Scholarly article on topic 'Future landscapes of Switzerland: Risk areas for urbanisation and land abandonment'

Future landscapes of Switzerland: Risk areas for urbanisation and land abandonment Academic research paper on "Earth and related environmental sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Applied Geography
Keywords
{"Land use change" / Scenarios / Dyna-CLUE / Modelling / "Spatially explicit"}

Abstract of research paper on Earth and related environmental sciences, author of scientific article — Bronwyn Price, Felix Kienast, Irmi Seidl, Christian Ginzler, Peter H. Verburg, et al.

Abstract In Switzerland, the decreasing significance of agriculture has led to prominent processes of land abandonment in mountainous areas where the maintenance of open land relies on human intervention. At the same time, urbanisation in Switzerland is increasing at a rapid rate at the expense of other land use types, particularly open land agriculture. In spite of these observed trends, the extent and location of anticipated land-use changes for the coming 20 years remain unknown, as does the impact on landscape services. This research defines 5 scenarios of future land-use for Switzerland along axes of Globalisation to Regionalisation and Market-driven developments to high policy intervention. Using the Dyna-CLUE land use modelling framework we incorporate socio-economic and bio-geographical variables to model scenarios of land change for 2035. By identifying locations for key land use transitions which occur across several scenarios, we find that unless large scale policy interventions are made, large areas of the Swiss Plateau and Alpine valley bottoms face strong urbanisation and much of the mountainous pasture agriculture continues to face risk of abandonment.

Academic research paper on topic "Future landscapes of Switzerland: Risk areas for urbanisation and land abandonment"

ELSEVIER

Contents lists available at ScienceDirect

Applied Geography

journal homepage: www.elsevier.com/locate/apgeog

Future landscapes of Switzerland: Risk areas for urbanisation and land abandonment

Bronwyn Price a *, Felix Kienasta, Irmi Seidl b, Christian Ginzler a, Peter H. Verburg c, Janine Bolliger a

a Swiss Federal Institute for Forest, Snow and Landscape Research WSL, Landscape Dynamics Unit, Zurcherstrasse 111, CH-8903 Birmensdorf, Switzerland b Swiss Federal Institute for Forest, Snow and Landscape Research WSL, Economics and Social Sciences Unit, Zurcherstrasse 111, CH-8903 Birmensdorf, Switzerland

c Institute for Environmental Studies, VU University Amsterdam, De Boelelaan 1087, 1081 HV Amsterdam, The Netherlands

ARTICLE INFO

ABSTRACT

Article history: Available online

Keywords: Land use change Scenarios Dyna-CLUE Modelling Spatially explicit

In Switzerland, the decreasing significance of agriculture has led to prominent processes of land abandonment in mountainous areas where the maintenance of open land relies on human intervention. At the same time, urbanisation in Switzerland is increasing at a rapid rate at the expense of other land use types, particularly open land agriculture. In spite of these observed trends, the extent and location of anticipated land-use changes for the coming 20 years remain unknown, as does the impact on landscape services. This research defines 5 scenarios of future land-use for Switzerland along axes of Globalisation to Regionalisation and Market-driven developments to high policy intervention. Using the Dyna-CLUE land use modelling framework we incorporate socio-economic and bio-geographical variables to model scenarios of land change for 2035. By identifying locations for key land use transitions which occur across several scenarios, we find that unless large scale policy interventions are made, large areas of the Swiss Plateau and Alpine valley bottoms face strong urbanisation and much of the mountainous pasture agriculture continues to face risk of abandonment.

© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction

In many parts of Europe including Switzerland, the decreasing significance of agriculture has led to prominent processes of land abandonment in mountainous areas (Lambin & Meyfroidt, 2011; Melendez-Pastor, Hernandez, Navarro-Pedreno, & Gomez, 2014; Stoate et al., 2009). The maintenance of open land in these mountainous areas often relies on human intervention (Wu, 2006), funded through agricultural or conservation subsidies. The loss of open land habitat also results in gains in other (usually forested) habitat and likely has significant consequences for species diversity and composition, and can be seen as a threat or opportunity depending on the conservation focus of a region (Queiroz, Beilin, Folke, & Lindborg, 2014).

* Corresponding author. Tel.: +41 44 739 28 19. E-mail addresses: bronwyn.price@wsl.ch (B. Price), felix.kienast@wsl.ch (F. Kienast), irmi.seidl@wsl.ch (I. Seidl), christian.ginzler@wsl.ch (C. Ginzler), peter.verburg@vu.nl (P.H. Verburg), janine.bolliger@wsl.ch (J. Bolliger).

Urbanisation and urban sprawl are key land use change processes in much of the world, including Switzerland (Jaeger & Schwick, 2014), where urban and settlement area increased by 23.4% between 1984 and 2009 (SFSO, Swiss Federal Statistical Office 2013a). Increased urban area occurs at the expense of other land use types, particularly open land agriculture, and therefore has implications for provision of ecosystem services, biodiversity and functions of landscape (Jaeger & Schwick, 2014).

Modelling of potential future land use changes by means of explorative scenarios offers researchers, land managers and policy and decision makers the opportunity to examine what change processes could occur in a given system under a set of defined conceivable conditions (Verburg, Eickhout, & van Meijl, 2008). Such scenario studies allow identification of uncertainties of a system and highlight the direction, magnitude and possible spatial patterns of likely land use changes, and the base from which to examine implications for system functioning and service provision (Verburg et al., 2008).

While several studies have considered scenarios of land use change for Switzerland, these studies have either been for small case study areas with very specific land use changes (van Strien

http://dx.doi.org/10.1016/j.apgeog.2014.12.009

0143-6228/© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

et al., 2014), or only explicitly considered single land-use change processes, such as urban sprawl (Jaeger & Schwick, 2014), agricultural land abandonment (Bolliger, Kienast, Soliva, & Rutherford, 2007), or agricultural land use change (Rounsevell, Ewert, Reginster, Leemans, & Carter, 2005). In addition, in general, European-wide studies such as EURURALIS and VOLANTE have explicitly excluded Switzerland (Verburg et al., 2008; Westhoek, van den Berg, & Bakkes, 2006).

In this paper we describe storylines for five scenarios of land use change for Switzerland along axes of Globalisation to Regionalisation and Market-driven to high levels of policy intervention. Using the Dyna-CLUE framework (Verburg & Overmars, 2009) we model the scenarios in a spatially explicit manner and produce five

maps of potential future land use for Switzerland in 2035. We identify hotspots for key land use change processes of urbanisation and agricultural land abandonment. We explore the extent to which planning restrictions and policy intervention can have a controlling influence on land use change processes.

Methods

Study area

The study area of Switzerland is a landlocked European country of approximately 4.1milion ha, with altitudes ranging from 196 m to 4634 masl (SFSO, 1997). Switzerland can be divided into five

Fig. 1. a) Swiss biogeographical regions and land use in 2009 as per the Swiss land use statistics (SFSO, 2013a). b)—f) Modelled land use changes for scenarios Trend, Al, A2, B1, B2 showing location of agricultural abandonment, urbanisation, reforestation and agricultural change (from one agricultural use to another agricultural use).

Table 1

Class aggregation from Swiss land use statistics.

Aggregated class

Swiss land use statistics (2004 nomenclature)

Closed forest

Open forest Overgrown Pasture agriculture

Intensive agriculture Settlement and Urban

Normal dense forest, Forest strips, Felling areas, Damaged forest areas, Groves and hedges, Clusters of trees

Reforestations, Open forest on unproductive areas, Brush forest, Scrub vegetation Brush meadows and farm pastures, Brush alpine pastures

Meadows, Farm pastures, Alpine meadows, Favourable alpine pastures, Rocky alpine pastures, Sheep pastures, Open forest on agricultural areas, Clusters of trees on agricultural areas

Arable land, Intensive orchards, Field fruit trees, Vineyards, Horticulture Industrial and commercial buildings, Surroundings of Industrial and commercial buildings, One and two family houses, Surroundings of one and two family houses, Terraced houses, Surroundings of terraces houses, Blocks of flats, Surroundings of blocks of flats, Public buildings, Surroundings of public buildings, Agricultural buildings, Surroundings of agricultural buildings, Unspecified buildings, Surroundings of unspecified buildings, Motorways, Green motorway environs, Roads and paths, Green road environs, Parking areas, Sealed railway areas, Green railways environs, Airports, Airfields green airport environs, Energy supply plants, Waste water treatment plants, Other supply or waste treatment plants, Dumps, Quarries and mines, Construction sites, unexploited urban areas, Public parks, Sports facilities, Golf courses, Camping areas, Garden allotments, Cemeteries Lakes, Rivers, Flood protection structures, Unproductive grass and scrubs, Avalanche and rockfall barriers, Wetlands, Alpine sports facilities, Rocks, Screes and sand, Landscape interventions, Glaciers and perpetual snow

distinct biogeographical regions: The Plateau with low elevation oceanic climate, The Northern Alps and Jura with oceanic high elevation climate, Central Alps with intra-alpine and continental climate and the Southern Alps with an insubrian climate (Gonseth,

Wohlgemuth, Sansonnens, & Buttler, 2001) (Fig. 1a). Switzerland has a long history of human habitation which defines landscapes and land use patterns. Urban and settled area covers 7.5% of the surface area, agriculture 35.9%, Wooded areas 31.3% and unproductive areas 25.3% (SFSO, 2013a). The Swiss landscape is continually changing with trends over the last 25 years towards urban expansion, particularly in the lowlands, and, largely in Alpine areas, expansion of wooded areas. Both of these processes occurring mostly at the expense of agricultural land (SFSO, 2013a). Switzerland has a strong economy and close relations with the EU (although is not an EU member state) and a history of moving between liberalisation and regulation movements (Mach, Hausermann, & Papadopoulos, 2003). Switzerland has a system of Federalism and direct democracy, and is made up of 26 cantons as member states. Implementation of spatial planning is largely the responsibility of the individual cantons or communities, with federal level framework legislation and coordination and a general requirement for protection of agricultural cultivable land, conservation of natural landscapes and integration of settlements into the landscape (Swiss Federal Assembly, 2012).

Approach

A scenario-based approach is taken in order to investigate the magnitude and location of likely land-use changes in Switzerland, in consideration of key global-scale drivers with uncertain future development trajectories. Future land-use pattern is modelled with the spatially explicit land-use allocation modelling framework CLUE (Conversion of Land Use and its Effects), Dyna-CLUE version, (Verburg & Overmars, 2009) for the year 2035, a time frame relevant to planning and policy horizons and matching to current population growth scenarios for Switzerland (at the cantonal level). Local to regional scale drivers of land use are considered in determining suitability for a given land use type and spatial allocation of land use changes. National level socio-economic and demographics are over-arching driving forces. Dyna-CLUE has several input requirements: initial state of the study area, spatially explicit location suitability per land use type, land-use specific conversion settings, and total area demand for land use type. The initial state is defined by the Swiss land use statistics valid for the year 2009.

Table 2

Explanatory variables for land use suitability modelling. Variable Source Spatial resolution Abbreviation

Climate

Continentality index (Gams) CSD/DEM25 (Zimmermann & Kienast, 1999) 1 ha cont

Yearly moisture index CSD/DEM25 (Zimmermann & Kienast, 1999) 1 ha mois

March global solar radiation MeteoSwiss 2010 (Durr, Zelenka, Muller, & Philipona, 2010) 1.25° minutes radi

May average precipitation MeteoSwiss 2010 (Schwarb, Daly, Frei, & Schar, 2001) 1 km prec

Relief

Slope DEM100 (SFSO, 1992) 1 ha slope

Elevation DEM100 (SFSO, 1992) 1 ha elev

Cos of Aspect DEM100 (SFSO, 1992) 1 ha c_asp

Sine of Aspect Soil DEM100 (SFSO, 1992) 1 ha s_asp

Soil permeability Soil suitability maps, (SFSO, 1992) Vector 1:200 000 soil_p

Soil stoniness Soil suitability maps, (SFSO, 1992) Vector 1:200 000 soil_st

Soil suitability for agriculture/forestry Soil suitability maps, (SFSO, 1992) Vector 1:200 000 soil_suit

Infrastructure/neighbourhood

Distance to major roads Swiss TLM (swisstopo, 2013) Vector 1:25 000 road_d

Distance to forest Swiss land use statistics (SFSO, 2013a) 1 ha for_d

No. of closed forest neighbours Swiss land use statistics (SFSO, 2013a) NA for_nn

Socioeconomic

Taxable income per taxpayer (SFSO, 2013b) Municipality tax

Percentage working residents (SFSO, 2013b) Municipality p_emp

employed in the primary sector

Public Transport accessibility Swiss Federal Office for Spatial Development (ARE, 2013) 1 ha pt

Land use statistics

The Swiss land-use/land cover statistics are interpreted by the Swiss Federal Statistical Office (SFSO) from aerial photography obtained in three time periods with a periodicity of 12 years 1979—1985, 1992—1997 and 2004—2009 (SFSO, 2013a). We have aggregated the available 72 categories of land use/land cover (SFSO, 2013a) into 6 land use classes which we consider affected by key land use change processes in Switzerland: Closed forest, Open Forest, Overgrown areas, Pasture agriculture, Intensive Agriculture and Settlement and Urban areas, and have not considered the remainder ('Other') (Table 1). Aggregation for Settlement and Urban area and Unproductive (Other) followed the standard aggregation set by the SFSO (2013a). Using the detailed class description provided by the SFSO (2013a), distinction was made between scrub and shrub vegetation that could be considered the stable natural vegetation type and that which is likely a succession stage between abandoned land and forested. Only the later was included in the class 'Overgrown (the former in the no data class 'Other'). Distinctions between forest and agricultural land were defined according to land use, so that trees on agricultural land were defined as agriculture. The distinction between open forest and closed forest largely followed the SFSO (2013a) definition based on tree height and crown cover (closed forest is >3 m tall with >60% crown cover), with the exception of areas that would meet this criteria but for temporary damage, i.e. from logging or storm damages. These areas were defined as closed forest (the SFSO defines them as open forest).

Explanatory variables

A suite of biophysical, infrastructure and spatially explicit socioeconomic variables were used as the basis of suitability models for each land use type. All relevant data available Swiss-wide was examined and reduced to a final set of 17 explanatory variables (Table 2) based on previous work modelling influence factors for urbanisation in Switzerland (Willhauck, 2013), and land cover change transitions in Switzerland (Rutherford, Bebi, Edwards, & Zimmermann, 2008, Rutherford, Guisan, & Zimmermann, 2007), as well as univariate linear regression, and cross-correlation analysis to prevent inclusion of variable pairs with a cross-correlation coefficient higher than 0.7.

Sources of all explanatory variable are given in Table 2. Distance to closed forest and number of closed forest neighbours was determined from the Swiss land use statistics dataset using the 1985 data for the validation model and the 2009 data for the final model. The date of the socio-economic data and the spatial definition of the municipality boundaries matched the date (period end year) of the land use statistic data. Spatial zoning data designating 'building zones' which are areas that can be developed as urban areas dated from 2012 (ARE, Swiss Federal Office for Spatial Development, 2012).

Suitability modelling

To account for high variability across Switzerland in terms of topography, climate and socio-economics we divided our study area into four regions, corresponding the bioregions but with the Southern and Central Alps combined into one region (Fig. 1a), and modelled suitability for each land use type separately for each region as well as for the whole of Switzerland.

Binomial generalised linear modelling (glm) was used to define the relationship between the probability of presence of each land use type and the explanatory variables (Verburg, Veldkamp, Willemen, Overmars, & Castella, 2004). Random sampling was used to sample approximately 10% of the total sample points for

each land use type in each region. Sample points were a minimum of 1 km apart where possible to reduce spatio-autocorrelation issues (Dungan et al., 2002). In cases of low representation of a land use class in a region, minimum separation distance was reduced to 300 m. To determine a final model for each land use type and region we modelled all combinations of the explanatory variables and selected the set of variables with the best fit according to the AIC value (Akaike, 1973). A relative importance for each explanatory variable for each land use type and region was determined in a model averaging approach as the sum of Akaike weights over all models (Burnham & Anderson, 2002). All models were fitted using R ver. 3.0.1 www.r-project.org with the packages Multi-model inference (MuMIn) and glm. Explanatory power of the models was measured with area under the Receiver Operating Curve (AUC) (Pontius & Schneider, 2001). The final glm models were applied across the explanatory variable datasets per region. The resulting regional spatial models were combined into an overall probability model for all of Switzerland for each land use type. Probability of a given land use presence was used as a proxy for land use suitability. Although some of our explanatory variables are dynamic, we used a single static suitability model (using 1985 data to model suitability for model validation and 2009 data for the scenario models). We considered this approach reasonable partly for simplicity and data availability reasons, but also because over the relatively short modelling time period (26 years) changes in these variables or the relative spatial differences of these values are likely to be minimal. Any changes in climatic variables over the 26 year modelling period are considered unlikely to fundamentally affect the relative suitability for a given land use type, considering our coarse land use classes. The suitability models were the same across the scenarios, except for the urban suitability model which included a weighting in current regional centres, small towns and tourist areas in the regionalisation scenarios (A2 and B2).

Scenario development

Key external driving factors that influence land-use change in Switzerland over the time period 2009—2035 were identified as demography, economy and governance drivers (Hersperger & Burgi, 2009). To represent uncertainty in the development of these drivers, five scenarios of future land-use demands were defined for Switzerland. Storylines were developed describing plausible socio-economic development scenarios related to the IPCC Special Report on Emission Scenarios (SRES) defined along two axes: a globalisation vs regionalisation axis and an axis of market orientation versus a high degree of policy and planning intervention. Population growth is considered to be linked to economic growth and this and variation in governance were the key drivers of change within the scenarios. Switzerland has a history of moving between liberal and regulation movements. Stronger linkages to the EU since the 1990's and WTO accession in 1995 have led to more market liberalisation although Switzerland's system of federalism and direct democracy limits direct intervention by extraterritorial organisations (Mach et al., 2003) and can generate regulative policies. For instance, there has been a tradition of implementing controlling and planning policies on land use (Jaeger & Schwick, 2014). Four scenarios are defined along these axes (Fig. 2) plus a 'Trend' scenario (a linear projection of observed trends in land use transition from the past).

Scenario descriptions

Trend scenario: linear projection of observed land use trends. Based on the land use statistics from 1985, 1997 and 2009, a linear interpolation of total area of each land use type was conducted to obtain land use demand for the year 2035. Observed trends show

Globalisation

low Swiss growth, low concern for environmental issues, low support for

conservation and agricultural subsides

+3 c №

More global

Global cooperation

Low Swiss growth, Environmental concerns high, shift toward resource efficiency

f Driving forces^N.

( Population, ) Economy, ... v/

Heterogeneous world, regionally centered growth

High Swiss economic and population growth driven by immigration, High urban sprawl, low support for

conservation and agricultural subsides

More regional

o Self-sufficiency

£ e Moderate Swiss growth,

£ High levels of environmental

.2 concern, High support for

c subsides, new regional

centres

re o (B2)

Fig. 2. Scenario storyline axes.

an increase in urban area by 23.4%, and an increase in wooded area by 3.1%, alongside a decrease in intensive agriculture by 10% and pasture agriculture by 3% (SFSO 2013a).

Scenario A1: globalisation, low Swiss growth. Global economic growth is high but focused on emerging markets and in highly developed Switzerland economic growth and population growth remain low. Due to wage competition and out-sourcing of industry, Swiss technological innovation is lower than that of more competitive markets. Concern for environmental issues is low and, also due to low economic growth, the support for agricultural and conservation subsidies is very low. In addition, liberalised and global markets, and even closer ties to the EU result in land abandonment with less willingness to pay for higher priced local products. Urban sprawl is however restrained due to low population growth.

Scenario A2: high growth, pressure scenario. Within a heterogeneous world context, Switzerland is its own 'island' with high economic and population growth, largely driven by immigration.

Investment into Switzerland is high. Environmental concerns are low. A liberalised economy means little support for direct payments supporting agricultural or conservation. However, with focus on regionalisation and distancing from the EU, there is still demand for local agricultural products. In conjunction with increased agro-technological innovations this leads to high intensity land use on agricultural land. Urban sprawl is high, with increased population and increased per capita urban demand, and, with a regional focus, the introduction of new urban centres.

Scenario B1: global cooperation. In a globalised world market Switzerland faces relatively low economic and population growth. Worldwide there is a shift towards resource efficiency and environmental concerns are high so that despite the low economic growth, support for agricultural and conservation subsidies remains similar to current (high) levels. Low population growth coupled with low economic growth results in low urban sprawl.

Scenario B2: self-sufficiency. There is an emphasis on self-sufficiency and development at the regional level as well as

Table 3

Summary of scenario quantification, where SFSO population scenarios are those developed by the Swiss Federal Statistical Office (SFSO, 2010).

Scenario SFSO population scenario Per capita urban Agricultural land Interventions Regionalisation

demand demanda

A1 Medium 407 m2 pp -5% None None

A2 High 509 m2 pp -20% None Weighting to increase urban suitability in regional centres and villages

B1 No population growth 399 m2 pp No change Conversion of Agriculture to Overgrown not permitted above 900 masl Conversion to urban area only within the building zones None

B2 Medium 407 m2 pp +2% Conversion of Agriculture to Overgrown not permitted above 900 masl Weighting to increase urban suitability in regional centres and villages

a Relative to 2009 agricultural land area.

higher ecological awareness. Globally economic growth is slow but Switzerland has medium economic growth due to investment locally, and medium population growth thanks largely to immigration. Tourism is supported to encourage regional growth. Support for conservation and agricultural subsidies is high. Urban sprawl is also medium because of development of regional centres, and medium population growth.

Scenario quantification

Quantification of the storylines provides future land use area demands and relied heavily on the Swiss Federal Statistical Office (SFSO) population growth scenario models (SFSO, 2010). The SFSO has modelled three population growth scenarios — 'low', 'medium' and 'high', which incorporate economic development and immigration. We used the modelled population level for 2035 in combination with current data on the per capita urban area demand to calculate a total urban area demand per scenario (Table 3). Values for per capita urban area demand were determined based on the range of current values across the cantons, with 'medium' being the current mean value, 'low' being the lower 95% confidence interval bound and 'high' the upper 95% confidence interval bound. Tourism as a driver for settlement development was included in the improved suitability weighting for tourism areas as described in Section 2.4.

Agricultural area demand is based on our storylines, current trends and findings from the literature. Rounsevell et al. (2005) find that given technological advances, Europe can expect to have an

oversupply of agricultural land in the future under all SRES scenarios with decreases in agricultural land by 2040 by between ~9% (B1) and ~25% (A2). However, observed trends for Switzerland are considerably more moderate with a 5.4% decrease in overall agricultural land between 1985 and 2009 (SFSO, 2013a). Observed Swiss data also shows a greater decrease in arable and horticultural land area (10%) than in pasture area (3%), due to more intensive cattle farming practices (SFSO, 2013a). We assume this relative breakdown will continue in our market driven (A) scenarios. With low population growth and low support for subsidies, loss of agricultural land will continue at a similar rate to the past in the A1 scenario, and at a higher rate of almost 1% per year in the A2 scenario with greater urban growth. In each of these scenarios technological advances will allow more intensive agriculture (EEA, 2007; Kopainsky, Flury, Pedercini, Sorg, & Gerber, 2013; Rounsevell et al., 2005). Recent work commissioned by the Swiss Federal Office for Agriculture finds that in order to reach policy goals of improved self-sufficiency it will be necessary to maintain the current area and quality of agricultural land (Kopainsky et al., 2013). Therefore in our high intervention scenarios (B) agricultural demand is at least maintained at current levels, and in the case of the self-sufficiency/regionalisation scenario B2, a slight increase in agricultural area is foreseen. Demand for forest area and overgrown area are defined as a joint demand and equal to the remaining areas, once demand for urban and agricultural areas is taken into account.

A matrix of allowed land use class conversions is defined per scenario. It is considered likely that current legal 'no-net-loss'

Table 4

Regression modelling results for land use suitability modelling. Explanatory power/model fit is measured by area under the receiver operating curve (AUC). 'Full Swiss' designates the result for a single model (per land use type) for the whole of Switzerland, and 'combined' the results for the regional models combined into a whole of Switzerland model. + and - denote positive or negative regression coefficients.

Model Explanatory power AUC 5 most Important variables in order (see Table 2 for definition)

Full Switzerland Full Swiss combined

Closed Forest 0.9054 0.9173 for_nn+, soil_suit, radi+, soil_per, p_emp+

Open Forest 0.9102 0.8389 p_emp-, slope+, soil_suit,for_ nn+, prec+

Overgrown 0.9314 0.8889 for_d+, soil_suit, cont-, elev+, soil_ston

Pasture Agriculture 0.7342 0.7759 radi-, pt-, for_d+, elev+, soil_p

Arable Agriculture 0.8702 0.9012 cont+, for_d+, pt-, soil_suit, elev-

Urban 0.8655 0.8666 for_d+, elev-, soil_suit, soil_p, slope-

Closed Forest 0.9276 for_nn+, slope-, prec-, cont-, radi+

Open Forest 0.7434 slope+, prec+, elev+, for_nn-, soil_suit

Overgrown 0.6715 slope+, for_d+, elev+, prec+, radi-

Pasture Agriculture 0.7577 for_d+, elev+, soil_suit, slope+, radi-

Arable Agriculture 0.8866 for_d+, pt-, elev-, soil_suit, cont+

Urban 0.8326 for_d+, pt+, slope+, elev+, tax-

Plateau

Closed Forest 0.9727 for_nn+, slope-, tax+, road_d+, elev-

Open Forest 0.5912 for_nn+, soil_suit, prec+, elev-, p_emp-

Overgrown 0.5023 for_d+, slope+, tax+, elev-, cont-

Pasture Agriculture 0.6717 for_d+, elev+, soil_suit, radi-, pt-

Arable Agriculture 0.7439 pt-, for_d+, soil_suit, slope-, elev-

Urban 0.8312 pt+, for_d+, p_emp-, elev-, soil_suit

Northern Alps

Closed Forest 0.9248 for_nn+, soil_suit, elev+, radi+, slope+

Open Forest 0.634 slope+, soil_suit, elev+, pt-, for_nn+

Overgrown 0.8644 for_d+, soil_suit, elev+, cont-, pt-

Pasture Agriculture 0.736 for_d+, soil_suit, elev+, pt-, radi-

Arable Agriculture 0.9423 for_d+, elev-, soil_suit, p_emp+, cont+

Urban 0.8665 for_d+,elev-, soil_suit, slope-, pt+

Central and Southern Alps

Closed Forest 0.9454 for_nn+, soil_suit, elev+, radi+, slope+

Open Forest 0.6968 slope+, soil_suit, p_emp-, for_nn+, cont-

Overgrown 0.7728 for_d+, soil_suit, elev+, slope-, pt-

Pasture Agriculture 0.8501 for_d+, elev+, slope-, radi-, soil_suit

Arable Agriculture 0.9538 for_d+, elev-, prec-, soil_suit, soil_ston-

Urban 0.8984 for_d+, soil_suit, radi+, soil_p, p_emp+

restrictions for forest clearing will remain in place (FOEN, 2013), and as such in all scenarios no conversion from closed forest or open forest is allowed. Only overgrown areas may be converted to forest (to represent natural forest succession). Variable time lags for succession to forest were applied based on the work of (Tasser, Walde, Tappeiner, Teutsch, & Noggler, 2007) depending on the elevation (15 years at <1200 masl, 30 years at 1200—1600 masl, 45years at 1600—2000 masl, at over 2000 masl no forest succession

within the time span of our scenarios). Conversion to overgrown from the other land use types (urban, intensive agriculture and pasture agriculture) was permitted for all scenarios but with spatial restrictions in the high intervention B scenarios. In these scenarios, to represent cultural landscape conservation subsidies aimed at preventing land abandonment in mountainous areas, conversion of agriculture to overgrown areas was not allowed at elevations over 900 m. In the Trend and B1 scenarios, with high level of policy

Fig. 3. Hotspots for (a) Agricultural land abandonment and (b) Urbanisation which were identified by mapping the locations where transitions occurred across multiple scenarios.

intervention, conversion to urban land use was only permitted inside of the current building zones. All other conversions were allowed.

Model validation

In order to validate the modelling approach and the suitability models, we compared the fit of our model's simulated 2009 'land use model' to the independent 2009 actual land use data ('reference model') and to a 'random model' for 2009. The random model was obtained using the above modelling approach but where suitability for each land use class was a spatial dataset of randomly generated values between 0 and 1. The 1979/1985 land use data was used as the initial state at the start of each simulation. Goodness of fit for the modelled 2009 data ('land use model' and 'random model') compared to the 2009 actual land use statistics data ('reference model') was measured using the figure of merit (Pontius et al., 2008). The figure of merit is the ratio of the proportion of correctly predicted pixels to the union of all observed changing pixels and predicted changing pixels (Pontius et al., 2008; Pontius & Millones, 2011).

Results

Statistical modelling

Comparison of the suitability models showed that the combined and Swiss-wide models have greater explanatory power for all land use classes than the separate regional models, nevertheless there are differences in the explanatory variables of importance between regions (Table 4). In addition, for most land use types, the combined regional models performed better than the single all of Switzerland model, which demonstrates the difference in explanatory variables from region to region and the benefit of combining the regional models. Several of the models have relatively low AUC values, suggesting relatively poor fits to the data. This occurs for land use types that have relatively low presence within a region (i.e. Open Forest and Overgrown on the Plateau) and where within-class heterogeneity is likely to be great. Presence of these land use types is likely to be strongly influenced by historical socioeconomic factors that are not captured in our set of explanatory variables and for which quantitative data is not available or not at the extent of our Swiss-wide study.

Model validation

Comparison of the figure of merit for the reference model with the random model (0.895) and our land use model (0.902) showed that our modelling approach was better able to reproduce the observed land use changes over the period 1985 to 2009.

Scenarios

The A2 market driven/regionalisation scenario is defined as a population growth and urbanisation scenario and therefore, as expected, results in the most extreme land use pattern changes, in particular a strong trend toward urbanisation (Fig. 1). Although this is a regionalisation scenario, urbanisation is concentrated on the Plateau since this remains the area with greatest suitability for urban land use. However, the A1 market driven/globalisation scenario has higher levels of agricultural land abandonment (Figs. 1 and 4). The B1 scenario is closest to a 'no change' scenario, with very little urbanisation, land abandonment or reforestation (Figs. 1 and 4). The B2 scenario also has a distinct amount of transition back to pasture agriculture from overgrown land (over 46,000 ha), and at the same time urbanisation of the same magnitude (Figs. 1 and 4).

Hotspots for the key land use transition processes were identified by mapping the locations where transitions occurred across many of the scenarios. The process of urbanisation is widespread, and is most commonly occurring at the expense of agricultural land across all scenarios (Fig. 3). Key locations for urbanisation (Fig. 3) are determined by the suitability model and are often on the fringes of existing cities/urban areas with particular focus on the Plateau and in other areas with lower elevation, lower slopes, and good access to major roads and public transport, such as in the Central and Southern Alps. There are also hotspots for urbanisation in the eastern part of the Northern Alps, along major roads and in the area of St Gallen and the Austrian and Liechtenstein borders (Fig. 3).

Land abandonment is also widespread (Fig. 3). As anticipated, abandonment is most prevalent in areas of marginal favourability for agriculture in mountains areas of the Alps and the Jura, particularly the Southern Alps (Rutherford et al., 2008; Verburg et al., 2008). However, there are considerable areas of the Plateau and Jura likely to be subject to land abandonment, particularly areas of comparatively higher elevation or with soil less suited to agriculture. Key areas for reforestation are in similar parts of the Southern and Central Alps, but also spread throughout the Northern Alps area.

Discussion and conclusions

Incorporating socio-economic as well as bio-geographical variables into suitability modelling allowed us to improve on bio-geographical only approaches such as Bolliger et al. (2007) and Rutherford et al. (2008). While our results showed that key important variables for explaining presence of a given land use were similar across the regions (Table 4), there were some important differences, for example relative importance of slope, elevation or soil stoniness. We have shown that large areas of the Plateau, especially on the current urban fringe, as well as valley bottoms in the Central and Southern Alps are likely to be subject to strong urbanisation pressure (Fig. 3). These are areas of good accessibility, comparatively lower elevation and incline. However, there are

Urbanised 200000 160000 a 120000 80000 40000 0

Abandonned Agriculture

/innnn

Afforestation 40000

Trend A1

Fig. 4. Land use change transitions by scenario, where 'Urbanised' is the transition to urban area from any other land use type, 'Abandoned Agriculture' is the transition from any kind of agriculture to overgrown, open forest or closed forest, and 'Afforestation' is the transition from any land use to forest.

spatial differences in the key drivers of urbanisation (Table 4). While on the Plateau public transport connectivity and employment in the secondary and tertiary sectors play a bigger role, in the Central and Southern Alps employment in the primary sector, soil factors and solar radiation are more important, and in the Jura lower income tax plays and important role in urbanisation. Alpine areas that have been shown to be likely urbanisation hotspots (Fig. 3) include areas of the Northern and Central and Southern Alps with high tourism value and scenic beauty and might be considered to be subject to 'naturbanisation' pressure (Prados, 2005; Serra, Vera, Tulla, & Salvati, 2014). Although, due to lack of nation-wide data, we have not explicitly considered tourism value as a driver, this would be an important consideration for locally focussed studies. Without intervention to conserve mountain pasture, large areas of the Alps and Jura mountain are likely to be subject to agricultural land abandonment, particularly the southern Alps and the central and eastern part of the northern Alps (Fig. 3). Some areas of the Plateau are also shown to be likely to be subject to agricultural abandonment (Fig. 3) and here taxation levels and slope play an important role while in Alpine areas soil suitability (for agriculture) is a key factor. These results emphasise the importance of considering regional variations in drivers in land use modelling (Verburg et al., 2008), and demonstrates the varying importance of policy and management drivers versus environmental drivers.

The B1 globalisation, high intervention scenario, which has low economic and population growth for Switzerland, can be considered to be close to a no-change model (Figs. 1 and 4). In previous work on land use change in mountain areas of Switzerland, Bolliger et al. (2007) found that stakeholders prefer landscapes that resemble the current state over other greater change scenarios. However, the costs (social, economic and cultural) of a scenario that puts emphasis on biodiversity was considered unacceptable (Bolliger et al., 2007). Strong patterns of urbanisation, land abandonment and agricultural land loss as observed in the A scenarios (and sometimes the Trend scenario) are also unlikely to be favourable for many stakeholders. Scenario studies like ours are important in demonstrating to managers the potential implications of land management decisions.

With intervention to provide incentives for pasture agriculture (B scenarios), reinstatement of pasture could occur in Alpine regions, neighbouring current pasture. Nevertheless, as found by (Lawler et al., 2014), our results suggest that the policy interventions would need to be aggressive in order to change current underlying land use trends. Switzerland has a tradition of market liberalism and it can be considered that recent trends, with stronger connections to the European Union and international involvement, are toward more market liberalisation (El Benni & Finger,

2013). However, Switzerland also has a history of implementing policy with strong regulations on land use (Jaeger & Schwick, 2014). The Swiss population is concerned about land use change, particularly in the case of highly visible urbanisation, as demonstrated by several recent popular Federal initiatives such as that posing strict limits on the construction of second homes and that aiming to freeze current building zones (Hersperger, Franscini, & Kuebler,

2014). This demonstrates that assertive policy intervention managing land use change is politically possible in Switzerland. Knowledge of the key spatial locations and magnitude of likely land use changes under different trajectories of socio-economic development is key for enabling policy makers to make informed decisions considering consequences for the landscape. At the same time, knowledge of spatial differences in the drivers of land use change is important for effective planning policy.

Modelling future land use change is subject to a large amount of uncertainty (Verburg, Tabeau, & Hatna, 2013), not least of which in

future socio-economic and policy trajectories. Land use processes driven by management and policy decisions are generally more difficult to model accurately than those driven more by physical environmental processes (Moran-Ordonez, Suarez-Seoane, Calvo, & de Luis, 2011 ). We do not expect that any one of our modelled scenarios are a likely realisation for Switzerland in 2035. However, by identifying locations for key land use transitions which occur across several scenarios, we can identify likely hotspots or risk areas (Fig. 3). These areas will be most important to monitor and manage in terms of implications for landscape services, as they will be subject to change under a wide range of situations. While in some areas of the world, reforestation can be considered a positive process for biodiversity (Queiroz et al., 2014), previous work in Switzerland has found that loss of agricultural land in mountainous areas of Switzerland can have strong negative implications for open land and wetland species (Bolliger et al., 2007; Luetolf, Bolliger, Kienast, & Guisan, 2009; Maggini et al., 2014; van Strien et al., 2014). Agricultural change, land abandonment and associated reforestation has also been found to have major effects on carbon stocks and carbon sequestration, particularly in mountainous areas (Bolliger et al., 2008). In addition to the loss of areas of other land use types, urbanisation has a range of consequences including changes in soil conditions, loss of habitat and food production areas, increased noise and pollution, increased infrastructure pressure, neighbourhood effects on surrounding land uses etc (Jaeger & Schwick, 2014). These scenarios will provide key base information for future work including assessing conflicts and synergies in land-use planning or assessing impacts of land-use change on landscape function and ecosystem health.

Acknowledgements

This work was supported by the Swiss Federal Institute for Forest Snow and Landscape Research (WSL) internal project funding.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.apgeog.2014.12.009.

References

Akaike, H. (1973). Information theory and an extension of the maximum likelihood principle. In Second international symposium on information theory (pp. 267-281).

ARE, Swiss Federal Office for Spatial Development. (2012). Bauzonenstatistik Schweiz

2012, Statistik und Analysen. ARE, Swiss Federal Office for Spatial Development. (2013). OV-Giiteklassen —

Berechnungsmethodik ARE. Retrieved 20th December, 2013. Bolliger, J., Hagedorn, F., Leifeld, J., Bohl, J., Zimmermann, S., Soliva, R., et al. (2008). Effects of land-use change on carbon stocks in Switzerland. Ecosystems, 11(6), 895—907. http://dx.doi.org/10.1007/s10021-008-9168-6. Bolliger, J., Kienast, F., Soliva, R., & Rutherford, G. (2007). Spatial sensitivity of species habitat patterns to scenarios of land use change (Switzerland). Landscape Ecology, 22(5), 773—789. http://dx.doi.org/10.1007/s10980-007-9077-7. Burnham, K. P., & Anderson, D. R. (2002). Model selection and multi-model inference:

A practical information-theoretic approach (2nd ed.). New York: Springer. Dungan, J., Perry, J., Dale, M., Legendre, P., Citron-Pousty, S., Fortin, M., et al. (2002). A balanced view of scale in spatial statistical analysis. Ecography, 25(5), 626—640. http://dx.doi.org/10.1034/).1600-0587.2002.250510.x. Durr, B., Zelenka, A., Muller, R., & Philipona, R. (2010). Verification of CM-SAF and MeteoSwiss satellite based retrievals of surface shortwave irradiance over the Alpine region. International Journal of Remote Sensing, 31(15), 4179—4198. http:// dx.doi.org/10.1080/01431160903199163. EEA, European Environment Agency. (2007). Land-use scenarios for Europe: Qualitative and quantitative analysis on a European scale. EEA Technical report No 9/ 2007 (p. 78). Copenhagen. El Benni, N., & Finger, R. (2013). The effect of agricultural policy reforms on income inequality in Swiss agriculture — an analysis for valley, hill and mountain

regions. Journal of Policy Modeling, 35(4), 638-651. http://dx.doi.org/10.1016/ j.jpolmod.2012.03.005.

FOEN, Swiss Federal Office for the Environment. (2013). Forest policy 2020: Visions, objectives and measures for the sustainable management of forests in Switzerland (p. 66).

Gonseth, Y., Wohlgemuth, T., Sansonnens, B., & Buttler, A. (2001). Die biogeographischen Regionen der Schweiz. Erlauterungen und Einteilungsstandard. Bern: Bundesamt fuer Umwelt Wald und Landschaft.

Hersperger, A., & Burgi, M. (2009). Going beyond landscape change description: quantifying the importance of driving forces of landscape change in a Central Europe case study. Land Use Policy, 26(3), 640-648. http://dx.doi.org/10.1016/ j.landusepol.2008.08.015.

Hersperger, A. M., Franscini, M. P. G., & Kuebler, D. (2014). Actors, decisions and policy changes in local urbanization. European Planning Studies, 22(6), 1301-1319. http://dx.doi.org/10.1080/09654313.2013.783557.

Jaeger, J. A. G., & Schwick, C. (2014). Improving the measurement of urban sprawl: weighted Urban Proliferation (WUP) and its application to Switzerland. Ecological Indicators, 38, 294-308. http://dx.doi.org/10.1016/ j.ecolind.2013.11.022.

Kopainsky, B., Flury, C., Pedercini, M., Sorg, L., & Gerber, A. (2013). Ressourceneffizienz im Dienste der Ernährungssicherheit. Zürich/Washington: Teilprojekt Modellierung - Schlussbericht.

Lambin, E., & Meyfroidt, P. (2011). Global land use change, economic globalization, and the looming land scarcity. Proceedings of the National Academy of Sciences of the United States of America, 108(9), 3465-3472. http://dx.doi.org/10.1073/ pnas.1100480108.

Lawler, J. J., Lewis, D. J., Nelson, E., Plantinga, A. J., Polasky, S., Withey, J. C., et al. (2014). Projected land-use change impacts on ecosystem services in the United States. Proceedings of the National Academy of Sciences of the United States of America, 111(20), 7492-7497. http://dx.doi.org/10.1073/pnas.1405557111.

Luetolf, M., Bolliger, J., Kienast, F., & Guisan, A. (2009). Scenario-based assessment of future land use change on butterfly species distributions. Biodiversity and Conservation, 18(5), 1329-1347. http://dx.doi.org/10.1007/s10531-008-9541-y.

Mach, A., Hausermann, S., & Papadopoulos, Y. (2003). Economic regulatory reforms in Switzerland: adjustment without European integration, or how rigidities become flexible. Journal of European Public Policy, 10(2), 301-318. http:// dx.doi.org/10.1080/1350176032000059053.

Maggini, R., Lehmann, A., Zbinden, N., Zimmermann, N. E., Bolliger, J., Schroeder, B., et al. (2014). Assessing species vulnerability to climate and land use change: the case of the Swiss breeding birds. Diversity and Distributions, 20(6), 708-719. http://dx.doi.org/10.1111/ddi.122OT.

Melendez-Pastor, I., Hernandez, E., Navarro-Pedreno, J., & Gomez, I. (2014). Socioeconomic factors influencing land cover changes in rural areas: the case of the Sierra de Albarracin (Spain). Applied Geography, 52, 34-45. http://dx.doi.org/ 10.1016/j.apgeog.2014.04.013.

Moran-Ordonez, A., Suarez-Seoane, S., Calvo, L., & de Luis, E. (2011). Using predictive models as a spatially explicit support tool for managing cultural landscapes. Applied Geography, 31(2), 839-848. http://dx.doi.org/10.1016/ j.apgeog.2010.09.002.

Pontius, R., Boersma, W., Castella, J., Clarke, K., de Nijs, T., Dietzel, C., et al. (2008). Comparing the input, output, and validation maps for several models of land change. Annals of Regional Science, 42(1), 11-37. http://dx.doi.org/10.1007/ s00168-007-0138-2.

Pontius, R., & Millones, M. (2011). Death to Kappa: birth of quantity disagreement and allocation disagreement for accuracy assessment. International Journal of Remote Sensing, 32(15), 4407-4429. http://dx.doi.org/10.1080/ 01431161.2011.552923.

Pontius, R., & Schneider, L. (2001). Land-cover change model validation by an ROC method for the Ipswich watershed, Massachusetts, USA. Agriculture Ecosystems & Environment, 85(1-3), 239-248. http://dx.doi.org/10.1016/S0167-8809(01) 00187-6.

Prados, M.-J. (2005). Territorial recognition and control of changes in dynamic rural areas: analysis of the Naturbanization process in Andalusia, Spain. Journal of Environmental Planning and Management, 48(1), 65-83.

Queiroz, C., Beilin, R., Folke, C., & Lindborg, R. (2014). Farmland abandonment: threat or opportunity for biodiversity conservation? A global review. Frontiers in Ecology and the Environment, 12(5), 288-296. http://dx.doi.org/10.1890/120348.

Rounsevell, M., Ewert, F., Reginster, I., Leemans, R., & Carter, T. (2005). Future scenarios of European agricultural land use II. Projecting changes in cropland and

grassland. Agriculture Ecosystems & Environment, 107(2-3), 117-135. http:// dx.doi.org/10.1016/j.agee.2004.12.002.

Rutherford, G., Bebi, P., Edwards, P., & Zimmermann, N. (2008). Assessing land-use statistics to model land cover change in a mountainous landscape in the European Alps. Ecological Modelling, 212(3-4), 460-471. http://dx.doi.org/10.1016/ j.ecolmodel.2007.10.050.

Rutherford, G., Guisan, A., & Zimmermann, N. (2007). Evaluating sampling strategies and logistic regression methods for modelling complex land cover changes. Journal of Applied Ecology, 44(2), 414-424. http://dx.doi.org/10.1111/j.1365-2664.2007.01281.x.

Schwarb, M., Daly, C., Frei, C., & Schär, C. (2001). Mean annual and seasonal precipitation in the European Alps 1971-1990. Hydrological Atlas of Switzerland. University of Bern.

Serra, P., Vera, A., Tulla, A. F., & Salvati, L. (2014). Beyond urban-rural dichotomy: exploring socioeconomic and land-use processes of change in Spain (1991-2011). Applied Geography, 55, 71-81.

SFSO, Swiss Federal Statistical Office. (1992). GEOSTAT Benützerhandbuch. Neu-chatel: Bundesamt für Statistik.

SFSO, Swiss Federal Statistical Office. (1997). Statistisches Jahrbuch der Schweiz. Zürich: Verlag Neue Zürcher Zeitung.

SFSO, Swiss Federal Statistical Office. (2013a). Land use in Switzerland: Results of the Swiss land use statistics. Neuchatel.

SFSO. (2010). Swiss Federal Statistical Office, Szenarien zur Bevolkerungsentwicklung der Schweiz 2010-2060. Neuchatel.

SFSO, Swiss Federal Statistical Office. (2013b). STAT-TAB: Die interaktive Statistikdatenbank. Retrieved 10th July, 2013.

Stoate, C., Baldi, A., Beja, P., Boatman, N., Herzon, I., van Doorn, A., et al. (2009). Ecological impacts of early 21st century agricultural change in Europe - a review. Journal of Environmental Management, 91(1), 22-46. http://dx.doi.org/ 10.1016/j.jenvman.2009.07.005.

van Strien, M., Keller, D., Holderegger, R., Ghazoul, J., Kienast, F., & Bolliger, J. (2014). Landscape genetics as a tool for conservation planning: predicting the effects of landscape change on gene flow. Ecological Applications, 24(2), 327-339. http:// dx.doi.org/10.1890/13-0442.1.

Swiss Federal Assembly, Schweizerischer Bundesrat, KdK, BPUK, SSV, SGV. (2012). Raumkonzept Schweiz. Bern: Überarbeitete Fassung.

Swisstopo, Swiss Federal Office of Topography. (2013). Swiss topographical landscape model 3D.

Tasser, E., Walde, J., Tappeiner, U., Teutsch, A., & Noggler, W. (2007). Land-use changes and natural reforestation in the Eastern Central Alps. Agriculture Ecosystems & Environment, 118(1-4), 115-129. http://dx.doi.org/10.1016/ j.agee.2006.05.004.

Verburg, P. H., Eickhout, B., & van Meijl, H. (2008). A multi-scale, multi-model approach for analyzing the future dynamics of European land use. Annals of Regional Science, 42(1), 57-77. http://dx.doi.org/10.1007/s00168-007-0136-4.

Verburg, P., & Overmars, K. (2009). Combining top-down and bottom-up dynamics in land use modeling: exploring the future of abandoned farmlands in Europe with the Dyna-CLUE model. Landscape Ecology, 24(9), 1167-1181. http:// dx.doi.org/10.1007/s10980-009-9355-7.

Verburg, P. H., Tabeau, A., & Hatna, E. (2013). Assessing spatial uncertainties of land allocation using a scenario approach and sensitivity analysis: a study for land use in Europe. Journal of Environmental Management, 127, S132-SM4. http:// dx.doi.org/10.1016/j.jenvman.2012.08.038.

Verburg, P. H., Veldkamp, A., Willemen, L., Overmars, K. P., & Castella, J. C. (2004). Landscape level analysis of the spatial and temporal complexity of land-use change. In R. S. DeFries, G. P. Asner, & R. A. Houghton (Eds.), Ecosystems and land use change (Vol. 153, pp. 217-230).

Westhoek, H., van den Berg, M., & Bakkes, J. (2006). Scenario development to explore the future of Europe's rural areas. Agriculture Ecosystems & Environment, 114(1), 7-20. http://dx.doi.org/10.1016Zj.agee.2005.11.005.

Willhauck, R. (2013). Zersiedelung in der Schweiz - Explorative statistische Untersuchung wichtiger Einflussgrässen. Master thesis. Birmensdorf: Swiss Federal Research Institute ETH.

Wu, J. (2006). Landscape ecology, cross-disciplinarity, and sustainability science. Landscape Ecology, 21(1), 1-4. http://dx.doi.org/10.1007/s10980-006-7195-2.

Zimmermann, N., & Kienast, F. (1999). Predictive mapping of alpine grasslands in Switzerland: species versus community approach. Journal of Vegetation Science, 10(4), 469-482. http://dx.doi.org/10.2307/3237182.