the sensitivity of central european mountain forests to scenarios

Lexer, Hönninger, Scheifinger, Matulla, Groll and Kromp-Kolb The Sensitivity of Central European Mountain Forests ...

113

The Sensitivity of Central EuropeanMountain Forests to Scenarios of ClimaticChange: Methodological Frame fora Large-scale Risk Assessment

Manfred J. Lexer, Karl Hönninger, Helfried Scheifinger, Christoph Matulla,Nikolaus Groll and Helga Kromp-Kolb

Lexer, M.J., Hönninger, K., Scheifinger, H., Matulla, C., Groll, N. & Kromp-Kolb, H. 2000. Thesensitivity of central European mountain forests to scenarios of climatic change: methodo-logical frame for a large-scale risk assessment. Silva Fennica 34(2): 113–129.

The methodological framework of a large-scale risk assessment for Austrian forestsunder scenarios of climatic change is presented. A recently developed 3D-patch model isinitialized with ground-true soil and vegetation data from sample plots of the AustrianForest Inventory (AFI). Temperature and precipitation data of the current climate areinterpolated from a network of more than 600 weather stations to the sample plots of theAFI. Vegetation development is simulated under current climate (“control run”) andunder climate change scenarios starting from today's forest composition and structure.Similarity of species composition and accumulated biomass between these two runs atvarious points in time were used as assessment criteria. An additive preference functionwhich is based on Saaty's AHP is employed to synthesize these criteria to an overallindex of the adaptation potential of current forests to a changing climate. The presentedmethodology is demonstrated for a small sample from the Austrian Forest Inventory.The forest model successfully simulated equilibrium species composition under currentclimatic conditions spatially explicit in a heterogenous landscape based on ground-truedata. At none of the simulated sites an abrupt forest dieback did occur due to climatechange impacts. However, substantial changes occured with regard to species composi-tion of the potential natural vegetation (PNV).

Keywords climate change, alpine forests, risk assessment, patch model, multi-attributedecision making, potential natural vegetation.Authors' addresses Lexer & Hönninger: Institute of Silviculture, University of Agricul-tural Sciences, Peter-Jordanstrasse 70, A-1190 Vienna, Austria Fax 0043-1-47654 4092,E-mail [email protected] Internet http://www.boku.ac.at/waldbau/Groll, Kromp-Kolb, Matulla & Scheifinger: Institute of Meteorology and Physics,University of Agricultural Sciences, Türkenschanzstrasse 18, A-1180 Vienna, AustriaReceived 1 October 1999 Accepted 10 May 2000

Silva Fennica 34(2) research articles


114

1 Introduction

Central European forests comprise a wide rangeof ecological site conditions from high altitudeconiferous mountain forests to low elevation for-ests mainly dominated by broadleaved species.In particular mountain forests have to serve amultitude of functions and play an importantrole in maintaining alpine landscapes. Beyondproducing timber these forests protect settlementsand infrastructure from natural hazards such asavalanches and rockfall, they prevent soil ero-sion and play a key role in maintaining biodiver-sity. The importance of functional sustainableforests has been underlined by a number of inter-national cooperations and resolutions (CIPRA1997, Ministerial Conference for the Protectionof European Forests 1998). Discussions on alikely global climate change gave rise to con-cerns about possible impacts on forest ecosys-tems (Houghton et al. 1992).

It is important to recognize that to prevent animmediate loss of forest functioning under chang-ing climatic conditions at least the physiologicalrequirements of trees for the growth segment oftheir life cycle have to be met. However, tosupport decision making in forest resource man-agement, information on the synecological be-

haviour of tree species (i.e. the competitive rela-tionships between species) under a changing cli-mate is needed. Short-term experimental studieson physiological processes of small individualsaplings are not directly applicable to supportdecision making on temporal and spatial scalesrelevant for forest management. Simplistic con-ceptual models of general altitudinal shifts ofvegetation belts with increasing temperatures (e.g.Peters and Darling 1985, Starlinger 1998) arenot sufficient to support site-specific silvicultur-al planning. A promising approach to integrateavailable knowledge on vegetation-site interac-tions and to investigate the possible transientresponses of forest ecosystems to changing cli-matic conditions are individual-based models offorest dynamics models (Shugart et al. 1992). Avariety of forest models which are considered tobe useful for forest management decision sup-port under changing environmental conditionshave been developed during the last years. Amongexisting modelling approaches gap models (patchmodels sensu Shugart 1998) are considered to beparticularly useful for simulating forest develop-ment under changing climatic conditions. ForEuropean forests such models have been con-

Abbreviations

SMAP = Short- to midterm adaptation potential to a changing climateLAP = Longterm adaptation potential to a changing climateACSMAP(i) = Assessment criterion (i) with regard to SMAPACLAP(j) = Assessment criterion (j) with regard to LAPPACSMAP(i) = Preference function characterizing the contribution of assessment criterion (i) to SMAPPACLAP(j) = Preference function characterizing the contribution of assessment criterion (j) to LAPSWCm = Soil water content in month (m) [mm]AETm = Actual evapotranspiration in month (m) [mm]ROm = Runoff in month (m) [mm]SMm = Snow melt in month (m) [mm]WHC = Water holding capacity of the mineral soil [mm]SMI = Soil moisture index [0–1]Rsm = Species response to soil moisture availabilityRn = Species response to nutrient status of the siteRal = Tree response to available lightRgdd = Species response to growing degree days (heat sum > 5.5 °C)


115

structed by Kienast (1987), Kellomäki et al.(1992), Prentice et al. (1993), Kräuchi (1994)and Bugmann (1994).

Most model applications to assess the poten-tial impacts of climatic change have been con-fined to a very limited number of scattered siteswith more or less synthetic site characteristics(e.g. Kienast and Kuhn 1989, Bugmann 1994).Just recently Lindner et al. (1997) employed twogap models in a regional climate change impactassessment for the State of Brandenburg (Ger-many) where the site and climate data necessaryto initialize the model runs were lumped at aspatial resolution of 10 × 10 km2. However, thislumping procedure leads to synthetic “average”site conditions within each gridcell, thus blur-ring the effect of site characteristics on forestresponse to a changing climate. In all large-scalemodel applications so far, the simulations werestarted from bare ground (i.e. “clear cut condi-tions”) aiming at the steady state species compo-sition as the only assessment endpoint (e.g. Lind-ner et al. 1997). Despite the obvious relevance oftodays forest composition for future forest de-velopment very few model-based climate changeimpact studies considered today's forest compo-sition and structure. One of the rare exemptionswas provided by Lindner (1998) who simulatedapproximately 200 ha of managed forests undera climate change scenario. In a risk assessmentKienast et al. (1996) used the site data availablefor sample plots of the national forest inventoryof Switzerland to feed a static climate-sensitiveequilibrium model of the potential natural vege-tation. Recently Talkkari and Hypen (1996) pre-sented an outline of the application of gap mod-els based on spatially explicit data for borealconditions.

So far it can be summarized that most modelapplications to assess the potential impacts ofclimate change on alpine forests suffered fromthe following shortcomings: (i) simulation studiesfor individual “representative” more or less syn-thetic site conditions miss the spatial dimensionand coverage, (ii) in essentially all model appli-cations simulation runs were started from bareground, thus neglecting the composition andstructure of current forests, (iii) risk assessmentstudies with static vegetation-site equilibriummodels miss the individualistic nature in the

formation of vegetation composition. Further-more, static equilibrium models are not capableof indicating the transient response of currentforests to a changing climate.

To overcome these limitations and to providean estimate for the sensitivity of current Austri-an alpine forests to scenarios of climatic changea modified spatially explicit 3D-patch-model isemployed to simulate vegetation development atspatially explicit defined sites. To represent thevarious combinations of current vegetation andsite characteristics and to capture the effects ofchanging climatic conditions there on, all modelruns were initialized spatially explicit withground-true soil and vegetation data of the Aus-trian Forest Inventory (AFI).

The objectives of this contribution are three-fold: Firstly, we give a comprehensive outline ofthe employed risk assessment methodology ingeneral. Secondly, the patch model PICUS v1.2which is used to forecast forest vegetation devel-opment is briefly presented, and thirdly, the datarequired to initialize and to drive the model runsare described. Finally some demonstration re-sults are presented.

2 Methods

2.1 Risk Assessment Procedures

By definition risk is the chance that an unfavour-able event which is related to some kind of dam-age will occur (Holloway 1979). In case of for-ests the functioning of forest ecosystems underchanging climatic conditions, i.e. the ability offorests to sustainably secure societal needs, is atrisk. Regarding climate change impacts the worstcase scenario would be a rapid forest dieback.However, impact mechansisms may be more sub-tle. For instance, in low-elevation forests at thefoothills of the Alps secondary coniferous for-ests mainly consisting of Norway spruce (Piceaabies (L.) Karst.) may suffer from increaseddrought which in turn may increase susceptibili-ty to attacks by insects and fungi. This increasedvulnerability of forests may result in an imbal-ance in harvesting operations, thus increasingmanagement costs and causing perturbations oftimber markets by imbalancing demand and sup-


116

ply. Due to the involved complex interactionsessentially it is impossible to quantify the condi-tional probability of a particular event to occur.Rather qualitative risk assessments with aggre-gated more general risk descriptors are appropri-ate means (Hunsaker et al. 1990).

A crucial step in risk assessments is the defini-tion of appropriate assessment endpoints (Suter1993). According to Suter (1993) and Hunsakeret al. (1990) risk assessment endpoints have tomeet the requirements of societal and biologicalrelevance, unambiguous operational definition,accessibility to prediction and measurement aswell as susceptibility to the hazardous agent.Ecosystem-level characteristics such as speciescomposition or accumulated biomass are consid-ered useful parameters in evaluating the func-tioning of forests (Grottenthaler and Laatsch1973, Gundermann 1974, Ott and Schönbächler1986, Gordon 1994).

Beyond defining the entity which is at risk(i.e. a forest stand), ecological risk assessmentprocedures have to devise methods to define thecurrent state and quantify the expected changesof the assessment entities (Suter 1993). Modelsof forest vegetation development which are re-sponsive to the disturbance agent (i.e. changingclimatic conditions) are appropriate means toproject ecosystem changes. However, it must berecognized that the fate of a particular foreststand under a changing climate will be hardlyever directly predictable for several reasons:

(a) European forests have been and are subject tointensive forest management. There is no reasonto assume that this will change. Thus, knowledgeon future silvicultural treatments would be an in-evitable prerequisite for such a task. Due to theheterogenous ownership structure in Austria it isneither realistic to assume a single treatment pro-gramm for particular stand categories nor to dif-ferentiate between all possible stand treatmentparadigms.

(b) Stochastic events such as windthrow strongly de-termine stand development. Such natural distur-bances are stochastic events and quite frequentbut usually are not included in forest models.

(c) Finally it must be recognized that timber marketsare among the main factors driving the selectionof stand treatment programms and tree species.

Though there are attempts to link socio-economictimber market models and static vegetation mod-els (Sohngen et al. 2000), these attempts largelymiss to take into account the feedbacks betweentransient forest response to a changing climate(i.e. supply of timber) and timber markets.

Another problem is that the adaptation potentialof forests can not be measured directly. There-fore it was concluded that a set of indicators ofthe adaptation potential of current forests tochanging climatic conditions extended over short-to longterm temporal scales can provide a rea-sonable basis for the evaluation of climate changeimpacts. One approach very common in centralEuropean forestry is to utilize the divergence ofthe actual forest composition and the speciescomposition of the potential natural vegetation(PNV sensu Tüxen 1956) as a proxy for thesuitability of forest vegetation. According toTüxen (1956) PNV is the equilibrium speciescomposition of the expected forest communityat a particular site considering today's site condi-tions. No direct anthropogenic influences are as-sumed. The rationale for the use of PNV in re-source planning is, that PNV characterizes theecological potential of forest sites and thus indi-cates which species mixture is well adapted tothe prevailing site conditions (e.g. Kienast et al.1996). Moreover, minimizing the divergence be-tween PNV and the actual forest composition isexpected to minimize the input of effort andenergy (i.e. silvicultural treatments) required tosustain ecosystem function (Leibundgut 1981,Mayer 1984). The latter point derives from thefact, that PNV integrates information on the com-petitive relationships between tree species.

The newly developed patch model PICUS v1.2(Lexer and Hönninger 1998a, 2000a) was em-ployed to simulate vegetation development un-der current and future climate starting from cur-rent forest composition until the site-specific equi-librium species composition (which is equiva-lent to PNV) was reached. To allow a futureequilibrium species composition to establish thehypothetical future climate of the year 2050 ac-cording to a climate change scenario was keptconstant with temperatures and precipitation var-ying around their long-term means. Current for-est composition observed at a particular site can


117

then be matched with the species composition ofthis simulated future PNV. Beyond this long-term proxy of tree species suitability biomassaccumulation patterns as well as possible abruptchanges in species composition are evaluated ona decenial scale for the period 2000 to 2050 tocapture the short- to midterm sensitivity of cur-rent forests to a changing climate directly. Fig. 1shows the hierarchy of assessment criteria usedto synthesize an overall index for the adaptationpotential of current forests to scenarios of cli-matic change.

The short- to midterm sensitivity (SMAP) offorests to changing climatic conditions was quan-tified as the difference in species compositionand accumulated biomass between the simulatedvegetation development under current and chang-ing climate respectively at 10-year-intervalls forthe period from 2000 to 2050 (ACSMAP(1–5)).The forest simulator PICUS v1.2 is run in thePNV-mode, i.e. it is assumed that seeds of allspecies are potentially available in the simulatedstand, no forest management interventions areconsidered. As an indicator for changes in eco-logical site potential the divergence of PNV un-der current and the assumed future climate iscalculated (ACLAP(1)). Finally the current spe-cies set is matched with the equilibrium speciescomposition under the altered climatic condi-tions of the climate change scenario with regardto the relative abundance of species (ACLAP(2)).The latter indicator accounts for the longtermadaptation potential of current forests.

The assessment criteria ACSMAP(i) and

ACLAP(j) were quantified by the percentage sim-ilarity PS (Prentice and Helmisaari 1991) (1).

PS

a b

a b

i ii

n

i ii

n= −−

+( )=

=

∑

∑1 1

1

(1)

ai = proportion (absolute or relative) of species (i)in set a

bi = proportion (absolute or relative) of species (i)in set b

This measure not only evaluates differences inspecies proportions but also integrates the abso-lute amount of biomass a species is able to accu-mulate at a site. PS can take values between 0(totally different species composition) and 1(identical species composition).

2.1.1 A Multiple-attribute Preference Modelto Evaluate the Impact of ClimaticChange

To evaluate the overall adaptability of a givenstand (i.e. the severity of expected climate changeimpacts) an approach that borrows from multi-ple-attribute utility theory (MAUT) was adopt-ed. This method requires the mathematical char-acterization of preferences over a set of attributes(Goicoechea et al. 1982). According to Fig. 1 theoverall utility with regard to the adaptability of a

ACSMAP(1)2010

PS(c/cc)

ACSMAP(2)2020

PS(c/cc)

ACSMAP(3)2030

PS(c/cc)

ACSMAP(4)2040

PS(c/cc)

ACSMAP(5)2050

PS(c/cc)

SMAP

ACLAP(1)–

PS(PNVc/PNVcc)

ACLAP(2)–

PS(c/PNVcc)

LAP

AP

Fig. 1. Characterizing the adaptation potential (AP) of current forests to a changing climate by hierarchicaldecomposition into measurable assessment criteria. – SMAP = short-/midterm adaptation potential, LAP =longterm adaptation potential; PS(c/cc) = percentage similarity (PS) of simulated vegetation under current(c) and changing climate (cc); PS(PNVc/PNVcc) = PS of simulated PNV under current (c) and futureclimate (cc); PS(c/PNVcc) = PS of current vegetation and simulated future PNV; ACSMAP(i) = assessmentcriterion(i) for SMAP; ACLAP(j) = assessment criterion (j) for LAP.


118

forest to a climate change scenario is composedof partial utilities with regard to the short- tomidterm (USMAP) and the longterm adaptationpotential (ULAP) respectively (2)

U a U b USMAP LAP= ⋅ + ⋅ (2)

under the constraint that (a + b = 1). The partialutilities are calculated from preference functionsPACSMAP(i) (3a–b) and PACLAP(j) respectively (4a–b).

U

a P a P

a P a P

a P

SMAP

ACSMAP ACSMAP

ACSMAP ACSMAP

ACSMAP

=⋅ + ⋅ +⋅ + ⋅ +⋅

1 1 2 2

3 3 4 4

5 5

(3a)

aii=∑ =

1

5

1 (3b)

and

U b P b PLAP ACLAP ACLAP= ⋅ + ⋅1 1 2 2 (4a)

bjj =∑ =

1

2

1 (4b)

The preference functions calculate the “value”of each realization of an assessment criterionwith respect to its contribution to either SMAP orLAP. It is important to note, that it is not veryrealistic to assume a linear relationship betweenthe similarity of vegetation under current cli-mate and a climate change scenario and its mean-ing with regard to forest adaptability. The pref-erence functions in (3a) and (4a) were estimatedby the eigenvalue method as applied in the ana-lytic hierarchy process (Saaty 1977, 1996). Es-sentially the approach of Saaty requires a prioriknowledge of all elements to be compared. How-ever, the assessment criteria can take any valuebetween 0 and 1. Thus, the original method ofSaaty had to be modified. Instead of comparingalready realized outcomes for a criterion we par-titioned the entire possible range for a criterion(0–1) in eight categories of equal width and usedthe midpoints of these categories as elements inthe comparison matrix. The resulting prioritiesfrom the matrix were scaled to 1. Thus, we ar-

rived at a continuous normalized preference func-tion which subsequently could be employed tocalculate preference values for any realization ofACSMAP(i) and ACLAP(j), respectively (compareFig. 2).

2.2 The Forest Model

In this section a brief outline of PICUS v1.2 isgiven. For details see Lexer and Hönninger(1998a, 2000a). Conceptionally, PICUS v1.2 isbased on the following assumptions: (a) a 3-dimensional spatial resolution of 10 × 10 × 5 mis sufficient to track spatial effects such as influ-ence of gap size and orientation and allows forthe application of distance independent algo-rithms to model inter-tree competition, (b) theconsideration of light, of the thermal regime rep-resented by an effective temperature sum, of soilmoisture availability and of site nutrient statusprovides a reliable description of a species' fun-damental niche. In addition winter minimum tem-peratures are evaluated twofold with respect toregeneration processes: regarding frost damageas well as chilling requirements to induce dor-mancy and subsequently break of flower buds(Sykes et al. 1996). (c) Optimum unconstrainedtree growth can be derived from data on opengrown trees (OGTs).

The 10 × 10 × 5 m structural base elementscontain all information on the distribution of treebiomass in space. From this point of view PICUSis a descendant of the ZELIG-model (Urban1990). A similar approach can also be found inHuth et al. (1996). Thus, unlike in conventionalgap models which in fact can be considered as

Fig. 2. Continuous preference function for the valua-tion of ACSMAP(i) and ACLAP(j).

Pref

eren

ce

1.0

0.8

0.6

0.4

0.2

00 0.2 0.4 0.6 0.8 1.0

Assessment criterion[ACSMAP(i)/ACLAP(j)]


119

“point”- models, the simulated forest changesover time as an interactive unit rather than as aseries of independent plots. Because PICUS in-cludes a fairly detailed model of the above- andwithin-canopy light regime interactions amongneighbouring patches as well as the effect ofslope and orientation on incoming radiation areconsidered. Moreover, it allows to differentiatethe light conditions at different sites with respectto shielding effects of surrounding topography.The range of spatial interactions between patchesdepends on the characteristics of the vegetationon the simulated patches (i.e. tree height, crownlength), site characteristics (slope, orientation,latitude) and season (i.e. solar altitude, sun angleand direction). The spatial core structure of themodel is presented schematically in Fig. 3.

As in most gap models actual tree growth inPICUS is derived from a growth potential whichis modified by environmental constraints. Opti-mum diameter increment is derived from thecarbon budget approach given by Moore (1989).Beyond diameter at breast height (1.3 m) treedimensions are defined by tree height, height tothe live crown and leaf weight and leaf area,respectively. Tree height is updated by means ofa species specific second order polynomial func-tion. Leaf weight and leaf area are calculatedfrom dbh for species groups adopting the recent-ly refined equations from Bugmann (1994). Leafweight is distributed equally along the bole. De-termination of the height to the live crown isbased upon species-specific light requirements(light compensation point) and the radiation re-gime a tree experiences.

Tree Response to Environmental Factors

An index of available light for an individual treeis calculated from the available light a tree crownin the simulated stand actually experiences. Theresponse to available light for 9 species catego-ries is modelled by interpolating between theresponse functions for category 1 “very shadetolerant” and category 9 “very shade intolerant”.

The representation of the thermal environmentbasically follows other recent gap models wherea heat sum above the threshold for positive netphotosynthesis (GDD) and winter minimum tem-peratures (WT) are used to represent the temper-ature regime at a site. GDD are meant to charac-terize general thermal conditions for photosyn-thesis (Larcher 1995), WT may seriously limitthe regeneration of tree species which are vul-nerable to frost occurrence (Woodward 1987,Ellenberg 1996). The coldest month of a year istaken as an estimate for WT (Bugmann, 1994)and is also used to evaluate, whether the chillingrequirements for species requiring dormancy aresatisfied (Sykes et al. 1996). It is important tonote that in PICUS tree growth is not restrictedat super-optimal temperatures. Based on the con-sideration that tree growth at a species' southernrange limit is not necessarily constrained by tem-perature per se (compare Bugmann and Solo-mon 2000) we modified the heavily critizicedparabola (e.g. Schenk, 1996) and derived newtemperature response functions from the com-bined network of forest inventory data, soil andmeteorological data (Lexer and Hönninger1998b). In Fig. 4 these newly formulated tem-perature response functions are presented for se-lected species.

Canopylayer

Speciespotential

Leaf area

Environment

Individual

Canopy cell

Patch

Stand

Fig. 3. Spatial structure of PICUS v1.2.

0

0.2

0.4

0.6

0.8

1.0

0 50 100 150 200GDD

Gro

wth

res

pons

e

Picea abies

Larix deciduaFagus sylvatica

Quercus petraea

Fig. 4. Modified temperature response functions forselected species (Lexer and Hönninger, 2000).GDD = Heat sum above the threshold for positivenet photosynthesis.


120

To provide a more realistic representation ofsoil water dynamics we modified the one layerbucket model which is conventionally used ingap models (e.g. Pastor and Post 1985, Kienast1987, Bugmann 1994). The general logic to up-date the state variable soil water content (SWC)each month is given by (5)

SWC

SWC P SM AET ROm

m m m m m

+ =+ + − −1

(5)

where input of water to the soil is given byprecipitation (P) and snowmelt (SM). Actual eva-potranspiration (AET) and runoff (RO) representoutput of water from the soil. RO occures, ifSWC exceeds the water holding capacity of thesoil (WHC). SM is calculated as a function oftemperature according to Dunne and Leopold(1978). AETm represents the monthly proportionof the atmospheric evaporative demand (PETm)that can be satisfied by available soil water. Anestimate for PETm is computed applying the for-mulation of Thornthwaite and Mather (1957).Given unlimited water supply AET equals PET.When the soil moisture content decreases, PETis modulated depending on soil texture and theamount of water in the soil (6).

AET PET fSWC

WHCm m sm= ⋅

(6)

Formulations fs for different soil texture typeswere taken from Dunne and Leopold (1978). Toprovide a parameterized representation of thesimultaneous processes of water infiltrating intothe soil and water being depleted from the soilinput and output from the bucket were calculat-ed in 3 time steps per month where the sequenceof water input and output varies stochastically.The ratio of evaporative demand and supply in-tegrated over the growing season is considered areliable proxy of drought stress experienced byvegetation (7).

SMI

AET

PET

bgs

egs

bgs

egs= −∑

∑1 (7)

bgs = begin of growing season (temperature ex-ceeds 5.5 °C)

egs = end of growing season (temperatures below5.5 °C)

Similar to the derivation of the temperature re-sponse the data of the Austrian Forest Inventorywere utilized to derive soil moisture responsefunctions. PICUS v1.2 requires monthly temper-ature and precipitation data. The model drivingclimate can be provided as either time series dataor can be sampled stochastically from empiricaldistributions fitted to longterm time series ofclimate data. A normal distribution is used forgenerating monthly temperatures, whereas a 2-parameter gamma distribution is used for precipi-tation.

In contrast to other gap models PICUS v1.2doesn't rely on the nitrogen supply function aspresented by Pastor and Post (1985). Instead thewater holding capacity at a site (WHC), the pH-value (PH) as well as the C/N-ratio (CN) of thetop soil are employed to characterize the nutrientstatus of a site. In a fuzzy logic control unit theseindicators of site nutrient status are linked to treegrowth performance by means of a rule base.Rules were constructed for each of four responsecategories (tolerant, intolerant, PH-sensitive, CN-sensitive). By linear interpolation between thesefour main response categories a total of 9 speciesresponse groups were defined. Based on ecolog-ical literature (e.g. Ellenberg, 1996) each specieswas assigned to one of these response catego-ries. A detailed description is given in Lexer etal. (2000).

In modelling the combined effect of environ-mental factors on tree growth various approach-es have been proposed. Following Liebig's lawof the minimum Botkin et al. (1972) and Kienast(1987) employed the minimum-operator to de-rive the aggregate environmental response froman array of monocausal responses. In Botkin(1993) the multiplicative combination of all en-vironmental factors was used. Bugmann (1994)proposed a geometric mean which implicitelyconsiders compensation between environmentalfactors. In PICUS v1.2 compensation as well asintensification was considered explicitely. Thebelowground factors soil moisture (Rsm) and nu-trient supply (Rn) were combined in Rsoil. Thehypothesized interrelationship of Rsm and Rn wasmade explicit by (8)–(10).


121

R R f R Rsoil biom soil sm n= ⋅ ( ), ; .0 3 (8)

effectR Rsm n=

⋅ ( ) −( )0 3 0 5

0 5

. max , .

.(9)

fsoil

R R R R effect effect

R R R R R effect else

=

+ −( ) ⋅[ ] >+ − ⋅( ) ⋅[ ]

min , ; .

min , ;max min max min

min min min min max

0 0

(10)Rmin = min(Rsm, Rn)Rmax = max(Rsm, Rn)

Thus, if both the effects of soil moisture andnutrient supply are below 0.5, the effect of themost limiting factor is intensified. If one envi-ronmental factor satisfies species requirementswith Rmax 0.5, the effect of the limiting factorcould partly be compensated. The combined ef-fect of Rsm and Rn was further intensified by

R

biom

which decreases with increasing accumulated bio-mass at the simulated patches as a function ofthe maximum possible biomass. Based on theapproach of Fischlin et al. (1995) site-specificestimates of maximum biomass were calculatedfrom temperature, an index of soil water avail-ability and an index for site nutrient status. In asimilar approach the effects of temperature (Rgdd)and light (Ral) were combined. The final overallresponse to environmental factors was calculat-ed from the combined below and abovegroundeffects again utilizing the compensation algo-rithm from eqs. 8–10.

Similar to other patch models, tree mortality ismodelled as a stochastic process. The intrinsicrisk of death is increased under unfavourableconditions when a tree fails to realize a specifiedminimum diameter increment for a number ofsuccessive years. This approach is conceptuallybased on a hierarchical pattern of carbon alloca-tion where diameter increment is of low priorityand thus an indicator of reduced tree vigor (Koz-lowski et al. 1991). For Picea abies bark beetlemortality is considered explicitely by couplingthe patch model with a 2-stage-stand risk model(Lexer and Hönninger 1998a).

New trees are generated by a recruitment sub-model and are explicitely modelled beyond adiameter treshold of 1.0 cm ± 0.2 cm. In contrastto other patch models PICUS considers seed pro-

duction by overstorey trees and seed dispersalexplicitely. The seed production of each adulttree is modelled as a function of tree size, thespecies' seed production characteristics and cur-rent light consumption of the parent tree. Seeddispersal of each seed producing tree is mod-elled as a cone-shaped density function aroundthe center of each parent tree's patch where thecone is defined by tree height and maximumdispersal range. For selected species zoochorousseed dispersal is considered as well. PICUS v1.2was already successfully evaluated against spa-tially explicit expert reconstructions of PNValong ecological gradients in the Eastern Alps(Lexer and Hönninger 2000a, Lexer and Hön-ninger 2000b).

2.3 Model Input and Initialization

2.3.1 Site Data

To utilize available data and to further enhancethe information content of existing data basesthe study at hand was based on the sampling gridof the Austrian Forest Inventory (AFI). AFI sam-ples site and vegetation data on a systematic3.89 km grid of permanent plots. At each gridlocation a cluster of four plots is located at thevertices of a 200 m square. For this risk assess-ment a subsample of all AFI-plots was takenwhich is considered to represent the variety ofsite-vegetation combinations of Austria's forests.For each plot included in the risk assessmentstudy an array of site and soil parameters wasrequired to initialize the model runs (Table 1).Geographical location, aspect as well as slopeare standard site descriptors recorded by the Aus-trian Forest Inventory (AFI). The surroundingsuperelevated topography in the four main direc-tions (horizon) was retrieved from a digital ele-vation model of 250 m spatial resolution imple-mented in GIS ARC/Info. Lack of quantitativesoil data is a weak point of many large-scaleforest inventories. To circumvent this problemLexer and Hönninger (1998c) and Lexer et al.(1999) developed a routine for the estimation ofsoil parameters for sample plots of the AFI basedon Bayesian probability theory. This approachwas employed to calculate estimates for the soil


122

parameters WHC, PH and CN for all sites in-cluded in the impact study.

2.3.2 Vegetation Data

At each sample plot of the AFI vegetation data isrecorded. Using variable radius plot sampling(Bitterlich 1948) with a basal area factor of 4 m2/ha trees larger than 10.4 cm in diameter are re-corded at least by dbh and species. All trees withdiameters between 5 and 10.4 cm are recorded ona fixed area plot located at the plot center with aradius of 2.6 m. No trees smaller than 5 cm diam-eter are sampled. From this information a speciesspecific diameter distribution was generated foreach selected sample plot. The spatial resolutionof PICUS would allow for the consideration ofhorizontal species mixture patterns. However,AFI does not provide sufficient information onthis stand characteristic. Thus, trees larger than 5cm in dbh were assigned randomly to the patchesof the simulated forest starting from the largestdiameter class. Tree height and leaf weight werecalculated by the model-internal relationships.The initial height to the live crown was deter-mined by calculating the light regime in the ini-tialized stand with the radiation submodel ofPICUS. Initial leaf weight below the light com-pensation point of a species was pruned.

Established regeneration at a site requires par-

ticular consideration as it provides the potentialnext generation of trees. Unfortunately AFI pro-vides just semiquantitative information on theregeneration status at a sample plot (Instruktio-nen für die Feldarbeit der ... 1995). For AFI aminimum number of seedlings is required de-pending on the average height of the seedlings(minimum height = 10 cm) to proceed with datacollection. Thus, plots with existing sparse re-generation might occur in the data records withthe label “no regeneration”. From the attributeswhich are subsequently recorded three AFI-pa-rameters were utilized for the current study toquantify the regeneration status at a sample plot:(i) the distribution of seedlings (categories: wholearea, in groups, single), (ii) for a maximum offive species the percent cover (crown projectionarea) of the dominating height class of a species(6 categories) as well as (iii) the mixture type ofthat species (categories: pure, groups, random).From the cover percentages the number of seed-lings per hectar was calculated with (11)

a x dc= ⋅ (11)

where a is the average distance between neigh-bouring seedlings, dc is the average crown diam-eter of a species of a given height class and x is aconstant specific of a cover perentage class asgiven in Instruktionen für die Feldarbeit der ...(1995). The species specific crown diameters forgiven heights were calculated from formulae inHasenauer (1997). Once the average distance a(i)

between neighbouring individuals of a species isknown, the expected number of seedlings (n(i)/ha) was calculated from (12) assuming a trian-gular tree distribution.

n haa

ii

( )( )

/ =⋅

⋅

10000

3

46

2(12)

With this approach a pool of available seedlingsat a site was calculated and provided as input toPICUS. Recruitment of new trees in PICUS drawson this seedling pool as long as seedlings areavailable. Seedling numbers in the pool are ei-ther reduced by the recruitment process or byunfavourable environmental conditions charac-terized by the environmental drivers of PICUS(see section 2.2).

Table 1. Site and soil parameters required to initializethe simulation runs with PICUS v1.2.

Parameter Unit Characterization

Coordinates (°, ', '') Location of siteAspect (°) Eight categoriesSlope (%) Inclination of slopeHorizon (1, 2, 3, 4) (°) Angle to horizont

in directions north,east, south, west

Water holding capacity (cm) Uppermost 30 cmmineral soil

ph-value – Average of theuppermost 30 cmmineral soil

C/N-ratio – Average of theuppermost 30 cmmineral soil


123

2.3.3 Climate Data

For the risk assessment we required current cli-mate and climate change scenarios at all selectedAFI-plots. Current climate at the plots was de-rived from a network of more than 600 weatherstations. For this purpose Austria was dividedinto 8 subregions (compare Tabony 1985). Asubregion is comprised by a valley system withborders running along mountain ridges. Thus, itshould be guarantied that most of the subregionshare the same air mass with more or less thesame vertical temperature distribution. Temper-ature was supposed to show no significant hori-zontal gradients within one subregion so that asingle vertical temperature–altitude relationshipcould be deduced using data from all availableweather stations within a subregion. Tempera-ture values for each point of interest are calculat-ed with this polynomial temperature–altitude re-lationship for each individual month and subre-gion with the elevation of the point as input. In asubsequent step this temperature values are thenmodified with regard to topography (aspect,slope, horizon) utilizing the ratio of incomingshortwave radiation at the particular point andincoming radiation under the assumption of aflat horizon and zero slope. A more detailedapproach was applied to the spatial interpolationof precipitation. For each AFI- plot per individu-al month a precipitation–altitude relationship wasdeduced fitting a linear model to the precipita-tion of the nearest 20 weather stations (Lauscher1976). With this model residuals for each of the20 weather stations were calculated and a resid-ual was interpolated for the point of interest byan inverse distance weighting procedure. Withthe point-specific residual, the precipitation–al-titude relationship and the altitude of the point ofinterest a precipitation value can then be recal-culated for each AFI-plot.

For the present study it is planned to derivetransient climate change scenarios for AFI sam-ple plots from simulation runs of the global cir-culation model ECHAM4 under several scenari-os of the IPCC (Houghton et al. 1992). Since thespatial resolution of general circulation models(GCMs) is too coarse to directly derive reliableregional climate change projections statisticaldownscaling methods (e.g. Gyalistras et al. 1994)

are applied. Currently empirical orthogonal func-tions (EOF) are tested to establish statistical re-lationships between the behaviour of the mete-orological fields for temperature and humidity atthe grid points of the NCAR data set extendingfrom 1961 to 1995 and time series of monthlymean temperature and precipitation at the sam-ple plots of the AFI. In a further step this rela-tionship will then be used to deduce climatechange effects from GCM-output.

For a first application of the developed riskassessment procedure we defined a climatechange scenario in accordance with averageanomalies at GCM gridpoints over central Eu-rope. With regard to the year 2050 we assumedwinter temperature to linearly increase by 2 °C,summer temperatures by 2.5 °C. Precipitationduring summer was reduced by 15 percent. Sim-ilar assumptions had been made in earlier impactstudies (Bugmann 1994, Kienast et al. 1996).

5 Applying the Methodology:An Example

The described risk assessment methodology isdemonstrated for a small number of sites locatedin the ecoregion 4.2 (northern front range of theAlps, eastern part) according to Kilian et al.(1994). This ecoregion covers an altitudinal rangeof approximately 2000 m, thus including a varie-ty of forest types from oak-hornbeam forests atlow elevations to subalpine spruce forests. Fif-teen sample plots were randomly drawn fromthis AFI subset (Table 2). PICUS v1.2 was rununder current climate as well as under the de-fined climate change scenario for 1000 yearseach. This period was sufficient to reach equilib-rium species composition. Size of the simulatedforest was 1.0 ha. Model output was stored every10 years to increase computation speed. The com-parison of simulated steady state species compo-sition under current climate with the expectedPNV as reconstructed by vegetation experts yield-ed very plausible results (Table 3). This compar-ison was based on quantitative characterizationsof PNV-formations with regard to possible rang-es of species proportions for each of the 23 con-sidered forest communities (Table 4). The 15


124

Table 4. Examples for the quantitative characterization of possible species proportions for natural forestcommunities (PNV). Source: (Starlinger, unpublished).

Expert PNV Code Possible species proportions*

Larch/stone pine forest 01 (ld+pc) 0.6, pc 0.15, pa < 0.35, aba < 0.15, ps < 0.05, acp < 0.05,sum of other species 0.1

Spruce/fir forest 05 (pa+aba) > 0.5, aba 0.15, ld < 0.4, ps < 0.35, (acp+fs+fe+ug+ainc)< 0.2, qusp < 0.1, sum of other species < 0.1

* ld = Larix decidua, pc = Pinus cembra, pa = Picea abies, aba = Abies alba, ps = Pinus sylvestris, acp = Acer pseudoplananus, fs = Fagussylvatica, fe = Fraxinus excelsior, ug = Ulmus glabra, ainc = Alnus incana, qusp = Quercus spp.

Table 3. Comparison of simulated and expert PNV at selected sample plots of the Austrian Forest Inventory.Simulated equilibrium species composition classified according to Starlinger (unpublished).

Simulated equilibrium species composition

Expert PNV Subalpine Montane Spruce/fir/beech Beech forest Oak/hornbeamat AFI-plots (code) spruce forest spruce forest forest forest

Subalpine spruce forest (03) 2 – – – –Montane spruce forest (04) – – 1 – –Spruce/fir/beech forest (06) – 2 5 – –Beech forest (07) – – – 1 1Oak/hornbeam forest (08) – – – 1 2

Table 2. Site characteristics, soil and general climate parameters for selected sample plots of the AFI inecoregion 4.2 (Kilian et al. 1994).

ID Altitude Aspect Slope WHC PH CN Tav Precav (m) (°) (%) (cm) (°C) (mm)

246 1050 000 64 10.8 7.2 21.5 5.6 1393252 0950 045 90 19.8 3.7 17.1 6.1 1405304 0750 000 30 17.6 7.1 15.6 6.9 1439319 1250 000 72 12.9 6.4 13.5 4.8 1604258 0850 270 08 34.3 3.9 16.3 5.7 1323361 1050 270 16 21.1 5.1 18.9 5.6 1542441 0450 180 39 14.5 4.4 10.2 7.9 0919491 0450 180 12 22.6 5.3 13.4 7.9 0713458 0650 180 30 12.7 6.9 17.1 6.9 0930482 0350 000 19 12.6 4.4 15.0 8.5 0780492 0350 000 00 15.6 4.0 14.5 8.5 0683407 1350 000 61 12.8 4.4 12.0 4.2 1302445 1650 045 45 09.1 5.8 12.7 2.9 1227377 950 135 35 17.1 4.0 18.3 5.6 1001437 850 270 65 14.3 7.2 20.3 6.1 1175

selected plots covered the whole range from sub-alpine and montane spruce forests (2 plots and 1plot respectively), mixed spruce-fir-beech for-ests (7 plots), beech forests (2 plots) to oak-

hornbeam forests (3 plots). Compared to othergap model evaluations this test can be consid-ered rather restrictive and thus strengthened ourconfidence in the reliability of model results. It


125

is important to note that PICUS v1.2 also per-formed reasonably well at other temperate forestsites all over Europe (Badeck et al., in prepara-tion). Thus, it can be concluded that the modelrealistically captures synecology of European treespecies and is not primarily tuned to the sitesincluded in this study. In Fig. 5 selected assess-ment criteria which had been calculated from themodel output as well as the aggregated indicesfor the adaptation potential are presented by for-est types under current climate. From the figuresfor SMAP it can be seen that obviously at noneof the 15 sites an abrupt forest dieback did occurdue to the changing climate. However, the as-sessment criterion ACLAP(1) which characterizesshifts in PNV under a changed climate indicates,that major changes in the ecological site poten-tial are about to occur under the applied climatechange scenario. It is interesting to note, that atleast for the 15 selected sites this change seemsto be more severe at higher altitudes. This mayindicate that at the oak/hornbeam sites the droughttolerance limits of these species had not beenreached yet whereas at high elevation sites in-creasing temperatures favoured the immigrationof species with higher temperature requirements.Matching current species composition with fu-ture PNV (ACLAP(2)) yields generally low pref-erence values regarding the longterm adaptationpotential (LAP) with similar values for all foresttypes indicating a rather low longterm adapta-tion potential. Due to the high weight of theshort- to midterm index SMAP in eq. (2) the

overall rating for all 15 stands is fairly high withthe lowest values occuring at today's beech sites.This surely is due to the substantial portion ofplanted Norway spruce at these sites which areoutcompeted by broadleaves under a warmer anddrier climate, whereas the index AP at oak/horn-beam sites yields higher values due to the factthat at these sites current forests mainly consistof broadleave-dominated stands which are lessaffected by a changing climate.

6 Discussion

In this paper a risk assessment methodology forAustrian forests is presented. Core of the ap-proach is the application of a recently developed3D-patch model at sample plots of the AustrianForest Inventory. The main argument for thisapproach origines from the fact that large-scaleforest inventories which are installed in manyEuropean countries provide unique georeferenceddata on forest vegetation composition and struc-ture. To complete the set of site attributes neces-sary to initialize and drive forest simulation mod-els selected physical and chemical soil parame-ters had been estimated by means of empiricalprobabilistic relationships (Lexer and Hönnin-ger 1998c, Lexer et al. 1999). Current climatedata on a monthly basis were interpolated frommore than 600 weather stations. In the presenteddemonstration application of the risk assessmentsystem a general increase of temperature and a

Fig. 5. Selected assessment criteria and indices for the adaptation potential of current forests grouped by foresttypes (current climate). – Data base: 15 sample plots of the Austrian Forest Inventory.


126

decrease in summer precipitation respectivelyhad been assumed. This rather crude assumptionwill be replaced by transient climate change sce-narios which will be derived from GCM-outputby statistical downscaling techniques. The inte-grated data base of site, vegetation and climatedata allows for representative sampling proce-dures of the multi-dimensional space of site/veg-etation combinations.

In a first demonstration of the methodology itcould be shown that PICUS v1.2 was able torealistically reproduce equilibrium species com-positions at spatially explicit sites in the heterog-enous landscape of the Eastern Alps. It is impor-tant to note that simulated PNV was compared toanother – expert based – model of PNV. Never-theless, the results of this comparison in combi-nation with the modifications in model formula-tions strengthened our confidence in model reli-ability. The criticized parabolic temperature re-sponse included in earlier patch models (e.g.Bugmann 1994, Kräuchi 1994) has been replacedby an asymptotic response function which wasparameterized based on a broad empirical database (compare Lexer and Hönninger 1998b).Thus, in PICUS v1.2 trees are potentially able togrow well beyond their natural geographicalrange limits. This leads to a much more realisticmodel behaviour of PICUS compared to earliermodels. For instance, Norway spruce has beenheavily promoted at sites naturally supportingbroadleaved forest communities. A parabolic re-sponse to temperature parameterized from natu-ral range limits would result in unrealistic forestdieback events in such secondary spruce forests.Nevertheless, despite this modification PICUSv1.2 simulates broadleaved forest types at to-day's beech and oak sites due to the modelledcompetitive characteristics of tree species. Thus,we are confident that at least for the range oftemperatures covered in the present study modelbehaviour has been improved. However, we ad-mit that the problem of defining the fundamentalniche of tree species remains a highly relevantproblem (e.g. Austin 1992).

In contrast to other climate change impactstudies we employed an approach from the fieldof multiple-attribute utility theory to synthesize aset of indicators of forest sensitivity to changingclimatic conditions to an overall index for the ad-

aptation potential of current forests. The advan-tages of this approach are twofold. Firstly,through the modification of Saaty's method it waspossible to calculate consistent preference valuesfor any realization of the employed assessmentcriteria with regard to the adaptation potential offorests. Thus, we avoided abrupt and often notvery plausible discontinuities which often occur ifdiscrete risk rating matrices are employed (e.g.Kienast et al. 1996, Grabherr et al. 1998). Anotheradvantage of the presented approach is, that itallows sensitivity testing with regard to the sub-jective judgements involved in calculating therelative importance of individual assessment cri-teria. With the employed approach it is possible tocapture the transient response of currently existingforests to scenarios of climatic change. A decenialsequence of anomalies in composition and bio-mass accumulation under current climate and aclimate change scenario is condensed to an indi-cator for the short- to midterm adaptation poten-tial of current forests. No human interventions areassumed. The latter assumption may seem undulywhen dealing with managed forests. However,this “no management” assumption is a prerequi-site to identify forest conditions vulnerable to achanging climate. Species compositions whichare well adapted to the evolving site conditionswill show no signs of abrupt diebacks and rathermaintain forest functions without management in-terventions. This natural behaviour of forestswould be blurred by the effect of simulated man-agement operations. The “classical” approach inCentral European silviculture to derive the suita-bility of tree species at a given site by utilizing thespecies composition of the potential natural veg-etation (PNV sensu Tüxen 1956) as a benchmarkis employed to derive an estimate of a “longterm”adaptation potential of current forests (compareKienast et al. 1996). In accordance with Kienast etal. (1996) this approach alone would provide arather crude classification of a current forest's riskin case of climatic change because absence ofnaturally dominating tree species may not result ina forest breakdown. However, the greater the shiftin the ecological site potential and the larger thedivergence between current species compositionthe more likely is an adverse effect of climaticchange on forest functioning or the more intensethe required management to maintain forest func-


127

tions. It is important to note that with the present-ed approach the current forest is matched withPNV which is generated by a dynamic vegetationmodel for each site specifically. Thus, the postu-lation of fixed species proportions for syntaxo-nomic PNV-formations (e.g. Mucina et al. 1993,Ellenberg 1996) was avoided.

There are two important points to be adressedwhen interpreting the results from the demon-stration example. Firstly, it is important to rec-ognize the uncertainties involved with climatechange scenarios. Driving an ecosystem modelwith a climate change scenario by definition willresult in a scenario of forest ecosystem develop-ment. In addition there is no reason to assumethat climate change will come to a halt in 2050.Secondly, several sources of uncertainty in theemployed forest model have to be considered(model structure, model parameters) when inter-preting simulation results. From this we mightconclude that climate change impact studies donot have predictive value but rather are sensitiv-ity tests of how a forest ecosystem might re-spond under a range of specified conditions whichmay never occur in reality (Bugmann 1997).

The described methodology will be applied toan extensive sample of plots from the AustrianForest Inventory to represent the various vegeta-tion-site combinations of Austrias forests. Thepresented methodology seems particularly suita-ble to adress the first task of the “classical” se-quence of problem solving (Rauscher 1999):problem identification. Based on such findingsmanagement strategies can then be developed tomitigate the possible impacts of a changing cli-mate.

Acknowledgements

This work was conducted within the researchproject “The Adaptation Potential of AustrianForests to Scenarios of Climatic Change” fund-ed by the Federal Ministry of Agriculture andForestry, the Federal Ministry of Environment,Youth and Family Affairs and the Federal Envi-ronmental Agency. We are grateful to KarlSchieler, Klemens Schadauer, Franz Starlingerand Michael Englisch for making the data of theAustrian Forest Inventory and the Austrian For-

est Soil Survey available as well as for valuablecomments. Two anonymous reviewers providedhelpful comments on an earlier version of themanuscript.

References

Austin, M.P. 1992. Modelling the environmental nicheof plants:implications for plant community re-sponse to elevated CO2-levels. Australian Journalof Botany 40: 615–630.

Bitterlich, W. 1948. Die Winkelzählprobe. Allg. Forst-und Holzwirt. Zeitung Wien 59: 4–5.

Botkin, D.B. 1993. Forest dynamics. An ecologicalmodel. Oxford University Press. 309 p.

— , Janak, J.F. & Wallis, J.R. 1972. Some ecologicalconsequences of a computer model of forestgrowth. J. Ecol. 60: 849–872.

Bugmann, H. 1994. On the ecology of mountainousforests in a changing climate: A simulation study.Diss. ETH No. 10638. 258 p.

— 1996. A simplified forest model to study speciescomposition along climate gradients. Ecology 77:2055–2074.

— 1997. Gap models, forest dynamics and the re-sponse of vegetation to climate change. SpringerVerlag, NATO ASI Series, Vol. 47: 442–453.

— , Solomon, A.M. 2000. Explaining forest composi-tion and biomass across multiple biogeographicalregions. Ecological Applications 10(1): 95–114.

CIPRA. 1997. Protokoll zur Durchführung der Alpen-konvention von 1991 im Bereich Bergwald. Inter-nationale Alpenschutzkonvention. 17 p.

Dunne, Th. & Leopold, L.B. 1978. Water in environ-mental planning. W.H. Freeman, New York. 818 p.

Ellenberg, H. 1996. Vegetation Mitteleuropas mit denAlpen in ökologischer Sicht. 5. Auflage, UlmerVerlag. 1095 p.

Fischlin, A., Bugmann, H. & Gyalistras, D. 1995.Sensitivity of a forest ecosystem model to climateparametrization schemes. Env. Poll. 87: 267–282.

Goicoechea, A., Hansen, D.R. & Duckstein, L. 1982.Multiobjective decision analysis with engineeringand business applications. John Wiley & Sons,New York.

Gordon, R. 1994. Waldbauliche Planung im Schutz-wald. Österr. Forstztg. 1: 22–24.

Grabherr, G., Koch, G., Kirchmeir, H. & Reiter, K.1998. Hemerobie Österreichischer Waldökosys-


128

teme. Österreichische Akademie der Wissenschaf-ten 17. 493 p.

Grottenthaler, W. & Laatsch, W. 1973. Untersuchun-gen über den Hangabtrag im Lainbachtal bei Bene-diktbeuern. Forstw. Cbl. 84: 212–229.

Gundermann, N. 1974. Beiträge zur Quantifizierungder Sozialfunktionen des Waldes im bayerischenHochgebirge. Teil B: Schutzfunktionen. ForstlicheForschungsberichte München 21. 222 p.

Gyalistras, D., v. Storch, H., Fischlin, A. & Beniston,M. 1994. Linking GCM-simulated climatic chang-es to ecosystem models: case studies of statisticaldownscaling in the Alps. Clim. Res. 4: 167–189.

Hasenauer, H. 1997. Dimensional relationships ofopen-grown trees in Austria. For. Ecol. Manage-ment 96: 197–206.

Holloway, C.A. 1979. Decision making under uncer-tainty: Models and choices. Prentice Hall Inc. 522 p.

Houghton, J.T., Callander, B.A. & Varney, S.K. (eds.)1992. Climate change 1992 – the supplementaryreport to the IPCC scientific assessment. Reportprepared for IPCC by Working Group 1. Cam-bridge University Press. 200 p.

Hunsaker, C.T., Graham, R.L., Suter, G.J., O'Neill,R.V., Barnthouse, L.W. & Gardner, R.H. 1990.Assessing ecological risk on a regional scale. En-vironm. Management 14: 325–332.

Huth, A., Ditzer, T. & Bossel, H. 1996. Simulation ofthe growth of tropical rain forests. Center for En-vironmental Systems Research, University of Kas-sel. 180 p.

Instruktionen für die Feldarbeit der ÖsterreichischenWaldinventur 1992–1996. 1995. Forstliche Bun-desversuchsanstalt. 194 p.

Kellomäki, S., Väisänen, H., Hänninen, H., Kolström,T., Lauhanen, R. & Pajari, B. 1992. SIMA: Amodel for forest succession based on the carbonand nitrogen cycles with application to silvicul-tural management of the forest ecosystem. SilvaCarelica 22: 1–85.

Kienast, F. 1987. FORECE – A forest successionmodel for southern Central Europe. Oak RidgeNational Laboratory, Oak Ridge, Tennessee,ORNL/TM-10575, Publ. No. 2989.

— & Kuhn, N. 1989. Simulating forest successionalong ecological gradients in Southern CentralEurope. Vegetatio 79: 7–20.

— , Brzeziecki, B. & Wildi, O. 1996. Long-termadaptation potential of Central European moun-tain forests to climate change: a GIS-assisted sen-

sitivity assessment. For. Ecol. Management 80:133–153.

Kilian, W., Müller, F. & Starlinger, F. 1994. Dieforstlichen Wuchsgebiete Österreichs. Eine Natur-raumgliederung nach waldökologischen Gesichts-punkten. FBVA-Berichte 82. 60 p.

Kozlowski, T.T., Kramer, P.J. & Pallardy, S.G. 1991.The physiological ecology of woody plants. Aca-demic Press Inc. 657 p.

Kräuchi, N. 1994. Modelling forest succession as in-fluenced by a changing environment. Diss. ETHNo. 10479. 116 p.

Larcher, W. 1995. Physiological plant ecology. 3rd

edition, Springer Verlag. 506 p.Lauscher, F. 1976. Weltweite Typen der Höhen-

abhängigkeit des Niederschlags. Wetter und Leben28: 80–90.

Leibundgut, H. 1981. Die natürliche Waldverjüngung.Haupt, Bern.

Lexer, M.J. & Hönninger, K. 1998a. Simulated ef-fects of bark beetle infestations on stand dynamicsin Norway spruce stands: Coupling a patch modeland a stand risk model. In: Beniston, M. & Innes,J.L. (eds.). The impacts of climate variability onforests. Lecture Notes in Earth Sciences, SpringerVerlag. p. 289–308.

— & Hönninger, K. 1998b. Defining the physiologi-cal amplitude of alpine tree species using the com-bined network of forest inventory data, soil andmeteorological data. Ecologie 29(1–2): 383–387.

— & Hönninger, K. 1998c. Estimating physical soilparameters for sample plots of large scale forestinventories. For. Ecol. Management 111: 231–247.

— & Hönninger, K. 2000a. A modified 3D-patchmodel for spatially explicit simulation of vegeta-tion composition in heterogenous landscapes. For.Ecol. Management (in press).

— & Hönninger, K. 2000b. The sensitivity of Nor-way spruce forests to scenarios of climatic change(in German). FBVA-Berichte (in press).

— , Hönninger, K. & Englisch, M. 1999. Schätzungvon chemischen Bodenparametern für Waldstan-dorte am Beispiel der Österreichischen Wald-inventur. Forstw. Cbl. 118: 212–227.

— , Hönninger, K. & Vacik, H. 2000. Modelling theeffect of forest site conditions on the ecophysio-logical suitability of tree species: An approachbased on fuzzy set theory. Computers and Elec-tronics in Agriculture (in press).

Lindner, M. 1998. Wirkung von Klimaveränderungen


129

in mitteleuropäischen Wirtschaftswäldern. PIK-Report 46. 108 p.

— , Bugmann, H., Lasch, P., Flechsig, M. & Cramer,W. 1997. Regional impacts of climatic change onforests in the state of Brandenburg, Germany. Ag-ric. For. Meteorology 84: 123–135.

Mayer, H. 1984. Waldbau auf soziologisch-öko-logischer Grundlage. 3. neu bearb. Aufl. Fischer,Stuttgart.

Ministerial Conference for the Protection of EuropeanForests, 1998. Beschlüsse und Resolutionen derDritten Ministerkonferenz zum Schutz der Wälderin Europa. Bundesministerium für Land- und Forst-wirtschaft. 93 p.

Moore, A.D. 1989. On the maximum growth equationused in forest gap simulation models. Ecol. Model-ling 45: 63–67.

Mucina, L., Grabherr, G. & Wallnöfer, S. (eds.) 1993.Die Pflanzengesellschaften Österreichs, Teil III,Wälder und Gebüsche. Gustav Fischer Verlag.353 p.

Ott, E. & Schönbächler, D. 1986. Die Stabilitäts-beurteilung im Gebirgswald als Voraussetzung fürdie Schutzwaldüberwachung und -pflege. Schweiz.Z. f. Forstwesen 137(9): 725–738.

Pastor, J. & Post, W.M. 1985. Development of a linkedforest productivity – soil process model. ORNL/TM-9519.

Peters, R.L. & Darling, J.D. 1985. The greenhouseeffect and nature reserves. BioScience 35: 707–717.

Prentice, I.C. & Helmisaari, H. 1991. Silvics of northEuropean trees: Compilation, comparisons andimplications for forest succession modelling. For.Ecol. Management 42: 79–93.

— , Sykes, M.T. & Cramer, W. 1993. A simulationmodel for the transient effects of climate changeon forest landscapes. Ecol. Modelling 65: 51–70.

Rauscher, H.M. 1999. Ecosystem management deci-sion support for federal forests in the United States:A review. For. Ecol. Management 114: 173–197.

Saaty, T.L. 1977. A scaling method for priorities inHierarchical Structures. J. Math. Psychology 15:234–281.

— 1996. The Analytic Hierarchy Process. RWS Pub-lications, Pittsburgh.

Schenk, H.J. 1996. Modelling the effects of tempera-ture on growth and persistence of tree species: Acritical review of tree population models. Ecol.Modelling 92: 1–32.

Shugart, H.H. 1998. Terrestrial ecosystems in chang-ing environments. Cambridge University Press.537 p.

— , Leemans, R. & Bonan, G.B. (eds.) 1992. A sys-tems analysis of the global boreal forest. Cam-bridge University Press. 565 p.

Sohngen, B., Sedjo, R., Mendelson, R. & Lyon, K.2000. Determining long term structural shifts inregional timber supply resulting from the impactof climate change on forested ecosystems. Sev-enth Symposium on Systems Analysis in ForestResources, May 28–31, 1997, Shanty Creek, MI,USA, Conference Proceedings (in press).

Starlinger, F. 1998. Mögliche Auswirkungen einerKlimaänderung auf die Waldvegetation Öster-reichs. In: Müller, F. (ed.). Klimaänderung: Mög-liche Einflüsse auf den Wald und waldbaulicheAnpassungsstrategien, Österreichischer Forstver-ein. 80 p.

Suter, G.W. 1993. Ecological risk assessment. LewisPublishers. 538 p.

Sykes, M.T., Prentice, I.C. & Cramer, W. 1996. Abioclimatic model for the potential distribution ofnorth European tree species under present andfuture climates. Journal of Biogeography 23: 203–233.

Tabony, R.C. 1985. Relations between minimum tem-perature and topography in Great Britain. Journalof Climatology 5: 503–520.

Talkkari, A. & Hypen, H. 1996. Development andassessment of a gap-type model to predict theeffects of climate change on forests based on spa-tial forest data. For. Ecol. Management 83: 217–228.

Thornthwaite, C.W. & Mather, J.R. 1957. Instructionsand tables for computing potential evapotranspi-ration and the water balance. Publ. Clim. 10(3).311 p.

Tüxen, R. 1956. Die heutige potentielle natürlicheVegetation als Gegenstand der Vegetationskartie-rung. Angewandte Pflanzensoziologie, Stolzenau/W. 13: 5–42.

Urban, D.L. 1990. A versatile model to simulate for-est pattern. A user's guide to ZELIG version 1.0.Environmental Sciences Department, The Univer-sity of Virginia, Charlottesville, Virginia. 107 p.

Woodward, F.I. 1987. Climate and plant distribution.Cambridge University Press. 174 p.

Total of 68 references

the sensitivity of central european mountain forests to scenarios

Documents