Regional Environmental Change

, Volume 17, Issue 1, pp 33–48 | Cite as

Can current management maintain forest landscape multifunctionality in the Eastern Alps in Austria under climate change?

  • Florian Irauschek
  • Werner Rammer
  • Manfred J. Lexer
Original Article

Abstract

In Central Europe, management of forests for multiple ecosystem services (ES) has a long tradition and is currently drawing much attention due to increasing interest in non-timber services. In face of a changing climate and diverse ES portfolios, a key issue for forest managers is to assess vulnerability of ES provisioning. In a case study catchment of 250 ha in the Eastern Alps, the currently practiced uneven-aged management regime (BAU; business as usual) which is based on irregularly shaped patch cuts along skyline corridors was analysed under historic climate (represented by the period 1961–1990) and five transient climate change scenarios (period 2010–2110) and compared to an unmanaged scenario (NOM). The study addressed (1) the future provisioning of timber, carbon sequestration, protection against gravitational hazards, and nature conservation values under BAU management, (2) the effect of spatial scale (1, 5, 10 ha grain size) in mapping ES indicators and (3) how the spatial scale of ES assessment affects the simultaneous provision of several ES (i.e. multifunctionality). The analysis employed the PICUS forest simulation model in combination with novel landscape assessment tools. In BAU management, timber harvests were smaller than periodic increments. The resulting increase in standing stock benefitted carbon sequestration. In four out of five climate change scenarios, volume increment was increasing. With the exception of the mildest climate change scenario (+2.6 °C, no change in precipitation), all other analysed climate change scenarios reduced standing tree volume, carbon pools and number of large old trees, and increased standing deadwood volume due to an intensifying bark beetle disturbance regime. However, increases in deadwood and patchy canopy openings benefitted bird habitat quality. Under historic climate, the NOM regime showed better performance in all non-timber ES. Under climate change conditions, the damages from bark beetle disturbances increased more in NOM compared with BAU. Despite favourable temperature conditions in climate change scenarios, the share of admixed broadleaved species was not increasing in BAU management, mainly due to the heavy browsing pressure by ungulates. In NOM, it even decreased and mean tree age increased. Thus, in the long run NOM may enter a phase of lower resilience compared with BAU. Most ES indicators were fairly insensitive to the spatial scale of indicator mapping. ES indicators that were based on sparse tree and stand attributes such as rare admixed tree species, large snags and live trees achieved better results when mapped at larger scales. The share of landscape area with simultaneous provisioning of ES at reasonable performance levels (i.e. multifunctionality) decreased with increasing number of considered ES, while it increased with increasing spatial scale of the assessment. In the case study, landscape between 53 and 100 % was classified as multifunctional, depending on number and combinations of ES.

Keywords

Mountain forests Ecosystem services Scale PICUS Climate change Forest management 

Introduction

Mountain regions provide a diverse range of ecosystem services (ES). In the Eastern Alps in Central Europe, mountain forests serve as a source of timber to support the needs of industry as well as of fuel wood for subsistence use. Forests have to protect slopes from landslides and soil erosion and protect settlements and infrastructure against gravitational natural hazards like snow avalanches and rockfall (Malin and Maier 2007; Dorren et al. 2004). In Austria, for instance, 31 % of forest area has been assigned such a protective role as top priority in forest spatial planning (Niese 2011). Regionally, this percentage may even be as high as 66 %. Due to close to nature status of many mountain forests, the share of nature protection areas, such as those under the EU Natura 2000 regulations, is particularly high in mountain areas. Recently, provisioning of drinking water and the dampening of run-off peaks for hydropower production as well as carbon sequestration have also been recognized as key ecosystem services. The importance of these ES in general and for specific stakeholder groups in particular may vary strongly from region to region (European Environment Agency 2010). This multitude of vital ES demands has led to the paradigm of multifunctional forestry where forests have to provide many ES simultaneously at relatively small spatial scale (Nijnik et al. 2010).

Forests are multifunctional by nature (Kaljonen et al. 2007) and when ecosystem services are complementary or neutral integration may be a feasible approach (Raudsepp-Hearne et al. 2010). In case of conflicting ES, trade-offs must be considered. Inefficient solutions may be the consequence (e.g. Jacobsen et al. 2013) if management concepts enforce integration at small scales (i.e. stand level, a few hectares). In such situations, zoning is then a useful approach to disentangle ES conflicts (Côté et al. 2010). While local practical solutions to balance actual ES demands have been established ever since by managers and stakeholders, the paradigm of multifunctionality in general has been rarely touched in policy making and governance (Suda and Pukall 2014).

In recent years, the paradigm of landscape-level planning in forest management has evolved in forest sciences (e.g. Fries et al. 1998). While in Scandinavian countries there is already ample experience in landscape-level planning (e.g. Lämås and Eriksson 2003), in Central European forestry it has been seldomly implemented in practice so far. The issue of scale in ES provisioning has only recently attracted more attention in landscape ecology and land use planning (e.g. Grêt-Regamey et al. 2014; Wu et al. 2002; Raudsepp-Hearne et al. 2010). Landscape-level planning implies that (1) multiscale processes such as disturbance regimes are considered in forest management, (2) different ES may require different spatial scales for quantification and monitoring and (3) different ES or portfolios of ES may be prioritized in different parts of the landscape. A prerequisite for such approaches to manage for portfolios of ES is sound knowledge about the interrelatedness of different ES and how this may depend on forest management regimes and other drivers such as climate change. Planning across ownerships imposes particular challenges (Nijnik et al. 2010). Thus, in landscapes with small-scale ownership structure integration of ES provisioning may be the only practical solution. Future climate change may impact ES differently (Seidl et al. 2007; Lindner et al. 2010; Hanewinkel et al. 2012), thus adding complexity to forest management decision making with the need to find new balances in ES provisioning. In forestry, the issue at which spatial scale the provision of specific ES portfolios is feasible and economically efficient is still a matter of debate and calls for focused research.

Here, we set out to assess a currently practiced uneven-aged forest management regime in a catchment in the Eastern Alps in Austria under climate change conditions and evaluate impacts on ES (timber production, carbon sequestration, nature conservation values and protection against snow avalanches, landslides and erosion). We used the forest ecosystem model PICUS version 1.5 in combination with a recently developed landscape assessment tool (Maroschek et al. 2015). We were furthermore particularly interested in (a) the effect of spatial scale on mapped ES indicators and (b) how the spatial scale of ES assessment (i.e. grain size) affects the simultaneous provision of several ES (i.e. multifunctionality) in a mountain forest landscape.

Materials and methods

Study area

The study area is located in the Province of Vorarlberg in Austria, close to the Swiss border in the Rellstal valley (47.08° N, 9.82° E). Landowner is the Stand Montafon Forstfonds (SMF), which owns about 6.500 ha forest land in total. Depending on bedrock, the soils are rendzinas, rankers, podzols and rich cambisols. The terrain is steep, with slope angles from 30° to 45°, which makes forest management difficult and underlines the protective function against gravitational natural hazards (snow avalanches, rockfall, landslides and erosion). The case study is a catchment of 250 ha total area (234 ha forest area) in the upper part of the valley at altitudes between 1060 and 1800 m a.s.l. The timber line that potentially may be as high as 2000 m a.s.l. has been strongly shaped by human activities such as livestock grazing and alpine pasturing. During the last decades, those activities have been widely regulated, and since then grazing has been abandoned in the study area (Malin and Maier 2007). As a consequence successional dynamics are moving the timberline upward. Forest management has been practiced since more than 500 years (Bußjäger 2007). The current management objectives of the owner are income generation from timber production and securing sustainable protection against snow avalanches and landslides (Malin and Lerch 2007). In addition, major shares of the forest area are under Natura 2000 regulations with a focus on bird habitat protection for black woodpecker (Dryocopus maritimus) and three-toed woodpecker (Picoides tridactylus) (Grabherr 2000).

Forest

The forests in the case study area are dominated by Norway spruce (Picea abies, 96 % of growing stock) with minor shares of Silver fir (Abies alba, 3 %), European beech (Fagus sylvatica, 1.6 %) and other broadleaved species (e.g. Acer pseudoplatanus, Fraxinus excelsior, 1 %). Historic forest management has led to mostly uneven-aged patchy stand structures with a considerable share of large old trees (Malin and Lerch 2007; Malin and Maier 2007). Game management has favoured high densities of ungulates, and consequently, the browsing pressure on Silver fir and broadleaves is high. According to internal records of the owner, productivity ranges from 3.5 to 12 m3 ha−1 year−1 depending on site and stand composition and structure (Malin and Maier 2007). The current mean standing stock in the case study area is 455 m3 ha−1. The largest part of the forest is located on steep slopes which are not accessible by forest roads but require skyline-based harvesting systems for timber extraction.

Climate data

A baseline climate represented by the historic climate of the period 1961–1990 (c0) and five transient climate change scenarios (c1–c5), each consisting of a 100-year time series covering the period 2010–2110 of daily temperature, precipitation, radiation and vapour pressure deficit, were prepared for the model simulations. The baseline climate was generated from available daily instrumental data of the historic period 1961–1990 from the meteorological station Feldkirch (9.6° long, 47.27° lat) and adjusted for representative site types within the case study area regarding altitude, slope and aspect using the algorithms in Thornton and Running (1999). The five climate change scenarios were based on regional climate simulations from the ENSEMBLES project (Hewitt and Griggs 2004; www.ensembleseu.org) which had been downscaled from the grid scale of the regional climate model simulations to local sites. For details on the downscaling approach see Bugmann et al. (this volume). Mean historic climate at 1000 m a.s.l. is characterized by 6.2 °C mean annual temperature and 1150 mm annual precipitation with 840 mm during summer season from May to September. In all climate change scenarios, temperature increased (+2.6 °C in c1, +3.0 °C in c2, +3.5 °C in c3, +4.3 °C in c4, +6.0 °C in c5). In all climate change scenarios except c1, there was a relative shift of precipitation from summer (May–September) to winter with a reduction in summer by −7 % in c2, −32 % in c3, −19 % in c4 and −14 % in c5. Climate anomalies were all related to the baseline climate and the period 2081–2110 of the climate change scenarios.

The PICUS forest ecosystem model

General

The model used for this study is the hybrid (sensu Peng 2000) forest ecosystem model PICUS version 1.5. The model is a hybrid of classical gap model components (PICUS v1.2, Lexer and Hönninger 2001) and process-based stand-level NPP algorithms (3PG, Landsberg and Waring 1997). A detailed description of the model is provided in Seidl et al. (2005). Here, just a brief overview on the core model concept is given.

PICUS simulates growth, regeneration and mortality of individual trees on a grid of 10 × 10 m2 patches. Tree biomass is arranged in cells with a vertical depth of 5 m. A three-dimensional light model, allowing for the explicit consideration of direct and diffuse radiation within the canopy, is used to estimate absorbed radiation for each tree. Stand-level productivity is estimated with a simplified model of light use efficiency (Landsberg and Waring 1997) which depends on temperature, radiation, vapour pressure deficit, soil water and nutrient supply. Redistribution of assimilates to individual trees, assuming fixed respiration rates (Landsberg and Waring 1997), is accomplished according to the relative competitive success (i.e. biomass increment in the preceding year) of the individuals (see Lexer and Hönninger 2001). The development of seedlings and saplings is modelled in a size class approach within five height classes (Woltjer et al. 2008). The PICUS model furthermore includes a descendent of the TRACE soil model as described in Seidl et al. (2008) which in the current study was used to simulate the decomposition of C from litter and deadwood as a prerequisite to monitor effects of management and climate on the soil C pools. Dead trees, if not removed through forest management, are transferred stochastically from snags to wood detrital pools on the forest floor. The PICUS model includes also a bark beetle disturbance module which (1) computes the stochastic infestation risk for simulated forest stands, (2) estimates the damage intensity if an infestation occurs and (3) distributes the resulting tree mortality within the simulated stand (Seidl et al. 2007). PICUS contains a flexible management module based on a scripting language allowing for spatially explicit harvesting interventions as well as planting operations at the level of the 100 m2 patches. The basic time step of the simulation is monthly with annual integration of the tree population dynamics processes. The model requires information about the soil water storage capacity, the pH value of the mineral soil as well as plant-available nitrogen as a proxy for nutrient supply as well as a number of parameters for the soil submodel. The PICUS model in the current study was driven by monthly values of temperature, precipitation, solar radiation and vapour pressure deficit of the atmosphere. With the current model version, stands of up to 25 ha can be simulated. PICUS has been tested intensively (e.g. Huber et al. 2013; Didion et al. 2009; Seidl et al. 2005).

A PICUS simulation can start from bare ground or with any defined stand structure. The initial state of a simulated stand (trees with diameter at breast height (DBH) > 2 cm) can be provided as a tree list and a related map with tree positions containing species, DBH and height for each individual or as a species-specific DBH distribution and a height-diameter model. If no tree coordinates are available individuals can be distributed randomly or based on qualitative information about the mixture form (i.e. small groups and patches). Please note that the population dynamics model and the NPP module do not distinguish the position of individual trees below the 100 m2 patch resolution. Regeneration as species-specific density (n ha−1) in 5 height classes can be initiated as patchy pattern (100 m2 resolution) or as a homogeneous regeneration layer throughout the simulated stand. For the calculation of spatial stand structural indices, tree maps can be exported and loaded into a landscape assessment tool (LAT). The LAT can visualize and analyse multiple single-tree maps on a digital terrain model. Tree and standing dead wood attributes (coordinates, species, and dimension) can be analysed in freely selectable subareas or with moving window approaches (see Maroschek et al. 2015) and exported for further analysis or enhanced visualization purposes.

Model calibration

PICUS has been developed as a generalized model of tree population dynamics and forest development aiming at a generic species parameterization (see Seidl et al. 2005). In the current study, the same species parameterization for European temperate forest ecosystems was used as established in Seidl et al. (2010, 2011) and later applied in Huber et al. (2013) and Maroschek et al. (2015). There was one exception related to the tree regeneration module. In trial runs, it became obvious that tree regeneration, particularly for pioneer species (Alnus sp., Betula pendula, Populus tremula), developed too fast, and density as well as height development on patch cut areas could not be matched with observations from the field. The main reason for this mismatch was competition by grass and herb species which were not considered in the model. To adjust establishment rate in the smallest height class (<10 cm) and early growth, the germination rate of seeds and height growth potential in the five height classes were reduced (see Table SM-1 in Supplementary Material).

Model evaluation

In order to evaluate the ability of PICUS to reproduce tree growth and development of stand structure in the case study area time-series data as collected by Neumann (1993) were used. Neumann (1993) reconstructed a 30-year time series of growth in three Norway spruce stands at altitudes of 950, 1230 and 1690 m a.s.l. (see Supplementary Material) in the adjacent valley close to the study site from a tree and stump inventory in 1991 and tree ring and stem analysis for selected trees. The stands at 1230 and 1690 m were uneven-aged with heterogeneous canopies and the stand at 950 m a.s.l even-aged. For the model evaluation, the stands were initialized according to the stand characteristics in 1961 and a random thinning was conducted in every 10-year period to mimic the stem numbers given in Neumann (1993). Historic climate to drive the simulations and soil parameters were provided as described in “Climate data” and “Forest initialization” sections. Simulated dominant height, average height, average DBH and basal area were compared with the data in Neumann (1993). Simultaneous F tests for regression models of observed versus predicted values for mean height and DBH and basal area (states in 1971, 1981 and 1991 as well as periodic increments) indicated good fit of the simulations. Simulated dominant height showed larger deviations, particularly at the high altitude site. For details, see Supplementary Material.

Forest initialization

Based on 53 polygons that had been derived manually from aerial images, a terrestrial inventory was carried out for the purpose of this study. At least eight inventory plots per polygon on a base raster of 50 × 50 m were measured using angle-count sampling to gather information about basal area shares of tree species, diameter distributions of tree species, a height-diameter regression model, a description of tree regeneration (density by species and height class, mixture form) and soil attributes. From Hollaus et al. (2006, 2007), a normalized crown model and a volume map derived from high-resolution LiDAR data were available. Based on these data, tree maps (size, species and location of individual trees) were generated for each of the polygons. For details of the approach, see Maroschek et al. (2015). All polygons had a spatially explicit position, and the 53 tree maps were subsequently mapped into the 250 ha landscape (Fig. 2).

The landscape was then structured into 18 harvesting units (HU). These 18 HUs were used as basic simulation entity (4–20 ha in size, Fig. 1) in the PICUS model. The main rationale for the delineation of the HUs has been topography which determines the efficient location of skyline tracks for timber harvesting. In prior work, based on interviews with local forest management staff and supported by GIS a total of 131 skylines had been located in the catchment area. The parameter values for the site types in the polygons and HU, respectively, were taken from the soil data base for Austrian forests as described in Seidl et al. (2009).
Fig. 1

Study landscape structured into 18 harvesting units as simulation entities and 131 skyline tracks to implement BAU management 2010–2110

Forest management

The currently practiced management regime (BAU; business as usual) is aiming at uneven-aged, structurally diverse forests. Due to steep terrain, timber harvesting is bound to motor-manual felling, delimbing and cutting the stems to length. The logs are extracted to forest roads at the base of the slopes by cable yarding with skyline systems. Skyline tracks typically extend diagonally across the slope to avoid vertical corridors which may favour avalanches and rockfall. The mean skyline length in BAU management over the 100 year analysis period is 534 m (minimum 110 m, maximum 955 m). Current management features patch cuts along the skyline track. Size and shape of the patches is variable with a typical maximum width of 50 m (i.e. maximum lateral skidding distance) and a mean length of 40-50 m along the skyline (compare Fig. 1). Spacing and timing of the skylines depend on the maturity of forests on one hand, and on the avoidance of negative visual impact by the implementation of too many locally clustered skylines and related intensive timber harvesting activities on the other hand. Current management relies fully on natural regeneration. No tending and thinning operations are carried out in the rejuvenated patches. The general silvicultural aim is to maintain and further develop the heterogeneous uneven-aged forest structure. According to management records and as implemented in the BAU regime in this analysis, each year approximately 0.4 % of the forest area is subject to felling operations (i.e. 0–2 skylines per year which corresponds to an average of 0.83 ha patch cut area per year). Overall, the implemented BAU management results in a complete area turnover of the case study catchment of 250 years. Based on Maroschek (unpublished data), the annual browsing probability in the model runs for A. alba seedlings was set to 0.78, for F. excelsior 1.0, A. pseudoplatanus 0.51 and F. sylvatica 0.70.

For comparison, a no-management regime (NOM) without any active silvicultural intervention has also been simulated. Browsing intensity in NOM was as in the BAU scenario.

Analysis

Ecosystem service indicators

A set of indicators is used to characterize the level of ES provisioning (Table 1). Timber production is represented by standing volume of life trees (V), the harvested stemwood volume (THV), the periodic mean increment (VI) and the timber volume killed by bark beetle infestations (BBD). CS includes carbon in tree biomass, standing deadwood, coarse woody debris and soil carbon. Biodiversity indicators are species diversity (D; Eqs. 1a, 1b) (Jost 2007), tree size diversity (H; Eqs. 2a, 2b), the volume in standing deadwood (SDVW; DBH > 20 cm) and the number of large living trees (LLTN; DBH > 50 cm).
$$ D = \exp (H) $$
(1a)
$$ H = - \sum\limits_{i = 1}^{S} {p_{i} \ln (p_{i} )} $$
(1b)
where S is the number of tree species and pi is the relative basal area share of species (i).
$$ H_{\text{size}} = \frac{{H_{\text{DBH}} + H_{H} }}{2} $$
(2a)
$$ H_{\text{DBH}} = - \sum\limits_{m = 1}^{{N_{\text{DBH}} }} {p_{m} \ln (p_{m} )} $$
(2b)
$$ H_{H} = - \sum\limits_{n = 1}^{{N_{H} }} {p_{n} \ln (p_{n} )} $$
(2c)
where NDBH is the number of 5 cm DBH classes for trees >5 cm DBH and NH is the number of 2 m height classes for trees taller than 4 m. pm is the relative basal area within a DBH class and pn is the relative basal area within a height class.
Table 1

Ecosystem service indicators for current “business-as-usual” management (BAU) and the no-management regime (NOM) under historic climate (c0) for three assessment periods: P1 = 2010–2043, P2 = 2044–2077, P3 = 2078–2110

Acronym

Explanation

Unit

BAU

NOM

P1

P2

P3

P1

P2

P3

Timber production

V

Standing volume living trees

m3 ha−1

436.5

450.4

490.1

466.4

525.6

599.6

THV

Annual volume harvested

m3 ha−1 year−1

1.9

1.6

2.8

VI

Annual net volume increment

m3 ha−1 year−1

4.9

5.6

6.5

5.0

5.6

5.9

BBD

Volume killed by bark beetles

m3 ha−1 year−1

0.6

0.6

0.5

0.6

0.6

0.5

Carbon sequestration

CS

Carbon (trees, standing deadwood, coarse woody debris and soil carbon)

t ha−1

220.2

218.2

222.9

237.6

244.4

253.1

Biodiversity

D

Tree species diversity

1.33

1.32

1.23

1.32

1.29

1.22

H

Tree size diversity (mean Shannon diversity of DBH and height)

5.75

5.41

5.60

5.72

5.21

4.96

AA

Basal area share of Abies alba

%

6.2

4.6

3.4

6.1

4.6

3.6

BL

Basal area share of broadleaves

%

3.4

3.3

2.6

3.3

2.9

2.0

SDWV

Standing deadwood volume (DBH < 20 cm)

m3 ha−1

22.9

20.5

23.8

24.4

23.5

28.8

LLTN

Large living trees (DBH > 50 cm)

n ha−1

56.1

54.5

49.7

60.4

65.0

66.3

BHQ

Bird habitat quality

%

[51/39/10]

[40/58/01]

[36/64/00]

[46/44/10]

[36/60/05]

[23/75/02]

Protection against gravitational hazards

API

Avalanche protection index

0–1

0.91

0.96

0.98

0.93

0.98

0.98

LPI

Landslide and erosion protection

%

[07/45/48]

[02/26/71]

[01/27/71]

[04/42/55]

[00/25/75]

[00/26/74]

Categories for BHQ and LPI: 1 = bad, 2 = moderate, 3 = good, provided in percentage of area for categories [1/2/3]. Indicator values for BHQ, API and LPI are based on 1 ha samples and all other indicators based on HU simulation entities

Furthermore, a bird habitat quality (BHQ) index on ordinal scale (BHQ1 = poor habitat quality, BHQ2 = moderate, BHQ3 = good habitat quality) characterizes habitat quality for black woodpecker (D. maritimus) and three-toed woodpecker (P. tridactylus) as key bird species in the study region. BHQ is a composite indicator based on structural attributes of the forest (standing deadwood with DBH > 30 cm, large living trees with DBH > 50 cm, canopy cover) and time since previous management activities as an indicator for anthropogenic disturbance (for details, see Bugmann et al. this issue).

The avalanche protection index (API) indicates protection against snow avalanche release. The index is calculated from mean slope, basal area and average diameter (Eq. 3).
$$ {\text{API}} = \hbox{min} \left[ {\frac{G}{{\left( {0.2901*{\text{mDBH}} + 1.494} \right) \times \left( {0.1333*{\text{slope}} - 3} \right)}};1} \right] $$
(3)
where G is stand basal area (m2 ha−1), mDBH is mean DBH (cm) and slope is related to the respective stand (°).

The indicator for landslide and erosion protection (LPI) builds on crown cover defined by projected crown area for trees with DBH > 5 cm (compare also Frehner et al. 2005). LPI is ordinally scaled with three categories (LPI1 < 30 % canopy cover, ≥30 % LPI2 < 60 %, LPI3 ≥ 60 %).

For details on indicator definition, see Bugmann et al. (this volume).

Assessment approach

BAU and NOM scenarios were each simulated under historic climate and five climate change scenarios, and model output of the simulated HUs (tree level) was mapped into a digital elevation model. Four spatial aggregation levels were defined for the calculation of the ES state indicators: (1) harvesting units (HU), (2) 1 ha, (3) 5 ha and (4) 10 ha grid cells (i.e. grain size; see Fig. 2).
Fig. 2

Three grain sizes for the assessment of ES indicators in the study landscape (indicated as black boxes for a 1 ha, n = 84; b 5 ha, n = 18; c 10 ha, n = 15). Dark grey area shows forested area and light grey non-forest area

The flow indicators TVH [total harvested volume (THV)], VI (mean periodic volume increment) and BBD (bark beetle damage) were only provided for the harvesting units and then aggregated at landscape level.

First, ES provisioning at landscape scale under historic climate and climate change conditions is presented. For calculation of ES indicators, the model output for the 18 HU was used. The indicators BHQ, API and LPI required a local spatial context and were therefore calculated as mean value from 1 ha samples (see Fig. 2).

Second, we tested the effect of grain size in estimating ES provisioning the landscape. The different grain sizes were also used to test for effects of analysis period and climate scenarios. ANOVA and Tukey tests were employed for continuous indicators. Shapiro–Wilks and Levene tests were used to test normality and heteroscedasticity of indicators. For ordinal indicators (BHQ, LPI), nonparametric Friedman and Wilcoxon tests were employed. To provide identical sample sizes for the Friedman test, 100 samples of 15 grid cells each were randomly drawn from the 1 ha and 5 ha pixels and compared to the 15 10 ha cells.

Third, the effect of grain size on ES indicator calculation was analysed to explore the joint provisioning of ES at different spatial scales (i.e. multifunctionality).

Results

Ecosystem service provisioning under historic climate and climate change scenarios

Under historic climate (c0) and BAU management, standing stock increased from 436.5 to 490.1 m3 ha−1 at the end of the simulation period (periodic mean in period P3) because (1) harvests remained clearly below the periodic increment [mean TVH of 1.9 m3 ha−1 year−1 over the entire analysis period versus a mean increment (VI) of 5.7 m3 ha−1 year−1] and (2) bark beetle induced tree mortality remained at relatively low level (0.5–0.65 m3 ha−1 year−1). In the NOM scenario without harvests, volume increased to 599.6 m3 ha−1 in P3 (Table 1). When forests are managed according to BAU, the increment increased under climate change scenarios c1, c3 and c4, while in c5 productivity decreased (Fig. 3) depending on the interplay of precipitation and temperature. In warmer climates, damages from bark beetle disturbances increased in BAU (448 % under c5; see Fig. 3) and in NOM (522 % under c5; not shown). Consequently, in combination with harvests and other natural tree mortality, standing stock decreased under all climate change scenarios (−15 % in P3 under BAU management in climate c5) except under climate scenario c1, which resulted in slight increases in P3 under both BAU and NOM (Fig. 3; Table 2).
Fig. 3

Impact of five climate change scenarios on ES indicators under BAU management in period P3 (2078–2110) in relation to historic climate (c0)

Table 2

Ecosystem service indicators for “business-as-usual” management (BAU) and the no-management regime (NOM) under climate change scenarios c1, c3 and c5 in period P3 (2078–2110)

Acronym

Explanation

Unit

BAU

NOM

c1

c3

c5

c1

c3

c5

Timber production

V

Standing volume living trees

m3 ha−1

497.3

488.1

418.44

609.8

599.2

520.8

THV

Annual volume harvested

m3 ha−1 year−1

2.1

2.1

1.8

0

0

0

VI

Annual net volume increment

m3 ha−1 year−1

7.4

7.5

5.9

6.7

6.8

5.4

BBD

Volume killed by bark beetles

m3 ha−1 year−1

1.4

1.6

2.9

1.4

1.6

2.9

Carbon sequestration

CS

Carbon (trees, standing deadwood, coarse woody debris and soil carbon)

t ha−1

223.8

221.4

206.1

254.3

251.4

234.2

Biodiversity

D

Tree species diversity

1.31

1.35

1.37

1.27

1.31

1.34

H

Tree size diversity (mean Shannon diversity of DBH and height)

5.51

5.48

5.56

4.87

4.83

4.95

AA

Basal area share of Abies alba

%

4.0

4.3

4.7

4.2

4.6

5.6

BL

Basal area share of broadleaves

%

3.7

4.3

4.4

2.7

3.2

3.7

SDWV

Standing deadwood volume (DBH > 20 cm)

m3 ha−1

27.6

29.2

35.9

32.2

34.1

42.1

LLTN

Large living trees (DBH > 50 cm)

n ha−1

48.4

47.1

40.7

65.06

63.4

55.4

BHQ

Bird habitat quality

%

[32/63/05]

[24/69/07]

[17/62/21]

[19/69/12]

[15/73/12]

[11/68/21]

Protection against gravitational hazards

API

Avalanche protection index

0–1

0.98

0.98

0.97

0.98

0.98

0.98

LPI

Landslide and erosion protection

%

[01/26/73]

[00/25/75]

[01/30/69]

[00/25/75]

[00/25/75]

[02/25/73]

Classification categories for BHQ and LPI: 1 = bad, 2 = moderate, 3 = good, provided in percentage of area for categories [1/2/3]. Indicator values for BHQ, API and LPI are based on 1 ha samples and all other indicators based on HU simulation entities

Closely correlated with volume are in situ carbon pools which increased in BAU under historic climate (c0). In the final assessment period P3, the NOM regime holds an additional amount of 30.2 t ha−1 compared with BAU (Table 1). Carbon storage was fairly insensitive to climate change scenarios c1–c4 under both management regimes (BAU, NOM). However, under the severe climate change scenario c5 the carbon pools decreased by −7.5 % in BAU as well as in NOM (Table 2).

Under historic climate, the shares of Silver fir and admixed broadleaved species were decreasing over time under BAU as well as NOM, whereas under all climate change scenarios the opposite trend occurred. In relative terms, Silver fir and broadleaves could benefit from more favourable growing conditions under the warming scenarios. As a consequence, while tree species diversity (D) for the entire landscape was declining in BAU and NOM under historic climate, it was increasing slightly under all climate change scenarios, strongest under scenarios c4 and c5 (Table 1, response under BAU management in Fig. 3). Under historic climate, tree size diversity was declining under BAU as well as NOM management regime (Table 1). Under climate change conditions, tree size diversity increased under BAU management due to newly regenerated trees in the patch cuts (see Fig. 3), while the effect of bark beetle damages alone was not sufficient to increase size diversity in the NOM scenario (Table 2).

Under historic climate, the bird habitat quality (BHQ) index improved and showed a shift towards categories BHQ2 (“moderate”) and BHQ3 (“good”) under BAU as well as NOM. The combined area shares in BHQ2 and BHQ3 increased under BAU from 49 % in P1 to 64 % in P3, under the NOM regime even more to 77 % (Table 1). Under the climate change conditions of c1 and c5, the area shares in P3 increased to 68 and 83 % (under BAU) and to 81 and 89 % (NOM), respectively, in period P3 (Table 2; Fig. 4). Climate change scenarios, particularly scenario c5, favoured habitat quality because of more standing deadwood of large dimensions and canopy openings due to bark beetle damages. In NOM without active management, standing deadwood volume (SDWV) increased to 28.8 m3 ha−1 (P3, historic climate; Table 1). Under climate change scenarios, SDWV increased strongly (compare Fig. 3 and Table 2). Under NOM, the number of large trees (LLTN) increased under all climate scenarios except c5 where natural tree mortality and bark beetle infestations resulted in a decrease in LLTN. Under BAU, harvests and bark beetle damages reduced LLTN (Table 1; Fig. 3), the latter particularly under conditions of climate change (Table 2).
Fig. 4

Effect of grain size on bird habitat quality (BHQ) and landslide and erosion protection index (LPI) in period P3 (2078–2110) under historic climate (c0) and strong warming (c5)

Landslide protection at the beginning of the simulation period was already well developed. In P1 under historic climate and BAU management, 93 % of the landscape area met the crown cover criterion for medium (LPI2) or good protection (LPI3) (Table 1). This share rose to 98 % in P3. Under the NOM regime, these shares were even higher, and in P3 100 % of the landscape area were classified as LPI3. The development was similar for protection against avalanches (API; see Table 1). Interestingly, both API and LPI were almost insensitive to climate change (compare Table 2).

Scale dependency of ecosystem service indicators

With three grain sizes, we explored the effect of the assessment scale on indicator estimates and spatial heterogeneity of indicators. Levene tests for heteroscedasticity of the ES indicators were non-significant and Shapiro–Wilks tests were significant for API, size diversity H and partly the large standing deadwood volume SDVW. After visual inspection of the data, we decided against transformation of variables because ANOVA is fairly insensitive to slight deviations from normality (McDonald 2014). ANOVAs for effects of grain size within periods and management regimes under historic climate were not significant (α = 0.05) for V, CS, D, SDVW, LLTN and API (compare data in Table 2). Figure 4 shows the area shares for the BHQ and LPI categories for each of the three grain sizes. For bird habitat, the share of categories BHQ1 (“poor”) decreases with increasing grain size, and for BHQ3 (“good”) it is the opposite. This pattern is independent of the used climate scenario and the management regime. For LPI, however, this effect of grain size is similar for the NOM regime but not as apparent under BAU management. In general, for ordinal BHQ and API indicators the share of significant Friedman tests (α = 0.05) for effects of grain size was small, but larger under the NOM regime compared with BAU (see Table 3). The percentage of significant Friedman tests under historic climate c0 increased from P1 (BAU: 5 % for BHQ, 1 % for LPI; NOM: 6 % for BHQ, 1 % for LPI; not shown) to P3 (BAU: 11 % for BHQ, 1 % for LPI; NOM: 19 % for BHQ, 8 % for LPI).
Table 3

Effect of grain size on ES indicators in BAU and NOM management under historic climate (c0) and climate change scenarios (c1, c5)

Indicator

Grain size

BAU

NOM

1 ha

5 ha

10 ha

1 ha

5 ha

10 ha

V

c0

406.1 ± 118.9

386.4 ± 80.1

398.6 ± 58.8

500.3 ± 142.3

471.5 ± 86.2

490.4 ± 58.7

c1

411.4 ± 123.8

387.6 ± 83.1

398.7 ± 64.8

508.1 ± 150.5

476.3 ± 94.1

493.2 ± 69.4

c5

345.7 ± 122.2

319.7 ± 83.1

331.9 ± 68.6

434.0 ± 154.7

397.6 ± 100.3

415.7 ± 81.0

CS

c0

226.6 ± 34.8

225.8 ± 23.6

227.1 ± 19.0

259.3 ± 43.0

252.2 ± 27.6

255.7 ± 21.2

c1

228.2 ± 36.1

226.5 ± 24.7

227.7 ± 20.8

261.2 ± 45.0

253.5 ± 30.1

256.7 ± 24.0

c5

212.4 ± 35.4

210.0 ± 25.2

212.1 ± 21.6

243.4 ± 45.0

234.6 ± 31.9

238.6 ± 26.6

D

c0

1.26 ± 0.39

1.32 ± 0.47

1.37 ± 0.41

1.25 ± 0.38

1.30 ± 0.43

1.35 ± 0.40

c1

1.34 ± 0.44

1.42 ± 0.53

1.47 ± 0.48

1.31 ± 0.42

1.37 ± 0.48

1.43 ± 0.45

c5

1.40 ± 0.47

1.50 ± 0.56

1.57 ± 0.54

1.37 ± 0.46

1.46 ± 0.53

1.53 ± 0.52

SDWV

c0

24.6 ± 7.3

23.5 ± 4.4

24.1 ± 2.9

29.5 ± 8.5

28.2 ± 4.8

28.8 ± 3.6

c1

28.1 ± 9.6

27.2 ± 5.6

27.8 ± 4.4

32.8 ± 10.2

31.4 ± 5.5

32.5 ± 4.6

c5

36.0 ± 12.6

35.1 ± 6.8

35.6 ± 5.2

42.8 ± 14.3

42.1 ± 7.8

42.2 ± 6.3

LLTN

c0

48.2 ± 21.9

46.3 ± 13.3

46.6 ± 12.6

64.9 ± 28.8

61.4 ± 16.4

62.2 ± 15.4

c1

47.7 ± 22.7

45.2 ± 13.8

45.4 ± 13.5

65.0 ± 29.9

60.8 ± 18.0

61.5 ± 17.1

c5

40.2 ± 21.2

37.3 ± 13.5

38.2 ± 13.0

55.5 ± 28.8

50.8 ± 17.9

52.0 ± 17.0

BHQ

c0

[36/64/00]1

[22/78/00]1

[13/87/00]1

[23/75/02]3

[11/89/00]3

[00/00/00]3

c1

[32/63/05]2

[06/94/00]2

[00/100/00]2

[19/69/12]4

[00/89/11]4

[00/93/07]4

c5

[17/62/21]

[06/83/11]

[0.00/0.87/0.13]

[11/68/21]5

[00/67/33]5

[00/60/40]5

API

c0

0.98 ± 0.06

1.00 ± 0.00

1 ± 0.00

0.98 ± 0.07

1.00 ± 0.00

1.00 ± 0.00

c1

0.98 ± 0.07

1.00 ± 0.00

1 ± 0.00

0.98 ± 0.06

1.00 ± 0.00

1.00 ± 0.00

c5

0.97 ± 0.10

0.99 ± 0.05

1 ± 0.00

0.98 ± 0.08

1.00 ± 0.01

1.00 ± 0.00

LPI

c0

[01/27/71]

[00/39/61]

[00/27/73]

[00/26/74]6

[00/17/83]6

[00/07/93]6

c1

[01/26/73]

[00/28/72]

[00/27/73]

[00/25/75]7

[00/11/89]7

[00/07/93]7

c5

[01/30/69]

[00/39/61]

[00/33/67]

[00/25/73]8

[00/28/72]8

[00/13/87]8

Indicator values are mean values for period P3 (2078–2110). BHQ and LPI provided in percentage of area for classification categories [1 = bad/2 = moderate/3 = good]. Bold entries indicate share of significant Friedman tests ≥5 % and share of significant Friedman tests: 1 11 %, 2 38 %, 3 19 %, 4 5 %, 5 12 %, 6 8 %, 7 9 %, 8 11 %

The aggregate BHQ index responded sensitive to the assessment scale because the relatively rare structural elements required for good habitat quality (large snags, large live trees) are covered much better with larger grain size. LPI, in contrast, did not show this sensible response as it builds on canopy cover as such and is independent of tree size distribution. Please note that landscape structure (forest, non-forest area) remains constant throughout the simulation. Small 1 ha samples covering non-forest area shares are rather insensitive to changes in forest attributes.

In general, the variability of indicators expressed as coefficient of variation (CV) decreased with increasing grain size (see Fig. 5). However, magnitude and temporal development of the variability differed among indicators. Not unexpectedly, the number of large living trees (LLTN) varied greatly when sampled with 1 ha squares, but the CV decreased in P2 and P3. Estimates of total carbon storage showed a decreasing CV over the three assessment periods almost independent of climate scenario. Tree species diversity showed very low contrasts between grain sizes and climate scenarios which is mainly due to the dominating role of Norway spruce. Admixed tree species are rare and do occur in spatial clusters depending on site conditions and regeneration fellings. Particularly, the fellings along the skyline tracks initiate regeneration of Silver fir, mountain maple and beech which become effective in later decades of the assessment period. With small grain size these rare events are obviously missed in most samples which leads to low variation in D, while with larger sample size admixed species were more frequently hit, thus increasing the CV in D, particularly under warming scenarios. Similarly, API values are favourable throughout the landscape, resulting in a very low variation which even decreased further along time (Fig. 5).
Fig. 5

Coefficient of variation (CV) for selected ES indicators over the three assessment periods P1–P3 in dependence of grain size and three climate scenarios (historic climate c0, climate change scenarios c1 and c5)

Grain size and multifunctionality

We decided to use carbon storage (CS), bird habitat provision (BHQ) and protection against avalanches (API) and landslides (LPI) to analyse the simultaneous provision of ecosystem services as these ES are represented unambiguously by one specific ES indicator. We categorized total carbon storage (category CS1: < 203 t ha−1, CS2: 203–332 t ha−1, CS3: >332 t ha−1) and API (API1: 0–0.33, API2: 0.34–0.66, API3: 0.67–1.0) in three performance classes (spread of simulated indicator values divided by 3) so that they were comparable with the ordinally scaled indicators BHQ and LPI. For each grain size, the share of samples with two, three and four ES indicators being simultaneously rated at least “moderate” (categories 2 or 3) was determined to represent the “multifunctional” share of the landscape.

Figure 6 shows several distinct patterns for period P3 where effects of management as well as climate change scenarios became visible: (a) the tendency that the multifunctional area in the case study landscape decreased with increasing number of considered ES. This effect was particularly visible when moving from portfolios of two to three ES; (b) under the NOM regime the multifunctional area shares were not as strongly decreasing with increasing number of considered ES; (c) with increasing scale (i.e. grain) the multifunctional area increased in general; and (d) the increase in multifunctional area under the moderate warming of scenario c1 (also having a favourable distribution of summer precipitation) was higher in BAU compared with NOM. In both BAU and NOM, the multifunctional area was smaller under the severe climate change scenario c5 compared with c1, and sometimes even smaller than under the historic reference climate c0.
Fig. 6

Multifunctional share of landscape area [bundles of ES in performance category 2 (moderately good) or 3 (good) in period P3 (2078–2110)] under historic climate (c0) and two climate change scenarios (c1, c5) under BAU (“business-as-usual”) management and the no-management regime (NOM). Symbols represent mean area percentage, and whiskers indicate worst and best case within given situation. Considered ES: carbon storage, bird habitat, protection against avalanches and landslides

The multifunctional share of the landscape under BAU management in period P3 when two ES were combined varied between 53 % (combination of CS and API under c5 at 5 ha grain size) and 100 % (several ES combinations, particularly the combined protective services against avalanches and landslides and erosion). In the NOM scenario, these shares increased to 77 % (CS & BHQ, LPI and BHQ under historic climate c0 at 1 ha grain size) and 100 % (all possible combinations of 2 ES, in different climates and grain size combinations).

The multifunctional share of the landscape combining three ES ranged from 53 (API, CS and BHQ under climate scenario c5 with 5 ha grain size) to 86 % (LPI, CS, BHQ under climate scenario c1 and 10 ha grain size) under BAU management. In unmanaged conditions (NOM), the least compatible ES combination increased the multifunctional area to 76 % (LPI, CS, BHQ) and the most complementary ES combination to 100 % of the landscape (all ES triplets under historic climate and grain sizes larger 1 ha), respectively.

When four ES had to be integrated at small scale, a minimum of 53 % (5 ha grain size, climate scenario c5) and a maximum of 86 % (10 ha grain size, climate scenario c1) of the managed landscape (BAU) were considered as multifunctional. NOM generated between 76 % (under c0 and c5 at 1 ha grain size) and 100 % (under historic climate c0 at 10 ha grain size) of multifunctional landscape area.

Analysis of the temporal development of multifunctionality in the study landscape under historic climate c0 revealed that from periods P1 to P3 the area with sufficient ES provisioning increased with sample sizes of 1 and 10 ha and decreased or showed no trend at all with 5 ha grain size (not shown).

Discussion and conclusion

In the presented study, we analysed the current “business-as-usual” (BAU) management regime and its effect on the provisioning of timber production, carbon sequestration, nature and habitat conservation and protection against gravitational hazards (avalanches, landslides) under historic climate and five climate change scenarios in a catchment of 250 ha in the Eastern Alps in Austria. Here, we scrutinize the analysis approach, the ability of BAU management to provide the demanded ES, the spatial heterogeneity of ES indicators within the case study landscape and finally the level of ES integration in providing multiple ecosystem services from the same parcel of forest.

Analysis approach

The PICUS forest ecosystem model has already proven its ability as a valuable tool for climate change impact studies (Maroschek et al. 2015; Seidl et al. 2011; Lexer et al. 2002). The model has been evaluated in several validation experiments and has built up substantial credibility for applications in European mountain forests (e.g. Huber et al. 2013; Seidl et al. 2011; Didion et al. 2009; Seidl et al. 2005). Given the uncertainty in the data given by Neumann (1993) and the complex stand structures in two of the three validation stands, the validation against three 30-year records of stand development in the current study context further supported the credibility of the model. Essentially the generic parameterization from earlier studies generated 30-year stand development trajectories which matched reconstructed time series of key stand attributes well. Also the simulated volume productivity in the application matched internal records of the forest owner closely. Model adjustments for the current study were confined to establishment and height increment potentials of tree seedlings. Due to intense competition by grass and herb species, regeneration dynamics were much slower in reality compared with initial model simulations. In a generic model, not every ecosystem component and each process can be explicitly covered. However, assuming that the vitality of grass and herb species will not change under future climatic conditions, the implicit parameterization via adjusted germination rates seemed to be a reasonable approach.

In the current contribution, the size of a simulation unit approximately 20 ha simultaneously simulated forest area. This extent of the simulation entities allowed a consistent implementation of felling regimes and disturbances by bark beetles (see Pasztor et al. 2014, 2015). Tree mortality caused by disturbance agents is highly influential regarding tree regeneration dynamics, habitat quality and protection against gravitational hazards. Storm damages have not been an important disturbance factor so far, and this is not expected to change in the future (Nikulin et al. 2011).

A further key feature for the analysis was the spatially explicit availability of a 250 ha forest landscape at the resolution of individual trees and 100 m2 patches. This imposes the challenge of realistic model initialization. Roces-Díaz et al. (2015) point at the importance of remote sensing data to (1) overcome the data bottleneck for ES assessments and (2) to improve the spatial representation of ES supply. The employed forest initialization and projection approach in the current study supports both issues. For larger assessment areas, it would otherwise be impossible to provide realistic initial states for model simulations. The ability to map the trees from a set of simulation units into a digital terrain model offers huge potential to analyse structural and compositional features of a forested landscape (Maroschek et al. 2015). In the current study, we utilized this model feature to assess the effect of different grain sizes on the estimation of ES indicators and thus to shed light on the level of ES integration in a Central European mountain forest landscape.

According to Villa et al. (2014), ES assessment methods must be quantitative and scalable. The ES indicators used in the presented study are transparent and quantitative and allow the analysis of trade-offs among ES. The spatial simulation and assessment approach can be scaled continuously from tree to landscape level. To isolate the effect of management as such a non-intervention regime (NOM) has also been implemented and used for comparison with the “business-as-usual” (BAU) regime. We thus concluded that the assessment approach was well suited for the research objectives at hand.

Landscape-level ES provisioning under BAU management

BAU management and the related cutting intensity have led to increasing volume stocks. From the mean annual increment of 5.7 m3 ha−1 year−1, about 33 % were harvested. Another 10 % were lost to bark beetle induced tree mortality which, at mortality rates of 0.58 m3 ha−1 year−1, could not be salvaged at reasonable costs and therefore largely remained in the forest. At the end of the 100-year simulation period, the mean standing volume stock of living trees was over 600 m3 ha−1 with the clear tendency to increase further. On the one hand, this increased the in situ carbon pools. On the other hand, however, it also increased the risk of damage through disturbances (Pasztor et al. 2014) which may then in turn negatively feedback on the climate change mitigation effect via in situ carbon storage. Due to the low felling intensity, avalanche (API) and landslide protection (LPI) improved over the 100-year simulation period to sufficient levels at almost the entire landscape area. This did not change significantly under climate change conditions. However, the potential for negative climate change impacts has been revealed under the severe climate change scenario c5 and the related increase of bark beetle damage to more than 500 % compared with historic climate under the NOM regime. It should be noted that in the current study the avalanche protection index did not include gap formation by disturbances or management in explicit form (compare Maroschek et al. 2015) and may therefore overestimate the protection effect. However, the 1 ha grain size used to calculate the API and LPI indicators can be considered as suitable for the detection of protection deficiencies.

The bird habitat quality in initial phases of the analysis period is insufficient (bird habitat quality category BHQ1) on about one-third of the landscape. However, it must be noted that this is mostly due to missing large snags which could not be realistically initialized from available information. From period P2 onward, simulation results indicated that simulated tree mortality generates realistic numbers of snags in the forest.

Effects of scale and multifunctionality

As a general pattern, our analysis revealed the largest variation for 1 ha, smallest for 10 ha grain size and a decrease in spatial variation over time for most ES indicators, independent of grain size. The fine-grained cutting pattern of BAU management homogenized most ES indicators in course of the twenty-first century, after an initial increase in spatial variation. In the NOM regime, this homogenizing effect is less visible. Of course, these results cannot be generally extrapolated to any other landscape because they largely depend on the interplay of harvesting pattern and major ecosystem processes (growth, regeneration, mortality). However, such studies are rarely reported in the literature (compare Grêt-Regamey et al. 2014) but can shed light on the importance of spatial scale in ES assessments.

The set of ES considered in the presented study is typical for European mountain forests: timber production, nature and habitat conservation and protection against gravitational hazards. In situ carbon sequestration may be a general public interest as well. In our case study area, BAU management sacrificed timber harvests in favour of carbon sequestration, protection against gravitational hazards and bird habitat provisioning. To which extent the protective services would be affected by intensified harvests is not easy to guess and would require the extended analysis of alternative silvicultural regimes. In our study, the most severe trade-offs seem to be related to timber harvesting. In general, the NOM regime showed better performance in all non-timber ES throughout the entire simulation period. The reduction in multifunctional area when the number of demanded ES is increased from two to four is surprisingly low (from about 80 to 60 % under BAU management and historic climate). Under the NOM regime, this area reduction is even smaller and may show no effect at all. However, what must be noted is that the share of admixed species is further reduced in NOM and that no tree regeneration is initialized. Thus, the mean tree age is increasing and in the long run NOM may enter a phase of reduced ES provisioning and lower resilience compared to BAU.

Compared with the BAU regime, the “multifunctional” area share was in general larger in NOM. Whether the non-timber ES are neutral or even complementary over longer time periods depends on several issues: first, the thresholds that define “sufficient level of ES provisioning”. In our analysis, we used categories of equal width between minimum and maximum values of each ES indicator. In a real decision-making situation, this will depend mainly on stakeholder preferences. Second, silvicultural regimes can be tailored locally to meet the requirements of demanded ES portfolios. For instance, Boncina (2011) discussed approaches to integrate nature conservation values in forest management. Fuhr et al. (2015) show that protection against gravitational hazards can be provided also by ageing forests as long as forest structure is irregular and patchy.

With no explicit benchmarks available, it is difficult to qualify BAU management in the study area. It is obvious that timber production could be intensified. Protective effects against avalanche release, landslides and erosion appear as sufficient for large shares of the landscape. Habitat quality for various protected bird species is already maintained by current management on vast shares of the analysed landscape. The remaining question is whether it is possible to intensify timber harvests without jeopardizing (1) the other demanded ES and (2) the adaptation of the forest to conditions of climatic changes. Whether such a management approach can be identified will be the focus of further work. In the long run, increasing the shares of broadleaves and conifers which are less vulnerable under warmer climates would be a useful means to increase the resistance against disturbance agents. Detrimental to this adaptive management strategy was the high browsing pressure in the area which actually renders any significant shift in species composition as highly unrealistic.

Notes

Acknowledgments

We are grateful to the Forstfonds Stand Montafon for making internal information and data available and to Hubert Malin and Bernhard Maier for support and their interest in the study. The presented work was financially supported by the EU FP7 ARANGE project under Grant No. KBBE-289437.

Supplementary material

10113_2015_908_MOESM1_ESM.pdf (456 kb)
Supplementary material 1 (PDF 455 kb)

References

  1. Boncina A (2011) Conceptual approaches to integrate nature conservation into forest management: a Central European perspective. Int For Rev 13:13–22. doi:10.1505/ifor.13.1.13 Google Scholar
  2. Bußjäger P (2007) zu Luxusbauten wird kein Holz verabfolgt! - Die Geschichte des Forstfonds des Standes Montafon. In: Malin H, Maier B, Dönz-Breuß M (eds) Montafoner Standeswald - Montafoner Schriftenr. 18. Heimatschutzverein Montafon, Schruns, pp 9–24Google Scholar
  3. Côté P, Tittler R, Messier C, Kneeshaw DD, Fall A, Fortin MJ (2010) Comparing different forest zoning options for landscape-scale management of the boreal forest: possible benefits of the TRIAD. For Ecol Manag 259:418–427. doi:10.1016/j.foreco.2009.10.038 CrossRefGoogle Scholar
  4. Didion M, Kupferschmid AD, Lexer MJ, Rammer W, Seidl R, Bugmann H (2009) Potentials and limitations of using large-scale forest inventory data for evaluating forest succession models. Ecol Model 220:133–147. doi:10.1016/j.ecolmodel.2008.09.021 CrossRefGoogle Scholar
  5. Dorren LK, Berger F, Imeson AC, Maier B, Rey F (2004) Integrity, stability and management of protection forests in the European Alps. For Ecol Manag 195:165–176. doi:10.1016/j.foreco.2004.02.057 CrossRefGoogle Scholar
  6. European Environment Agency (2010) Europe’s ecological backbone: recognising the true value of our mountains. doi:10.2800/43450
  7. Frehner M, Wasser B, Schwitter R (2005) Nachhaltigkeit und Erfolgskontrolle im Schutzwald. Wegleitung für Pflegemassnahmen in Wäldern mit Schutzfunktion. Bundesamt für Umwelt, Wald und Landschaft, BernGoogle Scholar
  8. Fries C, Carlsson M, Dahlin B, Lämås T, Sallnäs O (1998) A review of conceptual landscape planning models for multiobjective forestry in Sweden. Can J For Res 28:159–167. doi:10.1139/x97-204 CrossRefGoogle Scholar
  9. Fuhr M, Bourrier F, Cordonnier T (2015) Protection against rockfall along a maturity gradient in mountain forests. For Ecol Manag 354:224–231. doi:10.1016/j.foreco.2015.06.012 CrossRefGoogle Scholar
  10. Grabherr G (2000) Biodiversity of mountain forests. In: Price MF, Butt N (eds) Forests in sustainable mountain development: a state of knowledge report for 2000. Task Force on Forests in Sustainable Mountain Development. CABI, Wallingford, pp 28–51. doi:10.1079/9780851994468.0028 CrossRefGoogle Scholar
  11. Grêt-Regamey A, Weibel B, Bagstad KJ, Ferrari M, Geneletti D, Klug H, Schirpke U, Tappeiner U (2014) On the effects of scale for ecosystem services mapping. PLoS ONE. doi:10.1371/journal.pone.0112601 Google Scholar
  12. Hanewinkel M, Cullmann DA, Schelhaas M-J, Nabuurs G-J, Zimmermann NE (2012) Climate change may cause severe loss in the economic value of European forest land. Nat Clim Chang 3:203–207. doi:10.1038/nclimate1687 CrossRefGoogle Scholar
  13. Hewitt CD, Griggs DJ (2004) Ensembles-based predictions of climate changes and their impacts. EOS Trans AGU 85(52):566CrossRefGoogle Scholar
  14. Hollaus M, Wagner W, Eberhöfer C, Karel W (2006) Accuracy of large-scale canopy heights derived from LiDAR data under operational constraints in a complex alpine environment. ISPRS J Photogramm Remote Sens 60:323–338. doi:10.1016/j.isprsjprs.2006.05.002 CrossRefGoogle Scholar
  15. Hollaus M, Wagner W, Maier B, Schadauer K (2007) Airborne laser scanning of forest stem volume in a mountainous environment. Sensors 7:1559–1577. doi:10.3390/s7081559 CrossRefGoogle Scholar
  16. Huber MO, Eastaugh CS, Gschwantner T, Hasenauer H, Kindermann G, Ledermann T, Lexer MJ, Rammer W, Schörghuber S, Sterba H (2013) Comparing simulations of three conceptually different forest models with National Forest Inventory data. Environ Model Softw 40:88–97. doi:10.1016/j.envsoft.2012.08.003 CrossRefGoogle Scholar
  17. Jacobsen JB, Vedel SE, Thorsen BJ (2013) Assessing costs of multifunctional NATURA 2000 management restrictions in continuous cover beech forest management. Forestry 86:575–582. doi:10.1093/forestry/cpt023 CrossRefGoogle Scholar
  18. Jost L (2007) Partitioning diversity into independent alpha and beta components. Ecology 88:2427–2439. doi:10.1890/06-1736.1 CrossRefGoogle Scholar
  19. Kaljonen M, Primmer E, De Blust G, Nijnki M, Kulvik M (2007) Multifunctionality and biodiversity conservation—institutional challenges. In: Chmelievski T (ed) Nature conservation management: from idea to practical issues. PWZN Print6, Lublin, pp 53–69 Google Scholar
  20. Lämås T, Eriksson LO (2003) Analysis and planning systems for multiresource, sustainable forestry: the Heureka research programme at SLU. Can J For Res 33:500–508. doi:10.1139/x02-213 CrossRefGoogle Scholar
  21. Landsberg JJ, Waring RH (1997) A generalised model of forest productivity using simplified concepts of radiation-use efficiency, carbon balance and partitioning. For Ecol Manag 95:209–228. doi:10.1016/S0378-1127(97)00026-1 CrossRefGoogle Scholar
  22. Lexer MJ, Hönninger K (2001) A modified 3D-patch model for spatially explicit simulation of vegetation composition in heterogeneous landscapes. For Ecol Manag 144:43–65. doi:10.1016/S0378-1127(00)00386-8 CrossRefGoogle Scholar
  23. Lexer MJ, Hönninger K, Scheifinger H, Matulla C, Groll N, Kromp-Kolb H, Schadauer K, Starlinger F, Englisch M (2002) The sensitivity of Austrian forests to scenarios of climatic change: a large-scale risk assessment based on a modified gap model and forest inventory data. For Ecol Manag 162:53–72. doi:10.1016/S0378-1127(02)00050-6 CrossRefGoogle Scholar
  24. Lindner M, Maroschek M, Netherer S, Kremer A, Barbati A, Garcia-Gonzalo J, Seidl R, Delzon S, Corona P, Kolström M, Lexer MJ, Marchetti M (2010) Climate change impacts, adaptive capacity, and vulnerability of European forest ecosystems. For Ecol Manag 259:698–709. doi:10.1016/j.foreco.2009.09.023 CrossRefGoogle Scholar
  25. Malin H, Lerch T (2007) Schutzwaldbewirtschaftung im Montafon. In: Malin H, Maier B, Dönz-Breuß M (eds) Montafoner Standeswald - Montafoner Schriftenr. 18. Heimatschutzverein Montafon, Schruns, pp 115–128Google Scholar
  26. Malin H, Maier B (2007) Der Wald - Das grüne Rückgrat des Montafon. In: Malin H, Maier B, Dönz-Breuß M (eds) Montafoner Standeswald - Montafoner Schriftenreihe 18. Heimatschutzverein Montafon, Schruns, pp 91–114Google Scholar
  27. Maroschek M, Rammer W, Lexer MJ (2015) Using a novel assessment framework to evaluate protective functions and timber production in Austrian mountain forests under climate change. Reg Environ Change 15:1543–1555. doi:10.1007/s10113-014-0691-z CrossRefGoogle Scholar
  28. Mayer DG, Stuart MA, Swain AJ (1994) Regression of real-world data on model output: an appropriate overall test of validity. Agric Syst 45:93–104. doi:10.1016/S0308-521X(94)90282-8 CrossRefGoogle Scholar
  29. McDonald J (2014) Handbook of biological statistics, 3rd edn. Sparky House Publishing, BaltimoreGoogle Scholar
  30. Neumann M (1993) Increment research on spruce at different altitudes in the Austrian Central Alps. Cent für das gesamte Forstwes 110:221–274Google Scholar
  31. Niese G (2011) Österreichs Schutzwälder sind total überaltert. BFW Praxisinformation 24:29–31Google Scholar
  32. Nijnik M, Nijnik A, Lundin L, Staszewski T, Postolache C (2010) A study of stakeholders’ perspectives on multi-functional forests in Europe. For Trees Livelihoods 19:341–358. doi:10.1080/14728028.2010.9752677 CrossRefGoogle Scholar
  33. Nikulin G, Kjellström E, Hansson U, Strandberg G, Ullerstig A (2011) Evaluation and future projections of temperature, precipitation and wind extremes over Europe in an ensemble of regional climate simulations. Tellus A 63:41–55. doi:10.1111/j.1600-0870.2010.00466.x CrossRefGoogle Scholar
  34. Pasztor F, Matulla C, Rammer W, Lexer MJ (2014) Drivers of the bark beetle disturbance regime in Alpine forests in Austria. For Ecol Manag 318:349–358. doi:10.1016/j.foreco.2014.01.044 CrossRefGoogle Scholar
  35. Pasztor F, Matulla C, Zuvela-Aloise M, Rammer W, Lexer MJ (2015) Developing predictive models of wind damage in Austrian forests. Ann For Sci 72:289–301. doi:10.1007/s13595-014-0386-0 CrossRefGoogle Scholar
  36. Peng C (2000) Understanding the role of forest simulation models in sustainable forest management. Environ Impact Assess Rev 20:481–501. doi:10.1016/S0195-9255(99)00044-X CrossRefGoogle Scholar
  37. Raudsepp-Hearne C, Peterson GD, Bennett EM (2010) Ecosystem service bundles for analyzing tradeoffs in diverse landscapes. Proc Natl Acad Sci 107:5242–5247. doi:10.1073/pnas.0907284107 CrossRefGoogle Scholar
  38. Roces-Díaz JV, Díaz-Varela RA, Álvarez-Álvarez P, Recondo C, Díaz-Varela ER (2015) A multiscale analysis of ecosystem services supply in the NW Iberian Peninsula from a functional perspective. Ecol Indic 50:24–34. doi:10.1016/j.ecolind.2014.10.027 CrossRefGoogle Scholar
  39. Seidl R, Lexer MJ, Jäger D, Honninger K (2005) Evaluating the accuracy and generality of a hybrid patch model. Tree Physiol 25:939–951. doi:10.1093/treephys/25.7.939 CrossRefGoogle Scholar
  40. Seidl R, Rammer W, Jäger D, Currie WS, Lexer MJ (2007) Assessing trade-offs between carbon sequestration and timber production within a framework of multi-purpose forestry in Austria. For Ecol Manag 248:64–79. doi:10.1016/j.foreco.2007.02.035 CrossRefGoogle Scholar
  41. Seidl R, Rammer W, Jäger D, Lexer MJ (2008) Impact of bark beetle (Ips typographus L.) disturbance on timber production and carbon sequestration in different management strategies under climate change. For Ecol Manag 256:209–220. doi:10.1016/j.foreco.2008.04.002 CrossRefGoogle Scholar
  42. Seidl R, Rammer W, Lexer MJ (2009) Estimating soil properties and parameters for forest ecosystem simulation based on large scale forest inventories [Schätzung von Bodenmerkmalen und Modellparametern fur die Waldokosystemsimulation auf Basis einer Großrauminventur]. Allg Forst- und Jagdzeitung 180:35–44Google Scholar
  43. Seidl R, Rammer W, Bellos P, Hochbichler E, Lexer MJ (2010) Testing generalized allometries in allocation modeling within an individual-based simulation framework. Trees-Struct Funct 24:139–150. doi:10.1007/s00468-009-0387-z CrossRefGoogle Scholar
  44. Seidl R, Rammer W, Lexer MJ (2011) Adaptation options to reduce climate change vulnerability of sustainable forest management in the Austrian Alps. Can J For Res 41:694–706. doi:10.1139/x10-235 CrossRefGoogle Scholar
  45. Suda M, Pukall K (2014) Multifunktionale Forstwirtschaft zwischen Inklusion und Extinktion (Essay). Schweizerische Zeitschrift fur Forstwes 165:333–338CrossRefGoogle Scholar
  46. Thornton PE, Running SW (1999) An improved algorithm for estimating incident daily solar radiation from measurements of temperature, humidity, and precipitation. Agric For Meteorol 93:211–228. doi:10.1016/S0168-1923(98)00126-9 CrossRefGoogle Scholar
  47. Vanclay JK, Skovsgaard JP (1997) Evaluating forest growth models. Ecol Model 98:1–12. doi:10.1016/S0304-3800(96)01932-1 CrossRefGoogle Scholar
  48. Villa F, Voigt B, Erickson JD (2014) New perspectives in ecosystem services science as instruments to understand environmental securities. Philos Trans R Soc Lond B Biol Sci 369:20120286. doi:10.1098/rstb.2012.0286 CrossRefGoogle Scholar
  49. Woltjer M, Rammer W, Brauner M, Seidl R, Mohren GMJ, Lexer MJ (2008) Coupling a 3D patch model and a rockfall module to assess rockfall protection in mountain forests. J Environ Manag 87:373–388. doi:10.1016/j.jenvman.2007.01.031 CrossRefGoogle Scholar
  50. Wu J, Shen W, Sun W, Tueller PT (2002) Empirical patterns of the effects of changing scale on landscape metrics. Landsc Ecol 17:761–782. doi:10.1023/A:1022995922992 CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2015

Authors and Affiliations

  • Florian Irauschek
    • 1
  • Werner Rammer
    • 1
  • Manfred J. Lexer
    • 1
  1. 1.Institute of SilvicultureUniversity of Natural Resources and Life Sciences, Vienna (BOKU)ViennaAustria

Personalised recommendations