cover have been investigated at all plots, following a snow course
scheme.
At all five plots, lysimeters have been installed in order to obtain seepage water for water quality
analysis. The suction lysimeters (ceramic cups with 8 cm diameter) were installed 15 cm and 60 cm
below the soil surface at the three plots in the montane zone. In B1 this device has been set up with six
replications, in F1 and B2 with three replications. Due to the soil properties in the subalpine zone
(high rates of rock material in the deeper soil horizons), the ceramic cups have been installed there in a
depth of 15 cm and 30 cm. At the A1 and L1 plot the lysimeter device has been set up with two
replications. The seepage water samples were analysed regarding nitrate, ammonium, calcium,
potassium, magnesium, sodium, pH value and electric conductivity.

First results and discussion

Monitoring of precipitation, canopy throughfall and stemflow during three summer seasons was able
to show the high variation of these parameters according to the characteristics of precipitation events.
Rainfall in general is a phenomena, which shows high spatial variability (Lopes 1996, Hanson et al.
1989). Between the two experimental plots B1 (spruce-fir-beech plot) and F1 (spruce plantation) in the
montane zone, gross rainfall shows different annual values (see table 1, GR). Interception within the
stands (B1 and F1) shows a high variability (Durocher 1990) as well as each precipitation event shows
different interception rates, dependant on the intensity of rainfall (Price et al. 1997) and other
meteorological variables (like temperature, wind etc). Due to the high variability of the canopy
structure within the mixed stand B1 (spruce-fir-beech
Table 1: Overview for summer canopy throughfall and net precipitation for a mixed spruce-firbeech stand (B1) and a homogeneous spruce plantation (F1):

B1
1999

2000

2001

Ctf
StD
Ste
NeP
Ctf
StD
Ste
NeP
Ctf
StD
Ste
NeP

65.8 %
5.3 %
7.3 %
73.1 %
60.5 %
6.0 %
6.5 %
67,0 %
53.6 %
4.4 %
6.6 %
60.2 %

Ctf
StD
Ste
(GR: 1178 mm) NeP
Ctf
StD
Ste
(GR: 882 mm) NeP
Ctf
StD
Ste
(GR: 813 mm) NeP

F1
69.7 %
3.9 %
-69.7
(GR: 1226 mm)
61.6 %
0.8 %
-61.6 % (GR: 901 mm)
53.0 %
3.1 %
-53.0 % (GR: 891 mm)

Ctf is the canopy throughfall, calculated as the mean of all troughs on each plot, StD is the
standard deviation of canopy throughfall for each plot, Ste is the stemflow on beech, calculated
as the value for the whole plot area of B1, NeP is the net precipitation, which is identical with
canopy throughfall in F1 and which is the sum of canopy throughfall and stemflow in B1; GR is
gross rainfall (mm). Values are expressed as a percentage of GR.
150

plot) in relation to the homogeneous stand in F1 (spruce plantation), the variability of canopy
throughfall is also higher in B1 than in F1; the standard deviation of canopy throughfall is therefore
higher in B1 than in F1 (Table 1). Stemflow on the beech trees within B1 is the quantity, which is
responsible for the difference in net precipitation between the mixed stand (B1) and the spruce
plantation (F1). It also seems to be obvious, that higher values of gross precipitation like in 1999 lead
to higher net precipitation values in relation to years with lower values of gross precipitation (like
2001, Table 1). The variation of interception loss for each precipitation event is dependant on the
intensity of rainfall. Intensive rainfalls show lower interception rates than rainfalls with lower
intensity. In F1 (spruce plantation) interception loss (on an event base) during the measuring period of
the year 2000 ranged between 8,9 % and 81,3 %. During the measuring period 2001 the interception
loss in F1 ranged between 22,2 and 67,6 %. The years with higher rainfall amounts show also higher
net precipitation.
In the subalpine zone, interception loss within the krummholz plot (L1 Pinus mugo) is highly
variable. Interception rates constituted between 19 % and 65 % of gross rainfall, while at the same
time one trough received 118 % of gross rainfall. The variations of canopy throughfall within the
krummholz plot could be related to stemflow dynamics at the Pinus mugo plants, which have not been
measured because of technical obstacles due to the growing shape of this plant. Pinus mugo is growing
with a high variation of the angle of the branches, so that the place for mounting of stemflow gauges is
not detectable. In the desert of Arizona it was possible to mount stemflow gauges at the base of
creosote bushes (Larrea tridentata) (Whitford et al. 1997), a technique, which would not be applicable
for Pinus mugo. Another reason for the canopy throughfall variations could be the capability of Pinus
mugo to intercept moisture from fog and clouds (occult precipitation). At the edges of the krummholz
groups, a higher rate of moisture is likely to be intercepted. In other montane forests of the earth,
occult precipitation is an important quantity (Hunzinger 1997, Flemming 1993). Also wind-driven
rainfall can have the effect of inhomogeneous distribution of canopy throughfall (Herwitz & Slye
1995). The canopy throughfall troughs quantify occult precipitation only as a unit together with gross
rainfall, therefore it is not possible to measure the percentage of total precipitation, which is given as
occult precipitation, with this technique.
Soil temperature is influenced by vegetation cover. In the subalpine zone, soil temperature (5 cm
depth) on the open subalpine grassland (A1) reaches maxima, which are more than 7C warmer than
on the krummholz area (Pinus mugo plot L1- on June 10th 2000 - Fig.1). The monthly means during
the summer season 2000 were also higher in A1 than in L1; the difference, viewed over all measured
soil horizons, varied between 0,5C and 4C. The trend of warmer soil temperatures in A1 during the
summer season was inverted during the winter season, where L1 exhibited warmer soil temperatures
than A1. Within the krummholz the soil did not freeze during the winter season (like 1999/2000) or
froze only slightly in the upper horizon (2000/2001), while the soil in A1 froze significantly (less than
minus 1C) and over all measured soil horizons. During the snowmelt period, the warmer soil
temperatures under krummholz vegetation (L1) allowed the melting water to percolate into the soil,
while on the frozen soil in A1 (open grassland) percolation could not occur on the whole area during
the same time.

151

Figure 1: Soil temperature 5 cm below the soil surface: A1 (subalpine grassland) and L1
(krummholz area, Pinus mugo).
This effect of soil temperature during the winter season has been estimated by the comparative
analysis of soil moisture and soil temperature dynamics during the snowmelt period in spring time
2000 in the subalpine zone. Melting water or precipitation water percolates easier into soils without
ground frost than into frozen soils (Shanley & Chalmers 1999). Also in the montane zone, soil
temperature shows characteristics according to vegetation cover. Peaks in soil temperature in the upper
soil horizons are differing on radiation days by between 2C and 7C during the day time between the
mixed forest stand B1 and the open regeneration plot B2, with the higher soil temperatures occurring
in B2. The homogeneous spruce-plantation (F1) exhibits slightly higher soil temperatures than the
mixed stand in B1. Soil temperature is a key parameter for many geo-chemical processes, which start
to be evident after using the clearcutting technique of harvesting (Likens & Bormann 1995, Reynolds
et al 1992, v. Wilpert et al 2000, Martin et al 2000).
Soil moisture dynamics showed different aspects for each vegetation type. In the subalpine zone for
example, soil moisture remained higher under krummholz vegetation (Pinus mugo L1) than under
subalpine grassland vegetation (A1) after a winter season with a high level of snow accumulation like
in 1999/2000. First during long lasting dry spells, which occurred in the year 2000 during the month of
August, soil moisture dropped quicker at L1 than at A1. In June 2000, soil moisture was at every
measurement plot of L1 higher than in A1. In August the soil moisture levels in L1 have been lower
than in A1 at two plots. Soil moisture is, especially during dry periods of a season, controlled by the
soil type and vegetation (Gautam et al. 2000). During the dry spell period in August 2000, soil
moisture dropped quicker under the krummholz vegetation than under subalpine grassland vegetation,
which can be related to higher transpiration demand of Pinus mugo in relation to grassland vegetation
(Dirnbck & Grabherr 2000). The high amounts of soil moisture content within the krummholz area in
152

the summer months of 2000 can be related to high snow accumulation during the winter season of
1999/2000. The snowmelt water percolated with higher ease into the soils which were covered by
krummholz vegetation. Variation of soil moisture conditions are higher within the krummholz area
(L1) than within the subalpine grassland area (A1). This can be related to higher variability of
precipitation distribution due to the growing shape of the Pinus mugo plants. Variations of soil
moisture conditions due to the variation of soils are not the case within the subalpine zone of the Rax,
because the soils on A1 and L1 are identical (chromic cambisols). The variation from differing depth
of groundwater table can be excluded since groundwater is not relevant for karstic sites at this altitude,
but which occurs in watersheds with crystalline bedrocks (Beldring et al. 1999).
The analysis of the seepage water, which has been extracted by the lysimeters, showed the highest
nitrate concentrations on the spruce plot F1 (8,7 mg NO3-/l as the highest value), while the
regeneration area B2 had the lowest nitrate concentrations due to already occurred mobilization and
erosion and (or) leaching of the nutrients, which had been stored in humus and root biomass, after the
clearcut period. The clearcut was made after wind blowdown, which took place 25 years ago. On the
plot in B1 the seepage water shows varying nitrate concentrations, which are generally lower than in
F1. After clearcutting, the nitrate concentration in seepage water can reach high values (v. Wilpert et
al. 2000, Likens & Bormann 1995, Reynolds et al. 1992, Martin et al. 2000). In general, the nitrate
concentrations in seepage water in the Rax area are low, also in comparison with the karstic research
area in the Northern Calcareous Alps of Tyrol at the Mhleggerkpfl (Smidt 2001, Feichtinger et al.
2002), what also may be due to lower atmospheric N inputs in the Rax area. The pH value and the
electric conductivity of the seepage water are generally increasing with soil depth. The highest
concentrations of nitrate and ammonium in precipitation water (crown throughfall) have been found
beneath the crowns of spruce (in F1 and B1). The tendency of spruce to filter higher amounts of
pollutants is also highlighted in other studies (Rothe et al. 1998, v. Wilpert et al. 2000, Adamson et al.
1993, Robertson et al. 2000).

Conclusions
The installation of a comparable set of measuring instruments on five experimental plots in the water
protection zone of the City of Vienna allowed to monitor some hydrological processes within these
karstic headwaters. The five experimental plots represent characteristic vegetation types within the
water protection area. The setting of the instrumentation is suited for the operation of the sensors
during summer season as well as during winter season. The first results of this project show
hydrological differences between the monitored vegetation types, which can be used in order to
elaborate more specific forest- and land use management concepts for water protection purposes.

Acknowledgements
The project has been funded by the Viennese Water Works (Municipal Department 31), by the
Forestry Office and Urban Agriculture (Municipal Department 49) and by the University of
Agricultural Sciences, Vienna, Austria (Internal Research Stimulation Program). The continuation of
the monitoring processes has been funded by the Austrian Federal Ministry for Education, Science and
Culture and by the Municipal Departments 31 and 49 of the City of Vienna. The authors want to
express their acknowledgements to the representatives of these institutions.
153

References
Adamson J K, Hornung M, Kennedy V H, Norris D A, Paterson I S, Stevens P A (1993): Soil solution
Chemistry and Throughfall Under Adjacent Stands of Japanese Larch and Sitka Spruce at Three
Contrasting Locations. Britain Forestry Vol 66, No 1.
Beldring S, Gottschalk L, Seibert J, Tallaksen L M (1999): Distribution of soil moisture and
groundwater levels at patch and catchment scales. Agricultural and Forest Meteorology 98-99, 305324.
Brner T, Johnson M G, Rygiewicz P T, Tingey D T, Jarrell G D (1996): A two-probe method for
measuring water content of forest floor litter layers using time domain reflectometry. Soil
Technology, 9, 199-207.
Delta-T-Devices Ltd (1999): Theta Probe Soil Moisture Sensor, User Manual. Burwell, Cambridge,
England.
Dirnbck T, Grabherr G (2000): GIS Assessment of Vegetation and Hydrological Change in a High
Mountain Catchment of the Northern Limestone Alps. Mount. Research and Development Vol 20,
No 2, 172-179.
Durocher M G (1990): Monitoring spatial variability of forest interception. Hydrological Processes Vol
4, 215-229.
Feichtinger F, Smidt S, Klaghofer E (2002): Water and Nitrate Fluxes at a Forest Site in the North
Tyrolean Limestone Alps. Environmental Science and Pollution Research, Special Issue 2, 31-36.
Flemming G (1993): Grundstzliche Probleme bei der Bercksichtigung des Klimas in der forstlichen
Standortslehre und Forstkologie. Forstw. Cbl. 112, 370-375, Verlag Paul Parey, Hamburg und
Berlin.
Gautam M R, Watanabe K, Saegsa H (2000): Runoff analysis in humid forest catchment with artificial
neural network. Journal of Hydrology 235, 117-136.
Hager H, Holzmann H (1997): Hydrologische Funktionen ausgewhlter naturnaher Waldkosysteme in
einem alpinen Flusseinzugsgebiet. Projektendbericht. Hydrologie sterreichs, sterr. Akad. der
Wissenschaften.
Hanson C L, Osborn H B, Woolhiser D A (1989): Daily precipitation Simulation Model for
Mountainous Areas. Transactions of the ASAE, St. Joseph, USA.
Herwitz S R & Slye R E (1995): Three-dimensional modelling of canopy tree interception of winddriven rainfall. Journal of Hydrology, 168, 205-226.
Hunzinger H (1997): Hydrology of montane forests in the Sierra de San Javier, Tucuman, Argentina.
Mountain Research and Development, 17, 299-308.
Kck R, Weidinger H, Mrkvicka A (2002):Berichte zur Standorstkartierung der Quellenschutzwlder
der Stadt Wien, Teilgebiet I. Wiener Hochquellenwasserleitung. Forstamt und
Landwirtschaftsbetriebe der Stadt Wien, Wien.
Likens G E, Bormann F H (1995): Biogeochemistry of a forested ecosystem, 2nd edition, Springer
Verlag.
Lopes V L (1996): On the effect of uncertainty in spatial distribution of rainfall on catchment
modelling. Catena 28, 107-119.

154

Martin C W, Hornbeck J W, Likens G E, Buso D C (2000): Impacts of intensive harvesting on

hydrology and nutrient dynamics of northern hardwood forests. Can. J. Fish. Aquat. Sci. 57 (Suppl.
2), 19-29.
Price, A G, Dunham K, Carleton T, Band L (1997): Variability of water fluxes through the black spruce
(Picea mariana) canopy and feather moss (Pleurozium schreberi) carpet in the boreal forest of
Northern Manitoba. Journal of Hydrology 196, 310-323.
Reynolds B, Stevens P A, Adamson J K, Hughes S, Roberts J D (1992): Effects of clearfelling on
stream and soil water aluminium chemistry in three UK forests. Environmental Pollution 77, 157165.
Robertson S M C, Hornung M, Kennedy V H (2000): Water chemistry of throughfall and soil water
under four tree species at Gaisburn, northwest England, before and after felling. Forest Ecology
and Management, 129, 101-117.
Robinson D A, Gardner C M K, Cooper, J D (1999): Measurement of relative permittivity in sandy
soils using TDR, capacitance and theta probes: comparison, including the effects of bulk soil
electrical conductivity. Journal of Hydrology, 223, 198-211.
Rothe A, Klling C, Moritz K (1998): Der aktuelle Kenntnissstand: Waldbewirtschaftung und
Grundwasserschutz. AFZ/Der Wald 6, 291-295.
Shanley J B, Chalmers A (1999): The effect of frozen soil on snowmelt runoff at Sleepers River,
Vermont. Hydrological Processes 13, 1843-1857.
Smidt S (2001): Luft- Depositions- und Bodenwasseranalysen am Mhleggerkpfl / Nordtiroler
Kalkalpen. FBVA-Berichte 119, 61-72.
v. Wilpert K, Nell U, Lukes M, Schack-Kirchner H (1998): Genauigkeit von Bodenfeuchtemessungen
mit Time Domain-Reflektometrie in heterogenen Waldbden. Z. Pflanzenernhr. Bodenk., 161,
179-185, Wiley Vch Verlag, Weinheim.
v. Wilpert K, Zirlewagen D, Kohler M (2000): To what extend can silviculture enhance sustainability of
forest sites under the immission regime in Central Europe. Water, Air and Soil Pollution 122, 105120.
Whitford W G, Anderson J, Rice P M (1997): Stemflow contribution to the fertile island effect in
creosotebush, Larrea tridentate. Journal of Arid Environments, 35, 451-457.
Zukrigl K (1973): Montane und subalpine Waldgesellschaften am Alpenostrand. Mitteilungen der
Forstlichen Bundesversuchsanstalt No 101, Wien.

155

FACHTAGUNGEN:
Mrz 2002: Symposium of the International Association of Vegetation Science, Porto
Alegre/Brasilien.
Dirnbck, T. Dullinger, S. & Grabherr, G.: Equilibrium versus Non-Equilibrium in Alpine Vegetation
Dullinger, S., Dirnbck, T. & Grabherr, G.: Krummholz-dynamics at the treeline due to land use and
climate change

Juni 2002: sterreichische Botanikertagung, Bundesanstalt fr Landwirtschaft,

Gumpenstein
Dirnbck, T. Dullinger, S. & Grabherr, G: GIS-gesttzte Prognosemodelle alpiner Pflanzenarten unter
Nutzungs- und Klimanderung

September 2002: Jahrestagung der Gesellschaft fr kologie (Cottbus/Deutschland)

Dirnbck, T., Dullinger, S. & Grabherr, G.: Ecological consequences of Pinus mugo invasion in
subalpine pastures
Dullinger, S., Dirnbck, T. & Grabherr, G.: Effects of summer farming on subalpine plant species
diversity

Juni 2003: Symposium of the International Association of Vegetation Science,

Neapel/Italien.
Dullinger, S., Dirnbck, T. & Grabherr, G.: Modelling climate change driven treeline dynamics in a
karstic drinking water catchment

September 2002: Workshop "Ecological and Economic Benefits of Mountain Forests"

Kck, R., Dirnbck, T., Dullinger, S., Hrtel, E., Hochbichler, E., Holtermann, C., Grabherr, G.,
Kuschnig, G., Weidinger, H.: Assessing potential vegetation and hydrological changes in karstic
catchments
Kck, R., Hrtel, E., Holtermann, C., Hochbichler, E., Hager, H. (2002): Soil Moisture Dynamics
Related to Vegetation Cover in the Subalpine Zone of the Northeastern Calcareous Alps in Austria.
Results of Case Studies in the Rax Area .

September 2002: International Conference: Interdisciplinary approaches in small

catchment hydrology: monitoring and research. (ERB 2002), Demnovsk dolina,
Slovakia
Kck, R., Hrtel, E., Hochbichler, E., Hager, H.& Schnthaler, K.E.:
Monitoring hydrological processes in montane and subalpine karst regions: comparison between
different types of vegetation. Experimental design, techniques and first results

156