Biol. Rev. (2012), pp. 000–000. 1
doi: 10.1111/brv.12000
Individual dispersal, landscape connectivity
and ecological networks
Michel Baguette
1,2,∗
, Simon Blanchet
2
, Delphine Legrand
2
, Virginie M. Stevens
2
and Camille Turlure
3
1
Mus´eum National d’Histoire Naturelle, UMR 7205 CNRS-MNHN-UPMC Origine, Structure et Evolution de la Biodiversit´e, F-75005
Paris, France
2
USR CNRS 2936, Station d’Ecologie Exp´erimentale du CNRS `a Moulis, 2 route du CNRS, F-09200 Saint Girons, France
3
F.R.S.-FNRS, Universite Catholique de Louvain, Earth and Life Institute, Biodiversity Research Centre, Croix du Sud 4, B-1348
Louvain-la-Neuve, Belgium
ABSTRACT
Connectivity is classically considered an emergent property of landscapes encapsulating individuals’ flows across space.
However, its operational use requires a precise understanding of why and how organisms disperse. Such movements,
and hence landscape connectivity, will obviously vary according to both organism properties and landscape features.
We review whether landscape connectivity estimates could gain in both precision and generality by incorporating
three fundamental outcomes of dispersal theory. Firstly, dispersal is a multi-causal process; its restriction to an ‘escape
reaction’ to environmental unsuitability is an oversimplification, as dispersing individuals can leave excellent quality
habitat patches or stay in poor-quality habitats according to the relative costs and benefits of dispersal and philopatry.
Secondly, species, populations and individuals do not always react similarly to those cues that trigger dispersal, which
sometimes results in contrasting dispersal strategies. Finally, dispersal is a major component of fitness and is thus under
strong selective pressures, which could generate rapid adaptations of dispersal strategies. Such evolutionary responses
will entail spatiotemporal variation in landscape connectivity. We thus strongly recommend the use of genetic tools
to: (i) assess gene flow intensity and direction among populations in a given landscape; and (ii) accurately estimate
landscape features impacting gene flow, and hence landscape connectivity. Such approaches will provide the basic
data for planning corridors or stepping stones aiming at (re)connecting local populations of a given species in a given
landscape. This strategy is clearly species- and landscape-specific. But we suggest that the ecological network in a given
landscape could be designed by stacking up such linkages designed for several species living in different ecosystems. This
procedure relies on the use of umbrella species that are representative of other species living in the same ecosystem.
Key words: biodiversity, biological conservation, extinction, gene flow, population isolation, habitat selection, individual
fitness, ideal free distribution, linkage strategy, landscape, seascape, water basin, functional connectivity, structural
connectivity, landscape fragmentation, landscape genetics, umbrella species.
CONTENTS
I. Introduction ................................................................................................ 2
II. Individual dispersal and the linkage strategy ............................................................... 4
(1) Costs and benefits of dispersal .......................................................................... 5
(2) Individual variability in dispersal ....................................................................... 6
(3) Variation in dispersal and linkages in the landscape ................................................... 6
III. From individual dispersal to landscape connectivity ........................................................ 6
(1) Connectivity of terrestrial landscapes .................................................................. 7
(2) Connectivity of seascapes .............................................................................. 7
(3) Connectivity of pondscapes and riverscapes ........................................................... 8
(4) Structural connectivity estimates ....................................................................... 8
* Address for correspondence (Tel: ++33 561040380; Fax: ++33 561960851; E-mail: baguette@mnhn.fr).
Biological Reviews (2012) 000–000 © 2012 The Authors. Biological Reviews © 2012 Cambridge Philosophical Society
2 M. Baguette and others
(5) Functional connectivity estimates ...................................................................... 9
(6) Useful genetic tools to assess functional connectivity ................................................... 10
(7) Practical implementation of connectivity assessments .................................................. 11
IV. From landscape connectivity to ecological networks ........................................................ 11
(1) From single- to multiple-species networks .............................................................. 12
(2) Selection of umbrella species ........................................................................... 13
V. Conclusions ................................................................................................ 13
VI. Acknowledgements ......................................................................................... 14
VII. References .................................................................................................. 14
I. INTRODUCTION
Current massive species extinctions highlight how human
activities negatively impact biodiversity worldwide (Pimm et
al., 1995; Rockstrom et al., 2009; Pereira et al., 2010). Among
the manifold pressures inflicted by Homo sapiens on other living
organisms, the destruction of natural ecosystems is undoubt-
edly one of the major causes of biodiversity loss due to the
resulting habitat loss and fragmentation (Vitousek et al., 1997;
Pimm & Raven, 2000; Foley et al., 2005; Lawler et al., 2006).
Theory predicts and empirical studies confirm that both
habitat loss and fragmentation contribute to local population
extinctions (Fahrig, 2003; Ewers & Didham, 2006; Swift &
Hannon, 2010). The extinction of a species is indeed usually
preceded by the fragmentation and the shrinking of its dis-
tribution area, which reflects the progressive disappearance
of local populations (Ceballos & Ehrlich, 2002).
By removing suitable resources, habitat loss directly affects
the carrying capacity of a given area, and hence its ability
to sustain large populations, while small populations are
more vulnerable to genetic, demographic and environmental
accidents. Low effective population sizes decrease the genetic
variability in populations and hence their adaptability to
environmental changes [the extinction vortex (terms in italic
throughout this review are defined in Table 1); Gilpin &
Soule (1986) and Fagan & Holmes (2006)]. Both empirical
(e.g. Saccheri et al., 1998) and experimental studies (e.g.
Madsen et al., 1999) document the harmful interactions
between genetic diversity and demographic stochasticity,
dooming local populations to extinction. Some exceptions
to this rule have been documented, particularly regarding
the importance of the loss of genetic diversity associated
with inbreeding (Reed, 2010). However, a vast majority of
empirical studies confirm that the probability of extinction
of a local population is positively related to its isolation, and
negatively related to its size (e.g. Ouborg, 1993; Pimm et
al., 1993; Hanski, 1999b; Brook et al., 2002; Rodriguez &
Delibes, 2003). Besides such local processes, the loss of habitat
associated with fragmentation also increases the distances
among suitable habitat patches, which in turn decreases
the settlement probability of immigrants. The resulting
functional isolation of local populations reduces both the
rescue of imperiled populations (the rescue effect, Brown
& Kodric-Brown, 1977) and the rate of (re)colonization of
vacant habitats (Hanski, 1998, 1999b), which should result
in wider-scale species extinctions.
The best way to curb such extinctions would be to increase
the carrying capacity of local populations, by increasing
either the habitat area (Hodgson et al., 2011a)orthehabitat
quality. Implicit to the first possibility, the re-allocation of
large areas to nature is rarely an option in heavily human-
dominated landscapes. Improving habitat quality is feasible
for those few species for which ecological requirements
are sufficiently well known, but often demands extensive
man-power and hence high financial support. In addition,
as habitat quality is species-specific and even population-
specific (e.g. Turlure et al., 2009), targeted conservation
efforts may prove to be detrimental to other species of the
same community.
An alternative (or complementary) strategy would be to
increase the exchange of individuals among local popula-
tions, to reduce their functional isolation. These exchanges
would facilitate the maintenance of large metapopulations (e.g.
Levins, 1969; Hanski & Gilpin, 1991, 1997; Hanski, 1998,
1999b) defined as groups of local populations where the
movement of individuals among habitat patches is possible
(Hanski & Simberloff, 1997). In addition to their demo-
graphic effects [rescue and (re)colonization], movements of
individuals among local populations may increase the genetic
mixing among populations, hence reducing possible genetic
variability erosion and thereby genetic diversity within popu-
lations, in turn sheltering these populations from extinction.
The metapopulation concept thus provides a solid
framework for the conservation of species in heavily
fragmented landscapes. In our vision of such spatially
structured populations, local populations (demes) occupy
habitat patches more or less isolated from each other in
a matrix of more or less sub-optimal habitats. By explicitly
considering that the matrix is composed of different elements
with different quality, such structures do not correspond
to the binary representation of landscape composed of
suitable habitats embedded in an uniformly unsuitable matrix
typical of the classical metapopulation theory. Our vision of
metapopulations rather integrates insights from landscape
ecology into metapopulation theory, as advocated by Wiens
(1997).
The linkage strategy, corollary of the metapopulation theory,
is an appealing methodology in conservation planning that
aims to facilitate the displacements of individuals among local
populations, either by the creation of corridors or stepping
stones that bind local habitat patches into functional ecological
networks (e.g. Beier & Noss, 1998; Bennett, 1999; Jongman &
Biological Reviews (2012) 000–000 © 2012 The Authors. Biological Reviews © 2012 Cambridge Philosophical Society
Dispersal, connectivity and networks 3
Table 1. Definition of terms in italic in the main text
Term Definition
Dispersal Any movement of individuals or propagules with potential consequences for gene flow across space
(Ronce, 2007)
Dispersal kernel The probability density function that dispersing individuals move a certain distance
Ecological network Set of suitable habitats and linkages (corridors, stepping stones) that allows the persistence of a viable
metapopulation
Efficient connectivity Ultimate measure of landscape connectivity which evaluates the amount of gene flow across the
landscape
Extinction vortex Suite of insidious reinforcement among biotic and abiotic processes following population decline
(environmental and demographic stochasticity, inbreeding) driving population size downward to
extinction (Gilpin & Soule, 1986; Fagan & Holmes, 2006)
Friction map Layer in a geographical information system that indicates the costs that the different ecosystems of the
landscape will impose on a dispersing individual
Functional connectivity See landscape connectivity
Functional habitat Set of resources that allows the completion of the life cycle of a given organism (e.g. Dennis, Shreeve &
Van Dyck, 2003, Turlure et al., 2009)
Graph theory Mathematical structures used to model pairwise relationships between habitats. A ‘graph’ is created of a
collection of ‘nodes’ (habitat patches) and a collection of edges (corridors) that connect pairs of
habitat patches (Urban & Keitt, 2011)
Habitat quality Ability of the environment to provide conditions appropriate for individual and population persistence
(Hall et al., 1997)
Habitat selection Behavioural process by which a given individual selects its functional habitat (Stamps, 2001)
Hanski connectivity index For a landscape of i + j patches, S
i
, the connectivity of patch i, is computed as S
i
(
t
)
=
j=i
e
−αa
ij
A
j
where t is time, α is a constant setting the survival of dispersing individuals over a
ij
, the distance
between patch i and patch j,andA
j
is the area of the patch j (Hanski, 1999a)
Ideal free distribution Theoretical concept that assumes that individuals move freely between habitat patches so as to
maximise their fitness (Fretwell & Lucas, 1970). Dispersal has been proposed to distribute individuals
such that they achieve the same fitness (McPeek & Holt, 1992), thus leading to an ideal free
distribution of individuals across space
Landscape According to biogeography, an area showing homogenous geomorphological and climatic conditions
(Blondel, 1987)
Landscape connectivity Degree to which the landscape facilitates or impedes movement among resource patches. The
landscape connectivity includes both structural connectivity, i.e. the physical relationships between
habitat patches (physical distances), and functional connectivity, i.e. an organism’s behavioural response
to both the landscape structure and the landscape matrix (Taylor et al., 1993, 2006)
Landscape genetics Discipline that investigates the contemporary processes affecting patterns of genetic variation across
natural environments (Manel et al., 2003)
Least cost path modelling Method used for measuring the effective distance, rather than the Euclidian distance, between habitat
patches. Typically, a resistance map is the input to least-cost modelling. The algorithm computes the
route(s) with minimal costs that connect pairs of habitat patches (Adriaensen et al., 2003). Least cost
path models rely on the implicit assumption that dispersing individuals have total knowledge of the
landscape
Linkage strategy Methodology aimed at increasing the connectivity between patches and hence facilitating the
displacements of individuals among local populations (Bennett, 1999)
Matrix In classical metapopulation theory, all ecosystems in the landscape that are not habitat patches
Metapopulation Systems of local populations in discrete habitat patches that interact via dispersal of individuals moving
in the matrix. Such systems are buffered against extinction by gene flow among local populations,
rescue effects or recolonisation after local extinction
Pondscape Equivalent of landscape for lentic ecosystems
Phylogeography Discipline which investigates the historical processes affecting patterns of genetic variation across
natural environments (Knowles, 2009)
Resistance map See friction map
Riverscape Equivalent of landscape for lotic ecosystems
Seascape Equivalent of landscape for marine ecosystems
Stepping stones Small patches of habitat that are too small to support a viable population, but where dispersing
individuals can stop-over
Structural connectivity See landscape connectivity
Umbrella species Species selected on the assumption that they are representative of the ecosystem in which they live. The
conservation actions that promote the persistence of umbrella species in the landscape must also
promote the persistence of (many, if not all) other species of the ecosystem (Caro et al., 2005)
Biological Reviews (2012) 000–000 © 2012 The Authors. Biological Reviews © 2012 Cambridge Philosophical Society
4 M. Baguette and others
Pungetti, 2004; Crooks & Sanjayan, 2006; Hilty, Lidicker &
Merenlender, 2006; Baguette & Van Dyck, 2007; Sawyer,
Epps & Brashares, 2011). The efficiency of the linkage
strategy in increasing metapopulation persistence has been
questioned repeatedly (e.g. Simberloff et al., 1992; Burkey,
1997; Hodgson et al., 2011b). However, theory predicts
(Hanski, 1999b), and empirical studies, reviews and meta-
analyses confirm, that movements of individuals among local
populations increase metapopulation persistence (e.g. Beier
& Noss, 1998; Griffen & Drake, 2008; Stevens & Baguette,
2008; Gilbert-Norton et al., 2010; Doerr, Barrett & Doerr,
2011a).
As conservation biologists, building functional ecological
networks that shelter the metapopulation of a given species
from extinction in a given landscape is our ultimate goal,
a goal that needs a subtle blend of two ingredients:
habitat patches of sufficient high-quality and, simultaneously,
efficient linkages allowing individual transfers among these
habitats. Definitions of habitat quality and linkages mainly
depend on habitat selection and dispersal, respectively, which
unfortunately are markedly separated fields in the scientific
literature, despite being conceptually strongly related
(Chetkiewicz, Clair & Boyce, 2006; Clobert, De Fraipont &
Danchin, 2008). Here, our discussion will mostly be centred
on dispersal and the linkage strategy. Habitat selection will
not be a focus, however, we acknowledge that the study of
habitat selection, i.e. the preference of individuals for certain
habitats, is essential to most conservation strategies, both by
allowing the precise definition of habitat quality (e.g. Turlure
et al., 2009), and by determining how individuals will move
in the landscape. We suggest that habitat selection should
not be considered only as a species-specific feature. Indeed,
dispersing individuals of the same species will select different
places to settle according to their particular phenotypes;
this ‘habitat-matching’ process clearly influences how and
where individuals disperse within metapopulations (Edelaar,
Siepielski & Clobert, 2008). There is thus a need to integrate
habitat quality and linkages in future research.
For terrestrial ecosystems, the appropriate spatial scale
for the deployment of functional ecological networks is
the landscape. We introduce here the corresponding seascape,
pondscape and riverscape in marine and freshwater (lentic and
lotic) environments, respectively. For the sake of concision,
we will use landscape as a generic term covering these
four appellations, but we will address the particularities
of each of these environments when relevant. There is a
long history of controversies regarding the suitability and
accuracy of landscape as a biological scale of investigation
in ecology, leading to some paroxysmal declaration [e.g.
‘The landscape level is dead: persuading the family to take
it off the respirator’ (Allen, 1998)]. We think that these
controversies reflect the existence of two extreme conceptions
of the landscape that are rooted either in biogeography
or in behavioural ecology. According to biogeography,
the landscape is a clearly defined level of organization,
like regions or continents. The landscape is an area of
space showing homogeneous geomorphology and climate
(including water currents and flow regimes for seascapes,
pondscapes and riverscapes), and its spatial scale is thus
delineated using criteria external to the biota (e.g. Pickett
& White, 1985; Blondel, 1987). In behavioural ecology,
the landscape is defined following the individual’s own
perception of its environment, and its spatial scale depends
on the lifetime track of the organism under investigation
(Baker, 1978; Nathan et al., 2008). As a result, with this
definition, the spatial scale of landscapes is variable from one
organism to the next. Here, we adopt the first conception,
i.e. landscapes defined using geomorphological and climatic
criteria correspond to mosaics of habitats organized along
environmental gradients including ecological successions,
which offer discrete patches with similar environmental
conditions. In such landscapes, individuals will select habitats
according to their ecological needs, local populations will thus
establish in more or less discrete patches, and metapopulation
functioning will emerge on a tractable scale. In addition to
this, biogeographical landscapes most often correspond to
homogeneous areas or zones regarding human activities like
land use (residential, industrial, etc.), shipping, harvesting
practices (agro-pastoralism, forestry, fisheries, etc.), and
hence can be translated easily into administrative entities
to facilitate the implementation of the linkage strategy.
Improving linkages among habitats and local populations
should be based on a detailed knowledge of the dispersal
process in the species of interest, an essential but often
neglected issue. Both dispersal and habitat selection involve
individual variation in performances and in decisions,
especially in mobile species. Identification of critical features
of linkage habitat should hence be based on data from
large samples of individuals, to cover the range of individual
variation, and accurately estimate both the mean and the
variance of dispersal and habitat selection. This approach
thus requires multiple population-centred studies (Morris &
Diffendorfer, 2004; Morris, Diffendorfer & Lundberg, 2004;
Schtickzelle, Mennechez & Baguette, 2006).
Here we start by reviewing whether current advances
in dispersal theory could assist in the implementation of the
linkage strategy. We then investigate the relationship between
individual dispersal and landscape connectivity. Finally, we
investigate how ecological networks could emerge from the
linkage strategy within landscapes and the corresponding
seascapes, pondscapes and riverscapes. We have adopted a
general coverage of these issues, potentially applicable to both
sessile and mobile organisms, and pointed out the differences
between these two kinds of organisms when relevant.
II. INDIVIDUAL DISPERSAL AND THE LINKAGE
STRATEGY
Dispersal, the movements of individuals or propagules that
can sustain gene flow (Ronce, 2007), is a complex, multi-
causal process (see reviews in Clobert et al., 2001; Clobert,
Ims & Rousset, 2004; Matthysen, 2012), potentially leading
to both fitness costs and benefits for dispersing individuals
Biological Reviews (2012) 000–000 © 2012 The Authors. Biological Reviews © 2012 Cambridge Philosophical Society
Dispersal, connectivity and networks 5
Patch size ↓
Patch isolation ↑
Matrix previsibility ↓ or ↑
Matrix quality ↓
Edge effects ↑
Philopatry
Orientation ability
(mapping,…)
Movement rules
Responsiveness to dispersal
cues
Perceptual range
Settlement rules
Dispersal
phenotype 1
Emigration
Transf er
Immig ratio n
Philopatry
Philopatry
Information acquisition and
management
Ability to sustain under
unsuitable conditions
Inbreeding ↑
Isolation by distance ↑
Landscape processes
Genetic processes
Genetic diversity ↓
-
+
-
+
-
+
Dispersal
phenotypes
Phase-specific phenotypic
variability in dispersal
traits
Examples of selected
dispersal traits
Examples of landscape
and genetic processes
consecutive to the loss of
functional connectivity
Dispersal
phenotype 2
Dispersal
phenotype 3
Fig. 1. Landscape/dispersal interactions can generate complex eco-evolutionary feedbacks. A loss of functional connectivity has
consequences on both the structure of the landscape and the genetic structure of the local populations. Dispersal phenotypes 1, 2
and 3 result from these interactions, with modifications on traits associated with each of the three phases of the dispersal process
(emigration, transfer and immigration). Each of the three dispersal phenotypes may have a particular dispersal strategy.
(Clobert, De Fraipont & Danchin, 2008; Bonte et al.,
2012). To appraise the costs and benefits of dispersal fully,
a convenient approach is to disentangle the process into
three successive, but inter-related phases: departure out of
a habitat, transfer within the landscape, and settlement and
reproduction in a new habitat, which may or may not
be occupied by conspecifics (Stenseth & Lidicker, 1992;
Ims & Yoccoz, 1997; Bowler & Benton, 2005; Baguette &
Van Dyck, 2007; Clobert et al., 2009; Bonte et al., 2012).
Both the capacity and the decision to disperse are shaped
by particular selective pressures potentially independent of
each other (Fig. 1), while others, like parental effects during
ontogeny, will constrain the whole dispersal process (Bonte
et al., 2012; Ducatez et al., in press.
(1) Costs and benefits of dispersal
Among the multiple benefits of dispersal in heterogeneous
environments, the most prominent are the avoidance of
conspecific individuals (i.e. avoidance of kin competition,
limitation of inbreeding) and the avoidance of variation
in reproductive success associated with deteriorating
environmental conditions, both with obvious direct
consequences on individual fitness (Clobert et al., 2001).
Density of kin or conspecific individuals is thus a sensible cue
that may help mobile individuals to make the appropriate
decision to leave a habitat patch before competition reaches
a critical threshold threatening their fitness (e.g. Travis,
Murrell & Dytham, 1999). Sessile individuals are by
definition immobile, but mothers can adapt the dispersal
abilities of their offspring to the density of kin and conspecific
individuals (e.g. Allen, Buckley & Marshall, 2008).
Landscape fragmentation gives rise to dispersal costs by
increasing the distances among habitat fragments (Fahrig,
2003; Kokko & Lopez-Sepulcre, 2006; Schtickzelle et al.,
2006; Bonte et al., 2012). Dispersing individuals have to travel
longer distances across unsuitable parts of the landscape (the
matrix), which requires time and energy and increases the
risk of unsuccessful dispersal (Bonte et al., 2012). These
costs often generate phenotypic responses that decrease
dispersal propensity (the probability that an individual
leaves a habitat), or that increase dispersal efficiency in
decreasing the time spent in the matrix by changing
morphological, behavioural or physiological attributes. This
would eventually reduce dispersal costs either through a
reduced search time or through the selection of relatively
safe dispersal routes (e.g. Baguette & Van Dyck, 2007;
Schtickzelle et al., 2007; Delattre et al., 2010; Turlure et al.,
2011). In passively dispersing organisms, where sensory-
motor adaptations are obviously more difficult, the increasing
dispersal costs due to landscape fragmentation may decrease
the rate of successful dispersal, with the potential negative
Biological Reviews (2012) 000–000 © 2012 The Authors. Biological Reviews © 2012 Cambridge Philosophical Society
