A wolf habitat suitability prediction study in Valais (Switzerland)

https://doi.org/10.1016/S0169-2046(01)00119-0Get rights and content

Abstract

In recent years, the European wolf (Canis lupus) population has expanded its southern range from the Italian Peninsula to the Maritime Alps (Italy and France) and to Piemonte (Italy); establishing small sub-populations. Hence re-colonisation of the Swiss Alps is now likely to occur. In 1995–1996 the wolf reached the southern part of Switzerland (Canton of Valais) from where he got extinct 150 years ago. Actual conflicts of interests between livestock breeders, local political authorities and nature conservation parties, as well as federal authorities defending the protected status of wolf, require serious management investigations.

In order to check wolf habitat suitability of an alpine landscape, like the Valais, subjected to dynamic landscape–ecology processes since the extinction of wolf, we present herein an application of a predictive wolf habitat model, using a stochastic model involving logistic regression. As no data were available in the Canton of Valais, the regression coefficients for the retained variables such as urban area, population density, arable land, minimal altitude, northwest exposure and wild ungulate diversity index, were derived from data collected in the northern Apennine (Northern Italy), where habitat variables were related to data of wolf presence. The selection of the parameters for the Canton of Valais has been performed in respect of their predictive power, as well as their availability and geo-morphological importance for the alpine landscape under consideration. Using the geographic information system (GIS), the simulation pointed out that 19% (1142 km2) of the total grid surface (5821 km2) are suitable for wolf presence. Moreover, it reveals that especially areas at lower altitudes (minimum altitude<800–900 m a.s.l.), due to the high anthropic activity, and areas at high altitudes (minimum altitude>1800–2000 m a.s.l.), due to lack of prey and severe geo-morphological conditions, present a reduced habitat suitability. The geomorphological and demographic situation of the alpine area lead to a wolf habitat of a partially fragmented and linear aspect, affecting overall habitat suitability.

The strengths of the application is not only the visualisation of the present habitat quality of an alpine landscape re-colonised by wolves, but also that it allows to make investigations in order to manage the different conflicts of interest.

Introduction

In central and western Europe wolf (Canis lupus) has nearly been exterminated 150 years ago. Expansion of human settlements, threats to livestock, competition for game species (Breitenmoser, 1998) as well as the negative image partially mediated by the clergy, caused a considerable persecution and wide range killing of wolves in central and western Europe. In spite of these factors, several isolated wolf populations remained in the Iberian Peninsula, in the Balkans and in Italy (Boitani and Ciucci, 1993, Promberger and Hofer, 1994). In the early 1970s the Italian wolf population was estimated with about 100 individuals (Boitani and Zimen, 1975). As a result of the legal protection of the wolf in Italy in 1976 and the reintroduction of wild ungulates, wolf population increased from 100 to about 500–1000 individuals in the end of 1990; expanding from the central and southern Peninsula to the Maritime Alps (Italy and France) and to Piemonte (Italy). The recolonisation of the Swiss Alps is now likely to occur — the Canton of Valais representing the first step.

In 1995–1996 first wolf observations have been made in the south of Switzerland (Canton of Valais) where about 100 sheep have been killed. During the winter of 1998–1999 another 40 sheep have been killed and two wolves were found dead because of poaching and traffic incident. These events initiated a fierce contention between nature conservation parties, federation authorities and livestock breeders in the Canton of Valais. Partly supported by local political authorities claiming that their return being not accordable with the actual way of stock-breeding (e.g. free-grazing sheep) and outlying that the alpine landscape would not be suitable for wolves, because of the increased anthropic activity (infrastructure, tourism, etc.). Indeed, since the eradication of wolves 150 years ago fragmentation of landscape increased but, opposed to public opinion, habitat conditions for large carnivores improved as a result of recovered wild ungulate populations and expansion of woodland (Breitenmoser, 1998). Despite illegal hunting new wolves immigrations will occur from established wolf populations in France and Italy.

In this paper, we will try to answer to the question raised especially by public and political parties whether the alpine landscape, which since the eradication of wolf has undergone different human-induced landscape–ecological processes, does still present suitable habitat for the re-colonising wolves. Therefore, we present herein a wolf habitat analysis of the Canton of Valais using a stochastic model based upon logistic regression. The model has initially been fitted on data collected in the northern Apennines (Massolo and Meriggi, 1998).

Up to now only a few studies included spatial factors determining wolf presence. Fuller et al. (1992), Thiel (1985) and Mech et al. (1988) revealed the effect of road density by simple correlation analysis. Mladenoff et al. (1995) used, as Massolo and Meriggi (1998) in the northern Apennines, a multivariate logistic regression approach to make predictions about wolf habitat suitability. Mladenoff et al. (1995) demonstrated the influence of road density and fractal dimension of the landscape on the presence of wolf in the northern Great Lakes region. Finally, Corsi et al. (1999) developed a large-scale model of wolf distribution in Italy using multivariate analysis.

Section snippets

Study region

The Canton of Valais is situated in the southern part of Switzerland, flanked by Italy and France; see Fig. 1. It is a typical alpine region, with a main valley (Rhone-valley) and altitudes ranging from 372 (Lake Geneva) to 4634 m a.s.l. (Dufourspitze). The climate is continental with an average yearly temperature fluctuating from −0.9°C (2479 m a.s.l) to 10.2°C (482 m a.s.l.). Of the total of 5224 km2, 22.2% are covered with forest, 21.7% are agriculture land (pastures included), 2.5% are urban

Initial model

The habitat analysis was performed using a stochastic model based upon logistic regression fitted to northern Apennines data by Massolo and Meriggi (1998). They used the ‘Universal Transverse Mercator’ (UTM) system to spot 143 sample squares of 23 km2 each — a scale considered as appropriate (15–20% of the estimated wolf home range). In each sample square 58 variables were assessed of which 10 for vegetation types, 5 for environmental complexity, 8 for anthropic variables, 11 for large prey

Model result

The Chi-square test indicates an overall validity of the formulated dichotomous logistic regression model (P-value=0.0000). Referred to the 143 samples squares of the Northern Apennines, the model classified 90.32% of the squares without wolf presence correctly and 70.00% of the squares with regular wolf presence (cut-level at P=50%). It yields that the reformulated model reveals a percentage of correct classifications of 83.22% of the Apennine data, compared to 93.00% of the model of Massolo

Discussion

The results of the habitat composition of suitable and non-suitable wolf areas, as well as their altitudinal distribution reflect well the strong landscape–ecological transformations, which especially the valley bottom had undergone since the extinction of wolf 150 years ago. The increased interest of industry (e.g. hydroelectricity) engendered by making the Alps accessible to transportation (e.g. trains, trucks) and the increased economical competition of traditional agriculture products from

Conclusion

Identifying areas suitable for wolves by spatial extrapolation through a habitat model can provide a useful tool not only for resource planners charged with managing restoration of wolves (Mladenoff et al., 1999), but also for other public or political parties, which intend to decrease man–wolf competition for resources and space. In this sense the model of Massolo and Meriggi (1998) applied to an alpine region prevailed, although based on Apennine data, an useful way to conceive a first

Acknowledgements

This study would not have been possible without the technical support of the ‘Laboratoire de Système d’Information à Référence Spatiale’ (SIRS/IGEO/EPFL), especially M. Gilgen and M. Riedo. Moreover, we want to thank U. Breitenmoser for his support, as well as I. Iorgulescu for his critical reading of the paper and the fruitful discussions.

Glenz Christian obtained a grade in Biology from the University of Lausanne (UNIL) in 1997. He participated at the Postgraduate Course in Environmental Sciences (PCES) at the Swiss Federal Institute of Technology in Lausanne (EPFL) from 1997 to 1999. In this context, he did his research work at the Institute of Ecosystem Management (GECOS), analysing wolf habitat suitability; especially for the region of the Canton of Valais. Currently he is working as a scientific collaborator in an

References (29)

  • U. Breitenmoser

    Large predators in the Alps: the fall and rise of man’s competitors

    Biol. Conserv.

    (1998)
  • K. Cozza et al.

    The damage–conservation interface illustrated by predation on domestic livestock in central Italy

    Biol. Conserv.

    (1996)
  • Bibikov, D.I., 1988. Der Wolf. Verlag AZ: Die Neue Brehm-Bücherei, Wittenberg...
  • Boitani, L., Ciucci, P., 1993. Wolves in Italy: critical issues for their conservation. In: Promberger, C., Schröder,...
  • L. Boitani et al.

    Number and distribution of wolves in Italy

    Z. Säugetierkd

    (1975)
  • Bowman, A.W., Azzalini, A., 1997. Applied Smoothing Techniques for Data Analysis: The Kernel Approach with S-Plus...
  • A. Branggi et al.

    Predation by wolves (Canis lupus L.) on wild and domestic ungulates in Northern Italy

    Ongulés/Ungulates

    (1991)
  • F. Corsi et al.

    A large-scale model of wolf distribution in Italy for conservation planning

    Conserv. Biol.

    (1999)
  • B.W. Dale et al.

    Functional response of wolves preying on barren-ground caribou in a multiple-prey ecosystem

    J. Animal Ecol.

    (1994)
  • Farina, A., 1998. Principles and Methods in Landscape Ecology. Chapman & Hall,...
  • Fischesser, B., 1982. La vie de la montagne: Chêne/Hachette,...
  • Forman, R.T., Godron, M., 1986. Landscape Ecology. Wiley, New...
  • Franz, H., 1994. Gefährdung und Schutz der Alpen. Wissenschaften OAd. Veröffentlichung der Kommission für...
  • Fritts, S.H., Mech, L.D., 1981. Dynamics, Movements and Feeding Ecology of a Newly Protected Wolf Population in...
  • Cited by (0)

    Glenz Christian obtained a grade in Biology from the University of Lausanne (UNIL) in 1997. He participated at the Postgraduate Course in Environmental Sciences (PCES) at the Swiss Federal Institute of Technology in Lausanne (EPFL) from 1997 to 1999. In this context, he did his research work at the Institute of Ecosystem Management (GECOS), analysing wolf habitat suitability; especially for the region of the Canton of Valais. Currently he is working as a scientific collaborator in an International project of forest quality evaluation at landscape level at the cited institute.

    Alessandro Massolo did his Master in Science from the University of Pisa examining factors affecting wolf habitat occupancy in the northern Apennines (northern Italy), supervised by Prof. A. Meriggi. Currently, he is working as research assistant in the Department of Evolutionary Biology, Ethology and Behavioural Ecology Group, where he is also performing his PhD.

    Diego Kuonen graduated from the Mathematics Department, Swiss Federal Institute of Technology in Lausanne (EPFL) in 1998, where he is currently doing his PhD in Statistics under the supervision of Prof. A.C. Davison. His main research interests include saddlepoint approximations, bootstrap, robust statistics, statistics in sports, data visualisation and datamining, which are subject to several scientific publications.

    Rodolphe Schlaepfer studied forestry from the Swiss Federal Institute of Technology in Zurich (1959–1964), obtained a Master in Science from the University Laval (Canada) in 1966 and a diploma in Statistics from the University of Edinburgh (UK) in 1968. He worked as a statistician for Ciba-Geigy from 1969 to 1975, was professor for statistics and physics at the Swiss School for Agricultural Engineering (1975–1982) and became professor for forest management at the Swiss Federal Institute for Technology in Zurich (ETHZ) in 1982. From 1987 to 1996 he was director of the Swiss Federal Institute for Forest, Snow and Landscape Research in Birmensdorf, while remaining a professor at the ETHZ. In 1996, he was visiting professor at the Oregon State University, College of Forestry. In 1997, he become professor for Ecosystem Management at the Swiss Federal Institute of Technology in Lausanne (EPFL) where he is head of the Laboratory of Ecosystem Management (GECOS).

    View full text