OIKOS 90: 7 –19.

Copenhagen 2000

MINI- Minireviews provides an opportunity to summarize existing knowledge of selected

ecological areas, with special emphasis on current topics where rapid and significant
REVIEW advances are occurring. Reviews should be concise and not too wide-ranging. All key
references should be cited. A summary is required.

On the usage and measurement of landscape connectivity

Lutz Tischendorf and Lenore Fahrig

Tischendorf, L. and Fahrig, L. 2000. On the usage and measurement of landscape

connectivity. – Oikos 90: 7 – 19.

This paper examines the usage and measurement of ‘‘landscape connectivity’’ in 33

recent studies. Connectivity is defined as the degree to which a landscape facilitates
or impedes movement of organisms among resource patches. However, connectivity
is actually used in a variety of ways in the literature. This has led to confusion and
lack of clarity related to (1) function vs structure, (2) patch isolation vs landscape
connectivity and, (3) corridors vs connectivity. We suggest the term connectivity
should be reserved for its original purpose. We highlight nine studies; these include
modeling studies that actually measured connectivity in accordance with the defini-
tion, and empirical studies that measured key components of connectivity. We found
that measurements of connectivity provide results that can be interpreted as recom-
mending habitat fragmentation to enhance landscape connectivity. We discuss rea-
sons for this misleading conclusion, and suggest a new way of quantifying
connectivity, which avoids this problem. We also recommend a method for reducing
sampling intensity in landscape-scale empirical studies of connectivity.

L. Tischendorf and L. Fahrig, Ottawa-Carleton Inst. of Biology, Carleton Uni6.,

Ottawa, ON, Canada K1S 5B6 (present address of LT: Busestrasse 76, D-28213
Bremen, Germany [[email protected]]).

What is landscape connectivity? Henein and Merriam 1990, Adler and Nuernberger
1994, Lindenmayer and Lacy 1995, Lindenmayer and
The effects of spatial structure (patchiness) on popu- Possingham 1996, Frank and Wissel 1998, Henein et
lation dynamics were first examined in patch-based al. 1998). Movement among habitat patches is, how-
population models beginning in the early 1970s (e.g., ever, not simply a function of an organism itself, but
Levins 1969, Reddingius and den Boer 1970, Levin also depends on the landscape through which it must
1974, 1976, Roff 1974). Further modeling studies move. To emphasize the interaction between species’
showed that assumptions about movement among attributes and landscape structure in determining
habitat patches greatly influence the predictions of movement of organisms among habitat patches, Mer-
such models (e.g., Lefkovitch and Fahrig 1985, riam (1984) introduced the concept of ‘‘landscape
Fahrig 1988, 1990, Fahrig and Paloheimo 1988, connectivity’’.

Accepted 31 January 2000

Copyright © OIKOS 2000
ISSN 0030-1299
Printed in Ireland – all rights reserved
OIKOS 90:1 (2000) 7
 Taylor et al. (1993) defined landscape connectivity as Current usage of connectivity
‘‘the degree to which the landscape facilitates or im-
pedes movement among resource patches’’. Similarly, Structure or function?
With et al. (1997) defined landscape connectivity as The literature review revealed that the term connectiv-
‘‘the functional relationship among habitat patches, ity is sometimes used as a functional concept and other
owing to the spatial contagion of habitat and the
times in a structural way. Structural connectivity is
movement responses of organisms to landscape struc-
equated with habitat contiguity and is measured by
ture’’. These definitions accentuate the dependence of
analyzing landscape structure, independent of any at-
movement on landscape structure, which suggests that
tributes of the organism(s) of interest (Collinge and
connectivity is species- and landscape-specific. One
Forman 1998).
must therefore describe landscape structure from a
The functional concept of connectivity explicitly con-
species’ point of view (Wiens and Milne 1989). This
siders the behavioral responses of an organism to the
starts with defining the species’ habitat. The next step is
various landscape elements (patches and boundaries).
to determine the scale at which the species responds to
Consequently, functional connectivity covers situations
landscape structure, through its fine-scale (grain) and
where organisms venture into non-habitat (matrix),
large-scale (extent) movement (Wiens 1997). This deter-
where they may (1) face higher mortality risks (e.g.,
mines the scale of habitat pattern as perceived by the
Lidicker 1975, Gaines and McGlenaghan 1980, Krohne
organism. Finally, one must determine how the species
and Burgin 1987, Henein and Merriam 1990, Schippers
responds to the different elements of a landscape. This
et al. 1996, Charrier et al. 1997, Poole 1997, Sakai and
comprises the species movement pattern and mortality
Noon 1997), (2) express different movement patterns
risk on landscape elements (patches) as well as reactions
(e.g., Baars 1979, Rijnsdorp 1980, Wallin and Ekbom
at boundaries. Note that all of these behavioral facets
contribute toward facilitating or impeding movement 1988, Wegner and Merriam 1990, Hansson 1991, John-
among resource patches. son et al. 1992a, Andreassen et al. 1996b, Matter 1996,
In summary, landscape connectivity encapsulates the Charrier et al. 1997, Collins and Barrett 1997), and (3)
combined effects of (1) landscape structure and (2) the cross boundaries (e.g., Mader 1984, Wiens et al. 1985,
species’ use, ability to move and risk of mortality in the Bakowski and Kozakiewicz 1988, Merriam et al. 1989,
various landscape elements, on the movement rate Duelli et al. 1990, Mader et al. 1990, Frampton et al.
among habitat patches in the landscape. 1995, Mauremooto et al. 1995, Charrier et al. 1997,
Sakai and Noon 1997).
Depending on the movement attributes of the organ-
ism, structural and functional connectivity can be syn-
onymous. This occurs when the organism’s movement
Objective and approach is confined to its preferred habitat, i.e., individuals do
We reviewed the literature covered by the Agriculture, not cross the habitat/matrix boundary, and the organ-
Biology & Environmental Sciences Edition of the Cur- ism moves freely within the preferred habitat (e.g.,
rent Contents database (CC 1998), from May 1993 to Bascompte and Solé 1996). This is the assumption
November 1998. We searched article titles and key behind most percolation-based connectivity measures
words for the term connecti6ity in combination with (Gardner et al. 1987, Gardner and O’Neill 1991, Green
landscape or patch or habitat. The search resulted in 49 1994).
papers. However, 17 of these papers did not use con- The fact that structural connectivity is relatively easy
nectivity at all. We omitted these from the review, and to measure could lead to the conclusion that connectiv-
included one other paper (Doak et al. 1992) leaving 33 ity is a generalized feature of a landscape. This would
papers, which are assembled in descending chronologi- be erroneous. In fact, the same landscape will have
cal and alphabetical order in Table 1, and classified in different connectivities for different organisms. Struc-
Fig. 1. turally connected habitat patches still may not be func-
Our objective was to examine the current usage and tionally connected and even non-contiguous habitat
measurement of landscape connectivity. We start with a patches may be functionally connected, depending on
critical discussion of the diverse usage of connectivity, the species (With 1997). For example, if the only two
followed by a description of modeling and empirical habitat patches in a landscape are structurally con-
studies that actually attempted to quantify connectivity nected by an inappropriate corridor for the species in
or key components of it. We then discuss crucial model- question (too narrow or too long), structural connectiv-
ing assumptions and reveal the deceptive paradox of ity would exist without successful movement (functional
patch-based connectivity measurements, and its poten- response) from one patch to the other. Likewise, non-
tial for misleading conclusions. We end by suggesting contiguous habitat patches may functionally be con-
ways to streamline and focus research on landscape nected if the species can cross the non-habitat area
connectivity. (matrix) successfully and move between habitat

8 OIKOS 90:1 (2000)

 Table 1. Chronological and alphabetical assemblage of the 33 reviewed connectivity studies.

No. Study Measurement/usage of connectivity Comments/study target Spatial scale Study type and duration

1 Andreassen et presence/absence of corridors effects on movement frequencies landscape experiment, radio-tracking, 13 wk, 3

OIKOS 90:1 (2000)

al. 1998 generations
2 Ault and summed influence of size and spatial effect of structural patch isolation on patch observational, 2 yr
Johnson 1998 arrangement of neighboring patches community structure and population
density
3 Bjornstad et al. presence/absence of corridors effect on dispersal distances and spatial landscape experiment, live-trapping, 13 wk, 3
1998 aggregation generation
4 Collinge and percent of equally spaced straight lines effect of structural connectivity landscape, 10 experiment
Forman 1998 covering habitat within a landscape measurement on insect density, richness, m×10 m
and community structure
5 Davidson 1998 structural patch isolation rationale on the quantification of landscape discussion
landscape fragmentation
6 Haig et al. 1998 movement of waterbirds among habitat rationale on the importance of functional landscape discussion
patches connectivity for waterbird conservation
7 Henein et al. presence, absence and quality of fencerows effects on population survival landscape spatially explicit simulation model, 25 yr
1998 in simulated landscape
8 Petit and Burel distances (euclidian, along hedgerows and, effect of structural and functional landscape experiment, one season
1998b weighted by movement intensity and connectivity on species’ local abundance
mortality in different habitat types)
between sample sites
9 Petit and Burel ‘‘functional distance’’ (weighted by effect of functional connectivity on landscape experiment, one season
1998a movement intensity and mortality in species’ local abundance
different habitat types)
10 Pither and movement ability of damselflies through effects of landscape structure on landscape manipulative mark-recapture experiment,
Taylor 1998 different habitat types movement frequencies one season
11 Root 1998 nearest neighbor distance (combined with effect of structural connectivity on landscape GIS based population dynamics model
dispersal frequency distribution) between (meta)population size (RAMAS)
habitat patches
12 Grashofbokdam amount of forest habitat around patches effect of structural patch isolation on patch vegetation survey
1997 within three zones up to 1000 m zoochorous and anemochorous plant
species
13 Hof and population abundance relationship connectivity measure assumed to be landscape static optimization and simulation model
Raphael 1997 between adjacent cells in a grid model spatial limitation factor
14 Metzger and connectivity components: a) degree of rationale on the effect of connectivity on landscape conceptual model, discussion
Décamps 1997 habitat percolation, b) corridor and biodiversity
stepping stone networks, c) matrix
permeability
15 Schmiegelow et corridors (riparian buffer strips) between effect on local bird community (species patch experiment, 1 yr
al. 1997 forest fragments abundances)
16 Spetich et al. proximity index-summarized (patch effects of simulated landscape changes on patch quantification of spatial pattern in GIS
1997 area/distance to focal patch) relationship structural patch isolation (proximity) maps
for all patches located within rectangular
buffer zone around focal patch
17 Tiebout and intrinsic (juxtaposition of similar habitat) estimation of effects on specialist landscape conceptual model
Anderson 1997 and extrinsic (corridor) connectivity colonizing ability
18 With et al. average distance between two sites of a effect of landscape spatial structure on landscape random walk simulation model on
1997 grid belonging to the same (percolation) percolation threshold and populations’ neutral (random and fractal) landscape

9
cluster, populations’ spatial distribution spatial distribution maps
10
Table 1. (Continued)

No. Study Measurement/usage of connectivity Comments/study target Spatial scale Study type and duration

19 Andreassen et al. structural discontinuities in corridors effects on movement rates patch (corridor) experiment, 3 mo
1996b
20 Andreassen et al. width of corridors effects on movement rates patch (corridor) experiment, 3 mo
1996a
21 Hess 1996 physical connection between patches effect on recolonization and extinction landscape metapopulation model
(corridors) rate
22 Hjermann and habitat isolation based on dispersal effect of functional patch isolation on patch observational, 2 yr
Ims 1996 distance distributions and negative species patch occupancy
exponential dispersal function
23 Hof and Flather between-patch movement probability effect on population means and patch optimization model
1996 dependent on a) species dispersal variances
capability, b) harshness of inter-patch
environment, c) inter-patch distance
24 Lecomte and simulated presence/absence of effects on inter-patch dispersal patch experiment, 1 yr
Clobert 1996 corridors in an experimental landscape
25 Paillat and Butet total length of hedges in a 0.5-km effect of structural patch isolation on patch observational, 1 yr
1996 radius around a sampling plot (patch) species abundance and fluctuations
26 Schippers et al. movement probabilities between all effect of real landscape structure on landscape GIS-related random walk simulation model
1996 pairs of habitat patches functional connectivity
27 Schumaker 1996 dispersal success rate, fraction of effect of landscape structure on landscape GIS-related random walk simulation model
individuals that located new territories functional connectivity
28 Swart and Lawes presence/absence of corridors in effect on metapopulation persistence landscape model
1996 metapopulation model
29 Demers et al. dispersal (colonization) success effect of real landscape structure on landscape GIS-related random walk and population dynamics
1995 functional connectivity simulation model
30 Arnold et al. 1993 presence/absence of corridors and/or effect on movement rates and distances landscape observational, 3 yr
stepping stones
31 Hof and Joyce probabilistic, distance dependent optimization of habitat placement landscape static optimization model
1993 isolation of a cell in a grid model
32 Taylor et al. 1993 degree to which the landscape definition of functional connectivity landscape conceptual discussion
facilitates or impedes movement
among resource patches
33 Doak et al. 1992 search time-number of movement effect of scale of clustering on landscape random walk simulation model on hierarchical, neutral
steps required to find a new habitat functional connectivity landscape maps
patch

OIKOS 90:1 (2000)

patches. Research is needed to determine what, if any, habitat in a circle surrounding the patch. Such studies
simple measures of landscape structure can be used as may reveal the relative importance of local patch vs
measures of landscape connectivity. surrounding landscape effects. However, they do not
directly contribute to determining landscape connectiv-
ity, because they do not actually determine rates of
Patch isolation or landscape connectivity? movement among patches.

Patch isolation is determined by the rate of immigration

into the patch; the lower the immigration rate, the more
Corridors or connectivity?
isolated is the patch. Immigration rate depends on (1)
the amount of occupied habitat surrounding the focal Corridors are narrow, continuous strips of habitat that
patch, (2) the number of emigrants leaving the sur- structurally connect two otherwise non-contiguous
rounding habitat, (3) the nature of the intervening habitat patches. The corridor concept (e.g., Forman
matrix, (4) the movement and perceptual abilities of the 1983, Bennett 1990, Merriam 1991, Saunders and
organism, and (5) the mortality risk of dispersers Hobbs 1991, Lindenmayer and Nix 1993, Merriam and
(Wiens et al. 1993). Since (1) and (3) are landscape Saunders 1993, Bonner 1994, Dawson 1994, Rosenberg
structural features and (4) and (5) are the organisms’ et al. 1997, Tischendorf 1997a) originated from the
responses to landscape structure, patch isolation de- generalized assumption that organisms do not venture
pends on ‘‘the degree to which the landscape facilitates into non-habitat. Under this assumption, addition of
or impedes movement...’’ (Taylor et al. 1993). Patch any habitat to a landscape increases the ability of
isolation is therefore imbedded within the concept of organisms to move. Corridors in a landscape may
landscape connectivity. In fact, landscape connectivity therefore be a component of its connectivity if they
is essentially equivalent to the inverse of the average promote movement among habitat patches, but they do
degree of patch isolation over the landscape; a land- not determine its connectivity. The degree to which
scape including mostly patches with a high degree of corridors contribute to landscape connectivity depends
isolation will be less connected than vice versa. on the nature of the corridors, the nature of the matrix
Five of the 33 studies we reviewed equated patch and the response of the organism to both (Rosenberg et
isolation with connectivity (Hjermann and Ims 1996, al. 1997, Beier and Noss 1998).
Paillat and Butet 1996, Grashofbokdam 1997, Spetich Six of the reviewed studies equated the term connec-
et al. 1997, Ault and Johnson 1998). Even though patch tivity with the presence/absence of corridors (Hess
isolation is clearly part of landscape connectivity 1996, Lecomte and Clobert 1996, Swart and Lawes
(above), none of these studies estimated immigration 1996, Schmiegelow et al. 1997, Andreassen et al. 1998,
rates into patches. Rather, they related a species’ abun- Bjornstad et al. 1998), and two studies associated con-
dance or presence/absence in a patch to structural nectivity with corridor width (Andreassen et al. 1996a)
attributes of the surrounding landscape, such as dis- or corridor continuity (Andreassen et al. 1996b). The
tance to the nearest occupied patch, or amount of studies investigated (1) what features of a corridor

Fig. 1. Classification of the

33 reviewed studies
according to study type (a),
year of publication (b), and
usage of the term
connectivity (c).

OIKOS 90:1 (2000) 11

determine its use by the organism, (2) space-use of by randomly designating habitat cells. Cells of the grid
organisms as a function of corridor presence/absence, represented territories. An individual-based correlated
and (3) population or community responses to corridors, random walk model was used to simulate movements
e.g., species richness, diversity or abundance. None of the across the landscape. Individuals were released in a
studies explicitly recognized that corridors are only a randomly selected 50% of habitat territories, and were
component of the concept of landscape connectivity; they allowed to settle in any unoccupied territory, which then
actually equated the connecting function of corridors became unavailable to subsequent immigrants. Land-
with connectivity. scape boundaries reflected approaching individuals.
Connectivity was measured as the mean fraction (over
several runs) of individuals that successfully dispersed
into new territories during the course of a simulation. The
Measurements of connectivity results revealed correlations between each of ten indices
In this section we review studies that quantified connec- of landscape structure and dispersal success (connectiv-
tivity or key components of it. Recall that connectivity ity).
is defined as the degree to which the landscape facilitates Demers et al. (1995) (no. 29 in Table 1) investigated
or impedes movement among resource patches. Only the relationship between colonization success of edge-
four of the studies (Doak et al. 1992, Demers et al. 1995, preferring organisms, and the amount and change of edge
Schippers et al. 1996, Schumaker 1996) measured move- habitat, in real agricultural landscapes. A vector-based
ments among resource patches over the entire landscape GIS data set containing fencerow and forest-edge cover-
and actually quantified connectivity in accordance with ages was used as a model landscape. Individuals were
its definition. All of these were modeling studies and were allowed to move only in suitable habitat after being
based on simulated movements across heterogeneous dropped at random points across the landscape. Individ-
landscapes. We also review five other studies which we uals could cross inhospitable habitat (matrix) up to a
think made an important contribution toward the con- maximum distance, after any edge habitat in the land-
cept of landscape connectivity (as explained below), even scape was successfully colonized. Occupied habitat could
though they did not measure movement among resource not be colonized by subsequent dispersers. The authors
patches directly (Arnold et al. 1993, With et al. 1997, Petit measured connectivity as the ‘‘total length and area of
and Burel 1998a, b, Pither and Taylor 1998). hedgerow and forest edge colonized by the offspring of
each successful virtual organism’’. The results showed
higher connectivity in landscapes with more and longer
overall edge habitat.
Modeling studies
Dispersal success Search time
Dispersal success is usually defined as the proportion of One paper (Doak et al. 1992) (no. 33 in Table 1) used
individuals that successfully immigrate into a new habitat search time to quantify connectivity. Search time is the
patch during the course of a simulation run. Three of the number of movement steps individuals require to find a
modeling studies quantified connectivity using dispersal new habitat patch.
success. Doak et al. (1992) examined the effect of spatial scale
Schippers et al. (1996) (no. 26 in Table 1) simulated on the success of dispersing individuals. An artificial
the badger’s (Meles meles) response (movement proba- landscape was modeled by a hierarchical grid of three
bility and mortality risk) to landscape structure using a layers (spatial scales). Clusters of habitat cells were
classified GIS grid map and empirical expertise. Move- created on different spatial scales. Virtual individuals
ment probabilities between cells were derived by compar- were released in the habitat and followed a random walk
ing the quality (for badger use) of adjacent cells. Higher until a new habitat patch (different from the origin) was
quality cells attracted moving individuals. Mortality found. Landscape boundaries acted as reflecting borders.
rates were higher in low-quality cells. The number of For each individual the number of movement steps
simulated movement steps corresponded to an estimated required to find a new habitat patch (search time) was
actual time of badger movement within a four-year recorded. The mean and standard deviation over all
period. The authors produced inter-patch transition individual search times were calculated and related to the
probabilities and movement frequency maps (visits per scale of clustering. Large-scale clustering (few large
grid cell), based on dispersal success. patches) induced longer search times than small-scale
Schumaker (1996) (no. 27 in Table 1) analyzed the clustering (more smaller patches) (see also Ruckelshaus
potential of indices of landscape structure to predict et al. 1997).
dispersal success. He created landscape models in two
ways: (1) sample landscapes were randomly drawn from Population spatial distribution
a GIS data set to cover a range of different landscape With et al. (1997) (no. 18 in Table 1) investigated the
configurations; (2) artificial landscape grids were created effects of landscape spatial structure on (1) the probabil-

12 OIKOS 90:1 (2000)

ity of habitat contiguity (percolation), and (2) a popu- lands containing patches of trees or fencelines with
lation’s spatial distribution. These are not direct mea- native vegetation. They concluded that short-term
sures of landscape connectivity but are components of refuges, such as stepping stones, may be important for
landscape and population structure (respectively) that landscape connectivity (see also Schultz 1998).
the authors assumed to be related to connectivity. Although these movement studies did not actually
Artificial landscape grid maps represented either ran- measure landscape connectivity, they measured the re-
dom (spatially independent) or fractal (spatially depen- sponse of movement to changes in landscape structure.
dent) distributions of three different habitat types. These studies therefore demonstrate the dependence of
Virtual individuals followed a random walk, starting connectivity on both landscape structure and move-
from random locations within the modeled landscape. ment attributes of the organism.
Habitat type-specific residence probabilities controlled
the movement probabilities at each simulation step.
Habitat abundance, scale (for fractal maps), and the
Assumptions and methods of connectivity
spatial arrangement of habitat (random vs fractal)
turned out to be important for a population’s spatial models
aggregation. Here we discuss critical modeling assumptions for mea-
suring connectivity. We emphasize the importance of
compatibility between model parameters and empirical
data, and we discuss the potential for misleading con-
Empirical studies clusions from connectivity measures.

Functional distances
Petit and Burel (1998a, b) (nos 8, 9 in Table 1) intro-
Landscape representation
duced functional distance as a way to extrapolate an
organism’s known responses to landscape elements. In the reviewed modeling studies, landscapes were rep-
The functional distance between two points in a land- resented either by GIS data sets or by artificial habitat
scape was calculated as the sum of weighted distances. distribution maps. The advantage of artificial maps is
The weight or ‘‘cost of displacement’’ of each land- that the simulated spatial pattern is adjustable, which
scape element was a function of movement intensity allows for a systematic investigation of the effect of
and mortality rate for that element, which were spatial pattern on connectivity (With and King 1997).
quantified based on preliminary radio-tracking. High Two studies (Schumaker 1996, With et al. 1997)
movement intensity and low mortality decreased the conducted comparisons between landscape representa-
cost of displacement for a landscape element. The tions. Schumaker (1996) concluded that the ‘‘study of
authors found functional distance to be a good predic- simulated habitat pattern may provide little insight
tor of local abundance of the forest carabid Abax into the extent to which habitat fragmentation actually
parallelepipedus. Although functional distance is not alters connectivity’’. However, we disagree with this
landscape connectivity, in principle it could be inte- conclusion. Schumaker showed that his landscape in-
grated over all pairs of points in the landscape to give dex, ‘‘patch cohesion’’, responded differently to habitat
a measure of landscape connectivity. amount for maps of real vs artificial landscapes (see
also Gustafson 1998). However, he did not actually
Measuring mo6ement examine dispersal success in the artificial maps. There-
Pither and Taylor (1998) (no. 10 in Table 1) performed fore, Schumaker did not demonstrate a difference in
a manipulative mark-recapture experiment on two structure – connectivity relationships for real vs artificial
sympatric species of damselfly. Observations were landscape representations.
made in five replicates of two different landscape types. With et al. (1997) found that a population’s distribu-
Each landscape had a single type of habitat (either tion was determined mainly by the spatial arrangement
forest or pasture) between the release point of marked of habitat in random maps, and was scale-dependent in
individuals and a stream. The number of individuals fractal maps. This is supported by Doak et al. (1992),
re-observed at the stream was equated with the move- who found the ‘‘scale of clustering to be the most
ment abilities of the species through each habitat type. important feature in determining disperser perfor-
The authors found that the forest species moved sig- mance’’. In these studies the scale of clustering of the
nificantly more readily through pasture habitat, while habitat was changed without adjusting the scale at
the open habitat species moved equally well through which the organism responds to landscape structure
pasture and forest. (Wiens 1997), so the results reflect the fact that the
Arnold et al. (1993) (no. 30 in Table 1) radio-tracked range of scales at which clusters are created may ex-
movements of kangaroos in a fragmented landscape. ceed the organism’s scale (extent) of response to land-
They recorded longer movement distances across farm- scape structure.

OIKOS 90:1 (2000) 13

Movement description patch movements (which should contribute to connec-
tivity). Below we discuss an alternative way to measure
Movement was usually modeled as a simple or corre-
connectivity that avoids this problem.
lated random walk based on probabilistic jumps into
Another source for misleading conclusions is disre-
the adjacent cells of a grid. The rules differed greatly
gard of the effect of mortality on the success of move-
among the modeling studies and were only once sup-
ment across a landscape (Krohne and Dubbs 1984,
ported with empirical expertise (Schippers et al. 1996;
Krohne and Burgin 1987). Dispersal success has re-
see also With and Crist 1995). The diversity of ap-
cently been shown to be highly sensitive to mortality en
proaches to modeling movement impedes cross-com-
route (Ruckelshaus et al. 1997). Consequently, ignoring
parisons between simulation results. We therefore argue
mortality in modeling studies or interpreting empirical
for more consistency in describing movement in con-
findings without the potential effects of mortality can
nectivity-related models.
lead to false results and misleading recommendations.
To be compatible with empirical measures of move-
Two examples should clarify this. First, mean search
ment (Kareiva and Shigesada 1983, Turchin et al.
time, i.e., the average number of movement steps re-
1991), models should use vector-based movement rules
quired to find a new habitat patch, can not reflect
(e.g., Tischendorf 1997a, b, Fahrig and Johsen 1998,
mortality-induced changes in the number of successful
Tischendorf et al. 1998). Length and direction of the
immigrations. This measure is solely based on counting
movement steps determine the grain of the simulated
time steps instead of individuals, so individuals that die
organism’s response to the landscape. In addition, the
would not contribute to a lack of connectivity, even
relationship between the extent of movement and the
though they should. Second, empirical studies (e.g.,
scale of landscape pattern (e.g., inter-patch distance
Baars 1979, Rijnsdorp 1980, Liro and Szacki 1987,
relative to dispersal distance; see Fahrig (1992)) must
Garrett and Franklin 1988, Wallin and Ekbom 1988,
always be addressed when examining landscape
Kozakiewicz 1993, Charrier et al. 1997, Collins and
connectivity.
Barrett 1997, Wiens et al. 1997, Pither and Taylor 1998)
As has frequently been stated (Turchin 1991, John-
reveal movement rates or distances to be higher
son et al. 1992a, Tischendorf 1997a, Wiens et al. 1997),
through inhospitable than through hospitable habitat
movement rules need to be specific to landscape ele-
(but see Wolff et al. (1997) for the opposite effect). This
ments (e.g., Baars 1979, Rijnsdorp 1980, Wallin and
increased movement rate may result from a perceived
Ekbom 1988, Hansson 1991, Johnson et al. 1992b,
increase in predation risk. Studies that assume higher
Andreassen et al. 1996b, Matter 1996, Charrier et al.
movement, but do not include higher mortality in ma-
1997, Collins and Barrett 1997). In addition, mortality
trix, can lead to the erroneous conclusion that removal
en route may influence the inter-patch movement suc-
or fragmentation of habitat may enhance interactions
cess and should be taken into account when investigat-
among local populations at broad spatial scales. Such
ing landscape connectivity (e.g., Lidicker 1975, Gaines
conclusions ignore the crucial tradeoff between move-
and McGlenaghan 1980, Krohne and Burgin 1987,
ment and mortality and may be fatal for species that
Henein and Merriam 1990, Schippers et al. 1996, Char-
face higher mortality risks outside their habitat.
rier et al. 1997, Poole 1997, Ruckelshaus et al. 1997,
Sakai and Noon 1997).
The future of connectivity research
The role of models and empiricism
Measurements of connectivity
The effort required to measure connectivity empirically
Measuring connectivity based on patch immigration likely exceeds any feasible project. In fact, we cannot
leads to the counter-intuitive result that connectivity is rely on independent data sets on which to verify con-
zero (no successful dispersal, or infinite search time) nectivity models. For this reason, the traditional evalu-
when there is only one habitat patch in a landscape. ation of model predictions against empirical data
This goes counter to the assumption that a landscape becomes difficult, if not impossible. Therefore, empiri-
containing a single contiguous habitat patch should cal studies and models must complement each other.
have higher connectivity than a landscape with the Models should focus on revealing the relative impor-
same amount of habitat occurring in many disjoint tance of parameters and assumptions in determining
patches. Thus, conclusions drawn from these measure- connectivity. This information can be used to direct
ments indirectly advocate fragmentation to enhance empirical research if model parameters are compatible
connectivity. This would have negative consequences with empirical data. For example, a model could indi-
for conservation. The problem with these measurements cate high relative importance of a particular movement
of connectivity is that they only count inter-patch parameter to connectivity, implying that it would be
movements (which become increasingly successful the useful to collect this type of movement information in
more patches there are). They completely ignore within- empirical studies.

14 OIKOS 90:1 (2000)

Fig. 2. Principal data – source
relationships for studies
replicated at the landscape scale
(a), patch-scale studies in single
landscapes (b), and hybrid
patch-scale landscape-scale
studies (c).

Empirical studies should provide data that either measurement would not depend on the number of
reinforce or question our modeling assumptions. For patches in the landscape, which would prevent mislead-
example, the observation that many organisms move ing conclusions. Using grid cells as the basis of immi-
more in inhospitable habitat than hospitable habitat gration may also provide grounds for determining
should be included in models of connectivity. under which conditions the simplified structural view-
point can substitute for actual connectivity.
Note that this approach is different from the mea-
surement of connectivity based on percolation theory.
Immigration into equal-sized areas
Although both methods represent the landscape as a
We suggest that measurement of connectivity should be grid of cells, in the percolation approach connectivity is
based on immigration rates into equal-sized areas (terri- defined as the probability of the presence/absence of a
tories in some cases), to eliminate the counter-intuitive percolating cluster of habitat cells on the landscape
conclusion, based on patch-based measures of connec- (Gardner et al. 1987, Gardner and O’Neill 1991, Green
tivity, that habitat fragmentation is predicted to in- 1994). Using our proposed method, connectivity can
crease connectivity (as described above). This could be have a range of values, depending on the rate of
done by superimposing a grid of equal-sized cells on the successful immigration into cells. Our method does not
landscape and measuring immigration into habitat cells assume that contiguous cells are always functionally
(see Schumaker (1996) for a similar approach). This connected, as is assumed in percolation models.

OIKOS 90:1 (2000) 15

 In applying a cell-based measurement of connectivity of the term connectivity, we conclude that the field
it is important to note that the resolution of the super- currently lacks focus. The issue is not just semantics;
imposed grid implies an artificial scale of observation. erroneous conclusions and recommendations are made
The grid scale is likely to affect the value of connectivity due to inconsistent usage and paradoxical measurement
calculated. Therefore in comparing connectivities of of connectivity. We suggest the following to streamline
different landscapes for a particular species, the same future connectivity research:
grid scale should be used for all landscapes. 1. The term landscape connectivity should be reserved
for its original meaning, i.e., the degree to which the
landscape facilitates or impedes movement among re-
Hybrid patch-landscape scale empirical studies source patches.

Landscape connectivity is, by definition, an attribute of 2. Any measure of connectivity must be based on move-
an entire landscape, where the scale of the landscape is ment of an organism through a landscape. Measures of
determined by the movement scales of the species of landscape structure as well as demographic indicators,
interest (Goodwin and Fahrig 1998). Empirical studies such as species abundance and distribution, while poten-
investigating key components of connectivity must there- tially related to connectivity, are not measures of connec-
fore be replicated at the landscape scale, i.e., each single tivity.
data point in the study must be obtained from a separate 3. Relationships between measures of landscape struc-
landscape (Fig. 2a). However, studies that use the land- ture and connectivity need to be examined.
scape as the unit of observation are generally rare (e.g., 4. We recommend measuring connectivity using immi-
McGarigal and McComb 1995, Pedlar et al. 1997, Pither gration into equal-sized territories or grid cells, rather
and Taylor 1998, Trzcinski et al. 1999). than variable-sized patches. This resolves the paradox of
In contrast, patch-scale studies, in which many patches zero connectivity in single-patch landscapes, and avoids
within a single landscape are compared, are common the problematic conclusion that habitat fragmentation
(e.g., Opdam et al. 1985, van Dorp and Opdam 1987, increases landscape connectivity.
Laan and Verboom 1990, van Apeldoorn et al. 1992, 5. For field studies on connectivity we recommend the
Fitzgibbon 1993, 1997, Dunning et al. 1995, Vos and hybrid patch-landscape scale approach in which non-
Stumpel 1995, Kinnunen et al. 1996, Luiselli and Capicci overlapping landscapes are the study units, but in each
1997, Fahrig and Jonsen 1998, Delin and Andrén 1999, study unit only a single patch or sample area (at the
Hokit et al. 1999). In a patch-scale study each patch in center of the landscape) is sampled. Immigration to that
the landscape represents an individual data point (Fig. patch or area is used as an index of connectivity for the
2b). Patch-scale studies can only produce a single mea- surrounding landscape.
sure of connectivity for the single landscape of the study.
Therefore, such studies are unreplicated with respect to 6. A consistent framework for modeling assumptions
landscape connectivity. regarding movement, mortality and boundary reactions
We suggest a hybrid approach between patch-scale and is necessary to allow cross-comparisons of simulation
landscape-scale studies. In such studies, sampling would and empirical results. For example, to be compatible with
be conducted in single patches (or equal-sized areas) in empirical measures of movement, models should use
the centers of non-overlapping landscapes (Fig. 2c). The vector-based movement rules.
structure of the landscape surrounding the patches would Acknowledgements – We particularly thank Kringen Henein,
be measured and related to the rates of immigration into Darren Bender and Brett Goodwin for their comments on a
draft of this paper. A stimulating discussion with students and
the focal patches. Rate of immigration into the focal faculty members of the Landscape Ecology Laboratory at the
patches would be used as an indicator of connectivity for Ottawa-Carleton Institute of Biology also helped us to clarify
the whole landscape surrounding the patch. Such a study and relate the different viewpoints on landscape connectivity.
design allows for sampling in multiple landscapes, but This work was supported by a postdoctoral scholarship from
the Deutsche Forschungsgemeinschaft (DFG) to LT, and a
only a single patch is sampled within each landscape, thus Natural Sciences and Engineering Research Council of
reducing sampling intensity. Note that for any one Canada grant to LF.
landscape, immigration to the selected patch may not be
representative of connectivity for this landscape. How-
ever, a simulation study (Tischendorf and Fahrig un- References
publ.) suggests that this approach should generally work
well for multi-landscape comparisons. Adler, F. R. and Nuernberger, B. 1994. Persistence in patchy
irregular landscapes. – Theor. Popul. Biol. 45: 41 – 75.
Andreassen, H. P., Halle, S. and Ims, R. A. 1996a. Optimal
width of movement corridors for root voles: not too nar-
row and not too wide. – J. Appl. Ecol. 33: 63 – 70.
Conclusions Andreassen, H. P., Ims, R. A. and Stenseth, N. C. 1996b.
Discontinuous habitat corridors: effects on male root vole
From our review of the current usage and measurement movements. – J. Appl. Ecol. 33: 555 – 560.

16 OIKOS 90:1 (2000)

Andreassen, H. P., Hertzberg, K. and Ims, R. A. 1998. Space- Fahrig, L. 1992. Relative importance of spatial and temporal
use responses to habitat fragmentation and connectivity in scales in a patchy environment. – Theor. Popul. Biol. 41:
the root vole Microtus oeconomus. – Ecology 79: 1223 – 300 – 314.
1235. Fahrig, L. and Paloheimo, J. 1988. Effect of spatial arrange-
Arnold, G. W., Steven, D. E., Weeldenburg, J. R. and Smith, ment of habitat patches on local population size. – Ecol-
E. A. 1993. Influences of remnant size, spacing pattern and ogy 69: 468 – 475.
connectivity on population boundaries and demography in Fahrig, L. and Jonsen, I. D. 1998. Effect of habitat patch
Euros Macropus robustus living in a fragmented landscape. characteristics on abundance and diversity of insects in an
– Biol. Conserv. 64: 219–230. agricultural landscape. – Ecosystems 1: 197 – 205.
Ault, T. R. and Johnson, C. R. 1998. Spatially and temporally Fitzgibbon, C. D. 1993. The distribution of grey squirrel dreys
predictable fish communities on coral-reefs. – Ecol. in farm woodland: the influence of wood area, isolation
Monogr. 68: 25–50. and management. – J. Appl. Ecol. 30: 736 – 742.
Baars, M. A. 1979. Patterns of movement of radioactive Fitzgibbon, C. D. 1997. Small mammals in farm woodlands:
carabid beetles. – Oecologia 44: 125–140. the effects of habitat, isolation and surrounding land-use
Bakowski, C. and Kozakiewicz, M. 1988. The effect of forest patterns. – J. Appl. Ecol. 34: 530 – 539.
road on bank vole and yellow necked mouse populations. Forman, R. T. T. 1983. Corridors in a landscape: their ecolog-
– Acta Theriol. 33: 345–353. ical structure and function. – Ekológia (CSSR) 2: 375–
Bascompte, J. and Solé, R. V. 1996. Habitat fragmentation 387.
and extinction thresholds in spatially explicit models. – J. Frampton, G. K., Cilgi, T., Fry, G. L. A. and Wratten, S. D.
Anim. Ecol. 65: 465–473. 1995. Effects of grassy banks on the dispersal of some
Beier, P. and Noss, R. F. 1998. Do habitat corridors provide carabid beetles (Coleoptera: Carabidae) on farmland. –
connectivity? – Conserv. Biol. 12: 1241–1252. Biol. Conserv. 71: 347 – 355.
Bennett, A. F. 1990. Habitat corridors and the conservation of Frank, K. and Wissel, C. 1998. Spatial aspects of metapopula-
small mammals in a fragmented forest environment. – tion survival – from model results to rules of thumb for
Landscape Ecol. 4: 109–122. landscape management. – Landscape Ecol. 13: 363 – 379.
Bjornstad, O. N., Andreassen, H. P. and Ims, R. A. 1998. Gaines, M. S. and McGlenaghan, L. R. 1980. Dispersal in
Effects of habitat patchiness and connectivity on the spa- small mammals. – Annu. Rev. Ecol. Sys. 11: 163 – 196.
tial ecology of the root vole Microtus oeconomus. – J. Gardner, R. H. and O’Neill, R. V. 1991. Pattern, process and
Anim. Ecol. 67: 127–140. predictability: the use of neutral models for landscape
Bonner, J. 1994. Widlife’s roads to nowhere? – New Scientist analysis. – In: Turner, M. G. and Gardner, R. H. (eds),
143: 30 – 34. Quantitative methods in landscape ecology. Springer, pp.
CC 1998. Current Contents. – Institute for Scientific Informa- 289 – 307.
tion, Inc. (ISI), Philadelphia, PA. Gardner, R. H., Milne, B. T., Turner, M. G. and O’Neill, R.
Charrier, S., Petit, S. and Burel, F. 1997. Movements of Abax V. 1987. Neutral models for the analysis of broad-scale
parallelepipedus (Coleoptera, Carabidae) in woody habitats landscape pattern. – Landscape Ecol. 1: 19 – 28.
of a hedgerow network landscape: a radio-tracing study. – Garrett, M. G. and Franklin, G. L. 1988. Behavioral ecology
Agric. Ecosyst. Environ. 61: 133–144. of dispersal in the black-tailed prairie dog. – J. Mammal.
Collinge, S. K. and Forman, R. T. T. 1998. A conceptional- 69: 236 – 250.
model of land conversion processes – predictions and Goodwin, B. J. and Fahrig, L. 1998. Spatial scaling and
evidence from a microlandscape experiment with grassland animal population dynamics. – In: Peterson, D. and
insects. – Oikos 82: 66–84. Parker, V. T. (eds), Ecological scale: theory and applica-
Collins, R. J. and Barrett, G. W. 1997. Effects of habitat tion. Columbia Univ. Press, pp. 193 – 206.
fragmentation on meadow vole (Microtus pennsyl6anicus) Grashofbokdam, C. 1997. Forest species in an agricultural
population dynamics in experiment landscape patches. – landscape in the Netherlands – effects of habitat fragmen-
Landscape Ecol. 12: 63–76. tation. – J. Veg. Sci. 8: 21 – 28.
Davidson, C. 1998. Issues in measuring landscape fragmenta- Green, D. G. 1994. Connectivity and complexity in landscapes
tion. – Wildl. Soc. Bull. 26: 32–37. and ecosystems. – Pac. Conserv. Biol. 1: 194 – 200.
Dawson, D. 1994. Are habitat corridors conduits for animals Gustafson, E. J. 1998. Quantifying landscape spatial pattern:
and plants in a fragmented landscape? A review of the what is the state of the art? – Ecosystems 1: 143 – 156.
scientific evidence. – English Nature Research Report 94. Haig, S. M., Mehlman, D. W. and Oring, L. W. 1998. Avian
Delin, A. E. and Andrén, H. 1999. Effects of habitat fragmen- movements and wetland connectivity in landscape conser-
tation on Eurasian red squirrel (Sciurus 6ulgaris) in a forest vation. – Conserv. Biol. 12: 749 – 758.
landscape. – Landscape Ecol. 14: 67–71. Hansson, L. 1991. Dispersal and connectivity in metapopula-
Demers, M. N., Simpson, J. W., Boerner, R. E. J. et al. 1995. tions. – Biol. J. Linn. Soc. 42: 89 – 103.
Fencerows, edges, and implications of changing connectiv- Henein, K. and Merriam, G. 1990. The elements of connectiv-
ity illustrated by two contiguous Ohio landscapes. – Con- ity where corridor quality is variable. – Landscape Ecol. 4:
serv. Biol. 9: 1159–1168. 157 – 170.
Doak, D. F., Marino, P. C. and Kareiva, P. M. 1992. Spatial Henein, K., Wegner, J. F. and Merriam, G. 1998. Population
scale mediates the influence of habitat fragmentation on effects of landscape model manipulation on two be-
dispersal success: implications for conservation. – Theor. haviourally different woodland small mammals. – Oikos
Popul. Biol. 41: 315–336. 81: 168 – 186.
Duelli, P., Studer, M., Marchand, I. and Jakob, S. 1990. Hess, G. R. 1996. Linking extinction to connectivity and
Population movements of arthropods between natural and habitat destruction in metapopulation models. – Am. Nat.
cultivated areas. – Biol. Conserv. 54: 193–207. 148: 226 – 236.
Dunning, J. B., Borgella, R., Clements, K. and Meffe, G. K. Hjermann, D. O. and Ims, R. A. 1996. Landscape ecology of
1995. Patch isolation, corridor effects, and colonization by the wart-biter Decticus 6erruci6orus in a patchy landscape.
a resident sparrow in a managed pine woodland. – Con- – J. Anim. Ecol. 65: 768 – 780.
serv. Biol. 9: 542–550. Hof, J. and Joyce, L. A. 1993. A mixed integer linear program-
Fahrig, L. 1988. A general model of population in patchy ming approach for spatially optimizing wildlife and timber
habitats. – Appl. Math. Comp. 27: 53–66. in managed forest ecosystems. – For. Sci. 39: 816 – 834.
Fahrig, L. 1990. Interacting effects of disturbance and disper- Hof, J. and Flather, C. H. 1996. Accounting for connectivity
sal on individual selection and population stability. – and spatial correlation in the optimal placement of wildlife
Comm. Theor. Biol. 1: 275–297. habitat. – Ecol. Model. 88: 143 – 155.

OIKOS 90:1 (2000) 17

Hof, J. and Raphael, M. G. 1997. Optimization of habitat Mader, H. J., Schell, C. and Kornacker, P. 1990. Linear
placement – a case study of the Northern Spotted Owl in barriers to arthropod movements in the landscape. – Biol.
the Olympic Peninsula. – Ecol. Appl. 7: 1160–1169. Conserv. 54: 209 – 222.
Hokit, D. G., Stith, B. M. and Branch, L. C. 1999. Effects of Matter, S. F. 1996. Interpatch movement of the red milkweed
landscape structure in Florida scrub: a population perspec- beetle, Tetraopes tetraophthalmus: individual responses to
tive. – Ecol. Appl. 9: 124–134. patch size and isolation. – Oecologia 105: 447 – 453.
Johnson, A. R., Milne, B. T. and Wiens, J. A. 1992a. Diffu- Mauremooto, J. R., Wratten, S. D., Worner, S. P. and Fry, G.
sion in fractal landscapes: simulations and experimental L. A. 1995. Permeability of hedgerows to predatory cara-
studies of tenebrionid beetle movements. – Ecology 73: bid beetles. – Agric. Ecosyst. Environ. 52: 141 – 148.
1968 – 1983. McGarigal, K. and McComb, W. C. 1995. Relationships
Johnson, A. R., Wiens, J. A., Milne, B. T. and Crist, T. O. between landscape structure and breeding birds in the
1992b. Animal movements and population dynamics in oregon coast range. – Ecol. Monogr. 65: 235 – 260.
heterogeneous landscapes. – Landscape Ecol. 7: 63 – 75. Merriam, G. 1984. Connectivity: a fundamental ecological
Kareiva, P. M. and Shigesada, N. 1983. Analyzing insect characteristic of landscape pattern. – In: Brandt, J. and
movement as a correlated random walk. – Oecologia 56: Agger, P. (eds), Proceedings of the 1st international semi-
234 – 238. nar on methodology in landscape ecological research and
planning. Roskilde Univ., Denmark, pp. 5 – 15.
Kinnunen, H., Järveläinen, K., Pakkala, T. and Tiainen, J.
Merriam, G., Kozakiewicz, M., Tsuchiya, E. and Hawley, K.
1996. The effect of isolation on the occurence of farmland
1989. Barriers as boundaries for metapopulations and
carabids in a fragmented landscape. – Annu. Zool. Fenn. demes of Peromyscus leucopus in farm landscapes. – Land-
33: 165 – 171. scape Ecol. 2: 227 – 235.
Kozakiewicz, M. 1993. Habitat isolation and ecological barri- Merriam, G. 1991. Corridors and connectivity: animal popula-
ers – the effect on small mammal populations and commu- tions in heterogeneous environments. – In: Saunders, D.
nities. – Acta Theriol. 38: 1–30. and Hobbs, R. J. (eds), Nature conservation 2: the role of
Krohne, D. T. and Dubbs, B. A. 1984. An analysis of dispersal corridors. Surrey Beatty & Sons, pp. 133 – 142.
in an unmanipulated population of Peromyscus leucopus. – Merriam, G. and Saunders, D. A. 1993. Corridors in restora-
Am. Midl. Nat. 112: 146–156. tion of fragmented landscapes. – In: Saunders, D. and
Krohne, D. T. and Burgin, A. B. 1987. Relative success of Hobbs, R. J. (eds), Nature conservation 3: reconstruction
residents and immigrants in Peromyscus leucopus. – Hol- of fragmented ecosystems. Surrey Beatty & Sons, pp. 71–
arct. Ecol. 10: 196–200. 87.
Laan, R. and Verboom, B. 1990. Effects of pool size and Metzger, J. P. and Décamps, H. 1997. The structural connec-
isolation on amphibian communities. – Biol. Conserv. 54: tivity threshold: an hypothesis in conservation biology at
251 – 262. the landscape scale. – Acta Oecol. 18: 1 – 12.
Lecomte, J. and Clobert, J. 1996. Dispersal and connectivity in Opdam, P., Rijsdijk, G. and Hustings, F. 1985. Bird communi-
populations of the common lizard Lacerta 6i6ipara: an ties in small woods in an agricultural landscape: effects of
experimental approach. – Acta Oecol. 17: 585– 598. area and isolation. – Biol. Conserv. 34: 333 – 352.
Lefkovitch, L. P. and Fahrig, L. 1985. Spatial characteristics Paillat, G. and Butet, A. 1996. Spatial dynamics of the bank
of habitat patches and population survival. – Ecol. Model. vole (Clethrionomys glareolus) in a fragmented landscape.
30: 297 – 308. – Acta Oecol. 17: 553 – 559.
Levin, S. A. 1974. Dispersion and population interactions. – Pedlar, J. H., Fahrig, L. and Merriam, G. 1997. Racoon
Am. Nat. 108: 207–228. habitat use at 2 spatial scales. – J. Wildl. Manage. 61:
Levin, S. A. 1976. Population dynamic models in heteroge- 102 – 112.
neous environments. – Annu. Rev. Ecol. Syst. 7: 287 – 310. Petit, S. and Burel, F. 1998a. Effects of landscape dynamics on
Levins, R. 1969. Some demographic and genetic consequences the metapopulation of a ground beetle (Coleoptera, Cara-
of environmental heterogeneity for biological control. – bidae) in a hedgerow network. – Agric. Ecosyst. Environ.
Bull. Entomol. Soc. Am. 15: 237–240. 69 (3): 243 – 252.
Lidicker, W. Z. 1975. The role of dispersal in the demography Petit, S. and Burel, F. 1998b. Connectivity in fragmented
of small mammals. – In: Golley, F. B., Petrusewicz, K. and populations – Abax parallelepipedus in a hedgerow net-
Ryszkowski, L. (eds), Small mammals: their productivity work landscape. – Comptes Rendus de l’Académie des
and population dynamics. Cambridge Univ. Press, pp. Sciences Série III – Sciences de la vie – Life Sciences 321:
103 – 128. 55 – 61.
Lindenmayer, D. B. and Nix, H. A. 1993. Ecological principles Pither, J. and Taylor, P. D. 1998. An experimental assessment
of landscape connectivity. – Oikos 83: 166 – 174.
for the design of wildlife corridors. – Conserv. Biol. 7:
Poole, K. G. 1997. Dispersal patterns of lynx in the northwest
627 – 630.
territories. – J. Wildl. Manage. 61: 497 – 505.
Lindenmayer, D. B. and Lacy, R. C. 1995. Metapopulation
Reddingius, J. and den Boer, P. J. 1970. Simulation experi-
viability of Leadbeater Possum, Gymnobelideus leadbeateri, ments illustrating stabilization of animal numbers by
in fragmented old-growth forests. – Ecol. Appl. 5: 164 – spreading of risk. – Oecologia 5: 240 – 284.
182. Rijnsdorp, A. D. 1980. Pattern of movement in and dispersal
Lindenmayer, D. B. and Possingham, H. P. 1996. Modelling from a Dutch forest of Carabus problematicus Hbst.
the inter-relationships between habitat patchiness, dispersal (Coleoptera, Carabidae). – Oecologia 45: 274 – 281.
capability and metapopulation persistence of the endan- Roff, D. A. 1974. The analysis of a population model demon-
gered species, Leadbeater’s possum, in south-eastern Aus- strating the importance of a dispersal in a heterogeneous
tralia. – Landscape Ecol. 11: 79–105. environment. – Oecologia 15: 259 – 275.
Liro, A. and Szacki, J. 1987. Movements of field mice Apode- Root, K. V. 1998. Evaluating the effects of habitat quality,
mus agrarius (Pallas) in a suburban mosaic of habitats. – connectivity, and catastrophes on a threatened species. –
Oecologia 74: 438–440. Ecol. Appl. 18: 854 – 865.
Luiselli, L. and Capicci, D. 1997. Influences of area, isolation Rosenberg, D. K., Noon, B. R. and Meslow, E. C. 1997.
and habitat features on distribution of snakes in Mediter- Biological corridors: form, function and, efficacy. – Bio-
ranean fragmented woodlands. – Biodiv. Conserv. 6: Science 47: 677 – 687.
1339 – 1351. Ruckelshaus, M., Hartway, C. and Kareiva, P. M. 1997.
Mader, H. J. 1984. Animal habitat isolation by roads and Assessing the data requirements of spatially explicit disper-
agricultural fields. – Biol. Conserv. 29: 81–96. sal models. – Conserv. Biol. 11: 1298 – 1306.

18 OIKOS 90:1 (2000)

Sakai, H. F. and Noon, B. R. 1997. Between-habitat move- Turchin, P. B., Odendaal, F. J. and Rausher, M. D. 1991.
ment of dusky-footed woodrats and vulnerability to preda- Quantifying insect movement in the field. – Environ. Ento-
tion. – J. Wildl. Manage. 61: 343–350. mol. 20: 955 – 963.
Saunders, D. A. and Hobbs, R. J. 1991. Nature conservation van Apeldoorn, R. C., Oostenbrink, W. T., van Winden, A.
2: the rule of corridors. – Surrey Beatty & Sons. and van der Zee, F. F. 1992. Effects of habitat fragmenta-
Schippers, P., Verboom, J., Knaapen, P. and van Apeldoorn, tion on the bank vole, Clethrionomys glareolus, in an
R. C. 1996. Dispersal and habitat connectivity in complex agricultural landscape. – Oikos 65: 265 – 274.
heterogeneous landscapes: an analysis with a GIS-based van Dorp, D. and Opdam, P. 1987. Effects of patch size,
random walk model. – Ecography 19: 97–106. isolation and regional abundance on forest bird communi-
Schmiegelow, F. K. A., Machtans, C. S. and Hannon, S. J. ties. – Landscape Ecol. 1: 59 – 73.
1997. Are boreal birds resilient to forest fragmentations? Vos, C. C. and Stumpel, H. P. 1995. Comparison of habitat-
An experimental study of short-term community responses. isolation parameters in relation to fragmented distribution
– Ecology 78: 1914–1932. parameters in the tree frog (Hyla arborea). – Landscape
Schultz, C. B. 1998. Dispersal behavior and its implication for Ecol. 11: 203 – 214.
reserve design in a rare Oregon butterfly. – Conserv. Biol. Wallin, H. and Ekbom, B. S. 1988. Movements of carabid
12: 284 – 292. beetles (Coleoptera Carabidae) inhabiting cereal fields: a
Schumaker, N. H. 1996. Using landscape indices to predict field tracing study. – Oecologia 77: 39 – 43.
habitat connectivity. – Ecology 77: 1210–1225. Wegner, J. F. and Merriam, G. 1990. Use of spatial elements
Spetich, M. A., Parker, G. R. and Gustafson, E. J. 1997. in a farmland mosaic by a woodland rodent. – Biol.
Spatial and temporal relationships of old-growth and sec- Conserv. 54: 263 – 276.
ondary forests in Indiana, USA. – Nat. Areas J. 17: Wiens, J. A. 1997. Metapopulation dynamics and landscape
118 – 130. ecology. – In: Hanski, I. and Gilpin, M. E. (eds), Meta-
Swart, J. and Lawes, M. J. 1996. The effect of habitat patch population biology. Academic Press, pp. 43 – 62.
connectivity on samango monkey (Cercopithecus mitis) Wiens, J. A. and Milne, B. T. 1989. Scaling of ‘landscapes’ in
metapopulation persistence. – Ecol. Model. 93: 57 – 74. landscape ecology, or, landscape ecology from a beetle’s
Taylor, P. D., Fahrig, L., Henein, K. and Merriam, G. 1993. perspective. – Landscape Ecol. 3: 87 – 96.
Connectivity is a vital element of landscape structure. – Wiens, J. A., Crawford, C. S. and Gosz, J. R. 1985. Boundary
Oikos 68: 571–572. dynamics: a conceptual framework for studying landscape
Tiebout, H. M. and Anderson, R. A. 1997. A comparison of ecosystems. – Oikos 45: 412 – 427.
corridors and intrinsic connectivity to promote dispersal in Wiens, J. A., Stenseth, N. C., VanHorne, B. and Ims, R. A.
transient successional landscapes. – Conserv. Biol. 11: 1993. Ecological mechanisms and landscape ecology. –
620 – 627. Oikos 66: 369 – 380.
Tischendorf, L. 1997a. Corridors as conduits for small ani- Wiens, J. A., Schooley, R. L. and Weeks, R. D. 1997. Patchy
mals: attainable distances depending on movement pattern, landscapes and animal movements: do beetles percolate? –
boundary reaction and corridor width. – Oikos 79: 603 – Oikos 78: 257 – 264.
611. With, K. A. 1997. The application of neutral landscape models
Tischendorf, L. 1997b. Modeling individual movements in in conservation biology. – Conserv. Biol. 11: 1069 – 1080.
heterogeneous landscapes: potentials of a new approach. – With, K. A. and Crist, T. O. 1995. Critical thresholds in
Ecol. Model. 103: 33–42. species responses to landscape structure. – Ecology 76:
Tischendorf, L., Irmler, U. and Hingst, R. 1998. A simulation 2446 – 2459.
experiment on the potential of hedgerows as movement With, K. A. and King, A. W. 1997. The use and misuse of
corridors for forest carabids. – Ecol. Model. 106: 107 – 118. neutral landscape models in ecology. – Oikos 79: 219 – 229.
Trzcinski, M. K., Fahrig, L. and Merriam, G. 1999. Indepen- With, K. A., Gardner, R. H. and Turner, M. G. 1997.
dent effects of forest cover and fragmentation on the Landscape connectivity and population distributions in
distribution of forest breeding birds. – Ecol. Appl. 9: heterogeneous environments. – Oikos 78: 151 – 169.
586 – 593. Wolff, J. O., Schauber, E. M. and Edge, W. D. 1997. Effects
Turchin, P. B. 1991. Translating foraging movements in het- of habitat loss and fragmentation on the behavior and
erogeneous environments into the spatial distribution of demography of gray-tailed voles. – Conserv. Biol. 11:
foragers. – Ecology 72: 1253–1266. 945 – 956.

OIKOS 90:1 (2000) 19