Professional Documents
Culture Documents
1 s2.0 S0301479722024306 Main
1 s2.0 S0301479722024306 Main
Research article
* Corresponding author. Centre for Environment and Sustainability, University of Surrey, Guildford, GU2 7XH, UK.
E-mail addresses: [email protected], [email protected] (R. Martinez-Cillero).
https://doi.org/10.1016/j.jenvman.2022.116857
Received 25 August 2022; Received in revised form 18 November 2022; Accepted 20 November 2022
Available online 14 December 2022
0301-4797/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
behaviour, and (3) represent all the principal community types within home range area of the species as per equation (1) below (Shirk
the study area: the European hedgehog (Erinaceus europaeus), Hazel and McRae, 2013):
dormouse (Muscardinus avellanarius), Adder (Vipera berus), and Water √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅)
vole (Arvicola amphibius). The outputs have been used to discuss the mwr = (home range area)/π (1)
possibilities for this kind of algorithm to be widely used in the planning
sector, and suggest how this could be integrated into the Defra metric.
(2) Next, spatial data must be gathered and each environmental
feature (e.g., habitats, roads, etc.) contained in the spatial data
2. Methods
must be assigned a resistance value, which defines the traversal
cost for every environmental feature in the landscape. The com
2.1. Functional connectivity and circuit theory
bination of such data (spatial data and associated resistance
values) allows the production of the resistance raster that repre
Academic literature usually distinguishes between two types of
sents the landscape as a resistance surface. In this step, the scale
connectivity: structural connectivity, which is centred around landscape
of the analysis needs to be defined, which includes deciding on
mosaics (e.g. land cover) and their correlation with species occurrence,
the spatial extent and the cell or pixel size.
versus functional connectivity, which is centred around individual spe
(3) Finally, the sources and destinations of movement need to be
cies, recognising the specific food, shelter, territory and abiotic condi
identified; this means that the areas of suitable habitat that need
tions requirements (Fischer and Lindenmayer, 2007). Because a
to be connected must be specified. In Circuitscape this requires
landscape can be functionally connected for some species but not for
defining the habitat cores to be connected. In Omniscape, this is
others, functional connectivity is more meaningful as it recognises that
done through the source strength raster, which, for every pixel,
connectivity is essentially a species-based attribute that is based on the
defines the relative amount of current to be injected into that
habitat requirements and dispersal ability of particular species (Keeley
pixel.
et al., 2021; Watts and Handley, 2010).
Functional connectivity can be quantified by various methods, each
There are many methodological choices that a practitioner must
having a unique set of assumptions and best practices and producing
make to create these data inputs: the focal species, the scale of the
slightly different types of maps (Wade et al., 2015). One such method is
analysis, and the best way to indicate sources and destinations of
electric circuit theory, one of the most recent, yet extensively used,
movement. These decisions are not straightforward and can have a large
connectivity approaches (Dickson et al., 2019; Hall et al., 2021). It treats
impact on the results of the connectivity analysis. Because this research
the landscape as a surface of resistors (a resistance grid), where the
was an experimental implementation of the approach aimed at
current flows from source nodes to ground nodes (core areas of suitable
providing recommendations for model functional connectivity in the
habitat); it quantifies the spatial patterns of current flow and accumu
context of BNG, a series of experiments modelling the effects of different
lation, as higher resistance areas shift the flow into pathways with lower
methodological choices was undertaken, as described below.
resistance (Hall et al., 2021; McRae et al., 2008a). Circuitscape was the
first open software that allowed users to apply the logic and mathe
matics of electrical circuit theory to questions of how genes, animals, or 2.3. Experiment design
processes flow across heterogeneous landscapes (Hall et al., 2021;
McRae et al., 2008a; McRae et al., 2016a, 2016b). The underlying The following subsections explain the experiments with methodo
theoretical background of electric theory has been confirmed as suited to logical choices in detail. In these experiments one variable is changed
predict ecological connectivity and to identify important movement while leaving others fixed, to allow for comparison of the effects of such
routes (Dickson et al., 2019; McRae et al., 2016a, 2016b), making them variable. For example, we run one model per species while maintaining
realistic models in ecological terms (McRae et al., 2008b). other parameters fixed, such as cell size and buffer size; this produces
Circuitscape has commonly been used in a pairwise mode, meaning four maps (accounting for the four focal species) allowing to compare
that current flow is calculated between pairs of user-defined habitat cores the connectivity of the landscape attending only to the species’ behav
(Landau et al., 2021). This is appropriate for studies with defined source iour. The next subsections explain each of these experiments in detail,
and destination points (e.g., joining pre-defined natural reserves) and an overview of all experiments is given in Table 1.
(Phillips et al., 2021). However, when cores are not defined, for example
when they are not known or when the species are not isolated to discrete 2.3.1. The effect of the focal species
patches, omnidirectional approaches are useful for producing The first step in building functional ecological network models is
regional-scale maps of connectivity (Phillips et al., 2021). The Omni selecting the focal species for which the network is going to function
scape (Landau et al., 2021) algorithm builds on and expands Circuit since species differ in their sensitivity to changes in the landscape matrix
scape applications; one of Omniscape’s novelties is that it produces maps (Bierwagen, 2007).
of omni-directional connectivity, which provide a representation of In this experiment, we tested how the connectivity results differ
connectivity between every pair of start and endpoints in the landscape across species depending on their habitat requirements and dispersal
(Landau et al., 2021). This allows understanding and predicting how the capabilities. For this, a suite of species representative of the region’s
likelihood that an ecological process (e.g., animal movement) manifests habitat types, but with very different ecological needs, dispersal dis
itself in geographic space. tances and home ranges was selected. Comparison analysis was run for
all the species at the same spatial scales and source strength inputs to
2.2. Steps for modelling connectivity and methodological choices determine how the outputs varied due to the different species’ ecological
requirements.
In summary, the process of modelling connectivity follows these
general steps: 2.3.2. The effect of source strength inputs
This experiment was a comparative analysis of the effect of different
(1) Firstly, the focal species for which connectivity will be modelled methods for producing the source strength inputs, which specify the
must be selected. Then, its dispersal capabilities need to be sources and destinations of the individuals moving through the land
determined. In Omniscape, this value is used to indicate the scape. Despite one of the main advantages of Omniscape being that there
moving window radius (mwr), which represents the search distance is no need to pre-define habitat cores, the question of what the best
for suitable habitat (Landau et al., 2021) calculated from the approach is for defining the sources and destinations of individuals
2
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
a
Table 1 mrw = moving window radius; the obtention of these values is specified in
Summary of modelling experiments and parameters, inputs and settings used for the Supplementary Materials.
b
each model. base-case = 5 km buffer, 9 m × 9 m cell size.
Experiment Model Fixed Inputs and Settings
parametersb summary remains. Three approaches for producing the source strength input in
Effect of the Adder • base-case • Resistance raster Omniscape were examined:
focal species file for Adder
• Source strength (1) Coreless, where, for every pixel of the landscape, the source
input: core-based strength is calculated as the inverse of the resistance values (IRV).
approach
• mwra = 41 m
This option was introduced by Omniscape developers to reduce
Dormouse • base-case • Resistance raster the data requirements as there is no need to create a second raster
file Dormouse file with the sources of strength; this approach is referred to as
• Source strength coreless-IRV.
input: core-based
(2) Coreless, where, for every pixel of the landscape, the source of
approach
• mwr = 40 m strength is derived from the habitat values (HV). In this scenario,
Hedgehog • base-case • Resistance raster the source strength is equivalent to the suitability of a habitat to
file Hedgehog support a population. This approach requires the development of
• Source strength both a resistance raster and another raster of habitat values; this
input: core-based
approach
approach is referred to as coreless-HV.
• mwr = 262 m (3) Core-based approach, where the current is injected only in pre-
Water vole • base-case • Resistance raster defined core areas, representing the origin of individuals mov
file Water vole ing across the resistance surface. This approach requires the
• Source strength
development of both a resistance raster and another raster for
input: core-based
approach defining the habitat cores; this approach is referred to as core-
• mwr = 69 m based.
Effect of source Coreless-IRV: current • All species • Resistance raster
strength source strength is the • base-case layer 2.3.3. The effect of scale
inputs inverse of resistance • mwr
The scale of the analysis includes both the size of the cells of the
values • Source layer is
calculated as the raster maps and the spatial extent (area coverage) included in the
inverse of the analysis (McRae et al., 2008b). Defining the scale of the analysis is a
resistance layer; challenge because larger spatial scales and finer map resolutions can be
resistance cut-off =
computationally prohibitive (Koen et al., 2019), coarser resolutions can
60
• mwr cause small patches or barriers to be lost, and spatial extents that are too
Coreless-HV: current •All species • Resistance raster small may result in the exclusion of connectivity paths. Consequently,
source strength is • base-case file these decisions can alter the connectivity patterns of the landscape. In
habitat values • Source layer is the addition, because ecological processes and elements of biological di
habitat raster; only
versity occur at a variety of scales, a comprehensive strategy to conserve
habitats above 0.6
value are included these processes and elements must also encompass a diversity of scales.
as source Despite its relevance, little guidance is available in the literature to
• mwr help address how differences in scale can affect Omniscape’s results.
Core-based: only •All species • Resistance raster
Therefore, simulations were created with a range of cell sizes (3mx3m,
habitat cores are •base-case file
sources • Cores raster file,
9mx9m and 18mx18 m) and spatial extents (buffer areas of 1 km, 5 km,
where cores = 10 and 10 km around the development site) to explore this issue for one of
and surrounding the focal species. These cell sizes and buffers are meaningful for BNG
pixels = 0. land managers to implement site-specific conservation measures, yet
• mwr
different enough to affect connectivity patterns (e.g., 18 m will inevi
Effect of scale Cell size 3 m Hedgehog • Resistance raster
Buffer 1 km file tably lose smaller habitat patches compared to the 3 m cell).
• Source strength
input: core-based 2.3.4. Differences between Omniscape and Circuitscape
approach
This experiment compared the outputs between Omniscape and the
• mwr
Cell size 9 m Hedgehog • Resistance raster
traditional Circuitscape algorithm in a pairwise mode. This was done to
Buffer 5 km (base- file examine how these approaches differ, and to assess which might be
case) • Source strength more suitable depending on the study context. To our knowledge, no
input: core-based published literature has shown a comparison between these two
approach
modelling approaches. For this experiment, we run both algorithms for
• mwr
Cell size 18 m Hedgehog • Resistance raster two focal species using the same input files.
Buffer 10 km file
• Source strength 2.4. Study site
input: core-based
approach
• mwr These experiments were run on the case study of Blackwell Park
Differences • Omniscape: core- • Adder and • Resistance raster (Fig. 1), a potential housing development situated on what is currently
between based approach hedgehog file known as Blackwell Farm, an area of previous Green Belt land on the
Omniscape • Circuitscape: • Base-case • Core raster file
western outskirts of the town of Guildford (Surrey, UK). About 200 ha of
and pairwise mode • mwr
Circuitscape the current Blackwell Farm are proposed to be converted to the
residential-led, mixed-use, Blackwell Park development of about 1800
homes. The current Blackwell Farm comprises 269 ha of farmland with
3
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
Fig. 1. Blackwell Park development (red outline) will be located on what is currently known as Blackwell Farm (yellow outline), an extension of land, owned by the
University of Surrey, on the outskirts of Guildford primarily composed of agricultural and woodland areas. (For interpretation of the references to colour in this figure
legend, the reader is referred to the Web version of this article.)
mature hedgerows and is bounded by the ancient Blackwell Woodland of official databases, as well as through data kindly provided by Black
(70 ha) to the Northeast. well Farm Ltd and the Surrey Wildlife Trust. A detailed list of the GIS
data used is supplied as Supplementary Material. For obtaining habitat
and the resistance values associated with each GIS feature for each spe
2.5. Gathering data and running the models cies, we used expert elicitation; this expert knowledge was gathered via
an online survey. Next, the resistance, habitat and core habitat layers
A summary of the methodology followed for running the models is were created using ArcGIS Desktop 10.5 software (Redlands, 2011) and
presented here; an extended version is included as Supplementary the Gnarly Landscape Utilities toolbox (McRae et al., 2014; Shirk and
Material. McRae, 2013).
The focal species for this study were selected on the basis that they Finally, to run Omniscape, an INI file per model was prepared to
have significance in the conservation objectives for the study area and contain the model specifications. This file specifies file paths for raster
account for a wide range of dispersal behaviour: (1) European hedgehog inputs and user-specified options, such as the mwr values. Circuitscape
(Erinaceus europaeus); (2) Hazel dormouse (Muscardinus avellanarius); was run in pairwise mode. For all models, raw cumulative current maps
(3) Adder (Vipera berus); (4) Water vole (Arvicola amphibius). Their were produced. In addition, histograms displaying the distribution of the
dispersal values (mwr) were derived from a review of academic litera image pixels were created with the current flow values of the pixels on
ture (see Supplementary Material). the x-axis, and the pixel counts on the y-axis.
The spatial inputs (GIS data) were gathered through online searches
4
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
Fig. 2. Comparison of results between (1) Omniscape (coreless-IRV) and (2) and (3) Circuitscape (pairwise) for the adder and the hedgehog. (2) and (3) depict the
same Circuitscape outputs but using two different visualization techniques (histogram equalization and equal intervals respectively). The grey outline represents the
boundary of the Blackwell Farm potential development site.
5
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
Fig. 3. Comparison of cumulative current maps across four species (the adder, Hazel dormouse, hedgehog, and water vole) and three types of Omniscape inputs:
coreless-IRV (Inverse Resistance Values) coreless-HV (Habitat Value) and core-based. On the left, the whole spatial extent of the analysis is shown; the red outline
represents the boundary of the Blackwell Farm potential development site. The right-hand graphs are zoomed in to show primarily the Blackwell Farm site extent.
Maps are displayed using percentage clip visualization. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of
this article.)
for the dormouse is still highly concentrated within its patches (it is a was lower, and the current was more diffused since it is not dominated
forest-specialist species), but there is more abundance of suitable habitat by smaller high-conductivity areas. This allowed for better differentia
in the landscape matrix. This is reflected in the histogram having many tion of connectivity features such as hedgerows.
values close to zero but a larger standard deviation. Finally, the higher
mobility of, and the more general habitat suitability for the hedgehog is 3.4. Evaluating differences between species
well represented in the histogram’s much larger relative mean and
standard deviation. Note that pixel count has been log-transformed to Bearing in mind the effect of the methodological choices and visu
improve visualization, due to the extreme accumulation of close-to-zero alization techniques above the effect of the species ecology as repre
values for highly isolated species. sented through these techniques can be addressed (Fig. 6). The different
habitat needs and dispersal abilities of each of the focal species resulted
3.3. Evaluating the effect of spatial scales in vastly different connectivity patterns illustrating how the same
landscape can be highly connected for some species (e.g., hedgehog) and
Fig. 5 shows the results of the analysis scale comparison; determining disconnected for others (e.g., adders and water voles).
the appropriate scale is important because larger sizes allow for reduced Adder. Heathlands are rare, fragmented, priority habitats at risk due
software processing times, which may be critical for assessing larger to the recent decline in their extent (Natural England, 2011). Due to the
landscape scales. In this study, running the analysis at a 1 km buffer and adder’s low dispersal capabilities and fragmented habitat, the landscape
a 3 × 3m cell size took longer than running the same scenario with a 5 matrix is almost impenetrable resulting in high isolation for the species.
km buffer and a 9 × 9 cell size (12 h 12 min vs 10 h 35 min). In the Due to this, analysis in Omniscape reflects almost zero connectivity
extreme, running a 10 km buffer with a 3 × 3m cell size was estimated to values between habitat patches (Fig. 3). In contrast, Circuitscape was
take 50 days on the same laptop device. In addition, concerning running better able to show possible dispersal paths between potential habitat
times, the size of the mwr was the factor that seemed to have the greatest cores (Fig. 2a). However, it should be remembered that, although the
effect on the run times. For example, running the analysis at base-case paths reflected in Fig. 2 a suggest a degree of connectivity between
scale for the core-based approach, took 8 min for adder (mwr = 41 m), habitat patches, such connectivity is still extremely low.
10 min for dormouse (mwr = 40 m), 2 min for water vole (mwr = 69 m) Water vole. Similarly to the adder, the landscape resistance for the
and 10 h 35 min for hedgehog (mwr = 262 m). No differences were water vole is very high due to their specialization in water courses. In
found in the connectivity patterns based on buffer extent or cell size. addition, most of the watercourses in the study area cross urban areas,
However, at larger spatial scales, the maximum cumulative connectivity therefore interrupting the potential connectivity that water features
6
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
Fig. 4. Histogram representing the pixel count in logarithmic scale (y-axis), against the intensity of cumulative current in each pixel (x-axis), for the models pre
sented in Fig. 3. The continuous line represents the mean of current values; the striped line represents the standard deviation.
provide to water voles. The highest connectivity values for hedgehogs were found at the
Dormouse. Although the dormice are still isolated within their habitat southwest and northeast of the study area, as well as on the northern
patches, their potential habitat is more widespread and connected than edge of Blackwell Farm, corresponding with areas of broadleaved
the adder’s, resulting in a more permeable landscape matrix. Fig. 3 also woodland. This is consistent with the literature as hedgehogs forage in
shows that, similarly to the water vole, urban areas act as an impene pastures (which are abundant in earthworms) but mainly rest in wooded
trable barrier for the dormouse. The analysis shows the importance of areas (Driezen et al., 2007).
hedges for the dispersal of this species, since areas with higher hedge The models indicate how ground-dwelling mammals and forest
density appear more connected, which is fully consistent with the specialists, which rely on arboreal and bushy features to thrive and
literature (Büchner, 2008). disperse, can be encouraged through careful design of the development
Hedgehog. The hedgehog is the most generalist of these focal species, and associated green infrastructure (including BNG sites), whereas the
resulting in a landscape matrix that displays as far more permeable than site lacks connectivity significance for the other two species, the adder
any of the other three species, even within urban areas (Figs. 3 and 6). and the water vole. The next section describes the significance of these
7
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
Fig. 5. Cumulative current map for the hedgehog comparing the effects of buffer and cell size, using the core-based approach. Larger cell sizes result in better
visualization of intermediate connectivity areas, as the plots are not dominated by very high-connectivity areas. The upper maps display the total spatial extent of the
analysis, the lower ones are zoomed in to show primarily the Blackwell Farm site extent (grey outline).
results for Blackwell Farm and discusses the significance of these ex • Account for a wide range of dispersal behaviour (Kintsch et al.,
periments for the BNG context. 2005). This is important because using high-mobility species exclu
sively will fundamentally fail to represent the vast numbers of less
4. Discussion mobile species, for which connectivity may be a far more important
factor in their conservation. However, using low-mobility species
4.1. Main challenges for applying this connectivity modelling in a BNG exclusively may miss large-scale landscape connections.
context? • Achieve a balance of keystone, foundation, indicator, flagship and
umbrella species that represent all the principal community types
One of the objectives of this study was to investigate the challenges within the study area (Miller, 2003). In addition, particularly for
that practitioners working in non-academic planning and environmental BNG delivered close to urban areas, the inclusion of flagship species
fields can be expected to face in performing this analysis, and to suggest may grant public support for biodiversity conservation. It is impor
ways these may be overcome. Similar to the results found by Bergsten tant to highlight that a single species may fall under more than one
and Zetterberg (2013), this study faced similar difficulties when category, which emphasizes the need to define the purpose of each
implementing connectivity analysis techniques, and also defined new species carefully.
challenges. Each of these challenges is key for the successful application • Regarding the number of species to be selected, Meurant et al. (2018)
of connectivity analysis techniques in practice. found that the use of a higher number of species resulted in more
effective, comprehensive and congruent prioritization schemes. In
4.1.1. Focal species their study, 5 to 7 species per (priority) habitat type, with diverse
This challenge refers to the selection of a relevant suite of species for habitat needs and movement abilities was found to be optimal.
the connectivity analysis. This step is critical because the selection of • A workshop of regional experts, as described by (Kintsch et al.,
species will affect the prioritization outcomes (Meurant et al., 2018). 2005), could be the best approach for selecting a such suite of
Based on the literature and the results of this study, the following species.
guidelines and observations are proposed as a starting point for focal
species selection. The chosen focal species for a given scenario should: All in all, a careful assessment of the relevant focal species for in
clusion in the connectivity analysis is the first step for any practitioner
• Have significance in the conservation objectives for the study area, performing this analysis. The selection of a suite of species that represent
such as specialised, rare and/or endangered species (Ehlers Smith a range of behaviours, dispersal mechanisms and habitat requirements is
et al., 2019). For example, the creation of an ecological network for the best approach for ensuring that the modelling outputs reflect the
an endangered species will likely be advantageous for local biodi overall needs of the species in the landscape.
versity goals in the study area.
8
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
Fig. 6. Species connectivity within, and immediately around, the development site, using the core-based approach and the histogram equalize visualization. The grey outline
represents the boundary of the Blackwell Farm potential development site.
4.1.2. Model requirements for landscape data (maps) and model significance of the site for this chalk grassland species connectivity is not
parameters (habitat and resistance values) accounted for.
The lack of biological data is a well-known limiting factor for In addition, some caution is needed with the process of expert elic
calculating functional connectivity metrics (Keeley et al., 2021). The itation due to possible biases in this process (O’Hagan, 2019). In future
main input to Omniscape and Circuitscape is a comprehensive and applications, it is recommended that the order of the habitats included in
detailed habitat map of the area for analysis. This includes a land cover the questionnaire is randomized, and the dispersal for the average
map, but also other environmental features that can be strong barriers or habitat/resistance values should be evaluated and tested to reflect the
corridors for the species, such as rivers or roads. Some environmental very high levels of heterogeneity among expert opinions. If such het
features that could be meaningful for the species dispersal are not, erogeneity occurs, a more participative approach that looks for
however, well represented in the analysis due to the lack of available consensus judgement may be more appropriate (O’Hagan, 2019). In
data. For example, there is no database of Sustainable Urban Drainage addition, empirical approaches to obtaining these values can be
systems, which may be important for predicting water vole presence in considered as another option; they are the least subjective, yet they are
urban areas (Leivesley et al., 2021). Similarly, the data layer for hedges extremely time and resource-intensive (see (Braaker et al., 2014),
only includes those present in Biodiversity Opportunity Areas identified (Driezen et al., 2007)).
by the Surrey Wildlife Trust. In addition, to represent greenspaces in
urban areas, we could only obtain ORVal Parks database, which does not 4.1.3. Spatial scale
include private gardens that are suitable habitats for hedgehogs and A challenge faced by practitioners when mapping resistance-based
other urban species (Baker and Harris, 2007). connectivity across large areas is the computational power required to
The lack of model parameters for the selected species is another data run these models, which can limit the number of cells or the extent of the
availability challenge. Practitioners performing this type of analysis will study area that can be analysed (Koen et al., 2019). Solutions to over
need to produce or obtain habitat and resistance values for each habitat come this problem include the use of supercomputers (unlikely to be
and environmental feature included in the GIS data layers, which in this available for many practitioners) and/or using parallel processing (an
study was done via expert elicitation. A lack of such data for the model option that is available both in Circuitscape and Omniscape, but which
parameters was experienced for one of the focal species considered requires a suitable computer). To overcome these computational prob
initially for the analysis (the butterfly Adonis Blue, Polyommatus bel lems, practitioners can lower the spatial extent of the analysis and/or
largus, a focal species for calcareous grasslands), and this species increase cell size. However, small patches or barriers may be lost when
consequently had to be left out of the focal species set due to no expert selecting a coarser pixel size, and a smaller spatial extent may result in
survey responses. Therefore, as it stands presently, the strategic the exclusion of cores or connectivity paths that alter the connectivity
9
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
patterns of the landscape. It is well known that numerical landscape effort since more tools (e.g., Gnarly Core Mapper) are needed to produce
connectivity indices are affected by the scale (extent and pixel size) of a cores layer. However, this approach is arguably the most ecologically
the source spatial data (Pascual-Hortal and Saura, 2007) and that the grounded since the identification of habitat cores drives explicit
patterns displayed in connectivity maps when using Circuitscape in the consideration of the minimum home range size, which is necessary for
traditional pairwise mode vary depending on the scale (Koen et al., the conservation of the species population. This means that suitable
2019). habitat patches that are smaller than the minimum requirements for the
In the BNG context, large-extent connectivity maps can be useful for species are not included in the analysis, as they are not considered core
the creation of Nature Recovery Networks that may guide gains at sites.
regional scales. Additionally, long-term conservation planning should The coreless-HV approach uses habitat value as the source strength. It
cover the species conservation requirements at broad spatial scales, does not require pre-identification of cores but does require the creation
allowing connectivity to be planned for broader time horizons. At the of a habitat value raster layer. The output maps were very similar to the
same time, small pixel sizes will be useful at development scales to aid core-based approach.
developers in fine-grain decisions such as regarding how to minimize The coreless-IRV approach was produced by Omniscape’s developers
impacts on connectivity and how to maximize the connectivity onsite to reduce data needs as the resistance layer can be used as the source
through the best distribution of the BNG units (Rudnick et al., 2012). layer. This method requires obtaining a single resistance value for each
There is little guidance available in the academic literature on these class item in each data layer, as opposed to obtaining both resistance and
considerations. Consequently, a clear ‘best-practice’ framework is habitat value. It assumes that pixels with low resistance are equivalent to
lacking for practitioners to decide on study area extent or cell size. a high habitat value, which is true in many cases. For example, for a
Among the existing literature, McRae et al. (2016a, 2016b) recommend forest species, a pixel of a forest habitat will have a low resistance as well
running the models at various scales and comparing the results with as a high habitat value. However, this is not the case in several other
local permeability analysis. In the present study, we investigated how instances (e.g., a gravel path will have a low resistance value for a
different spatial scales and cell sizes affected Omniscape’s outputs for mammal, but it does not imply that it has a high habitat value). In other
the hedgehog. No changes in connectivity patterns were found based on words, habitat value will not be directly equivalent to the inverse of
the spatial extent (1, 5 and 10 km buffers around the development site) resistance value in many cases. Despite this, the connectivity patterns
nor cell size (3, 9, and 18 m cell size). At larger spatial scales, the resulting from this approach closely resembled those of the other
maximum cumulative connectivity was lower, and the current appeared methods.
more diffused, since it is not dominated by smaller superconductive Because no substantial differences were observed in this research
areas, allowing for better differentiation of connectivity features such as between the three connectivity modelling methods, future assessments
hedges. may opt to employ only the coreless-IRV approach, reducing analytical
This is contrary to the findings of Pascual-Hortal and Saura (2007) time and data needs. From an ecological point of view, this approach is
who, using Circuitscape in pairwise mode, found that the analysis of something of a pragmatic compromise because the identification of
connectivity at smaller map scales impeded the detection of habitat cores beforehand is highly appropriate as it includes more spe
larger-scaled patterns of connectivity. This contradiction can perhaps be cies parameters (for example, the size of cores that match the species’
attributed to the differences in modelling approaches between Omni home range needs).
scape and Circuitscape. In pairwise mode, Circuitscape finds the best
pathways between pairs of cores and adds up the resulting current maps 4.1.5. Dispersal distances and home ranges
for all pairs of cores; therefore, the inclusion of cores due to the use of In the connectivity modelling, the species’ ability to disperse is re
larger spatial extents of smaller pixel sizes would affect such patterns. flected in the mwr value, which reflects the capacity of an individual to
On the other hand, Omniscape produces omni-directional connectivity search for suitable habitats (Landau et al., 2021). Different studies have
using a moving window to develop an overall connectivity surface. selected mwr values using different criteria ((de Rivera et al., 2022),
Hence, we consider that Omniscape is more robust than Circuitscape for (Thorne et al., 2020), (Jennings et al., 2020)) and such values can be
mapping larger spatial extents and the tiling approach of Pascual-Hortal obtained from the existing literature (such as in this study), movement
and Saura (2007)—which allows producing connectivity maps at large data and/or expert judgement. This process was also found to be chal
spatial scales whilst maintaining small cell sizes—may be better suited lenging in the present study, since different studies report significantly
for use with Omniscape than Circuitscape for constructing high resolu different values for the same species (e.g., hedgehog), some species being
tion, large scale connectivity maps. understudied (e.g., adder) etc. It can be recommended that the chosen
Based on this and our experience in the case study, we recommend mwr for a given scenario should satisfy the following:
that future research should investigate the effect of map tiling with
Omniscape’s models. The ecological requirements of the target species • For species with several values reported in the literature, results of
clearly must also be considered when choosing the spatial scale. For studies undertaken close to the study area should be preferred to
example, for species with long dispersal distances, the spatial extent better reflect the particular behaviours of local populations.
must be larger to capture the connectivity of cores further away. Simi • It is worth distinguishing between migration distance, home range
larly, the spatial resolution can be lowered for species with large home values and daily movements. It makes ecological sense that, for
ranges, since losing small patches of suitable habitat will not affect the species with a clear dispersal phase in their life history, this value is
overall connectivity of the landscape for such species. used (e.g., the adder). However, for species that do not have a clear
dispersal phase, the radius for the home ranges or daily movements is
4.1.4. Source strength a good indicator of their ability to search for new habitats. For spe
Because this analysis is an early application of Omniscape, insuffi cies that show great variability in home ranges (e.g., the hedgehog),
cient literature could be found on the effects of using different Omni the largest value should be preferred because it will better reflect the
scape inputs for the source strength layer. Therefore, three approaches, minimum core size that is required for the species’ survival.
requiring increasing data and/or effort from practitioners were assessed. • According to the existing literature (Schloss et al., 2022), the use of
The core-based approach consisted of pre-identifying the habitat larger mwr (e.g., 50 km) allows accommodation of dispersal over
cores and using the pixels within the cores as the only current source. longer periods, while still being representative of the short distances
This is the most data-hungry option, as it requires a resistance raster that characterize less mobile species. However, we did not find an
surface, a habitat value raster surface, and dispersal distances and home ecological explanation for this assertion, and recommend using this
range values to calculate core areas, in addition to increased computing approach of large mwr to model connectivity across large spatial
10
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
scales (e.g. at the county level) only if the results are complemented habitat blocks (e.g., hedgehogs and dormice). Therefore, as previously
with smaller spatial scales where ecologically comprehensible mwr discussed, consideration of the species’ ecology and behavioural fea
(e.g., home range and dispersal distances) is used (McRae et al., tures should be the first step for selecting the best connectivity approach.
2016a, 2016b).
4.1.8. Software requirements
All in all, the best approach to selecting mwr depends on the ecology Both Circuitscape and Omniscape are open-source software and do
of the species assessed. There are many options to choose from, and a not need associated software to be run. However, creating the resistance,
careful justification based on the best evidence needs to be made. habitat and core layers in this study was done using an associated
software Gnarly Landscape Utilities (McRae et al., 2014; Shirk and
4.1.6. Interpretation of results McRae, 2013) that is only available for ArcGIS 10 with the Spatial An
Academic papers rarely indicate which visualization technique is alyst extension. ArcGIS is a licensed software and not all practitioners
used to display cumulative current flows. In addition, very little guid may have access to it. Although other approaches can be used for
ance was found on best practices for visualizing and interpreting Cir creating habitat and resistance raster, these tools were specifically
cuitscape or Omniscape outputs. However, visualization techniques created by Circuitscape developers to support connectivity analysis and
affect the interpretation of the results. For example, Fig. 2.2 and 2.3 enormously facilitate this task.
depicts the same results arising from the same input data but using two This finalises the recommendations for the challenges identified in
different visualization techniques. Whereas the landscape matrix in the literature and experienced while performing these analyses. These
Fig. 2.2 may be interpreted as highly connected, Fig. 2.3 and associated practical recommendations are intended for real-life practitioners who
species histograms in Fig. 4 indicate extremely low connectivity across need to inform important planning decisions on any landscape, devel
areas. Fig. 2.2 should therefore be interpreted as pathways of the higher opment site or area. Next, how these results can guide habitat loss
likelihood for movement of the adder, although the likelihood of using avoidance and creation is considered.
these pathways is very low. Therefore, caution must be used when
interpreting Circuitscape results because its map visualization may 4.2. Development schemes and contributions to creating/maintaining
compromise the interpretation for species that are highly isolated, such ecological networks
as the adder.
In addition, caution must be taken in some special cases. When This modelling approach allows evaluation of the effects of potential
interpreting the output maps, high current values typically represent future landscape changes on habitat connectivity, which can enable the
high movement potential, and the pattern of the current flow describes development of sustainable strategies in urban planning and the
the network of pathways with higher probabilities of movements. implementation of conservation measures that take into account habitat
However, for sections of the landscape that have little variation in connectivity, a fundamental requirement for maintaining and enhancing
resistance (e.g., highly intact landscapes), or where current can spread urban biodiversity (Braaker et al., 2014). Electric circuit models are
out in many different directions, current will not become highly particularly good at highlighting pinch points because in locations with
concentrated. This can result in valuable landscapes with moderate few options for movements, such as a green bridge over a highway,
current magnitudes incorrectly being interpreted as having low move current will concentrate into a small number of pixels (Hall et al., 2021).
ment potential (i.e., low connectivity value) if they are not subject to Therefore, the pinch points signal a high risk of connectivity loss
careful examination (Hall et al., 2021). through a relatively small amount of landcover change; something of
To avoid this, the following cautions are advised. Firstly, accompa particular importance concerning avoiding severing corridors during the
nying the connectivity maps with the histograms representing the pixel design of developments.
count against the intensity of cumulative current in each pixel (Fig. 4) is The models in Figs. 3 and 6 depict the likelihood for four focal spe
extremely valuable. The histograms depict the level of isolation of the cies to move through and around the development site; the yellow to red
species in the landscape, without accounting for the possible biases colours represent flow that is channelled around areas of high resistance
caused by the visualization techniques used in the map outputs. Sec or barriers, which can be natural or manmade. At the moment, such
ondly, careful consideration of moderate flow values is needed which, as areas are agricultural fields and urban areas. The high connectivity areas
explained above, can indicate diffuse yet effective, near-natural levels of vary depending on the species evaluated. For water voles, the area has
connectivity, with multiple redundant pathways for movement rather extremely low connectivity value. For the adder, there is suitable habitat
than weak or ineffective connectivity (Hall et al., 2021). To avoid corresponding to an area of neutral grassland, but this patch is highly
misinterpretation in these cases, Hall et al. (2021) recommend studying isolated from any other. Therefore, one could argue that the develop
the resistance and source strength surfaces alongside the model outputs, ment site and immediate surroundings cannot be considered ecologi
to discern whether the moderate current values represent large expanses cally relevant for habitat improvement or the creation of heathland and
of suitable habitat. water-related species. For the dormouse, the connected cores are highly
restricted to woodland areas that in most cases are too far apart to be
4.1.7. Circuitscape or Omniscape? properly connected in the current land configuration; existing hedges
When circuit theory is selected as the best approach to model con provide some degree of connectivity, but the strategic use of green
nectivity, practitioners will be faced with a decision between Circuit infrastructure within the new housing development may be able to
scape and Omniscape. Because Circuitscape is better known and has enhance such connectivity. For the hedgehog, current flow is more
been widely used for conservation work, practitioners may prefer to use diffuse across the whole area, but still concentrated in non-agricultural
Circuitscape as it is well accepted by the academic community, and there areas; existing hedges appear as highly important connectors. The loss of
is more literature on its use and interpretation. However, it is important habitat in high-connectivity areas could prove to be the most damaging
to guide the decision with an understanding of the strengths and limi to the connectivity of the urban planned environment for the species
tations of these models to evaluate connectivity (Rudnick et al., 2012). examined here.
Each algorithm is best suited to somewhat different applications and But in addition, these models can be used proactively to visualize the
so the choice of which to use should depend on the focal species. Cir effects of habitat creation and/or improvement, and how these could
cuitscape (in pairwise mode) is well suited to predicting movements facilitate movement for two of the focal species: the dormouse and the
between defined habitat blocks that act as obvious source and destina hedgehog. This allows developers and land planners to understand that
tion points (e.g., adders and water voles), whereas the Omniscape this area has high importance for ground-dwelling mammals and forest
approach can estimate connectivity in regions that do not have distinct specialists, and therefore how the enhancement of those habitats could
11
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
be prioritized and integrated into the development planning to enhance parameters) have been shared in a public, online data repository:
particular functional aspects of the overall landscape important for those https://data.mendeley.com/datasets/7ry6s7fxb4/1
species, and the associated BNG.
Acknowledgements
5. Concluding remarks
This research is part of the corresponding author’s Practitioner
The application of Omniscape in this research emphasizes how Doctorate in Sustainability degree funded by the Surrey Wildlife Trust
development sites are embedded in the wider landscape, and how spe and the University of Surrey. We are grateful for the contribution of the
cies connectivity modelling can form part of BNG considerations for following experts for their participation in the species surveys: Dan
such sites. Clearly, addressing habitat connectivity alone is not enough to Forman, Swansea University; Dr Angela Julian, Coordinator, ARG UK;
address BNG, since the effect of habitat loss is larger than the effect of Phill Morgan - Mammal Society; Derek Crawley, Staffordshire Mammal
habitat fragmentation per se (Fahrig, 2003). This means that, although Group; Derek Crawley Staffordshire Mammal Group; Andrew Rothwell,
interconnected habitats are essential for maintaining biodiversity over independent zoological surveyor and Mammal Recorder for Hampshire;
time, connectedness is not enough if the habitat patches are not able to Jonathan Pounder; Kathryn Killner, ACIEEM Surrey Dormouse Group;
sustain a population or an individual’s territory. Therefore, connectivity David O’Brien, NatureScot; David O’Brien, NatureScot; Abigail Gazzard,
enhancement should be incorporated as a valuable co-objective of BNG University of Reading; Carly Pettett (no current affiliation); Derek
strategies coupled with increasing the extent and quality of habitats, Crawley, The Mammal Society Staffordshire, Mammal Group; Ben Wil
which are the current objectives of BNG. liams, independent consultant; Lynn Whitfield, Secretary, Surrey Bat
Based on this research, it appears highly desirable and feasible to Group; and Sue Hooton Suffolk Bat Group.
incorporate connectivity modelling into BNG policy, for example
through its integration within Defra’s Biodiversity Metric. One way to do References
this would be by creating a single, multi-species map at an appropriate
scale by merging several individual species maps (such as the ones Baker, P., Harris, S., 2007. Urban mammals: what does the future hold? An analysis of
the factors affecting patterns of use of residential gardens in Great Britain. Mamm
generated here). This way, each cell will have a connectivity value that Rev. 37 (4), 297–315. https://doi.org/10.1111/j.1365-2907.2007.00102.x.
reflects the importance of that parcel for the overall connectivity of the Bergès, L., Avon, C., Bezombes, L., Clauzel, C., Duflot, R., Foltête, J.-C., Gaucherand, S.,
landscape. Such a value could be used as an additional multiplier within Girardet, X., Spiegelberger, T., 2020. Environmental mitigation hierarchy and
biodiversity offsets revisited through habitat connectivity modelling. J. Environ.
the BNG metric. This way, the connectivity value of the biodiversity Manag. 256, 109950 https://doi.org/10.1016/j.jenvman.2019.109950.
units within important parcels, both within the site and in the wider Bergsten, A., Zetterberg, A., 2013. To model the landscape as a network: a practitioner’s
landscape, could be reflected in the biodiversity units for the develop perspective. Landsc. Urban Plann. 119, 35–43. https://doi.org/10.1016/j.
landurbplan.2013.06.009.
ment and its BNG attributes. Further research is warranted to help guide Bierwagen, B.G., 2007. Connectivity in urbanizing landscapes: the importance of habitat
the identification of the best approach(es) to creating multi-species configuration, urban area size, and dispersal. Urban Ecosyst. 10 (1), 29–42. https://
networks for habitat connectivity within the BNG metric, as no perfect doi.org/10.1007/s11252-006-0011-6.
Bigard, C., Pioch, S., Thompson, J.D., 2017. The inclusion of biodiversity in
solution exists to maximize the benefit for all species (Santini et al.,
environmental impact assessment: policy-related progress limited by gaps and
2016). semantic confusion. J. Environ. Manag. 200, 35–45.
Overall, Omniscape multi-model habitat connectivity modelling for Braaker, S., Moretti, M., Boesch, R., Ghazoul, J., Obrist, M.K., Bontadina, F., 2014.
wildlife species has been shown here to offer a reasonably efficient and Assessing habitat connectivity for ground-dwelling animals in an urban
environment. Ecol. Appl. 24 (7), 1583–1595. https://doi.org/10.1890/13-1088.1.
cost-effective approach to representing this key aspect of biodiversity Scopus.
support in the context of development planning. This modelling pro Büchner, S., 2008. Dispersal of common dormiceMuscardinus avellanarius in a habitat
vides a strong foundation and qualitative and quantitative information mosaic. Acta Theriol. 53 (3), 259–262.
Cardinale, B.J., Duffy, J.E., Gonzalez, A., Hooper, D.U., Perrings, C., Venail, P.,
for planning decisions; this can be particularly useful for the crucial, but Narwani, A., MacE, G.M., Tilman, D., Wardle, D.A., Kinzig, A.P., Daily, G.C.,
often overlooked, maintenance of habitat networks across the land Loreau, M., Grace, J.B., Larigauderie, A., Srivastava, D.S., Naeem, S., 2012.
scape. In addition, this work also contributes to the more general field of Biodiversity loss and its impact on humanity. Nature 486 (7401), 59–67. https://doi.
org/10.1038/nature11148. Scopus.
the assessment of the movement of species due to external pressures, as Crooks, K.R., Sanjayan, M., 2006. Connectivity conservation: maintaining connections
the learnings from the experiments can be used in other environmental for nature. In: Crooks, K.R., Sanjayan, M. (Eds.), Connectivity Conservation.
management processes and the creation of adaptation plans for indi Cambridge University Press, Cambridge Core, pp. 1–20. https://doi.org/10.1017/
CBO9780511754821.001.
vidual species. Crosher, I., Gold, S., Heaver, M., Heydon, M., Moore, L., Panks, S., Scott, S., Stone, D.,
White, N., 2019. The Biodiversity Metric 2.0: Auditing and Accounting for
Credit author statement Biodiversity Value. User Guide (Beta Version, July 2019). Natural England. http://n
epubprod.appspot.com/publication/5850908674228224.
de Rivera, C.E., Bliss-Ketchum, L.L., Lafrenz, M.D., Hanson, A.V., McKinney-Wise, L.E.,
Rocio Martinez-Cillero: Conceptualization; Data curation; Formal Rodriguez, A.H., Schultz, J., Simmons, A.L., Taylor Rodriguez, D., Temple, A.H.,
analysis; Investigation; Methodology; Project administration; Resources; Wheat, R.E., 2022. Visualizing connectivity for wildlife in a world without roads.
Front. Environ. Sci. 10. https://www.frontiersin.org/article/10.3389/fenvs.2022.
Software; Visualization; Writing - original draft, review & editing. Ben
757954.
Siggery: Methodology; Resources. Prof Richard Murphy: Supervision; Defra, 2019. In: Environment Bill Summer Policy Statement: July 2019. Department for
writing – review & editing. Dr Alvaro Perez-Diaz: Formal analysis; Environment, Food & Rural Affairs, London. https://www.gov.uk/government/pub
Methodology; Resources; Writing - review & editing. Ian Christie: Su lications/draft-environment-principles-and-governance-bill-2018/environment-bill-
summer-policy-statement-july-2019.
pervision. Sarah Jane Chimbwandira: Supervision. Dickson, B.G., Albano, C.M., Anantharaman, R., Beier, P., Fargione, J., Graves, T.A.,
Gray, M.E., Hall, K.R., Lawler, J.J., Leonard, P.B., Littlefield, C.E., McClure, M.L.,
Declaration of competing interest Novembre, J., Schloss, C.A., Schumaker, N.H., Shah, V.B., Theobald, D.M., 2019.
Circuit-theory applications to connectivity science and conservation. Conserv. Biol.
33 (2), 239–249. https://doi.org/10.1111/cobi.13230.
The authors declare that they have no known competing financial Driezen, K., Adriaensen, F., Rondinini, C., Doncaster, C.P., Matthysen, E., 2007.
interests or personal relationships that could have appeared to influence Evaluating least-cost model predictions with empirical dispersal data: a case-study
using radiotracking data of hedgehogs (Erinaceus europaeus). Ecol. Model. 209
the work reported in this paper. (2–4), 314–322. https://doi.org/10.1016/j.ecolmodel.2007.07.002. Scopus.
Ehlers Smith, D.A., Ehlers Smith, Y.C., Downs, C.T., 2019. Promoting functional
Data availability connectivity of anthropogenically-fragmented forest patches for multiple taxa across
a critically endangered biome. Landsc. Urban Plann. 190, 103579 https://doi.org/
10.1016/j.landurbplan.2019.05.010.
The data suporting the models in this article (habitat and resistance
12
R. Martinez-Cillero et al. Journal of Environmental Management 328 (2023) 116857
Natural England, 2020. The biodiveristy metric 2.0—beta test version consultation McRae, B.H., Shah, V., Edelman, A., 2016b. Circuitscape: modeling landscape
response. http://publications.naturalengland.org.uk/publication/5850908674228 connectivity to promote conservation and human health. Nat. Conserv. 14.
224. Meurant, M., Gonzalez, A., Doxa, A., Albert, C.H., 2018. Selecting surrogate species for
Environment Act, 2021. https://www.legislation.gov.uk/ukpga/2021/30/part/7/enact connectivity conservation. Biol. Conserv. 227, 326–334. https://doi.org/10.1016/j.
ed. biocon.2018.09.028.
Fahrig, L., 2003. Effects of habitat fragmentation on biodiversity. Annu. Rev. Ecol. Evol. Miller, B., 2003. Southern Rockies Wildlands Network Vision: A Science-Based Approach
Syst. 34 (1), 487–515. to Rewilding the Southern Rockies. Colorado Mountain Club Press.
Fischer, J., Lindenmayer, D.B., 2007. Landscape modification and habitat fragmentation: Natural England, N., 2011. Lowland heathland—a cultural and endangered
a synthesis. Global Ecol. Biogeogr. 16 (3), 265–280. https://doi.org/10.1111/ landscape—IN86. Nat. England - Access Evidence. http://publications.naturaleng
j.1466-8238.2007.00287.x. land.org.uk/publication/81012.
Hall, K.R., Anantharaman, R., Landau, V.A., Clark, M., Dickson, B.G., Jones, A., Platt, J., Ng, C.N., Xie, Y.J., Yu, X.J., 2013. Integrating landscape connectivity into the evaluation
Edelman, A., Shah, V.B., 2021. Circuitscape in julia: empowering dynamic of ecosystem services for biodiversity conservation and its implications for landscape
approaches to connectivity assessment. Land 10 (3). https://doi.org/10.3390/ planning. Appl. Geogr. 42, 1–12. https://doi.org/10.1016/j.apgeog.2013.04.015.
land10030301. Opdam, P., Steingröver, E., Rooij, S. van, 2006. Ecological networks: a spatial concept for
IPBES, 2019. In: Díaz, S., Settele, J., Brondízio, E.S.E.S., Ngo, H.T., Guèze, M., Agard, J., multi-actor planning of sustainable landscapes. Landscapes Sustain. 75 (3), 322–332.
Arneth, A., Balvanera, P., Brauman, K.A., Butchart, S.H.M., Chan, K.M.A., https://doi.org/10.1016/j.landurbplan.2005.02.015.
Garibaldi, L.A., Ichii, K., Liu, J., Subramanian, S.M., Midgley, G.F., Miloslavich, P., O’Hagan, A., 2019. Expert knowledge elicitation: subjective but scientific. Am.
Molnár, Z., Obura, D., Pfaff, A., Polasky, S., Purvis, A., Razzaque, J., Reyers, B., Roy Statistician 73 (Suppl. 1), 69–81. https://doi.org/10.1080/
Chowdhury, R., Shin, Y.J., Visseren-Hamakers, I.J., Willis, K.J., Zayas, C.N. (Eds.), 00031305.2018.1518265.
Summary for Policymakers of the Global Assessment Report on Biodiversity and Pascual-Hortal, L., Saura, S., 2007. Impact of spatial scale on the identification of critical
Ecosystem Services of the Intergovernmental Science-Policy Platform on Biodiversity habitat patches for the maintenance of landscape connectivity. Landsc. Urban Plann.
and Ecosystem Services. IPBES secretariat, p. 56. 83 (2), 176–186. https://doi.org/10.1016/j.landurbplan.2007.04.003.
Jennings, M.K., Zeller, K.A., Lewison, R.L., 2020. Supporting adaptive connectivity in Phillips, P., Clark, M.M., Baral, S., Koen, E.L., Bowman, J., 2021. Comparison of methods
dynamic landscapes. Land 9 (9). https://doi.org/10.3390/land9090295. for estimating omnidirectional landscape connectivity. Landsc. Ecol. 36 (6),
Keeley, A.T.H., Beier, P., Jenness, J.S., 2021. Connectivity metrics for conservation 1647–1661. https://doi.org/10.1007/s10980-021-01254-2.
planning and monitoring. Biol. Conserv. 255, 109008 https://doi.org/10.1016/j. Redlands, C.E.S.R.I., 2011. ArcGIS Desktop: Release, 10.5.
biocon.2021.109008. Rudnick, D., Beier, P., Cushman, S., Dieffenbach, F., Epps, C.W., Gerber, L., Hartter, J.,
Kintsch, J., Crooks, K.R., Pague, C., Theobald, D.M., Vanderhoof, M., Wostl, R., 2005. In: Jenness, J., Kintsch, J., Merenlender, A.M., Perkle, R.M., Preziosi, D.V., Ryan, S.J.,
Linking Colorado’s Landscapes: A Statewide Assessment of Wildlife Linkages Phase I. Trombulak, S.C., 2012. The Role of Landscape Connectivity in Planning and
Southern Rockies Ecosystem Project. https://www.codot.gov/programs/en Implementing Conservation and Restoration Priorities. Issues in Ecology. Report No.
vironmental/wildlife/wildlife-transportation-summit/background-documents/lin 16. Ecological Society of America, Washington, DC.
king-co-landscapes.pdf. Santini, L., Saura, S., Rondinini, C., 2016. A composite network approach for assessing
Koen, E.L., Ellington, E.H., Bowman, J., 2019. Mapping landscape connectivity for large multi-species connectivity: an application to road defragmentation prioritisation.
spatial extents. Landsc. Ecol. 34 (10), 2421–2433. https://doi.org/10.1007/s10980- PLoS One 11 (10), e0164794.
019-00897-6. Schloss, C.A., Cameron, D.R., McRae, B.H., Theobald, D.M., Jones, A., 2022. No-regrets”
Landau, V.A., Shah, V.B., Anantharaman, R., Hall, K.R., 2021. Omniscape. Jl: software to pathways for navigating climate change: planning for connectivity with land use,
compute omnidirectional landscape connectivity. J. Open Source Software 6 (57), topography, and climate. Ecol. Appl. 32 (1), e02468 https://doi.org/10.1002/
2829. eap.2468.
Leivesley, J.A., Stewart, R.A., Paterson, V., McCafferty, D.J., 2021. Potential importance Shirk, A.J., McRae, B.H., 2013. In: Gnarly Landscape Utilities: Core Mapper User Guide.
of urban areas for water voles: Arvicola amphibius. Eur. J. Wildl. Res. 67 (1) https:// The Nature Conservancy, Fort Collins, CO. https://circuitscape.org/gnarly-l
doi.org/10.1007/s10344-021-01467-5. Scopus. andscape-utilities/.
Maxwell, S.L., Fuller, R.A., Brooks, T.M., Watson, J.E., 2016. Biodiversity: the ravages of Tarabon, S., Bergès, L., Dutoit, T., Isselin-Nondedeu, F., 2019. Maximizing habitat
guns, nets and bulldozers. Nat. News 536 (7615), 143. connectivity in the mitigation hierarchy. A case study on three terrestrial mammals
McDonald, R., Colbert, M., Hamann, M., Simkin, R., Walsh, B., Ascensão, F., Barton, M., in an urban environment. J. Environ. Manag. 243, 340–349. https://doi.org/
Crossman, K., Edgecomb, M., Elmqvist, T., Gonzalez, A., Guneralp, B., Haase, D., 10.1016/j.jenvman.2019.04.121.
Hillel, O., Huang, K., Maddox, D., Mansur, A., Paque, J., Pereira, H.M., et al., 2018. Tarabon, S., Calvet, C., Delbar, V., Dutoit, T., Isselin-Nondedeu, F., 2020. Integrating a
Nature in the Urban Century, A Global Assessment of where and how to conserve landscape connectivity approach into mitigation hierarchy planning by anticipating
nature for biodiversity and human wellbeing. Nat. Conserv. urban dynamics. Landsc. Urban Plann. 202, 103871 https://doi.org/10.1016/j.
McKinney, M.L., 2002. Urbanization, biodiversity, and conservation. Bioscience 52 (10), landurbplan.2020.103871.
883–890. https://doi.org/10.1641/0006-3568. Scopus (2002)052[0883:UBAC]2.0. Thorne, J.H., Choe, H., Boynton, R.M., Lee, D.K., 2020. Open space networks can guide
CO;2. urban renewal in a megacity. Environ. Res. Lett. 15 (9), 094080 https://doi.org/
McRae, B.H., Dickson, B.G., Keitt, T.H., Shah, V.B., 2008a. Using circuit theory to model 10.1088/1748-9326/ab9fad.
connectivity in ecology, evolution, and conservation. Ecology 89 (10), 2712–2724. UNPD, 2018. In: World Urbanization Prospects: the 2018 Revision. United Nations
McRae, B.H., Dickson, B.G., Keitt, T.H., Shah, V.B., 2008b. Using circuit theory to model Population Division, New York. https://esa.un.org/unpd/wup/publications/Files/W
connectivity in ecology, evolution, and conservation. Ecology 89 (10), 2712–2724. UP2018-KeyFacts.pdf.
McRae, B.H., Shirk, A.J., Platt, J.T., 2014. In: Gnarly Landscape Utilities: Resistance and Wade, A.A., McKelvey, K.S., Schwartz, M.K., 2015. Resistance-surface-based wildlife
Habitat Calculator User Guide. The Nature Conservancy, Fort Collins, CO. https://cir conservation connectivity modeling: summary of efforts in the United States and
cuitscape.org/gnarly-landscape-utilities/. guide for practitioners. In: Gen. Tech. Rep. RMRS-GTR-333, vol. 93. US Department
McRae, B.H., Popper, K., Jones, A., Schindel, M., Buttrick, S., Hall, K., Unnasch, R.S., of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, CO,
Platt, J., 2016a. In: Conserving Nature’s Stage: Mapping Omnidirectional p. 333.
Connectivity for Resilient Terrestrial Landscapes in the Pacific Northwest. The Watts, K., Handley, P., 2010. Developing a functional connectivity indicator to detect
Nature Conservancy, Portland Oregon. https://conservationgateway.org//Con change in fragmented landscapes. Ecol. Indicat. 10 (2), 552–557. https://doi.org/
servationByGeography/NorthAmerica/UnitedStates/oregon/science/Pages/Resili 10.1016/j.ecolind.2009.07.009.
ent-Landscapes.aspx.
13