(Joly et al. 2003). Most species occupy an aquatic habitat for breeding and during
the larval period, and a terrestrial habitat after breeding and during hibernation.
Daily movements and seasonal migrations across the landscape matrix connect
these two types of habitat. Furthermore, many species are structured into
metapopulations, in which several subpopulations occupy spatially distinct habitat
patches separated by a more or less unfavorable matrix. Dispersal events allow
individuals to colonize new ponds or to recolonize sites where the species is nearing
extinction. A literature review about amphibian dispersal (Smith and Green 2005)
showed that the median distance is less than 400 m but 7% of observed species may
reach 10 km.
The major causes of landscape fragmentation are farming practices, urban
development, and the construction of transportation infrastructures (Forman and
Alexander 1998). Apart from direct loss of suitable habitats and road-kills, linear
infrastructures cause the loss of landscape connectivity (Forman and Alexander
1998; Geneletti 2004), which is recognized as a key functional factor for the viability
of species and their genetic diversity (Fahrig et al. 1995). Major infrastructures such
as motorways or high-speed railway lines act as barriers to the movement of animals
and isolate organisms in small subpopulations which become more sensitive to the
risk of extinction (Forman and Alexander 1998). This is especially the case for
populations of amphibians whose daily movements, seasonal migrations and dispersal events mean they regularly cross the landscape matrix (Allentoft and O’Brien
2010; Cushman 2006; Fahrig et al. 1995; Scherer et al. 2012).
Several case studies have contributed to identifying and quantifying the effects
of linear infrastructures on species distribution in many regions of the world, using
various methods. Authors have related data describing species (e.g. abundance,
collisions) to proximity of infrastructures (Brotons and Herrando 2001; Fahrig et al.
1995; Huijser and Bergers 2000; Kaczensky et al. 2003; Li et al. 2010) and to the
degree of habitat fragmentation (Fu et al. 2010; Serrano et al. 2002; Vos and
Chardon 1998). These studies measure the real impact of the infrastructure using
data on species collected after its construction. However, before the construction
phase, an impact prediction stage is also necessary to compare alternative infrastructure routes (Fernandes 2000; Geneletti 2004; Vasas et al. 2009) or to guide the
mitigation measures from the beginning of the project (Clauzel et al. 2015a, b;
Girardet et al. 2016; Mörtberg et al. 2007; Noble et al. 2011).
Reviews by Geneletti (2006) and Gontier et al. (2006) show that the effects of
landscape fragmentation are more difficult to predict than the direct loss of habitat.
According to these authors, current assessment methods are often restricted to protected areas or to a narrow strip on either side of the infrastructure. However, landscape fragmentation may have consequences on a far broader scale (Forman 2000).
To assess the long-distance effects of linear infrastructures on species distributions,
models must include connectivity metrics that take into account both structural (arrangement of habitat patches) and functional (behavior of the organisms) aspects. With
this aim in mind, the development of methods based on graph theory in landscape
modelling is promising (Dale and Fortin 2010; Urban et al. 2009). For our purposes, a
graph is a set of habitat patches of a given species (called “nodes”) potentially
216
C. Clauzel
the larval period, and a terrestrial habitat after breeding and during hibernation.
Daily movements and seasonal migrations across the landscape matrix connect
these two types of habitat. Furthermore, many species are structured into
metapopulations, in which several subpopulations occupy spatially distinct habitat
patches separated by a more or less unfavorable matrix. Dispersal events allow
individuals to colonize new ponds or to recolonize sites where the species is nearing
extinction. A literature review about amphibian dispersal (Smith and Green 2005)
showed that the median distance is less than 400 m but 7% of observed species may
reach 10 km.
The major causes of landscape fragmentation are farming practices, urban
development, and the construction of transportation infrastructures (Forman and
Alexander 1998). Apart from direct loss of suitable habitats and road-kills, linear
infrastructures cause the loss of landscape connectivity (Forman and Alexander
1998; Geneletti 2004), which is recognized as a key functional factor for the viability
of species and their genetic diversity (Fahrig et al. 1995). Major infrastructures such
as motorways or high-speed railway lines act as barriers to the movement of animals
and isolate organisms in small subpopulations which become more sensitive to the
risk of extinction (Forman and Alexander 1998). This is especially the case for
populations of amphibians whose daily movements, seasonal migrations and dispersal events mean they regularly cross the landscape matrix (Allentoft and O’Brien
2010; Cushman 2006; Fahrig et al. 1995; Scherer et al. 2012).
Several case studies have contributed to identifying and quantifying the effects
of linear infrastructures on species distribution in many regions of the world, using
various methods. Authors have related data describing species (e.g. abundance,
collisions) to proximity of infrastructures (Brotons and Herrando 2001; Fahrig et al.
1995; Huijser and Bergers 2000; Kaczensky et al. 2003; Li et al. 2010) and to the
degree of habitat fragmentation (Fu et al. 2010; Serrano et al. 2002; Vos and
Chardon 1998). These studies measure the real impact of the infrastructure using
data on species collected after its construction. However, before the construction
phase, an impact prediction stage is also necessary to compare alternative infrastructure routes (Fernandes 2000; Geneletti 2004; Vasas et al. 2009) or to guide the
mitigation measures from the beginning of the project (Clauzel et al. 2015a, b;
Girardet et al. 2016; Mörtberg et al. 2007; Noble et al. 2011).
Reviews by Geneletti (2006) and Gontier et al. (2006) show that the effects of
landscape fragmentation are more difficult to predict than the direct loss of habitat.
According to these authors, current assessment methods are often restricted to protected areas or to a narrow strip on either side of the infrastructure. However, landscape fragmentation may have consequences on a far broader scale (Forman 2000).
To assess the long-distance effects of linear infrastructures on species distributions,
models must include connectivity metrics that take into account both structural (arrangement of habitat patches) and functional (behavior of the organisms) aspects. With
this aim in mind, the development of methods based on graph theory in landscape
modelling is promising (Dale and Fortin 2010; Urban et al. 2009). For our purposes, a
graph is a set of habitat patches of a given species (called “nodes”) potentially
216
C. Clauzel