SYSTEMATIC CONSERVATION PLANNING IN THE WILD HARTS STUDY
AREA
by
Ian Curtis
B.Sc., University of Northern British Columbia, 2015

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
IN
NATURAL RESOURCES AND ENVIRONMENTAL STUDIES

UNIVERSITY OF NORTHERN BRITISH COULUMBIA
October 2018
© Ian Curtis, 2018

ABSTRACT
The Wild Harts Study Area (WHSA) supports a wide range of ecological diversity and
connects a network of protected areas spanning the length of the Rocky Mountain Cordillera.
The WHSA occurs within a region of northeastern British Columbia that is poorly
represented by protected areas. Industrial expansion threatens to fragment the contiguous
habitats found in the WHSA — reducing the ability of the area to perform important
landscape functions at local and continental scales. For this research, I employed a systematic
conservation planning approach to prioritize lands for conservation action in the WHSA. The
software MARXAN with integer linear programming (ILP) was used to produce optimal
solutions for conservation, at the lowest cost, and to enforce principles of protected area
design. Priority lands for conservation action were those that met the science-based targets
for a selection of ecological surrogates, displayed low edge-to-area ratios, and represented
intact landscapes outside the influence of resource development. The finalized conservation
portfolio produced in this research is meant to inform protected area planning in the WHSA.

ii

TABLE OF CONTENTS
ABSTRACT ............................................................................................................................ ii
TABLE OF CONTENTS ....................................................................................................... iii
LIST OF TABLES ................................................................................................................. vii
LIST OF FIGURES .............................................................................................................. viii
GLOSSARY ........................................................................................................................... xi
AKNOWLEDGEMENTS ................................................................................................... xvii
1.0 INTRODUCTION ..............................................................................................................1
1.1 Overview of Protected Areas in Canada .........................................................................1
1.2 The Ecology of the Anthropocene ..................................................................................2
1.3 Commitments and Challenges ........................................................................................4
1.4 The Wild Harts Study Area: An Opportunity for Conservation .....................................5
1.5 Research Purpose and Associated Questions ..................................................................8
2.0 LITERATURE REVIEW ...................................................................................................9
2.1 From Singles to Systems: A Brief History of Protected Area Design in Canada ...........9
2.2 Deciding What to Conserve ..........................................................................................14
2.3 Deciding How Much to Conserve .................................................................................16
2.4 The Right Tool for the Job ............................................................................................18
3.0 CASE STUDY ..................................................................................................................21
3.1 Overview of the Peace River Break ..............................................................................21

iii

3.2 The Wild Harts Study Area ...........................................................................................22
3.2.1 Geography & Natural History ................................................................................24
3.2.2 The Human Element ...............................................................................................26
3.2.3 Resource Development...........................................................................................26
3.2.4 Conservation in the Wild Harts Study Area ...........................................................27
4.0 METHODS .......................................................................................................................31
4.2 Step 1 Selection and Development of Conservation Features ......................................32
4.2.1 Coarse-filter Conservation Features .......................................................................32
4.2.2 Fine-filter Conservation Features ...........................................................................37
4.3 Step 2 Identify Conservation Goals and Targets for the WHSA ..................................43
4.4 Step 3 Review Existing Protected Areas within the WHSA.........................................46
4.5 Step 4 Use MARXAN-ILP to Generate a Portfolio of Priority Lands for
Conservation .......................................................................................................................46
4.5.1 Defining MARXAN-ILP Inputs .............................................................................46
4.6 Running MARXAN-ILP ...............................................................................................52
4.7 Scenario Development ..................................................................................................56
4.8 Statistical Analysis ........................................................................................................57
5.0 RESULTS .........................................................................................................................58
5.1 Conservation Feature Layers ........................................................................................58
5.1.1 Coarse-filter Conservation Features .......................................................................58

iv

5.1.2 Fine-filter Conservation Features ...........................................................................64
5.2 Gap Analysis .................................................................................................................70
5.3 Cost Surfaces ................................................................................................................71
5.3.1 High-Sensitivity......................................................................................................71
5.3.2 Medium-Sensitivity ................................................................................................71
5.3.3 Low-Sensitivity ......................................................................................................72
5.4 MARXAN-ILP Scenario Outputs .................................................................................76
5.4.1 Observable Patterns ................................................................................................76
5.4.2 Influence of Cost Surface: ......................................................................................78
5.4.3 Patch Statistics........................................................................................................80
6.0 DISCUSSION ...................................................................................................................81
6.1 What Areas Contain High Conservation Value for Select Coarse- and Fine-filter
Features? .............................................................................................................................81
6.2 What Areas Retain Conservation Value Despite the Presence of Resource
Development? .....................................................................................................................84
6.3 What is an optimal portfolio from which to direct conservation planning efforts? ......86
6.4 Next Steps .....................................................................................................................90
7.0 LIMITATIONS.................................................................................................................92
7.1 Limitations of Conducting Systematic Conservation Planning in the Wild Harts
Study Area ..........................................................................................................................92

v

7.2 Limitations of the MARXAN-ILP Model ....................................................................94
8.0 RECOMMENDATIONS..................................................................................................96
9.0 CONCLUSION ................................................................................................................98
10.0 REFERENCES .............................................................................................................100
APPENDIX 1: Spatial information used for MARXAN-ILP model ...................................108
APPENDIX 2: Cost Surface Geoprocessing Steps ..............................................................110
APPENDIX 3: Solution spaces produced in trial scenarios prioritizing areas for
conservation in the Wild Harts Study Area. .........................................................................111

vi

LIST OF TABLES
Table 1. The four steps of systematic conservation planning used to prioritize lands for
conservation in the Wild Harts Study Area. ...........................................................................31
Table 2. Nomenclature of the forest pattern and process conservation features used in
the MARXAN-ILP analysis to prioritize areas for conservation in the Wild Harts Study
Area.........................................................................................................................................36
Table 3. Targets assigned to each conservation feature for application in MARXAN-ILP
scenarios. ................................................................................................................................45
Table 4. Buffer distances used for the construction of the high-sensitivity cost surface to
represent the zone of influence in the Wild Harts Study Area. ..............................................49
Table 5. Description of the MARXAN-ILP scenario inputs used to prioritize lands for
conservation in the Wild Harts Study Area. ...........................................................................56
Table 6. Location and size of each forest pattern and process layer in relation to the
Wild Harts Study Area (WHSA). ...........................................................................................63
Table 7. Representation of conservation features achieved by existing protected areas in
the Wild Harts Study Area. .....................................................................................................70
Table 8. Patch statistics of four MARXAN-ILP scenarios used to prioritize lands for
conservation in the Wild Harts Study Area. ...........................................................................80

vii

LIST OF FIGURES
Figure 1. Boundaries of the Peace River Break and Wild Harts Study Area, British
Columbia...................................................................................................................................7
Figure 2. The Wild Harts Study Area juxtaposed against the Eastern System
Physiographic Region and linear development in the Peace River Break..............................23
Figure 3. The position of the Wild Harts Study Area in relation to the Muskwa-Kechika
Management Area, Central Rocky Mountains, and Yellowstone to Yukon Corridor............28
Figure 4. Existing protected area representation in the Wild Harts Study Area.....................30
Figure 5. Cross-section of the MARXAN-ILP model used to prioritize lands for
conservation in the Wild Harts Study Area. ...........................................................................53
Figure 6. Cross-section of the MARXAN-ILP conservation value calculations used to
prioritize lands for conservation in the Wild Harts Study Area. ............................................55
Figure 7. Spatial extent of the land facet diversity layer used by MARXAN-ILP to
prioritize lands for conservation in the Wild Harts Study Area; inset represented by red
frame. ......................................................................................................................................60
Figure 8. Spatial extent of the land facet rarity layer used by MARXAN-ILP to
prioritize lands for conservation in the Wild Harts Study Area; inset represented by red
frame. ......................................................................................................................................61
Figure 9. Spatial extent of all forest pattern and process layers used by MARXAN-ILP
to prioritize lands for conservation in the Wild Harts Study Area; inset represented by
red frame. ................................................................................................................................62

viii

Figure 10. Spatial extent of the woodland caribou layers used by MARXAN-ILP to
prioritize lands for conservation in the Wild Harts Study Area; inset represented by red
frame. ......................................................................................................................................66
Figure 11. Spatial extent of the bull trout layer used by MARXAN-ILP to prioritize
lands for conservation in the Wild Harts Study Area; inset represented by red frame. .........67
Figure 12. Spatial extent of the grizzly bear layer used by MARXAN-ILP to prioritize
lands for conservation in the Wild Harts Study Area; inset represented by red frame. .........68
Figure 13. Spatial extent of the special features layer used by MARXAN-ILP to
prioritize lands for conservation in the Wild Harts Study Area; inset represented by red
frame. ......................................................................................................................................69
Figure 14. Spatial extent of the high-sensitivity cost surface used to constrain
MARXAN-ILP when prioritizing lands for conservation in the Wild Harts Study Area;
inset represented by green frame. ...........................................................................................73
Figure 15. Spatial extent of the medium-sensitivity cost surface used to constrain
MARXAN-ILP when prioritizing lands for conservation in the Wild Harts Study Area;
inset represented by green frame. ...........................................................................................74
Figure 16. Spatial extent of the low-sensitivity cost surface used to constrain
MARXAN-ILP when prioritizing lands for conservation in the Wild Harts Study Area;
inset represented by green frame. ...........................................................................................75
Figure 17. Four portfolios produced by MARXAN-ILP to prioritize lands for
conservation in the Wild Harts Study Area; A: high-sensitivity cost surface used; B:
medium-sensitivity cost surface used; C: low-sensitivity cost surface used; D: no cost
surface used. ...........................................................................................................................77

ix

Figure 18. Overlay of four portfolios produced by MARXAN-ILP to prioritize lands for
conservation in the Wild Harts Study Area; inset represented by pink frame........................79
Figure 19. Finalized portfolio for conservation planning in the Wild Harts Study Area. ......89
Figure 20. Trial scenarios at 50% target for all conservation features; E. no cost surface
used; F. low-sensitivity cost surface used; G. medium-sensitivity cost surface used; H.
high-sensitivity cost surface used. ........................................................................................111

x

GLOSSARY
These definitions reflect how terms are referred to and applied within the context of this
research.
Aichi Target 11: An initiative established by the Convention on Biological Diversity to have
17% of terrestrial and inland waters, and 10% of coastal and marine areas, represented by
some form of conservation tenure by 2020.
ArcGIS: A geographic information system used to produce maps and complete
geoprocessing tasks (ESRI, 2011).
Avoidance Buffers: Spatial representations of distances at which woodland caribou avoid
permanent development.
Biodiversity: Refers to the complete range of species and genetic variation that comprises a
biological community, as well as the interactions that take place amongst these variables
(Primack, 2010).
Biogeoclimatic Zone: An ecological classification system representing geographic areas
with similar patterns of energy flow, vegetation, and soils as a result of a broadly
homogenous macroclimate (Province of British Columbia, 1995).
Biophysical Processes: Physical vectors that regulate, organize, and maintain biological
communities.
Biodiversity Surrogate: A component of a given ecosystem that is used as a proxy to
represent the biodiversity of the whole ecosystem (Grantham et al., 2010).

xi

Climate Change Resilience: The ability of an ecosystem to maintain self-organizing
processes and structures in the face of shifting climatological regimes. (Morecroft et al.,
2012).
Coarse-filter Conservation Feature: An element within a given planning region that is
thought to represent ecological patterns and processes of that region.
Compactness: A dimensionless measure used to describe clusters of selected planning units
with low edge-to-area ratios.
Complementarity: The extent to which selecting for one conservation feature results in
adequate representation of another by default.
Constraint: Reflective of the combined cost inflicted upon a MARXAN-ILP scenario by
user-defined inputs.
Cost Surface: A spatial representation of those elements in a planning region that have been
identified as having a negative impact on the conservation features being selected for.
Data-free Conservation Targets: Commonly used targets in systematic conservation
planning designed to represent a percentage of a given planning region’s overall area. These
targets are loosely based on theories of landscape ecology and conservation biology, but are
rarely informed by empirical data (Svancara et al., 2005).
Edge-to-area Ratio: The length of boundary surrounding a cluster of selected planning units
in relation to the combined surface area of those planning units.
Edge-effects: Ecological impacts that occur along the transitionary boundary between
resource development and contiguous habitats. These include barriers to species dispersal,

xii

altered temperature regimes within protected area boundaries, and the creation of entry points
for invasive species (Primack, 2010).
Fine-filter Conservation Feature: An element that is thought to represent components of a
given planning region that are endemic, rare, threatened, or characteristic of multiple other
values in that region.
Flagship Species: A species with a high public or political profile (Primack, 2010).
High Conservation Value: Areas containing multiple overlapping values predetermined to
be effective surrogates for biotic and abiotic diversity in a planning region.
Integer Linear Programming: Computational algorithm that increases the processing speed
of MARXAN and identifies the optimal solution for conservation at the lowest cost.
Integrated Lands: Planning units of disproportionately high conservation value that occur
within the integrated resource-use matrix.
Integrated Resource-use Matrix: Transitionary area, containing varying levels of resource
development, that occurs between fully-converted landscapes and those placed under
protection.
Keystone Species: A species whose presence within a given ecosystem is critical for the
ongoing maintenance and function of that ecosystem (Primack, 2010).
Land Facet: An area containing distinct combinations of abiotic variables such as slope,
aspect, elevation and landform.
Landscape Function: The ability of a planning region to maintain focal ecosystems, species,
and ecological processes within the natural range of variability (Poiani et al., 2000).
xiii

MARXAN: Computational software that prioritizes areas for conservation based on userdefined targets and constraints.
Natural Disturbance Regime: An area that has a distinct pattern of both frequency and
extent of natural disturbance (e.g. fire, insect epidemics, wind, landslides, flood).
North American Cordillera: A band of contiguous, mountainous landscapes that spans the
length of the western portions of the North American Continent.
Permanent Development: Areas in the Wild Harts Study Area where resource use has
resulted in complete removal of vegetation, alteration of soil structure, and manipulation of
topography. These areas contain little-to-no resemblance of their natural state.
Planning Unit: A parcel of land that is made available for selection by MARXAN-ILP. The
user defines the size, shape, and configuration of the unit.
Planning Unit Cost: A penalty metric that is assigned to all planning units and is meant to
serve as a baseline acquisition cost that increases efficiency of selection by MARXAN-ILP.
Portfolio: A spatial solution for conservation in a given planning area that is designed to
inform decision-making processes relating to protected area creation.
Principles of Good Reserve Design Include, but are not limited to, portfolios exhibiting
clusters of planning units that avoid resource development and have low edge-to-area ratios.
Priority Lands: Those planning units of disproportionately high conservation value in
comparison to the rest of a given planning region. Also representative of planning units that
adhere to principles of good reserve design.

xiv

Protected Area: A clearly defined geographical space, recognised, dedicated, and managed,
through legal or other effective means, to achieve long term conservation of nature with
associated ecosystems services and cultural values (Dudley, 2008).
Protected Area Design: The method of proactively identifying suitable areas for
conservation action using spatial tools and the principles of conservation biology and
landscape ecology.
Representation: The extent to which the spatial footprint of given conservation feature is
captured within the bounds of a protected area.
Resource Development: Refers to the spatial footprint of anthropogenic activity on the land
base. This includes both industrial and urban forms of development.
Rocky Mountain Corridor: Contiguous band of high elevation peaks and ridges that run
down the center of the WHSA on a southeast-northwest line.
Scenario: An iteration of prioritization completed by MARXAN-ILP that has a distinct
combination of user-defined inputs and constraints.
Science-based Conservation Target: Specific representation targets assigned to individual
conservation features based on what the scientific literature suggests is necessary to ensure
the persistence of those features (Svancara et al., 2005).
Simulated Annealing: Refers to the algorithm at the core of MARXAN that identifies a near
optimal solution for conservation at the lowest cost.
Solution Space: Used interchangeably with the term ‘portfolio’.

xv

Special Features: Those ecosystem components that are sensitive, spatially-limited, or of
high biodiversity value (Heinemeyer et al, 2004).
Systematic Conservation Planning: A structured approach for identifying candidate
protected areas that represent the values, goals, and constraints of a given planning process
(Margules & Pressey, 2000).
Temporary Development: The spatial extent of areas in the WHSA that have been logged
within the last 20 years.
Traditional Ecological Knowledge: A cumulative body of knowledge and beliefs, handed
down through generations by cultural transmission, about the relationship of living beings
with one another and with their environment (Berkes, 1993).
User-defined Inputs: Refers to the various mechanisms used to place constraint on a
MARXAN-ILP solution to achieve a desired result. These include features, planning units,
targets, and cost surfaces.
Umbrella Species: A species whose life history requirements captures the needs of multiple
other species (Primack 2010).
Zone of Influence: Areas containing resource-related impacts that extend past the spatial
footprint of the development (e.g. noise and light pollution) (Wilson, 2016).

xvi

AKNOWLEDGEMENTS
First and foremost, I wish to thank my supervisor, mentor, and friend Dr. Pamela Wright. Dr.
Wright’s commitment to her graduate students is unmatched and I am grateful to have
worked under such a leader in conservation. Special thanks go to Stafford Read for her
geospatial contributions and support throughout the project. I would also like to thank Dr.
Richard Schuster for his technical contributions during the construction of the MARXANILP model. I would like to acknowledge my committee of Dr. Katherine Parker, Dr. Chris
Johnson, and Dr. Tara Martin. The committee provided timely edits that were constructive
and instrumental for maintaining the momentum of this project. I would like to thank
everyone at the Ministry of Forests, Lands, Natural Resource Operations and Rural
Development who contributed knowledge in the early stages of the thesis. Special thanks go
to my team at the Crown Lands Division for supporting my educational pursuits. The
Yellowstone to Yukon Conservation Initiative must also be acknowledged for pioneering the
vision that is at the foundation of this thesis. Lastly, I wish to thank family for supporting me
throughout this endeavour.

xvii

1.0 INTRODUCTION
1.1 Overview of Protected Areas in Canada
Protected areas (PAs) provide refugia for biotic communities and serve as ecological
benchmarks from which rates of ecological change can be measured. These areas also
provide ecosystem services that contribute to human health and economic prosperity.
Examples of such services include forests that regulate climate through the sequestration and
storage of atmospheric carbon or the filtering and purifying of freshwater by wetland areas.
PAs provide a setting for people to engage in recreational and tourism pursuits. Benefits that
occur when people use these areas to conduct leisure include the generation of tourism
revenues, increases in physical wellbeing, and the awakening of an ecological consciousness
spurred on by meaningful experiences in nature.
Setting aside natural areas for the purpose of conservation is not a novel concept
unique to the western world. PAs are considered to be a cultural artifact with a history dating
back two millennia (Holdgate, 1999). The sacred groves of ancient India, for example, were
specifically set aside for the protection of natural resources (Eagles et al., 2002). In medieval
Europe, vast swaths of land were protected to serve as hunting grounds for the affluent
(Eagles et al., 2002). Indigenous societies the world over have, and continue to, engage in the
protection of natural areas for a myriad of reasons including providing food and serving as
areas for spiritual practice.
The emergence of a contemporary or Western PA movement began in the United
States with the establishment of Yellowstone National Park in 1872. The momentum of this
movement spread northward to Canada with the creation of Cave and Basin Hot Springs in
1885 — the first manifestation of what would eventually become known as Banff National
1

Park. Following the establishment of these parks, the importance and associated benefits of
PAs materialized within the political and social consciousness of Western society. This
resulted in the growth of the PA system within Canada that continues to expand. The
Canadian PA system currently encompasses 10.6% of the terrestrial land base (1.05 million
km2) and 0.9% of marine territories (55,000 km2) (Environment and Climate Change Canada,
2016).
During the early 20th century, PAs were created and managed primarily by the
Federal Government. Today, Canada’s PA system has diversified to include a multitude of
jurisdictions and approaches to management. Types of PAs in Canada include private land
trusts, tribal parks, municipal/regional parks, federal national parks, territorial parks and
provincial parks, conservancies and ecological reserves. Each jurisdiction operates under its
own mandate that outlines a purpose and objectives for the PA.
Private conservation areas, which are owned and operated independent of
government, can be effective in satisfying conservation niches such as the preservation of
species-at-risk or critical habitats. However, these areas comprise a very small percentage of
the overall protected land base and have a limited distribution across the northern latitudes of
the country. In Canada, the overwhelming majority of lands set aside for protection are held
by the Crown and managed under a mandate rooted both in biodiversity conservation and
visitor experiences.
1.2 The Ecology of the Anthropocene
We are currently living in a time of ecological crisis. The present geological period is
more biodiverse than any other time in Earth’s history — yet the current rate of extinction
surpasses that of any era in the past million years (Primack, 2010). Between 1970 and 2012, a
2

time scale which geologists would consider to be infinitesimal in relation to Earth’s history,
there has been a 58% decline in overall global vertebrate population abundance (World
Wildlife Fund for Nature, 2016). The most pronounced declines have occurred in freshwater
environs (68–89% reduction in aquatic vertebrate abundance) (World Wildlife Fund for
Nature, 2016). As of 2016, 21% of the Earth’s biomes have had their natural habitats
converted to some degree — a quarter of which have seen over 50% natural habitat
conversion (Oakleaf et al., 2015). If current trends continue, by 2020 global vertebrate
populations are predicted to have declined by 67% from their 1970 levels (World Wildlife
Fund for Nature, 2016).
This unprecedented loss in biodiversity can largely be attributed to the degradation of
natural systems that has come at the hands of humans (Dirzo et al., 2014; World Wildlife
Fund for Nature, 2016). Since the mid-20th century, human activity and resource
consumption have increased exponentially. The demands we have placed on the biosphere
have reached such levels that, by 2012, the equivalent of 1.6 Earths was needed to sustain
human consumption for a single year. So great has our impact on the natural world become
that we have now entered an entirely new geological age known as the Anthropocene — a
period during which human activity has been the dominant influence on climate and the
environment (Crutzen & Stoermer, 2000; Waters et al., 2016).
Even those areas set aside for protection have been subjected to human-induced
stressors. As of 2016, nearly half (46%) of national park ecosystems monitored by Parks
Canada were given an ecological integrity rating of fair-to-poor (Parks Canada, 2016). Of
particular concern are forest and grassland parks which are experiencing disproportionate
declines in biodiversity. These declining trends can largely be attributed to anthropogenic
3

disturbances, such as fire suppression, invasive species introduction, climate change, and
overuse of PAs for tourism/recreation (Parks Canada, 2016). In order to counteract these
trends, there is a growing need in Canada to improve management strategies and establish
additional PAs.
1.3 Commitments and Challenges
In keeping with their mandate of using PAs as tools for biodiversity conservation, the
Federal Government has committed, on behalf of the provinces and territories, to an
international target of conserving at least 17% of terrestrial and inland waters, and 10% of
coastal and marine areas, by the year 2020 (Environment and Climate Change Canada,
2016). This goal of 17%, otherwise known as the Aichi Target 11 under the Convention on
Biological Diversity (Secretariat of the Convention on Biological Diversity & United Nations
Environment Programme, 2001), has been criticized as having little grounding in the
discipline of conservation biology and is considered to be a politically-derived target
(Mackinnon et al., 2016). The conservation community has suggested that a target of 30–
70% is required to prevent further loss of global biodiversity and is more in keeping with
evidence-based studies (Locke, 2014). The current percentage of protected land in Canada is
well under the internationally-agreed upon target of 17%, and the country is faced with the
interim challenge of identifying and designating additional lands suitable for conservation
prior to 2020.
Adding to this challenge is the fact that Canada’s system is currently comprised of
relatively small (<10 km2) and disconnected PAs. Few Canadian PAs satisfy minimum size
thresholds (>3000 km2) deemed necessary for ecological persistence (Wright, 2016). These
inadequacies can be attributed to flawed historical approaches to PA design.
4

In the past, PAs were selected based on opportunity rather than ecological principles.
In keeping with the utilitarian mindset of the day, PAs were placed first in those areas
deemed undesirable or unprofitable to humans and, therefore, would go unmissed if tenured
for conservation purpose. Later strategies targeted areas for protection that would be
desirable tourism attractions. In efforts to increase visitation, PAs were established near
population centres, or along travel routes, instead of in the most biologically important areas.
This approach resulted in the sporadic placement of disjointed PAs in aesthetically-pleasing
landscapes that typically contained low levels of biodiversity (i.e., rock and ice parks).
Over time, planning strategies evolved to gradually embrace more ecologically-rooted
criteria for identifying PAs. For example, practitioners adopted a ‘one and done’ approach
whereby a single PA representing broad landscape patterns was deemed sufficient protection
for an ecoregion (Kavanagh et al., 1995). British Columbia applied ecological representation,
but the provincial approach was biased towards selecting for biodiversity hotspots and
proved to be insufficient at representing those ecological processes necessary to maintain
biodiversity itself (Province of British Columbia, 1993). By focussing on representativeness
alone, these approaches failed to fully incorporate principles of landscape ecology and
conservation biology that are at the forefront of contemporary approaches to PA design.
These rudimentary design criteria have produced a system of isolated PAs, selected to
capture standalone values, rather than a network of PAs containing multiple overlapping
values.
1.4 The Wild Harts Study Area: An Opportunity for Conservation
When compared to the populated areas that comprise the majority of the Global
South, northern Canada serves as a suitable candidate for more advanced conservation
5

planning practices given its low human population density and relatively intact landscapes
(Wiersma & Canadian Council on Ecological Areas, 2006). However, the global demand for
resources has resulted in increased development of fossil fuels and hydroelectric potential
across the northern extent of most provinces in the country (Wiersma & Canadian Council on
Ecological Areas, 2006). The Peace Region in northeastern British Columbia, referred to
throughout this study as the Peace River Break (PRB), is an example of one such area that
has been subjected to intensive resource development to satisfy global demands. A recent
analysis of linear disturbances noted that the PRB contains more than 29,000 km of pipelines,
45,000 km of roads, and 117,000 km of seismic lines that, if strung together, would encircle
the earth four times (Werring, 2015). Despite this development, a relatively intact band of
functioning ecosystems bisect the PRB on a southeast-northwest line. These areas comprise
what will hereinafter be referred to as the Wild Harts Study Area (Fig. 1).
The Wild Harts Study Area (WHSA) serves as a corridor joining PA complexes in the
Central Rocky Mountains to the large PAs in the Northern Rocky Mountains. In addition to
facilitating critical ecological exchanges between these two reserve systems, the WHSA
provides refugia for species living in a landscape pressured by competing resource interests.
The majority of the WHSA is not protected under any conservation tenure and is, therefore,
susceptible to encroaching development (Apps, 2013). The WHSA offers one of the last
opportunities to conduct systematic conservation planning (SCP) within the Peace Region.
The placement of additional PAs within the WHSA would ensure the North American
Cordillera remains a contiguous PA network and would assist Canada in reaching the Aichi
Target 11 before the year 2020.

6

Figure 1. Boundaries of the Peace River Break and Wild Harts Study Area, British Columbia.

7

1.5 Research Purpose and Associated Questions
The purpose of this research was to prioritize lands for conservation in the WHSA
using a systematic conservation planning approach. To achieve this, I was guided by the
following questions:
(1) What areas contain high conservation value for select coarse- and fine-filter
features?
(2) What areas retain conservation value despite the presence of resource
development?
(3) What is an optimal portfolio (i.e., spatial solution for conservation) from which to
direct conservation planning efforts?

8

2.0 LITERATURE REVIEW
2.1 From Singles to Systems: A Brief History of Protected Area Design in Canada
In the late 1800s, railway employees stumbled upon a mineral hot spring tucked away
in the Canadian Rockies. The Federal Government soon heard about this discovery and the
aesthetically-pleasing landscapes that made up the surrounding area. The hot spring was
thought to have high potential as a popular tourism and recreation destination. In 1885, the
Federal Government acted upon this opportunity and established a 25 km2 reserve boundary
to encompass the hot springs. The area was soon marketed by the Deputy Minister of the
Interior as being “the greatest and most successful health resort on the continent” (Lothian,
1976, p. 23). The establishment of this reserve, now known as Banff National Park, marked
the beginning of the contemporary PA movement in Canada.
Early motivations for PA establishment in Canada had little to do with conserving
areas of inherent ecological value (Dearden & Rollins, 2009). Instead, the tendency was to
select areas for protection based on their potential to generate economic stimulus through
tourism (Dearden & Rollins, 2009). Much like the creation of Banff National Park, early
planners targeted areas that would provide visitors with the opportunity to gaze upon features
of the greatest beauty or engage in recreational activities. The utilitarian mindset of the early
1900s was reflected in the creation of national parks like Banff, Yoho, Jasper, Waterton
Lakes, and Glacier. Each of these national parks contained world-class scenery, was
reasonably proximal to large human settlements, and was accessible by major transportation
corridors making them suitable tourism attractions. These early national parks were strictly
created out of opportunity and their boundaries were not designed using ecological principles
(Dearden & Rollins, 2009).
9

From 1911–1960, the economic stimulus that could be generated by tourism began to
materialize within the minds of bureaucrats in Ottawa and, as a result of this realization, the
era saw significant growth in the national parks system (Dearden & Rollins, 2009). The
growth of provincial and territorial park systems within Canada mirrored that of national
parks during this time. Like the federal system, many of the earliest provincial parks were
established primarily out of the opportunity to showcase the natural splendour of the
provinces (Stevens & Darling, 2004).
In the west, British Columbia’s first provincial parks were established along major
railway lines as a means to increase tourism and outdoor recreation (Stevens & Darling,
2004). This was followed by the creation of multiple small roadside provincial parks — each
equipped with campsite facilities to address the demand for car-based family recreation
opportunities that increased in the province following the Second World War. By the end of
the 1950s, British Columbia’s PA system was comprised of small, disconnected provincial
parks concentrated around southern population centers. In addition, the majority of provincial
parks were placed within snow and ice ecosystems that provided visually-appealing
scenescapes but contained low biodiversity value (Stevens & Darling, 2004). As was the case
with the national park planning process, the majority of these early provincial parks were
designed using criteria that had limited or no ecological foundation.
The 1960s saw the emergence of an environmental ethic within the Canadian
consciousness (Dearden & Rollins, 2009). As a result of this enlightenment, nongovernmental environmental organizations began to place pressure on parliament to expand
Canada’s PA system. This new paradigm placed an increased emphasis on establishing PAs
that would serve as refugia for ecosystems and biotic communities. These pressures incited a
10

shift in policy away from an exclusive focus on tourism and towards targeting those areas
higher in conservation value (Hummel, 1989).
In 1964, policy was established that deemed the preservation of significant natural
features as being the most fundamental and important obligation to national park designation
and management (Dearden & Rollins, 2009). In 1965, the Province of British Columbia
developed the first Park Act that placed importance on managing provincial PAs for
conservation alongside recreation (Stevens & Darling, 2004). These policy shifts coincided
with the emergence of conservation sciences — leading to broader understandings of the
need for a systematic approach to resource planning and management (Eagles et al., 2002).
Building on this new ecological understanding, PA design in the 1970s began to
embrace the concept of planning for representation of biodiversity (Wiersma & Nudds,
2009). For example, national parks were now designed to encapsulate a representative
sample of each of Canada’s unique landscapes. The nation was divided into national park
natural regions (NPNRs) based on prevailing vegetation types and landscape appearance
(Government of Canada, 1972). Candidate areas for protection were selected based on how
well they represented an NPNR’s natural features. The premise of this approach was that the
designation of a single national park, within an identified NPNR, would be sufficient to
achieve representation of said NPNR (Government of Canada, 1972).
The NPNR approach provided limited direction on where national parks should be
placed, how large they should be, or how they should be configured. In addition, the process
of delineating NPNRs, based on landscape appearance and prevailing vegetation, was
considered too coarse and ecologically inadequate (Noss et al., 1995). Recognizing these
shortcomings, non-government environmental organizations developed frameworks for
11

selecting PAs, founded on ecological principles, with hopes that they would be applied in
decision making within government.
An example of such an initiative was the introduction of the gap analysis planning
methodology. Developed by the World Wildlife Fund of Canada (WWF), and the Canadian
Council on Ecological Areas (CEAA), gap analysis is a method for identifying areas of
natural diversity that are not represented by an existing PA system (Kavanagh et al., 1995).
This is achieved by overlaying spatial data describing the surficial deposits and quaternary
geography of a landscape. Landscape units, known as enduring features, are then selected
based on a combination of topography-relief, parent material origin, and parent material
texture. The final stage of the method requires the identification of ‘gaps’, where existing
PAs don’t represent identified enduring features. Though data on individual species were
included when available, the gap analysis methodology was primarily coarse in that enduring
features were considered to be foundational elements of ecological diversity (Kavanagh et
al., 1995).
This gap analysis approach was the first design methodology aimed at creating an
inter-related system of PAs rather than a collection of standalone PAs. Some of the
suggestions contained within the WWF/CEAA approach to gap analysis were incorporated
into government PA planning processes. For example, the Province of British Columbia’s
1993 Protected Area Strategy (Stevens & Darling, 2004) drew upon gap analysis as a method
for identifying new PAs that would help address underrepresentation of mid-low elevation
ecosystems in the province (Province of British Columbia, 1993). Although the planning
intent of the Protected Areas Strategy was to incorporate a gap analysis approach, established
policy targets for conservation, and the existing legacy of PAs that were contrary to
12

representation goals prevented government from fully applying gap analysis methodology.
The combination of these factors resulted in limited progress towards building a functioning
system of PAs in British Columbia (Doyle & Province of British Columbia, 2010).
Despite the efforts of non-government environmental organizations, the Canadian
system remained comprised of small and disjointed PAs at the turn of the 21st century
(BEACONS, 2006; Doyle & Province of British Columbia, 2010). It does not resemble a
connected network designed to promote biodiversity, maintain biophysical processes, and
accommodate shifting ecological communities. Though planning for these values has since
become an initiative at the federal and provincial levels (Stevens & Darling, 2004; Parks
Canada, 2016), both levels of government continued to employ rudimentary representation
approaches for directing PA establishment.
The growing fields of conservation science, and accompanying technological
innovations, are ushering in a new era of PA design rooted in ecological principles. The
disciplines of landscape ecology, disturbance ecology, and conservation biology have coevolved with the development of large-scale data visualization, mapping, and management
techniques to facilitate the emergence of new planning approaches. With the assistance of
radio collars and satellite transmitters, researchers now understand that animals are moving
far greater distances than was previously understood (Chadwick et al., 2000). Remotely
sensed data, and the ability to process/analyze this information using various spatial tools and
metrics, enables researchers to grasp the extent of resource development on the landscape.
The advent and accessibility of these data and tools is allowing for more analytical planning
processes whereby practitioners model conservation values at multiple spatial scales.

13

In Canada, First Nation groups and non-government environmental organizations
have been leading the alternative PA design movement. For example, in 2003, the Taku
River Tlingit (TRT) developed a PA design for their 40,000 km2 traditional territory in
northwestern British Columbia (Province of British Columbia & Taku River Tlinget First
Nation, 2011). The design process combined a number of methodologies including: the
development of habitat models for several focal species; coarse-filter ecological community
classification and representation analysis; regional connectivity analyses; and spatial
optimization procedures at multiple scales. Traditional ecological knowledge was used to
supplement scientific data and verify computer-generated models. Through a combination of
identified core and connectivity areas, the TRT now have 55% of their traditional territory
protected under the plan (Province of British Columbia & Taku River Tlinget First Nation,
2011).
Additional examples of contemporary PA designs include the Muskwa-Kechika
Management Area (Heinemeyer et al, 2004) and a number of ecoregional assessments
conducted by the Nature Conservancy (see for example Horn, 2011). These design
approaches, which focus on building an interconnected system of PAs using multiple
representation lenses, require careful consideration of a number of key variables. These
variables include selection of features, targets, and tools for conservation area design. Each
of these PA design elements will be discussed below.
2.2 Deciding What to Conserve
Identifying what to conserve, and how to define it, is one of the first, and most critical
decisions in PA planning. Using representation of a single valued ecosystem component as
the sole criteria for a design will result in the establishment of PAs that are inadequately
14

equipped to support biodiversity and the biophysical processes needed to maintain it (Noss et
al., 1995). Therefore, it is crucial to employ multiple lenses when selecting features for
conservation. Hierarchies of conservation features used in a design can range from specieslevel assemblages to biogeoclimatic landscape units, or the natural disturbance regime of a
given planning area (Noss et al., 1995; Margules & Pressey, 2000; Delong & Province of
British Columbia, 2011). The use of multiple features, at varying spatial scales, will increase
the likelihood of the resultant PA system supporting a broad range of biodiversity and
maintaining ecological processes (Noss et al., 1995).
The complexity of ecological systems makes it nearly impossible to quantify all
facets of biodiversity within a proposed PA (Margules & Pressey, 2000). Therefore,
practitioners must be targeted in their approach and select those features that will best serve
as surrogates for biodiversity. Similar to how enduring features are utilized to represent a
region’s geofeatures, a surrogate is a component of an ecosystem that is used as a proxy to
represent the biodiversity of the whole ecosystem (Grantham et al., 2010). Biodiversity
surrogates are often categorized as being either taxonomic or environmental (Margules &
Pressey, 2000; Groves et al., 2002). Examples of taxonomic surrogates include rare,
umbrella, flagship, and keystone species; whereas environmental surrogates include, but are
not limited to, watersheds, natural disturbance regimes, land facets, or climatic zones
(Margules & Pressey, 2000).
The use of surrogates carries with it the assumption that one component of
biodiversity can be considered to be representative of another — an assumption criticized by
many as being an oversimplification of complex natural systems (Lambeck, 1997;
Simberloff, 1998; Molles, 1999; Margules & Pressey, 2000; Carroll et al., 2001; Bonn et al.,
15

2002; Lindenmayer et al., 2002). Therefore, a comprehensive PA design would include a
number of surrogates, selected from both taxonomic and environmental categories, in order
to account for the limitations and assumptions behind single surrogate use. Surrogate
selection is largely dictated by data availability and comprehensiveness, as well as the
specified objectives for a given planning process (Margules & Pressey, 2000). For the
purposes of this study, environmental and taxonomic surrogates are referred to as coarse- and
fine-filter conservation features respectively.
2.3 Deciding How Much to Conserve
Once the suite of conservation features has been selected, the next step is determining
how much of each feature should be protected. Typically, this target is expressed
quantitatively for use at the operational level (Margules & Pressey, 2000). Canada has a long
history of setting quantitative targets to help drive progress towards expanding the country’s
PA system. In 1992, Canada was a signatory of the UN Convention on Biological Diversity
(CBD), and through this agreement committed to expand the Canadian PA system to
encompass 12% of the nation’s terrestrial ecosystems (Canadian Parks and Wilderness
Society, 2016). This action came after the World Commission on Sustainable Development
(i.e., Bruntland Commission) called for a tripling of PAs globally. In 2010, the CBD
determined the target of 12% had not reduced declines in global biodiversity. Based on this
analysis, the CBD recommended countries commit to protecting at least 17% of terrestrial
and inland water areas by the year 2020 (i.e., Aichi Target 11). In 2015, the Canadian
government committed to achieving Aichi Target 11 in the report 2020 Biodiversity Goals
and Targets for Canada (Canadian Parks and Wilderness Society, 2016).

16

Though based on good intentions, global targets for biodiversity conservation are
criticized as being political abstractions, or data-free initiatives with little scientific
foundation (Schmiegelow et al., 2006; Locke, 2014; Wiersma & Sleep, 2018). The 12%
target, recommended by the Bruntland Commission, is considered by the conservation
community to be grossly inadequate and it is predicted that 50% of all species would be
committed to extinction if PA expansion was restricted to this target (Schmiegelow et al.,
2006; Locke, 2014). It has been argued that more realistic targets of 30–70% protection of
terrestrial environs are necessary to prevent further losses in biodiversity (Margules et al.
1988; Saetersdal et al. 1993; Soulé and Sanjayan 1998; Warman 2001; Rumsey et al. 2003;
Venter et al., 2014).
The policy-based targets described above are premised on the assumption that
selecting for a fixed percentage of a given planning region’s total area will effectively
represent the ecology of that region. In a review of 159 studies on conservation targets,
Svancara et al. (2005) found that the simplistic approach of targeting x% of a planning region
often leads to inadequate representation of individual species and their respective habitats.
Instead of blanketing entire planning regions with a single data-free target, it has been
suggested that a more effective approach is to employ a range of science-based targets in
SCP processes (Wiersma & Sleep, 2018). Under a science-based approach, each conservation
feature should be assigned an individualized target that is the product of an in-depth
understanding of the ecology and spatial extent of that feature within the context of a given
planning region. In this way, the overall percentage of a planning region’s total area needed
to achieve conservation is representative of the combined spatial requirements of individual
conservation features (Svancara et al., 2005).
17

Irrespective of the approach, the setting of conservation targets is plagued by
assumption and uncertainty (Schmiegelow et al., 2006). To ensure best practices when
selecting targets for a PA design, Margules and Pressey (2000: 246) suggest planners do the
following: “focus on scales that are much finer than whole countries or regions; deal with
natural processes as well as biodiversity patterns; reflect the relative needs of species and
landscapes for protection; recognize that reserves must be complemented by off-reserve
management; and leave options open for revision as social and economic conditions change”.
In order to achieve these requirements, planners must have an understanding of the
ecological and social dimensions of their planning area prior to establishing targets.
However, possessing knowledge alone will not result in the design of a PA system that is
effective at ensuring the preservation of biodiversity. Planners need to employ tools that
allow them to manipulate targets and assess possible solutions for conservation at multiple
scales.
2.4 The Right Tool for the Job
Given the complexity attendant to a system with multiple conservation features and
varying targets, optimization tools are increasingly employed by conservation planners to
assist in PA selection (Schuster et al., 2016). Through the use of these tools, planners are able
to explore the extent to which varying PA configurations achieve conservation targets and
minimize cost. Optimization software identifies a portfolio of high-value conservation lands
that can contribute to a transparent, inclusive, and defensible decision-making process
relating to PA creation (Schuster et al., 2016). The majority of these optimization tools are
strictly computational and have no graphical interface. The outputs of these programs can be

18

uploaded into ArcGIS (Environmental Systems Research Institute, 2011) to be viewed
spatially (Morrell et al., 2015).
To date, the most commonly used decision support software for conservation
planning is MARXAN (Ball et al., 2009). MARXAN is designed to select planning units
based on the presence/absence of identified conservation features (Morrell et al., 2015). The
user can manipulate how aggressively MARXAN works to select conservation features by
adjusting each feature’s associated target. Planners can also adjust the software to align with
additional goals for the PA design. For example, MARXAN can be manipulated to minimize
boundaries (edge) or avoid highly fragmented portions of the planning landscape — as these
would be considered as being costs to the overall effectiveness of the PA design. The
software is designed to maximise targets for conservation while minimizing costs (Morrell et
al., 2015).
The traditional MARXAN platform uses the heuristic method of simulated annealing
(SA) to produce portfolios of conservation lands (Morrell et al., 2015). SA is simply an
algorithm that produces ‘near optimal solutions’ for a problem given user-defined parameters
(i.e., planning units, features, targets, and costs) (Morrell et al., 2015). Planners have
recognized the problem-solving abilities of this software and, as a result, MARXAN has been
employed in multiple contemporary PA designs (Airamé et al., 2003; Evans, 2003; Leslie et
al., 2003; Stewart et al., 2003; Banks et al., 2005; Geselbracht & Torres, 2005, Loos &
Canessa, 2006; Horn, 2011; Venter et al., 2014). However, the traditional MARXAN model
has recently come under criticism for producing solutions of questionable quality (Beyer et
al., 2016). This is because the heuristic method of SA is limited to selecting PA designs that
are near optimal and, thus, may exclude important features from a prioritization. In addition,
19

SA can take hours/days to calculate a solution and processing time increases with the
complexity of datasets used and the size of a reserve planning area (Beyer et al., 2016).
Recent advances in algorithms and computing power have now increased the
capabilities of decision support software (Beyer et al., 2016). Integer linear programming
(ILP) is an example of a state-of-the art approach to optimization problems that has
superseded simulated annealing (Schuster et al., 2016). When used in conjunction with
MARXAN, ILP is capable of analyzing large, complex datasets to find an ‘optimal solution’
to identified biodiversity goals based on user-defined parameters (Schuster et al., 2016). ILP
is superior to SA in computational speed (seconds-minutes vs hours-day) and solution quality
(optimal vs near-optimal) (Beyer et al., 2016). These advantages allow planners to explore
trade-offs, assess the complementarity of selected surrogates, and evaluate effectiveness of
targets in near real time. When compared to the static portfolio of hypothetical PA designs
produced by MARXAN with SA, MARXAN with ILP offers an interactive platform that can
facilitate conversations with stakeholders and decision makers regarding the establishment of
PAs (Beyer et al., 2016; Schuster et al., 2016).

20

3.0 CASE STUDY
3.1 Overview of the Peace River Break
The Peace River Break (PRB) is a geographic designation coined by the Yellowstone
to Yukon Conservation Initiative to describe a 135,400 km2 section of the North American
Cordillera that straddles the Peace River in northeastern British Columbia (Fig. 1) (Apps,
2013). Ten ecoregions merge here to create a hotspot of biotic and abiotic diversity. The
Peace River itself has the unique distinction of being one of the only watercourses in Canada
that travels eastward through the Rocky Mountains. Where arctic air typically dominates this
region, the Peace River provides a channel for warm Pacific air to flow through, resulting in
a moderated climate and unique ecological conditions (Apps, 2013).
This west-east pathway, and the temperate climate it provides, made the PRB a
logical area for multiple First Nations groups to inhabit (Apps, 2013). The arable soils of the
Peace River valley later drew agricultural interests to the region at the turn of the 20th
century. By the 1950s, two major transportation corridors (i.e., Alaska Highway and John
Hart Highway) had been constructed, providing access to what was previously a remote and
isolated area. This led to the development of a natural resource industry in the area and,
consequently, the establishment of several communities to support the growing sector (Apps,
2013).
In recent times, the PRB has become one of British Columbia’s most prominent
regions for resource-related industry and extraction (Apps, 2013).With construction of the
Site C Dam underway, and other large-scale development proposals pending approval, such
as TransCanada’s Prince Rupert Gas Transmission Project, the area could see considerable
growth in population and resource-related infrastructure. Considering the PRB is currently
21

underrepresented by PAs (approximately 9% of the land base is conserved), landscape
connectivity and associated biodiversity values in the area are quickly becoming vulnerable
to the cumulative effects of industrial expansion (Apps, 2013).
3.2 The Wild Harts Study Area
A preliminary analysis of industrial activity in the PRB revealed the spatial footprint
of linear development on the land base (Fig. 2). Industrial expansion has formed an
ecological bottleneck in the region — whereby the mountain ranges that bisect the PRB on a
southeast-northwest line represent the only ecologically functional landscapes within this
section of the North American Cordillera. When compared to the rest of the PRB, these areas
remain relatively intact and form a natural corridor for species moving amongst existing PAs
(Fig. 2).
Given their ecological significance within local and regional contexts, the lands that
comprise the abovementioned corridor were selected to be the focus of this research. The
borders of the Eastern System Physiographic Region capture the corridor and, thus, served as
a logical reference from which to delineate longitudinal study area boundaries. Horizontal
boundaries were established in areas where lands beyond were thought to have sufficient
representation by existing PAs. All lands that fell within the finalized boundary comprised
what was referred to in this study as the Wild Harts Study Area (WHSA) (Fig. 2).

22

Figure 2. The Wild Harts Study Area juxtaposed against the Eastern System Physiographic Region and linear
development in the Peace River Break.

23

3.2.1 Geography & Natural History
The lands that make up the WHSA occupy 70,000 km2 of the Eastern System
Physiographic Region. This section of the Eastern System is predominately comprised of the
Northern and Central Canadian Rocky Mountain ecoregions — with small portions of the
Eastern Continental Ranges and Southern Alberta Upland ecoregions occupying the
southeastern corner (Demarchi, 2011). Elevation in the WHSA increases in a southerly to
northerly direction and decreases from west to east (Demarchi, 2011).
At the ecosection scale, the portions of the WHSA that fall below the Peace River are
largely characterized by the steep-sided, round-topped mountains of the Hart Ranges
(Demarchi, 2011). Moving eastward, the Hart Ranges subside into the low, rounded
mountains and wide valleys that make up the Hart Foothills. The remaining southeasterly
portions are comprised of the ridges and valleys of the Front Ranges and rolling uplands of
the Kiskatinaw Plateau (Demarchi, 2011).
The central portions of the WHSA contain rounded mountains characteristic of the
Peace Foothills and Misinchinka Ranges ecosection (Demarchi, 2011). Moving northward,
above the Peace River, these rounded mountains are disrupted by the deep valleys and rugged
peaks of the Muskwa Ranges which represent the highest mountains in the WHSA. The
Muskwa Foothills make up the remaining northeastern portion of the study area and are
characterized by rounded mountains and wide valleys (Demarchi, 2011).
The climate in the WHSA consists of cold winters and warm summers (Demarchi,
2011). Extreme cold events are frequent during winter and early spring when Arctic air
moves unhindered from north to south across the area. The rugged topography of the WHSA
results in unique patterns of surface heating at high-elevation slopes and cold-air drainage in
24

low valleys. Precipitation is strongly influenced by warm Pacific air that is brought across the
central interior by prevailing westerly winds. The Rocky Mountains serve as a barrier that
causes said warm air to rise, resulting in high precipitation in windward areas and the
occurrence of rain shadows in leeward areas. In general, precipitation levels are relatively
constant throughout the year (Demarchi, 2011).
The WHSA is unique in that its boundary captures portions of ten major watersheds
that serve as tributaries to both the Pacific and Arctic oceans (Demarchi, 2011). From north
to south, these include the Kechika, Fort Nelson, Finlay, Halfway, Peace, Pine, Parsnip,
Smokey, Kiskatinaw, and McGregor watersheds (Demarchi, 2011; Apps, 2013). The
complex relief of the WHSA causes drainage to radiate out in all directions. However, the
majority of streams are relatively short in length and either flow westward into the
Parsnip/Fraser watersheds, or eastward into the Peace (Demarchi, 2011).
The area’s disturbance regime is predominately comprised of stand-replacing fire
events and insect epidemics (Delong & Province of British Columbia, 2011). Vegetation
pattern in the WHSA is defined by six prevailing biogeoclimatic zones (Demarchi, 2011).
These include the Sub-boreal Spruce, Englemann Spruce – Subalpine Fir, Boreal Altai
Fescue Alpine, Interior Cedar – Hemlock, Boreal White and Black Spruce; and Spruce –
Willow – Birch zones (Demarchi, 2011; Apps, 2013).
The unique combination of topography, climate, hydrology, disturbance, and
vegetation pattern enables the WHSA to support a high diversity of native wildlife and fish
species. Characteristic mammalian species include wolverine, moose, caribou, mountain
goat, mule deer, elk, lynx, marten, black bear, Stone’s sheep, and grizzly bear (Demarchi,
2011; Apps, 2013). The area is said to support 57% of all avifauna known to occur in British
25

Columbia. Bull trout and Arctic grayling are the dominant fish species (Demarchi, 2011;
Apps, 2013).
3.2.2 The Human Element
The WHSA boundary captures traditional territories of the Halfway River First
Nation, McLeod Lake Indian Band, West Moberly First Nation, Saulteau First Nation,
Simpcw First Nation, Blueberry River First Nation, Horse Lake First Nation, and Lheidli
T’enneh First Nation. Four significant urban settlements (i.e., population > 1000) are within,
or directly proximal to, the study area. These include the towns of Mackenzie, Tumbler
Ridge, Chetwynd, and Hudson’s Hope. Residents of these communities rely heavily upon
natural resources for recreation and employment. The WHSA boasts some of the best
hunting, fishing and snowmobiling opportunities in British Columbia. Main employers in the
area include the oil, natural gas, coal, hydro-electric, forest, and resource-related-tourism
industries (Apps, 2013).
3.2.3 Resource Development
Where oil and gas dominate in the larger PRB area, forestry is the most prolific
resource-based industry in the WHSA (Apps, 2013). Harvesting activities within the WHSA
supply major pulp, lumber, and particle board mills situated in the communities of Chetwynd
and Prince George. The study area captures the majority of TFL 48 — a 640,000 ha tree farm
licence managed by Canadian Forest Products Limited that has an allowable annual cut of
1,550,000 m3 (Province of British Columbia, 2015a).
Mining is another major industry in this area. There are two major coal mines (i.e.,
Sukunka and Murray River) that occur within the bounds of the WHSA along with multiple

26

small placer and aggregate operations. There is some agricultural activity in the far southwest
and northeast portions of the study area. Several wind power proposals occur within the area
but remain in the investigative phase due to the potential energy surplus that the Site C Clean
Energy Project is expected to create. The WHSA is also bisected by John Hart Highway, two
railroads, and a myriad of access roads that are associated with the abovementioned
industries.
When compared to the rest of the Peace River Break, there is limited oil and natural
gas extraction/infrastructure in the WHSA. However, TransCanada’s Prince Rupert Gas
Transmission proposal is currently under review. If approved, this initiative would increase
the rate of natural gas extraction from the Peace River Break which would then be sent
overland by pipeline to Prince Rupert. There is a high likelihood that, in order to keep up
with increasing demands associated with this project, and others like it, natural gas
exploration and development will encroach upon the WHSA in the near future.
3.2.4 Conservation in the Wild Harts Study Area
At the North American scale, the WHSA joins large PA systems in the Central Rocky
Mountains to the Muskwa-Kechika Management Area (M-KMA) and territorial parks in the
north. In this way, the WHSA serves as an integral section of the Yellowstone to Yukon
corridor — an initiative centered on establishing a contiguous PA network stretching from
Wyoming to the Yukon Territory (Fig. 3). Fragmentation of the WHSA may create barriers
to species movements and effectively split the trans-continental corridor into two
ecologically disjointed PA networks. This limits the ability of biological communities to
adapt to shifting climate regimes by migrating along latitudinal gradients. It may also isolate

27

Figure 3. The position of the Wild Harts Study Area in relation to the Muskwa-Kechika Management Area,
Central Rocky Mountains, and Yellowstone to Yukon Corridor.

28

populations and reduce the amount of biological/genetic exchange occurring along the North
American Cordillera.
Similar to the Peace River Break, there is relatively little PA representation within the
WHSA (Fig. 4).Total PA coverage equals approximately 1,067,000 ha or 17.4% of the
WHSA’s total area. Northern Rocky Mountains Provincial Park and Kakwa Provincial Park
occupy the terminal ends of the study area and comprise approximately half of the total area
under protection in the WHSA. The remaining half is comprised of small and disjointed PAs
(Fig. 4). The creation of new PAs within these poorly-represented sections would limit
fragmentation of the WHSA and facilitate connectivity across the northern portion of the
North American Cordillera. In addition, PA creation in the WHSA could assist Canada in
achieving its international agreement of protecting 17% of the nation’s terrestrial land base
by 2020.

29

Figure 4. Existing protected area representation in the Wild Harts Study Area.

30

4.0 METHODS
To address each of my research questions, I used a variation of the methods for
conservation area design described in Margules and Pressey’s (2000) systematic conservation
planning (Table 1). The four steps involved in my SCP methods were: (1) select and develop
conservation features; (2) identify conservation goals and targets for the WHSA; (3) review
existing conservation areas; and (4) use MARXAN-ILP to identify lands suitable for
protection in the WHSA.
Table 1. The four steps of systematic conservation planning used to prioritize lands for conservation in the Wild
Harts Study Area.

1. Select and Develop Conservation Features
• Review literature and existing data to identify multi-spatial surrogates for biodiversity in
the WHSA.
• Select those datasets that have sufficient rigor and consistency to support the construction
of spatial layers for the WHSA.
• Develop spatial layers in ArcGIS that represent each conservation feature’s extent across
the WHSA.
2. Identify Conservation Goals and Targets
• Employ knowledge gained in the literature review to set goals for conservation in the
WHSA centered on promoting biodiversity, maintaining natural disturbance regimes, and
increasing resilience to climate change.
• Translate goals into a series of quantifiable targets for operational use.
3. Review Existing Protected Areas
• Determine the extent to which existing protected areas achieve my targets for conservation
in the WHSA.
• Identify gaps in the current WHSA protected areas system.
4. Use MARXAN-ILP to Identify Lands Suitable for Protection in the WHSA
• Apply the conservation software MARXAN with ILP to identify gaps in the current
protected areas system in order to spatially prioritize lands for conservation in the WHSA.
• Select a portfolio of priority lands that achieve optimal targets for conservation while
minimizing user-defined costs.

31

4.2 Step 1 Selection and Development of Conservation Features
I conducted a review of existing literature on the biodiversity and biophysical
processes present in the WHSA in order to identify a suite of possible surrogates for use in
the SCP process. Expert opinion was drawn upon informally to assess the extent to which
select surrogates represented other facets of biodiversity in WHSA. I analyzed databases to
identify which surrogates had geospatial layers that were most available, reliable, and
consistent across the WHSA (Appendix 1). Multiple datasets were combined, using
geospatial analysis in ArcGIS to create full layers for surrogates.
The combination of literature, expert opinion, and data availability resulted in the
selection of coarse-filter and fine-filter surrogates, hereinafter referred to as conservation
features, for use in the subsequent ILP scenarios. The conservation feature layers described
below were developed in binary form (0, 1). There is no continuous gradient that would
suggest one portion of a conservation feature is of higher value than another.
4.2.1 Coarse-filter Conservation Features
In SCP, coarse-filter conservation features are typically representative of ecological
patterns and the biophysical processes that define those patterns (Schneider et al., 2011). The
coarse-filter approach is grounded in the assumption that, by representing a wide array of
ecosystems and their constituent processes, the habitat needs of multiple species will be
accommodated in a resultant PA design (Schneider et al., 2011). In this way, coarse-filter
conservation features are an efficient means by which to compensate for a lack of data on the
ecology of individual species when prioritizing lands for conservation across a large region
such as the WHSA. Coarse-filter conservation features used in this study included land facet

32

diversity, land facet rarity, and a set of features displaying an intersection of natural
disturbance, biogeoclimatic zone, and forest age.
4.2.1.1 Land Facets
Ecological pattern and process is largely determined by underlying abiotic variables
such as slope, aspect, elevation, and landform (Beier & Brost, 2010). A land facet is a
recurring landscape unit with uniform combinations of these abiotic variables. In
contemporary conservation planning, land facets are used as a coarse-filter conservation
feature to build climate change resilience into PA designs. This is because topography and
geology represent those components of an ecosystem that are relatively stable and unlikely to
undergo abrupt reorganization in response to a shifting climate regime. Essentially, land
facets serve as the static stage for temporary biological assemblages to pass across and for
evolutionary processes to play out upon (Beier & Brost, 2010).
I used datasets developed by Michalak et al. (2015) and Carroll et al. (2017) to
construct a land facet diversity (LFD) and land facet rarity (LFR) layer for application in
MARXAN-ILP. Land facets were defined in these datasets as raster cells containing unique
combinations of 4 abiotic variables: elevation, landform, modified heat load index (HLI), and
soil order. Landform was referred to as coarse topographic features such as headwaters,
hilltops, ridges, valleys, and canyons. Modified heat load index was a measure of radiation
exposure relative to aspect and slope. Soil order was described as patterns in surficial
deposits at a scale of 1 km2. Michalak et al. (2015) and Carroll et al. (2017) delineated
individual land facets by categorizing the four abiotic variables into discrete classes for
application in the following equation:
Land Facet ID = (Landform + HLI + Elevation)*100 + Soil Order
33

4.2.1.1.1 Land Facet Diversity
Areas of high diversity occurred where clusters of land facet cells had dissimilar
combinations of abiotic variables relative to neighboring cells. Details on the formulas used
to generate diversity metrics can be found in the R script of the land facet diversity dataset
(see Carroll et al., 2017). In ArcGIS, I scaled the dataset produced by Carroll et al. (2017) to
the WHSA and ranked diversity on a scale of 1 (high LFD) to 10 (low LFD). The
mountainous terrain of the WHSA made it so that the majority of the area received a
diversity value of <3. Therefore, only those areas containing a diversity value of 1 were
selected to serve as a surrogate for LFD — as these areas represented the most diverse
combinations of abiotic variables in an already highly diverse landscape.
4.2.1.1.2 Land Facet Rarity
Using the dataset produced by Michalak et al. (2015), I assessed rarity based on the
total area that each land facet covered in the WHSA. The data were scaled to the WHSA and
each land facet type was assigned a value from 1 (low area) to 10 (high area). Those land
facets assigned a value of 1 occupied <1000 ha of the WHSA total area and were considered
to have a rare distribution. Accordingly, the LFR layer reflects those cells containing unique
combinations of elevation, landform, and HLI that occupy <1000 ha of the WHSA area.
4.2.1.2 Forest Pattern and Process
As previously stated, coarse-filter conservation features should represent the full array
of ecosystems that occur within a given planning region, as well as the biophysical processes
which create the conditions necessary for the persistence of those ecosystems. To satisfy
those criteria, I constructed a series of layers that incorporated data on vegetation types,
forest age, wildfire occurrence, and climatological patterns in the WHSA. Information on
34

vegetation types was included to ensure adequate representation of dominant tree species.
Old forests were included in the selection as these areas are synonymous with high-value
habitat features (e.g., coarse-woody debris, wildlife trees, stratified canopy, shrubs, etc.).
Recently burned areas (≤40 years) were selected as they represent suitable habitat for those
species that are heavily dependent upon burned, standing forests. Lastly, natural disturbance
regimes were drawn upon because of their ability to reorganize biological communities and
initiate ecological succession.
Provincial datasets from DataBC were used to construct the multiple layers
representing forest pattern and process in the WHSA (Appendix 1). In ArcGIS, I intersected
data describing natural disturbance type (NDT), biogeoclimatic ecosystem classification
(BEC) zone, and stand age to create polygons that represented unique combinations of these
three input datasets (Table 2). I used the seral stage definitions by NDT/BEC, as described in
the Biodiversity Guidebook (Province of British Columbia, 1995), to select for mature-old
stands. Additional datasets describing historical wildfire and harvesting activity were used to
select stands within each NDT/BEC that were recently burned and free of any harvest
activity since time of disturbance.

35

Table 2. Nomenclature of the forest pattern and process conservation features used in the MARXAN-ILP
analysis to prioritize areas for conservation in the Wild Harts Study Area.

Label*
NDT3-SBSMature/Old
NDT3-SBSBurned
NDT3-BWBSMature/Old
NDT3-BWBSBurned
NDT2-SWBMature/Old
NDT2-SWBBurned
NDT2-SBSMature/Old
NDT2-SBSBurned
NDT2-ESSFMature/Old
NDT2-ESSFBurned
NDT1-ESSFMature/Old
NDT1-ESSFBurned
NDT1-ICHMature/Old
NDT1-ICHBurned

Description
NDT3: ecosystems with frequent stand initiating events
SBS: Sub-boreal Spruce biogeoclimatic zone
Mature/Old: stand age >100 years
NDT3: ecosystems with frequent stand-initiating events
SBS: Sub-boreal Spruce biogeoclimatic zone
Burned: exposed to wildfire within last 40 years
NDT3: ecosystems with frequent stand-initiating events
BWBS: Boreal White and Black Spruce biogeoclimatic zone
Mature/Old: stand age >80 years
NDT3: ecosystems with frequent stand-initiating events
BWBS: Boreal White and Black Spruce biogeoclimatic zone
Burned: exposed to wildfire within last 40 years
NDT2: ecosystems with infrequent stand-initating events
SWB: Spruce - Willow - Birch biogeoclimatic zone
Mature/Old: stand age >120 years
NDT2: ecosystems with infrequent stand-initating events
SWB: Spruce - Willow - Birch biogeoclimatic zone
Burned: exposed to wildfire within last 40 years
NDT2: ecosystems with infrequent stand-initiating events
SBS: Sub-boreal Spruce biogeoclimatic zone
MO: stand age >100 years
NDT2: ecosystems with infrequent stand-initiating events
SBS: Sub-boreal Spruce biogeoclimatic zone
Burned: exposed to wildfire within last 40 years
NDT2: ecosystems with infrequent stand-initiating events
ESSF: Englemeann Spruce - Subalpine Fir biogeoclimatic zone
Mature/Old: stand age >120 years
NDT2: ecosystems with infrequent stand-initiating events
ESSF: Englemeann Spruce - Subalpine Fir biogeoclimatic zone
Burned: exposed to wildfire within last 40 years
NDT1: ecosystems with rare stand-initiating events;
ESSF: Engelmann Spruce - Subalpine Fir biogeoclimatic zone
Mature/Old: stand age >120 years
NDT1: ecosystems with rare stand-initiating events
ESSF: Engelmann Spruce - Subalpine Fir biogeoclimatic zone
Burned: exposed to wildfire within last 40 years
NDT1: ecosystems with rare stand-initiating events;
ICH: Interior Cedar - Hemlock biogeoclimatic zone
Mature/Old: stand age >100 years
NDT1: ecosystems with rare stand-initiating events
ICH: Interior Cedar - Hemlock biogeoclimatic zone
Burned: exposed to wildfire within last 40 years

*NDT: natural disturbance type; mean disturbance intervals described in Province of British Columbia (1995)

36

4.2.2 Fine-Filter Conservation Features
A coarse-filter SCP approach will likely identify lands that meet the habitat
requirements of the majority of species present in a planning region; however, it is inevitable
that the needs of some species will be missed and excluded from a given portfolio (Tingley et
al., 2014). To account for those species that pass through the wide mesh of the coarse-filter, it
is good practice to also incorporate fine-filter conservation features (Tingley et al., 2014).
The fine-filter approach addresses the individual habitat requirements of those species
considered to be endemic, threatened, or representative of multiple organisms in a planning
region (i.e., umbrella). Habitats that are unique, or rarely distributed across a planning region,
can also be used as fine-filter conservation features. I selected 4 fine-filter conservation
features to complement the previously discussed coarse-filter features. These included grizzly
bear, bull trout, woodland caribou, and a special features layer.
4.2.2.1 Grizzly bear
I selected grizzly bear to serve as a fine-filter conservation feature because they are a
wide-ranging omnivore that utilizes a number of different food sources and habitat types
(Apps, 2013). Grizzly bears that inhabit mountainous terrain of the WHSA have home ranges
of approximately 300 km2 and are capable of travelling ±10 km/day (Ciarniello, 2006). It is
because of these generalist life-history traits that grizzly bear is often referred to as an
umbrella species whose habitat requirements cover the needs of multiple other species by
default (Noss and Cooperrider, 1994; Noss et al., 1996; Apps, 2013). Accordingly, I mapped
and selected for suitable grizzly bear habitat in an effort to capture large, intact landscapes
that represented a multitude of ecosystems.

37

Grizzly bears are blue-listed as a ‘species of special concern’ in British Columbia.
Population declines are the result of low fecundity rates and direct/indirect mortality caused
by habitat fragmentation (Apps, 2013). Aside from capitalizing on their utility as an umbrella
species, seeking to capture large portions of grizzly bear habitat in this prioritization exercise
would help to promote the persistence of this species.
I used methodology described in the Draft Cumulative Effects Framework Grizzly
Bear Value Summary (Province of British Columbia, 2015b) as a baseline to construct a
grizzly bear layer for use in MARXAN-ILP. In ArcGIS, datasets describing BEC zone and
stand age/composition (Appendix 1) were used to remove mid-seral, conifer-dominated
forests, with >30% closed canopy as these stand characteristics are not ideal for grizzly bear
forage production. Of the remaining stands, only those identified in the Broad Ecosystem
Inventory (Appendix 1) as ‘capable/suitable’ grizzly habitat were selected for inclusion in the
layer and all others deleted. To refine the layer further, I only included those areas with >10
km 2 of roadless habitat to accommodate the daily range of an adult female grizzly bear
(Province of British Columbia, 2015b). Lastly, I added all Wildlife Habitat Areas (Appendix
1) set aside for grizzly bears in the WHSA to the finalized layer.
4.2.2.2 Bull Trout
Bull trout was selected for use as a fine-filter conservation feature that represented a
broad diversity of aquatic values across the WHSA. These fish are associated with cold,
high-elevation streams that have clean gravel beds and undisturbed riparian vegetation.
Mapping the known extent of the species across the WHSA allowed me to select for large,
healthy watersheds that contained relatively little fragmentation.

38

At the provincial scale, suitable habitat has been reduced by large-scale hydroelectric
projects and forestry-related riparian degradation. These forms of development have
contributed to bull trout being blue-listed as a ‘species of special concern’. The recent
approval of the Site C Dam Clean Energy Project may perpetuate bull trout declines within
the WHSA and provided further justification for prioritizing the habitat needs of this species.
I developed a bull trout layer for the WHSA using 1:50,000 watersheds as selectable
units (Appendix 1). To be included in this layer, watersheds were required to satisfy one of
the following criteria: overlap with a Fisheries Sensitive Watershed – Bull Trout (legal or
proposed), overlap with Wildlife Habitat Area – Bull Trout (legal or proposed), identified by
experts as critical rearing habitat (Williamson 2017, pers. comm), or identified in Appendix 4
of Hagen and Decker (2011) as being a tributary large enough to support local populations.
Provincial bull trout specialists provided supplemental information on which watersheds
should be included and/or excluded from the resultant layer (Williamson & Peck 2017, pers.
comm).
4.2.2.3 Caribou
Woodland caribou were selected to serve as a fine-filter conservation feature because
of their association with alpine and subalpine parkland habitats. These two habitat types were
not represented by any of the other conservation features used in this study. Caribou also
utilize old, mid-elevation stands containing arboreal lichens (Johnson et al., 2004). Therefore,
mapping and selecting for caribou habitat would not only capture high-elevation habitats, but
also provide redundant representation of old-forest alongside the forest pattern and process
layers.

39

The WHSA contains portions of Southern, Central, and Northern Mountain caribou
populations — 3 Designatable Units1 identified by the Committee on the Status of
Endangered Wildlife in Canada (COSEWIC) and recognized under the federal Species-atRisk Act (SARA). The Northern Mountain population is currently blue-listed as a ‘species of
special concern’ in the province. Those animals which comprise the Central and Southern
Mountain populations are red-listed (i.e., threatened or endangered). The status of these
animals provided further justification for their inclusion in this conservation planning
exercise.
I used a core habitat dataset, developed by Jones (2008) to represent mountain
caribou herds occurring in areas of the WHSA that fell within the South Peace region (i.e.,
Central and Southern Mountain populations). Jones (2008) identified suitable summer/winter
habitats using resource selection functions (RSFs) and ranked habitat quality based on
probability of selection from 1 (very low) to 5 (very high). Those areas with a ‘very high’
probability of selection were considered to be core habitat. Point locations of GPS-collared
animals were used to further refine the core habitat polygons and recognise the recent
distribution of caribou — potentially excluding those high-value habitats that were not
occupied. For this reason, I decided to merge core habitat polygons, reflective of ‘very high’
selection probability and recent telemetry, with those polygons assigned a ‘high’ probability
of selection. In this way, the overall selection area for South Peace herds was increased and
represented those habitats currently occupied by caribou, as well as those areas that are
unoccupied but contain high-value habitat features.

1

The term Designatable Unit (DU) is described by COSEWIC as a significant unit of a taxonomic species that
is important to the evolutionary legacy of the species as a whole and, if lost, would likely not be replaced
through natural dispersion.

40

A similar approach to that employed in the South Peace was used by Gustine and
Parker (2008) to delineate suitable habitat for herds occurring in the northern portions of the
WHSA (i.e., Northern Mountain population). Gustine and Parker (2008) completed RSFs
across all seasons and assigned a quality rank by binning probability of selection into deciles
(1 = low; 10 = high). Where Jones (2008) used telemetry points to refine core polygons,
Gustine and Parker (2008) used telemetry points to test the predictive capacity of the RSF
modelling. They found the modelling did well in predicting the habitat selection of collared
animals across seasons but that increasing the bin size to quintiles would absorb some of the
error present in RSFs (Gustine & Parker, 2008). Based on this observation, I rescaled Gustine
and Parker’s (2008) RSF dataset into quintile bins and extracted those habitats given a
ranking >4. The extracted selections from each season were then merged to create a single
layer representing the full range of year-round habitats selected by woodland caribou in the
northern portions of the WHSA.
I used an RSF model produced by Johnson et al. (2004) to represent those animals in
the Southern Mountain Population that were not captured by the Jones (2008) data (e.g., Hart
Range herd). The RSF produced by Johnson et al. (2004) differed from the abovementioned
datasets in that it only represented winter habitats. The RSF displayed probability of
occurrence based on observed distribution of caribou (i.e., survey data) relative to vegetation
and topography. Johnson et al. (2004) scaled probability of occurrence into four broad classes
of 1 (rare) to 4 (high). For consistency, I rescaled RSF dataset into quintile bins and extracted
those habitats given a ranking >4.
The multi-year collaring programs and RSF modelling described above were
relatively consistent in methodology and assisted in the development of a robust layer
41

depicting caribou habitat across the majority of the WHSA. There remained some gaps in the
layer where caribou were known to occur but for which a comparable level of analysis on
habitat quality and point locations had not been completed. Rather than attempt to extrapolate
local RSF modelling to herds with potentially dissimilar ecology, it was suggested that
Ungulate Winter Ranges 7-003 and 7-025 be used to fill these gaps in the layer. These special
management areas are relatively compatible with the mapping done by Jones (2008) and are
largely representative of high-elevation summer range as well (Seip 2017, pers. comm). In
addition to filling in gaps, these data would represent summer habitats utilized by the Hart
Range herd that were not captured in the RSF produced by Johnson et al. (2004).
All components described in this section were merged into a single dataset and the
resultant layer divided into individual herds based on a provincial layer describing legal herd
boundaries (Appendix 1). The finalized woodland caribou layer thus represented year-round
habitats that are currently occupied, as well as those areas that are unoccupied but contain
high-value habitat features. Before being applied in this study, the layer was validated by the
researchers whose data contributed to its creation (Seip & Parker 2017, pers. comm.).
4.2.2.4 Special Features
In SCP, a fine-filter conservation feature layer is usually developed to represent the
extent of special features across a given planning region. Special features are described as
those ecosystem components that are sensitive, spatially-limited, or of high biodiversity value
(Heinemeyer et al, 2004). For this study, I developed a single layer that represented wetlands,
karst deposits, and mineral licks across the WHSA.
Wetlands were included because they are an aquatic value in the WHSA whose full
distribution may not be adequately represented by the bull trout layer as wetlands are not
42

necessarily associated with high-elevation watersheds. Wetlands also provide habitat for
migratory waterfowl and perform valued ecosystem functions such as water filtration. Karst
formations were targeted as these ecosystems provide habitat for plant and animal species
that utilize caves for some portion of their life history. Mineral licks were incorporated as
they are an important nutrient source for ungulates and serve as hunting areas for carnivores.
Minimal geoprocessing was needed to construct the special features layer. Complete
and reliable datasets were available for each of the elements described above. In ArcGIS, I
merged datasets depicting karst deposits, mineral lick locations, and wetlands to produce a
consolidated layer for use in MARXAN-ILP (Appendix 1).
4.3 Step 2 Identify Conservation Goals and Targets for the WHSA
The overarching goal of this study was to prioritize lands for conservation in the
WHSA that were effective at promoting biodiversity, maintaining disturbance regimes, and
increasing climate change resilience. The conservation features described in the previous
section were intentionally selected to serve as mechanisms for meeting this goal. For
example, land facets build resilience by representing those abiotic elements that are
unaffected by a shifting climate and provide a foundation for evolutionary processes. The
forest pattern and process layers represent three distinct disturbance regimes in combination
with forest types containing high levels of biodiversity. The individual species used for finefilter conservation features each represent a wide array of ecosystems and the habitat needs
of species-at-risk making them effective surrogates for promoting biodiversity across the
WHSA.
Though integral to the SCP process, simply identifying these conservation features on
the landscape would not allow me to achieve the study goal. When conducting SCP in areas
43

such as the WHSA, there is an increased emphasis on using the area efficiently to achieve
desired conservation outcomes so that opportunities for resource use are maintained
(Wiersma & Sleep, 2018). With this in mind, I identified a suite of targets for use in this
study that were informed by relevant literature. These targets were meant to be
uncompromising in capturing the amount of area needed to achieve desired conservation
outcomes, but also efficient in that they would reach these outcomes while still allowing
some lands to be used for non-conservation purposes.
I drew upon the best available knowledge to assign each conservation feature with a
target that was reflective of that feature’s ecology and spatial extent within the WHSA (Table
3). These targets instructed MARXAN-ILP on how much area to include for each
conservation feature when conducting prioritizations. For example, if a conservation feature
was assigned a target of 10%, MARXAN-ILP would capture 10% of that feature’s spatial
extent across the WHSA in the resultant solution space. Effort was made to select those
targets that would provide each conservation feature enough area to effectively fulfill the
purpose for which it was designed (see Section 4.2 for purpose descriptions).
Of note, these targets were drawn from studies whose planning region(s) differed in
area from that of the WHSA. The percentage of area required for a feature to be adequately
represented at the scale of one study area may not constitute proportionate representation at
the scale of another. However, when conducting SCP it is often the case that targets specific
to a given study area are unavailable and practitioners must extrapolate from those used in
comparable research. The research used to inform target selection was conducted within, or
directly adjacent to, the WHSA and represented the best-available data at the time of this
study.
44

Table 3. Targets assigned to each conservation feature for application in MARXAN-ILP scenarios.

Coarse-filter
Land Facet Diversity
Land Facet Rarity
NDT3-SBS-Mature/Old
NDT3-SBS-Burned
NDT3-BWBS-Mature/Old
NDT3-BWBS-Burned
NDT2-SWB-Mature/Old
NDT2-SWB-Burned
NDT2-SBS-Mature/Old
NDT2-SBS-Burned
NDT2-ESSF-Mature/Old
NDT2-ESSF-Burned
NDT1-ESSF-Mature/Old
NDT1-ESSF-Burned
NDT1-ICH-Mature/Old
NDT1-ICH-Burned

Target (%)
50
50
76
100
46
100
83
100
66
100
75
100
74
100
75
100

Fine-filter
Burnt Pine Caribou Herd
Finlay Caribou Herd
Gataga Caribou Herd
Graham Caribou Herd
Hart Ranges Caribou Herd
Kennedy Siding Caribou Herd
Moberly Caribou Herd
Muskwa Caribou Herd
Narraway Caribou Herd
Pink Mountain Caribou Herd
Quintette Caribou Herd
Scott Caribou Herd
Grizzly Bear
Bull Trout
Special Features

Target (%)
90
90
90
90
90
90
90
90
90
90
90
90
60
60
60

In efforts to optimize conservation outcomes, targets assigned to individual woodland
caribou herds (i.e., retention of 90% of preferred habitats) met those objectives outlined in
Implementation Plan for the Ongoing Management of South Peace Northern Caribou in
British Columbia (Ministry of Environment, 2013). The Quinette Caribou Herd is assigned
an 80% target under the implementation plan but was increased up to 90% for purposes of
this study. The target of 50% assigned to land facet diversity/rarity was consistent with that
used to represent physical variety/rarity in the Muskwa-Kechika Management Area
Biodiversity Conservation and Climate Change Assessment (Yellowstone to Yukon
Conservation Initiative, 2012). I used the ‘higher biodiversity emphasis’ values, described in
the Biodiversity Guidebook (Province of British Columbia, 1995), to define targets for each
forest pattern and process layer. For bull trout, I used the aquatic focal species representation
45

target of 60% outlined in Conservation Area Design for the MKMA (Heinemeyer et al, 2004).
When no consistently referenced target was available, a conservative default target of >60%
of the feature’s extent was assigned (i.e., grizzly bear, special features).
4.4 Step 3 Review Existing Protected Areas within the WHSA
I combined all PAs within the WHSA into a single polygon to serve as a conservation
feature in MARXAN-ILP. I then ran a scenario where 100% of the existing PAs were
selected and all other conservation features left without a target. The result of this scenario
revealed the extent to which existing PAs in the WHSA contributed towards achieving
targets for conservation and assisted in identifying gaps within the existing system. The
percentage of each conservation feature’s total area represented by existing PAs was then
subtracted from each of the target used in this study. For example, if 10% of the bull trout
layer fell within existing PAs, an adjusted target would be 50% to achieve a sum total 60% of
bull trout representation in the WHSA.
4.5 Step 4 Use MARXAN-ILP to Generate a Portfolio of Priority Lands for Conservation
I used MARXAN-ILP to prioritize lands for conservation in the WHSA. The tool
allowed me to assess the adequacy of the existing PA system across the WHSA, observe the
impact of resource development on conservation features, and identify portfolios of candidate
lands for protection that met the targets and goals of this study.
4.5.1 Defining MARXAN-ILP Inputs
In order for MARXAN-ILP to operate, the user must first define a number of inputs
that reflect the overall purpose of a given SCP process. Inputs typically required for

46

MARXAN-ILP include a definition of planning units and costs. All inputs used for this study
are described in the following sections.
4.5.1.1 Planning Units
Prior to running MARXAN-ILP, I divided the WHSA into planning units. A planning
unit is a parcel of land made available for selection by MARXAN-ILP. The size and shape of
a planning unit is user-defined. Commonly used planning units in SCP include uniform grids,
surveyed land parcels, administrative boundaries, and those units defined by various
ecological classification systems (e.g., watersheds, wildlife management units,
biogeoclimatic zones, etc.). Given the large area of the WHSA, I opted to use 1 km2 planning
units in this analysis. This was determined after observing that 1 ha planning units used in
earlier versions of MARXAN-ILP developed for this study made for longer computing times
and produced a solution space not superior in quality to that produced by the larger 1 km2
planning units.
4.5.1.2 Planning Unit Costs
Once planning unit size and shape has been established, MARXAN-ILP requires a
cost metric for all units in a planning region. Planning unit cost is representative of the
expense associated with including a given planning unit in a selection. In this way, planning
unit cost can reflect property values (acquisition cost), resource potential (opportunity cost),
or an arbitrary number designed to place constraints on the selection. Whatever the metric,
planning unit cost forces the tool to be efficient and select high-value areas because the
acquisition of planning units is expensive. There are minimal private holdings in the WHSA
that would have an assigned market value, and the available data on resource potential in the
region was inconsistent and forestry-centric; therefore, an arbitrary planning unit cost of 1
47

was assigned to all planning units in the WHSA. This would force the tool to be economical
in its selection and target those planning units containing multiple, overlapping conservation
features as inefficient selection would absorb unnecessary planning unit cost.
4.5.1.3 Cost Surfaces
In addition to basic planning unit cost, MARXAN-ILP allows the user to exert further
constraint on selection through the use of cost surfaces. A cost surface is a spatial
representation of those elements in a planning region that have been identified as having a
negative impact on the conservation features being selected for. A cost surface is mapped
across the landscape and any planning unit that contains a portion of that surface is given a
user-defined penalty. This penalty is typically weighted based on the extent to which the
form of land-use represented in the cost surface degrades the values that make up a
conservation feature. In this way, the use of a cost surface dissuades the tool from selecting
those planning units containing elements that may jeopardize the goals of a given SCP
process. I developed three separate cost surfaces in order to explore the influence that varying
levels of resource development had on the selection of priority lands for conservation in the
WHSA.
4.5.1.3.1 High-Sensitivity
Of the three cost surfaces produced for this study, the high-sensitivity cost surface
was designed to be the most conservation-oriented. The intent of this cost surface was to
make MARXAN-ILP ‘highly sensitive’ to resource development when prioritizing areas for
conservation in the WHSA. To build the layer, I first mapped the spatial extent of permanent
development in the WHSA using the layers described in Appendix 1 and the geoprocessing
steps outlined in Appendix 2. I defined ‘permanent development’ as areas where human-use
48

had resulted in complete landscape conversion that would require substantial restoration
effort to be returned to a natural state.
Using ArcGIS, I placed buffers on the various forms of development based on
distances described in Dyer et al. (2001) and Polfus et al. (2011). The distances were a
reflection of the extent to which woodland caribou avoid the respective forms of permanent
development at the foundation for the high-sensitivity cost surface (Table 4). Given their
large size, these avoidance buffers not only addressed the needs of that species, but also
increased the likelihood of accommodating the needs of species that exhibited less sensitivity
to resource development.
Table 4. Buffer distances used for the construction of the high-sensitivity cost surface to represent the zone of
influence in the Wild Harts Study Area.

Development
Roads (paved)
Roads (rough)
Seismic Lines
Urban/Agriculture
Mines
Outfitter Cabins
Gas Wells

Buffer*
2 km from centerline
1 km from centerline
250 m from centerline
9 km radius
2 km radius
1.5 km radius
1 km radius

*Buffer distances represent woodland caribou avoidance of resource development described in Dyer et al.
(2001) and Polfus et al. (2011).

Planning units containing the high-sensitivity cost surface were assigned a large
penalty of 10, making it 10x as costly to select a planning unit containing the high-sensitivity
cost surface than one where the surface was absent. The rationale behind the large penalty
and avoidance buffers was to dissuade the MARXAN-ILP from selecting planning units
within, or proximal to, any form of resource development. Accordingly, the high-sensitivity
cost surface would be used in scenarios to prioritize those high-value conservation lands

49

occurring outside the ‘zone of influence’ in the WHSA. The term zone of influence (ZOI) is
used in this context to describe the ability of industry to create impacts that extend past the
spatial footprint of development.
4.5.1.3.2 Medium-Sensitivity
The intent of this cost surface was to make MARXAN-ILP exhibit a ‘mediumsensitivity’ to resource development when prioritizing areas for conservation. The mediumsensitivity cost surface was built to represent the spatial extent of both permanent and
temporary development in the WHSA. For the purposes of this study, I defined temporary
development as harvested areas <20 years old, as this is a threshold commonly used by forest
managers in British Columbia to identify areas that haven’t recovered fully from harvesting
activities (Ministry of Sustainable Resource Management, 2005). The surface was
constructed by merging the data described in Appendix 1 with the permanent development
layer used in the high-sensitivity cost surface.
All planning units containing portions of the medium-sensitivity cost surface were
given a moderate penalty of 5. No avoidance buffers were placed on temporary or permanent
development. Similar to the high-sensitivity cost surface, the medium-sensitivity cost surface
would encourage MARXAN-ILP to select non-developed areas in efforts to produce the
optimal solution for conservation at the lowest cost. However, in medium-sensitivity
scenarios, the reduced penalty and absence of buffers would make available for selection
those high-value planning units that occur proximal to temporary and permanent
development. These are planning units that would have been avoided in scenarios where the
high-sensitivity cost surface was used on account of them falling within avoidance buffers
and carrying high costs. In this way, the medium-sensitivity cost surface would assist in
50

locating areas of high conservation value that fell within the ZOI, but outside the physical
footprint of development where possible.
4.5.1.3.3 Low-Sensitivity
The intent of this cost surface was to force MARXAN-ILP exhibit a ‘low-sensitivity’
to resource development when prioritizing areas for conservation in the WHSA. The lowsensitivity cost surface only represented the spatial extent of permanent development in the
WHSA. I built the low-sensitivity cost surface using the layers described in Appendix 1 and
geoprocessing steps outlined in Appendix 2. When constructing the low-sensitivity cost
surface I did not include spatial representations of temporary disturbance or avoidance
buffers. All planning units containing portions of the low-sensitivity cost surface were given
a menial penalty of 2.
Similar to the two cost surfaces described above, the low-sensitivity cost surface
would encourage MARXAN-ILP to select non-developed areas in efforts to produce the
optimal solution for conservation at the lowest cost. However, the minimal penalty, absence
of buffers, and omission of forestry data in low-sensitivity scenarios would make those highvalue planning units that occur within temporary and permanent development available for
selection. These are planning units that would have been avoided in high-to-medium
sensitivity scenarios as they occur within avoidance buffers, recently logged areas, or fully
converted landscapes. Accordingly, the low-sensitivity cost surface was intended to assist in
identifying those areas of the WHSA that retained conservation value despite having
undergone extensive resource development.

51

4.5.1.5 Protected Areas
I used the ‘locked out’ option to make existing PAs unavailable for selection when
running MARXAN-ILP. The contributions of existing PAs were incorporated through the
completion of Step 3 in the SCP process. Targets had been adjusted based on the extent to
which existing PAs already achieved them and scenarios run to prioritize remaining
percentages. The ‘locked out’ option simply ensured that MARXAN-ILP selected planning
units based on their conservation value and not primarily based on their adjacency to PAs.
4.6 Running MARXAN-ILP
Once all of the user-defined inputs had been selected, the model was run using
MARXAN-ILP software. MARXAN-ILP uses integer linear programming to spatially
identify a configuration of planning units that represent the optimal solution to the objective
function:
Minimize ∑ 𝐶𝑜𝑠𝑡𝑠 s𝑒𝑙𝑒𝑐𝑡𝑒𝑑 𝑝𝑙𝑎𝑛𝑛𝑖𝑛𝑔 𝑢𝑛𝑖𝑡𝑠;
Given that ∑ Conservation 𝑓𝑒𝑎𝑡𝑢𝑟𝑒𝑠 s𝑒𝑙𝑒𝑐𝑡𝑒𝑑 𝑝𝑙𝑎𝑛𝑛𝑖𝑛𝑔 𝑢𝑛𝑖𝑡𝑠 ≥ 𝑢𝑠𝑒𝑟 𝑑𝑒𝑓𝑖𝑛𝑒𝑑 𝑡𝑎𝑟𝑔𝑒𝑡𝑠;
and 𝑠𝑒𝑙𝑒𝑐𝑡𝑒𝑑 𝑝𝑙𝑎𝑛𝑛𝑖𝑛𝑔 𝑢𝑛𝑖𝑡𝑠 = (0,1)
Prior to running MARXAN-ILP, the extents of all conservation features in the area were
mapped across the landscape. The cost surface and conservation features were then blanketed
by a planning unit grid that contained a planning unit cost for each cell (Fig. 5).

52

Figure 5. Cross-section of the MARXAN-ILP model used to prioritize lands for conservation in the Wild Harts
Study Area.

53

Figure 6 provides a reductionist model showing the mechanics of MARXAN-ILP.
When a scenario is initiated, MARXAN-ILP hovers over each planning unit, drills down
through the layers present, and calculates the overall value of that planning unit. If a
conservation feature is present with the planning unit, it is assigned a value of (1). If the
planning unit contains no conservation features, it is given a score of (0). If a planning unit
contains a cost surface, it is assigned whatever penalty the user has determined is in line with
the objectives of the study. For the purposes of this example, an arbitrary penalty of 5 was
assigned to all planning units containing a cost surface (Fig. 6).
In fulfilling its objective function, MARXAN-ILP attempts to meet conservation
targets by amassing those planning units that avoid cost surfaces and have the highest amount
of overlapping conservation features. The fact that each planning unit has an acquisition cost
of 1 places even further emphasis on efficiency of selection and cost reduction. With this in
mind, the selected cell in example A has a higher likelihood of being included in a
prioritization output than does the cell selected in B (Fig. 6).

54

Figure 6. Cross-section of the MARXAN-ILP conservation value calculations used to prioritize lands for
conservation in the Wild Harts Study Area.

55

4.7 Scenario Development
I developed four MARXAN-ILP scenarios to address each of my research questions
(Table 5). Across all scenarios, MARXAN-ILP was instructed to meet conservation targets at
the lowest cost possible. This was meant to encourage the selection of areas containing
multiple overlapping features. In doing so, the resultant portfolios would address the first
research question (i.e., what areas contain high conservation value?)
Table 5. Description of the MARXAN-ILP scenario inputs used to prioritize lands for conservation in the Wild
Harts Study Area.

Scenario
A.
B.
C.
D.

Cost Surface
High-sensitivity
Medium-sensitivity
Low-sensitivity
None

Constraint
High

Low

Cost surfaces were used to help identify areas of the WHSA that retained
conservation value despite the presence of resource development (i.e., research question 2).
Scenario A was designed to produce a portfolio of high-value lands unaffected by resource
development. This was meant to provide a benchmark of intactness from which I could
observe deviations made by MARXAN-ILP in subsequent scenarios. Scenario B was
designed to increase selection of high-value planning units in the ZOI. Scenario C was meant
to increase selection in temporarily developed landscapes. Scenario D was designed to
increase selection of permanently developed landscapes. When overlaid, these portfolios
would provide zonation from which to base management objectives.
In addition to the avoidance of development, cost surfaces were expected to place
constraint on MARXAN-ILP by reducing the amount of penalty-free area available to the
tool for selection. I predicted that the more constrained a scenario became, the greater the
56

emphasis would be on achieving targets with compact clusters of high-value planning units to
reduce costs. Within the context of this study, a compact portfolio containing low levels of
resource development was thought to exhibit principles of good PA design. The portfolio that
met targets and best emulated those principles would be used to address the third research
question (i.e., what is an optimal conservation portfolio to direct planning processes?).
4.8 Statistical Analysis
Using the ArcGIS Patch Analyst tool, I examined the extent to which common
landscape ecology fragmentation statistics varied across scenario results. Statistics used to
compare portfolios included total number of patches, maximum patch area (km2), average
patch area (km2), average patch perimeter (km), mean shape index, and total core area index.
Mean shape index (MSI) was used to measure the relative edge-to-area relationship of
patches that comprised a given solution space. As MSI decreases, so too does the shape
complexity of a given solution space.
Total core area index (TCAI%) measures the percentage of core area present within a
solution space based on user-defined buffers. I calculated TCAI% at 200 m (2x average tree
height) to determine the percentage of core area unaffected by altered temperature regimes
along patch edges. A TCAI% was also calculated at 500m from edge as this is a general core
requirement for wildlife that prefer interior habitats. The purpose of applying these buffers
was less to determine whether solutions met ‘core criteria’ for any specific value so much as
it was to assess the extent to which compactness of solution space differed across scenarios
A–D.

57

5.0 RESULTS
This section reviews the results of completing the 4 steps of SCP (Table 1). These
steps included (1) the selection and development of conservation features; (2) the
identification of conservation goals and targets; (3) a review of existing PAs; (4) and the use
of MARXAN-ILP to identify lands suitable for protection in the WHSA. Inset maps have
been included for some figures and are not intended to highlight any specific result so much
as they are in place to provide the reader with an understanding of fine-scale details.
5.1 Conservation Feature Layers
A total of 31 conservation feature layers were constructed, overlain across the entire
planning region, and made available for selection by MARXAN-ILP. All conservation
features were given an equal value of (1) if present within a planning unit.
5.1.1 Coarse-filter Conservation Features
I developed 16 coarse-filter conservation features for this study. These features
included land facet diversity, land facet rarity, and a series of layers displaying an
intersection of natural disturbance, biogeoclimatic zone, and forest age.
5.1.1.1 Land Facet Diversity
The land facet diversity layer occupies 9.6% of the WHSA total area and is
distributed relatively evenly across the planning region (Fig. 7). Of particular note is the
extent to which areas containing high abiotic diversity have a tendency to straddle the band
of mountainous terrain that bisects the WHSA on a southeast-northwest line (Fig. 7). Aside
from a few exceptions, the peak-and-ridge dominated landscape that comprises the center of

58

that band is low in abiotic diversity when compared to the transitional landscapes that occur
along the outer edges of the Rocky Mountain corridor (Fig. 7).
5.1.1.2 Land Facet Rarity
Land Facet Rarity occupies 0.5% of the WHSA. An observable pattern in this layer is
the tendency of areas with rare abiotic attributes to occur in river valley bottoms (Fig. 8). In
keeping with this observation, there appears to a disproportionate amount of rare land facets
bordering the Peace Arm of the Williston Reservoir (see inset map) (Fig. 8). This is likely
due to the fact that the Peace River is the only eastwardly flowing river through the Rocky
Mountains in the WHSA — a distinction that carries with it unique geological history and
climatic conditions that have created the rare abiotic features present there.
5.1.1.3 Forest Pattern and Process:
Table 6 describes the distribution and amount of area in the WHSA occupied by 14
forest pattern and process layers shown in Figure 9. The inset map provides insights on the
scale and detail at which these features were typed out across the landscape (Fig. 9). Alpine
areas have been excluded as these layers were intended to only capture forested areas. It must
be noted that although the following figure show layers for mature/old and burned forest on
the same map, these layers were used as standalone conservation features in all MARXANILP scenarios. They have only been combined in this figure for efficiency’s sake.

59

Figure 7. Spatial extent of the land facet diversity layer used by MARXAN-ILP to prioritize lands for
conservation in the Wild Harts Study Area; inset represented by red frame.

60

Figure 8. Spatial extent of the land facet rarity layer used by MARXAN-ILP to prioritize lands for conservation
in the Wild Harts Study Area; inset represented by red frame.

61

Figure 9. Spatial extent of all forest pattern and process layers used by MARXAN-ILP to prioritize lands for
conservation in the Wild Harts Study Area; inset represented by red frame.

62

Table 6. Location and size of each forest pattern and process layer in relation to the Wild Harts Study Area
(WHSA).

Label

Distribution

Confined to riparian areas and watersheds in the
northwestern quarter of the WHSA
Confined to riparian areas and watersheds in the
NDT3-SBS-Burned
northwestern quarter of the WHSA
Occupies low-elevation valleys and riparian
NDT3-BWBS-Mature/Old
areas that straddle the Rocky Mountain corridor
Occupies low-elevation valleys and riparian
NDT3-BWBS-Burned
areas occurring along the outer extents of the
WHSA
Occupies high-elevation valleys and riparian
NDT2-SWB-Mature/Old
areas within northeastern quarter of the WHSA
Comprised of a few large burned watersheds in
NDT2-SWB-Burned
the northeastern quarter of the WHSA
Occurs in low-elevation valley bottoms
NDT2-SBS-Mature/Old
containing streams in the southern half of the
WHSA
Occurs sporadically throughout the central
NDT2-SBS-Burned
sections of the WHSA at low elevations
Occurs at mid-elevation across the entire WHSA
NDT2-ESSF-Mature/Old and is densely concentrated in sections that occur
above the Peace Arm of the Willison Reservoir
Primarily confined to the mid-lower elevations
NDT2-ESSF-Burned
on the eastern side of the Rocky Mountain
corridor
Dominates the forested area on the western
NDT1-ESSF-Mature/Old slopes of the Rocky Mountain corridor at midhigh elevation
Occurs sporadically along the western slopes of
NDT1-ESSF-Burned
the Rocky Mountain corridor at mid-high
elevation
Poorly distributed and confined to mid-high
NDT1-ICH-Mature/Old
elevation landscapes at the most southern extents
of the WHSA
Poorly distributed and confined to mid-high
NDT1-ICH-Burned
elevation landscapes at the most southern extents
of the WHSA
NDT3-SBS-Mature/Old

Percent (%)
of total area
0.4
0.02*
10
1*
7
0.2*
6
0.2*
11
1.4*
12
0.2*
0.3
0.1*

*Below the area required to meet high biodiversity target described in Province of British Columbia (1995)

63

5.1.2 Fine-filter Conservation Features
To complement the coarse-filter conservation features, an additional 15 layers were
built that displayed the spatial extent of select fine-filter conservation features across the
WHSA. Fine-filter conservation features used in this study included woodland caribou, bull
trout, grizzly bear, and a special features layer displaying a mixture of unique habitats in the
WHSA.
5.1.2.1 Woodland Caribou
With all herds consolidated, the woodland caribou layer occupies 45% of the WHSA
total area and is distributed across the majority of the planning region (Fig. 10). The layer is
confined to high-elevation alpine areas and is largely absent from valley bottoms (see inset
map) (Fig. 10). There is no woodland caribou representation in the upper-half of the planning
region’s southeastern quarter and the lower-half of the northwestern quarter as herds have
been extirpated from both these areas (Fig. 10). It must be noted that although woodland
caribou herds in the WHSA are being presented in the same figure, each herd was used as a
standalone conservation feature and made available for selection by MARXAN-ILP in all
scenarios. They have only been combined in this figure for efficiency’s sake.
5.1.2.2 Bull Trout
The bull trout layer is evenly distributed across the entire planning region and
occupies approximately 44% of the WHSA total area (Fig. 11). The majority of major
watersheds and rivers present in the WHSA are represented by this layer. Of particular note
is the exclusion of a major drainage at the center of the northeast quarter of the WHSA (Fig.
11). This blank space represents the Sikanni Chief River headwaters — a section of river
where bull trout are absent due to a downstream obstacle outside the bounds of the WHSA
64

(i.e., Sikanni Chief Falls). The Peace Arm of the Williston Reservoir supports bull trout
populations but was excluded from this layer on account of being a manmade structure that
represents sub-optimal habitat.
5.1.2.3 Grizzly Bear
The grizzly bear layer dominates the majority of the planning region — occupying
61% of the WHSA total area. The wide distribution of this fine-scale conservation feature is
likely due to grizzly bear being a generalist species that is able to inhabit a vast array of
ecosystems and habit types. Despite the wide distribution, the grizzly bear layer is largely
absent from the southeastern quarter of the WHSA (Fig. 12). The disproportionate levels of
resource-related disturbance that occur within that section of the WHSA would have
prevented these areas from being included on account of their inability to accommodate the
daily range of an adult female grizzly.
5.1.2.4 Special Features
The special features layer displays the spatial extent of karst deposits, wetlands, and
mineral licks across the WHSA. This fine-filter conservation feature occupies approximately
4% of the WHSA total area. Karst deposits occur in narrow striations within the Rocky
Mountain corridor and are most abundant in the mid-upper half of the planning region (Fig.
13). Wetlands are distributed evenly across the WHSA and become most concentrated in
those low-elevation areas and alongside streams (Fig. 13). Mineral licks have a scattered
distribution with no noticeable pattern (Fig. 13). In efforts to mask sensitive datasets, all three
layers were merged into a single, homogenous layer.

65

Figure 10. Spatial extent of the woodland caribou layers used by MARXAN-ILP to prioritize lands for
conservation in the Wild Harts Study Area; inset represented by red frame.

66

Figure 11. Spatial extent of the bull trout layer used by MARXAN-ILP to prioritize lands for conservation in the
Wild Harts Study Area; inset represented by red frame.

67

Figure 12. Spatial extent of the grizzly bear layer used by MARXAN-ILP to prioritize lands for conservation in
the Wild Harts Study Area; inset represented by red frame.

68

Figure 13. Spatial extent of the special features layer used by MARXAN-ILP to prioritize lands for conservation
in the Wild Harts Study Area; inset represented by red frame.

69

5.2 Gap Analysis
Representation of conservation features within PAs in the WHSA was highly variable
(Table 7). Conservation features that were well represented (>50%) within PAs included the
Muskwa Caribou Herd and recently burned areas within the Spruce-Willow-Birch
biogeoclimatic zone, natural disturbance type 2 (NDT2-SWB-Burned). The majority of
remaining conservation features had mid-to-low levels of representation within existing PAs
(5–50%). Of particular note are those conservation features with little-to-no (<1%)
representation. These include the Scott Caribou Herd, Burnt Pine Caribou Herd, and those
areas in the Sub-boreal Spruce biogeoclimatic zone and natural disturbance type 3 that
contain late-seral or recently burned stands (NDT3-SBS-Mature/Old & NDT3-SBS-Burned).
Table 7. Representation of conservation features achieved by existing protected areas in the Wild Harts Study
Area.

Coarse-filter
% Protected
Land Facet Diversity
14
Land Facet Rarity
28
NDT3-SBS-Mature/Old
0.86
NDT3-SBS-Burned
0
NDT3-BWBS-Mature/Old
10
NDT3-BWBS-Burned
24
NDT2-SWB-Mature/Old
26
NDT2-SWB-Burned
67
NDT2-SBS-Mature/Old
5
NDT2-SBS-Burned
5
NDT2-ESSF-Mature/Old
13
NDT-ESSF-Burned
2
NDT1-ESSF-Mature/Old
8
NDT1-ESSF-Burned
22
NDT1-ICH-Mature/Old
30
NDT1-ICH-Burned
23

Fine-filter
Burnt Pine Caribou Herd
Finlay Caribou Herd
Gataga Caribou Herd
Graham Caribou Herd
Hart Ranges Caribou Herd
Kennedy Siding Caribou Herd
Moberly Caribou Herd
Muskwa Caribou Herd
Narraway Caribou Herd
Pink Mountain Caribou Herd
Quintette Caribou Herd
Scott Caribou Herd
Grizzly Bear
Bull Trout
Special Features

70

% Protected
0
3
42
14
14
16
4
96
21
26
10
0.35
19
19
22

5.3 Cost Surfaces
High, medium, and low-sensitivity cost surfaces placed differential constraints on the
MARXAN-ILP tool to identify high-value conservation lands within varying intensities of
resource development.
5.3.1 High-Sensitivity
The high-sensitivity cost surface was the largest used in this study, occupying 44% of
the WHSA total area. The layer dominated the majority of the planning region’s southeastern
quarter — an area containing extensive oil and gas infrastructure (Fig. 14). The western
portions of the layer occupy those valleys and river bottoms that provide access for various
forms of resource extraction within an otherwise impassible and mountainous landscape (Fig.
14).
In general, the layer is more concentrated towards the lower half of the WHSA as
these areas are closer to population centers containing those resources needed to sustain
extraction activities (i.e., labour, equipment, mill sites, etc.) (Fig. 14). Of particular note is
the band of relatively intact landscapes running down the center of the WHSA on a
northwest-southeast line that are not covered by the high-sensitivity cost surface (Fig. 14).
The absence of the high-sensitivity cost surface in these areas is likely due to the rugged
topography and the challenges such landscapes create for industry in terms of access.
5.3.2 Medium-Sensitivity
The medium-sensitivity layer occupies approximately 3.8% of the study area and, like
the high-sensitivity layer, dominates the majority of landscapes in the southeastern quarter
(Fig. 15). Most of the watersheds in the planning region show evidence of harvesting and

71

associated road infrastructure (Fig. 15). The band of relatively intact landscapes mentioned in
the high-sensitivity results is even more noticeable with the absence of avoidance buffers
(Fig. 15). The northernmost portions of the WHSA are almost entirely void of resource
development save for what appears to be a small number of roads confined to the outer
bounds of the planning region (Fig. 15).
5.3.3 Low-Sensitivity
The low-sensitivity cost surface occupies approximately 2% of the WHSA total area.
Despite the reduction in area, the low-sensitivity cost surface still dominates the majority of
landscapes present in the southeastern corner of the planning region (Fig. 16). The presence
of all three cost surfaces here suggests the area is highly compromised and would require
substantial restoration effort to return to a natural state. Where the medium-sensitivity layer
occupied large portions of watersheds in the WHSA, the absence of forestry data in the lowsensitivity cost surface made it so that the only disturbance in these watersheds was the
spatial footprint of access roads. The band of intact landscapes down the center of the WHSA
remains noticeable and the northernmost extents are virtually free of this cost surface (Fig.
16).

72

Figure 14. Spatial extent of the high-sensitivity cost surface used to constrain MARXAN-ILP when prioritizing
lands for conservation in the Wild Harts Study Area; inset represented by green frame.

73

Figure 15. Spatial extent of the medium-sensitivity cost surface used to constrain MARXAN-ILP when
prioritizing lands for conservation in the Wild Harts Study Area; inset represented by green frame.

74

Figure 16. Spatial extent of the low-sensitivity cost surface used to constrain MARXAN-ILP when prioritizing
lands for conservation in the Wild Harts Study Area; inset represented by green frame.

75

5.4 MARXAN-ILP Scenario Outputs
The following sections describe the results of the 4 MARXAN-ILP scenarios
developed to prioritize high-value areas for conservation in the WHSA (Table 5).
5.4.1 Observable Patterns
For each scenario, MARXAN-ILP selected the configuration of planning units that
met conservation targets at the lowest cost. A noticeable pattern across all portfolios is the
pronounced corridor of selected lands that stretches from the southwestern extent of the
planning region to Northern Rocky Mountains Provincial Park in the northeast (Figs. 4 and
17). Below the Peace Arm, selection is largely concentrated to the western half of the Rocky
Mountains (Fig. 17). The opposite occurs above the Peace Arm where prioritization has a
tendency to concentrate in the eastern portion of the Rocky Mountains (Fig. 17).
Of particular note is the almost complete absence of selected planning units along the
western boundary of the WHSA above the Peace Arm (Fig. 17). A similar pattern is
observable in those blank spaces occurring along the easternmost extent of the WHSA below
the Peace Arm (Fig. 17). The lack of selection here is likely due to the absence of the
woodland caribou conservation feature and the ability of large conservation features to meet
targets for small conservation features through complementarity elsewhere.

76

Figure 17. Four portfolios produced by MARXAN-ILP to prioritize lands for conservation in the Wild Harts
Study Area; A: high-sensitivity cost surface used; B: medium-sensitivity cost surface used; C: low-sensitivity
cost surface used; D: no cost surface used.

77

5.4.2 Influence of Cost Surface:
Overlaying the four portfolios produced in this study highlights the influence of each
cost surface (Fig. 18). Portfolio A was placed at the top of this overlay to serve as a
benchmark of intactness. The high-sensitivity cost surface forced MARXAN-ILP to avoid
the southeastern quarter of the WHSA and heavily-developed watersheds (Fig. 18). The
portfolio is comprised of 28,018 km2 of intact planning units. However, the large
conservation targets resulted in 30% of that area falling within the ZOI (Fig. 18).
Portfolio B was placed underneath A to observe which planning units became
selected when the level of constraint on MARXAN-ILP was lowered. The mediumsensitivity cost surface produced a portfolio similar to A in that the southeastern quarter of
the WHSA and heavily-logged watersheds were avoided (Fig. 18). The reduced sensitivity to
industrial activities resulted in 3892 km2 of high-value planning units being selected in the
ZOI that were avoided in scenario A (Fig. 18).
Portfolio C was placed underneath B to observe the effect of the low-sensitivity cost
surface on selection. Low-elevation valleys containing high road density were avoided in this
scenario (Fig. 18). The lack of sensitivity to recent timber harvesting resulted in 81 km2 of
high-value planning units being selected in temporarily disturbed landscapes that were
avoided in scenario B.
The results of scenario D were placed at the bottom of the overlay to observe which
planning units were selected when no constraints were placed on MARXAN-ILP. The
absence of a cost surface resulted in Portfolio D capturing 175 km2 of high-value planning
units that were avoided in more constrained scenarios (Fig. 18). This is evident by the
increased selection of planning units within the southeastern quarter of the WHSA (Fig. 18).
78

Figure 18. Overlay of four portfolios produced by MARXAN-ILP to prioritize lands for conservation in the
Wild Harts Study Area; inset represented by pink frame.

79

5.4.3 Patch Statistics
As predicted, the level of constraint placed on the scenarios controlled the extent to
which compactness amongst selected planning units was achieved in the resultant portfolios.
Of the four scenarios, A was most constrained because of the reduction in penalty-free area
inflicted by the high-sensitivity cost surface. As a result, Portfolio A met targets with the
fewest patches and the largest average patch area of any scenario (Table 8). The patches that
comprised Portfolio A were the least-convoluted of any produced in this study as is made
evident by the low mean shape index and high total core area (Table 8).
The results of scenarios B through D were relatively uniform statistically (Table 8).
When compared to A, these portfolios were comprised of more small patches of planning
units which resulted in less core area and higher edge-to-area ratios (Table 8). Of particular
note is the extent to which the patches that comprised Portfolio D were less-convoluted than
those that made up B and C (Table 8). The increase in shape complexity displayed in
Portfolio B and C was likely the result of MARXAN-ILP selecting narrow bands of highvalue in-between linear developments that were demarcated by the low and mediumsensitivity cost surfaces (Fig. 17).
Table 8. Patch statistics of four MARXAN-ILP scenarios used to prioritize lands for conservation in the Wild
Harts Study Area.

Variable
Total Area (km2)
Number of Patches
Maximum Patch Area (km2)
Average Patch Area (km2)
Average Patch Perimeter (km)
Mean Shape Index (MSI)
Total Core Area Index @ 150m (%)
Total Core Area Index @ 500m (%)

Portfolio A Portfolio B Portfolio C Portfolio D
28018
27555
27577
27533
1140
1477
1471
1349
10077
6594
5499
5001
25
19
19
20
22
21
21
22
1.30
1.32
1.32
1.32
83
78
78
79
59
49
49
51
80

6.0 DISCUSSION
The WHSA in northeastern British Columbia is an area of high ecological importance
that is currently underrepresented by protected areas (PAs). The underlying purpose of this
research was to prioritize lands for conservation in the WHSA using a systematic
conservation planning (SCP) approach. This proved to be a complex exercise as the WHSA
is a large planning region with multiple biotic and abiotic values. The challenge was
increased further by the presence of competing land interests that threatened those values
being targeted for conservation. Employing the four steps of SCP allowed me to address
those challenges systematically and answer my three research questions.
6.1 What Areas Contain High Conservation Value for Select Coarse- and Fine-filter
Features?
In a multi-use landscape like the WHSA, planners must use area efficiently to achieve
conservation objectives so that opportunities for resource development are maintained
(Wiersma & Sleep, 2018). Recognizing this, I set out to identify sections of the WHSA that
were disproportionately high in conservation value. To be considered high in conservation
value, a planning unit needed to contain multiple values predetermined to be effective
surrogates for biotic and abiotic diversity. Assembling high-value planning units allowed
MARXAN-ILP to achieve conservation targets at the lowest cost. In meeting targets, each
portfolio captures the infrastructure that comprises and regulates natural systems in the
WHSA.
Each portfolio captures 50% of both the land facet diversity and land facet rarity
layers. These areas can provide a stable platform for evolutionary processes to play out as
biological communities reorganize in response to climate change in the WHSA (Beier and
81

Brost, 2010). The abiotic elements that comprise land facets also govern variation in local
temperature, precipitation, and disturbance regimes (Beier and Brost, 2010). Accordingly,
each portfolio captures a diversity of physical environments to maintain the processes that
shape current patterns of biodiversity in the WHSA.
All four MARXAN-ILP portfolios met the targets assigned to individual forest
pattern and process features. Sufficient late-seral forest was captured within each landscape
unit (i.e., NDT-BEC) to withstand the average fire size for that unit without experiencing a
collapse in biodiversity (Province of British Columbia, 1995). Although there was not
enough young natural forest to meet biodiversity objectives for early successional stands,
each portfolio contained all of the recently burned areas in the WHSA that have not
undergone salvage operations. As such, the portfolios accommodate habitat requirements of
fire-obligate species and contain unique features not found in young managed stands (i.e.,
burned snags) (Province of British Columbia, 1995).
All portfolios capture 90% of year-round habitat for each woodland caribou herd in
the WHSA. These areas are primarily comprised of intact alpine forests that provide refuge
from predators and a year-round food source in arboreal lichens. Conserving the majority of
caribou habitat in the WHSA may assist in the recovery of the species and, in turn, prevent
the extirpation of a valued component of biodiversity in the WHSA. Selecting for woodland
caribou habitat also resulted in large portions of the alpine and subalpine parkland
ecosystems being represented in each portfolio. These areas are not captured by any of the
other features used in this study and provide suitable habitat for Rocky Mountain bighorn
sheep, Stone’s sheep, and wolverine among other species.

82

The four MARXAN-ILP portfolios each contain 60% of suitable grizzly bear habitat
in the WHSA. This resulted in the prioritization of intact landscapes containing low road
density. Open canopy forests have been included in each portfolio because of their
association with Vaccinium spp. fruits — an important food source for grizzly bears in the
summer and fall (Apps, 2013). Similarly, avalanche chutes are captured in each portfolio as
these areas provide a spring food source in the form of emerging vegetation (Apps, 2013).
Accommodating the habitat requirements of this generalist species (i.e., contiguous
landscapes with high productivity rates) not only promotes the persistence of grizzly bears
but also addresses the needs of multiple species at lower trophic levels such as songbirds,
furbearers, and ungulates.
The portfolios each contain 60% of identified bull trout habitat in the WHSA. This
resulted in the inclusion of healthy watersheds as bull trout are highly sensitive to riparian
disturbance and pollution (Hagen & Decker, 2011). Streams identified as suitable bull trout
habitat contain cold water, clean gravels, natural flows, and stable banks with large amounts
of cover. Conserving streams with these characteristics not only favours the persistence of
bull trout, but also accommodates the needs of other valued aquatic species in the WHSA
such as Dolly Varden and brook trout (Hagen & Decker, 2011). Similarly, the inclusion of
intact watersheds can benefit those non-aquatic species that make use of riparian and
floodplain habitats in the WHSA (e.g., fisher).
MARXAN-ILP included 60% of the special features layer in each portfolio. This
captured karst ecosystems containing cavities that are utilized by multitude of wildlife for
nesting and hibernation. Also represented in each portfolio are wetland ecosystems that
accommodate the needs of migratory waterfowl and perform important ecosystem services
83

such as carbon sequestration and water filtration. Lastly, achieving conservation targets for
the special features layer resulted in the inclusion of mineral licks within each portfolio. The
supplementary nutrition provided by these areas can contribute to the persistence of elk,
moose, mountain goats, and Stone’s sheep, and Rocky Mountain bighorn sheep (Rea et al,
2004).
6.2 What Areas Retain Conservation Value Despite the Presence of Resource Development?
For the purposes of this study, a portfolio containing low levels of resource
development was thought to display principles of good PA design. However, to suggest
undeveloped landscapes were the only option for conservation would create the impression
that all other areas of the WHSA could be written off as sacrifice zones. In SCP, sacrifice
zones are those areas that, due to their modified state, are surrendered to accommodate
natural resource development (Pressey, 1999). In reality, there exist multiple areas in the
WHSA that retain high conservation value despite the presence of industrial activity. With
this in mind, I intentionally designed scenarios that would visualize conservation targets
being met across different landscape types in the WHSA — ranging from untouched
wilderness areas, to those situated within the integrated resource-use matrix, to areas heavily
modified by industrial and urban development.
Figure 18 shows high-value planning units in permanently developed landscapes that
were selected in scenario D. Within the context of future conservation planning in the
WHSA, planners could target these areas to assess whether the identified conservation values
are at risk of being lessened or lost should resource development continue. If a planning unit
is nearing a threshold where value may be lost, it may be a suitable candidate for restoration

84

efforts. If the values present are relatively stable, stewardship agreements with stakeholders
could be put in place so that conservation value can continue to persist alongside industry.
The low-sensitivity cost surface used in scenario C helped identify pockets of high
conservation value within landscapes recently affected by forest harvesting (Fig. 18). Given
their high conservation value, and temporarily disturbed state, these areas could be
considered suitable candidates for removal from the integrated resource matrix. Protection of
these areas would allow succession to occur naturally without active restoration. This may
result in the eventual recruitment of conservation features that are currently absent because of
habitat degradation from logging activity (e.g., bull trout). In addition to allowing for
recovery, this would prevent the authorization of conflicting land-use that may further
degrade the already disturbed landscape and result in the present value being lost.
Portfolio B identified areas where human activity, industrial light, and noise pollution
is likely present but the landscape remains virtually void of fragmentation (Fig. 18). In their
current state, these landscapes can accommodate those conservation features (e.g., land
facets, forest pattern and process layers) whose ecology is such that they remain largely
unaffected by stressors present in the ZOI. Given that these stressors are often short-lived,
such as the noise created by a passing vehicle or seasonal construction, high-value lands in
the ZOI may represent suitable habitat throughout the majority of the year for species that
exhibit a high sensitivity to industrial activity (e.g., woodland caribou) (Wilson, 2016).
Because of their intactness, these areas should be treated as buffers for high-value lands
outside the ZOI.
The high-sensitivity cost surface applied to scenario A identified high-value areas
containing little-to-no resource development. The results of this scenario are best suited to
85

anchor a finalized conservation portfolio. The attributes of this portfolio and the associated
management implications are discussed in the following section.
6.3 What is an optimal portfolio from which to direct conservation planning efforts?
High-value planning units in fully-converted or recently-logged landscapes contain
attributes that limit their utility in planning initiatives aimed at immediate protection
measures. These areas have sharp edges between ecotones, increased occurrences of invasive
species, and are often comprised of young managed stands that are low in biodiversity
(Bannerman & Province of British Columbia, 1998; Delong & Province of British Columbia,
2011). Identifying high-value sections of these landscapes helped answer the second research
question and prevent condoning sacrifice zones. However, I wanted to avoid only suggesting
long-term options for conservation that were predicated on restoration effort, special
management, or surrender of existing tenures. Conservation initiatives often lack the
resources needed to perform such interventions.
Part of the rationale for enforcing principles of good protected area PA design on
MARXAN-ILP was to identify lands that achieved conservation objectives in their current
state and provided space on the landscape for other uses. Aside from some areas that fell
within the ZOI, the majority of lands that make up Portfolio A exist in complete isolation
from resource development and are free of any land-use conflicts. These areas would require
the least amount of intervention prior to PA creation and were thus considered to be best
suited for immediate application in planning processes.
In addition to administrative utility, Portfolio A displayed desirable patch
characteristics that were consistent with principles of good PA design. Of all the scenarios,
Portfolio A had the highest percentage of core area and the lowest mean shape index (Table
86

8). This was likely due to the fact that the portfolio was comprised of few, large patches that
were the least-convoluted of any produced in this study (Table 8). The high percentage of
interior forested habitat makes Portfolio A the most compatible with conservation features
that are sensitive to fragmentation (e.g., woodland caribou, bull trout, grizzly bear). A low
mean shape index also minimizes the amount of transitionary area between patches where
harmful edge-effects can occur. Examples of edge-effects include barriers to species
dispersal, altered temperature regimes within PA boundaries, and the creation of entry points
for invasive species (Bannerman & Province of British Columbia, 1998).
The large area requirement of the conservation targets, combined with the reduction
in penalty-free area induced by the high-sensitivity cost surface, caused a natural corridor of
prioritized lands to form in Portfolio A. This corridor can be seen across all portfolios but is
most intact and pronounced in Portfolio A (Fig. 17). The corridor spans the entire length of
the WHSA on a southeast-northwest line and shows a network of high-value lands prioritized
amongst existing PAs (Fig. 4 and 17). Achieving connectivity amongst PAs in the WHSA
has important implications for those species migrating along latitudinal gradients in response
to shifting climate regimes and could prevent the isolation of populations within the planning
region.
In summary, Portfolio A not only achieves the goal of this study by meeting targets
for all features, but does so with a corridor of contiguous landscapes that contain little-to-no
conflicting land-use. As outlined above, these characteristics are thought to complement the
biology of select conservation features and reduce the amount of resources required to
establish PAs. It is because of these attributes that the lands identified in Portfolio A are

87

considered to be ‘priority lands’ for protection and have been placed at the center of a
finalized portfolio to direct conservation efforts in the WHSA (Fig. 19).
If all priority lands were to be conserved, and consolidated with the 17%
representation achieved by existing PAs, the result would be a network of protected lands
covering 63% of the WHSA. This scale of representation is consistent with emergent
concepts that suggest protecting over half of terrestrial environs is necessary to address the
current inadequacy of Canada’s PA system and prevent the collapse of global biodiversity
(Locke, 2014). Though it exceeds political targets, it could be argued that this scale of PA
representation is what is realistically required to ensure the persistence of conservation
features and maintenance of landscape function along the North American Cordillera.
Figure 19 shows a diffuse buffer on the outside of priority lands labelled ‘integrated
lands’. Integrated lands are an amalgamation of those areas that were selected in scenarios B
through D (Fig. 17). Integrated lands should not be written off as sacrifice zones because of
their compromised state, but rather be considered as high-value areas that soften the
transition zone between heavily developed landscapes and priority lands. Integrated lands
also increase connectivity amongst priority lands and existing PAs in the WHSA (Fig. 19).

88

Figure 19. Finalized portfolio for conservation planning in the Wild Harts Study Area.

89

6.4 Next Steps
The next steps in the SCP process would be to identify suitable PA boundaries in the
WHSA using the finalized conservation portfolio produced in this research. The portfolio
was designed to serve as a coarse-scale reference of priority lands from which finer-scale
analyses could be directed. The MARXAN-ILP model used to produce the portfolio is a live
interface that can be scaled-down and applied to specific areas of interest. Decision makers
could use the final portfolio to locate areas best suited for immediate conservation action and
conduct a downscaled MARXAN-ILP analysis within those areas to delineate PA
boundaries. A downscaled analysis would not need to adopt the MARXAN-ILP inputs
applied in this model, but rather could apply a different set of user-defined constraints that
are reflective of the goals for that revised planning region.
Integrated lands require further analysis to gauge whether the current conservation
value present is at risk of succumbing to resource development. The zonation achieved by the
various cost surfaces displayed in Figure 18 can assist in targeting areas most at risk. The
results of such an assessment would inform whether special management is required to
sustain conservation value in these areas. Areas most at risk might be best served by a strong
form of protection; whereas, those low-risk areas could be managed in such a way that
accommodates other forms of land-use (i.e., recreation, low impact resource extraction, etc.).
Equally important is the step of completing a fine-scale connectivity analysis to
complement the happenstance connectivity achieved by the WHSA conservation portfolio.
Further investigation is required to determine whether the connectivity achieved within the
priority lands is reflective of that which actually occurs on the land base. Of additional

90

interest is whether the resource development present within the integrated lands degrades the
ability of these areas to facilitate movement of species.

91

7.0 LIMITATIONS
There were several limitations that emerged during the course of this research. These
included limitations associated with applying systematic conservation planning to the WHSA
and limitations built into the MARXAN-ILP model used in this study.
7.1 Limitations of Conducting Systematic Conservation Planning in the Wild Harts Study
Area
Many of the limitations involved in the SCP process existed in the form of the
assumptions required to complete the process itself. Principle amongst these was the
selection of surrogates. I employed surrogates assuming that one component of biodiversity
was representative of another — an approach that could be criticized as being an
oversimplification of ecological complexity present in the WHSA. I attempted to address that
assumption through the use of multiple surrogates at varying spatial scales. However, it is
still possible that the conservation features did not constitute adequate representation of those
biological elements being selected for through surrogacy.
Other limitations arose in the form of data availability, or lack thereof, when building
the conservation feature layers to drive MARXAN-ILP. The WHSA is of such a large scale
that consistent and reliable spatial data for the entire area were difficult to obtain or generate.
Oftentimes those datasets that were available had a commercial element to them or were used
by licensees and the provincial government to measure the impact of resource use. For
example, the layers used to build the forest pattern and process layers have a primary purpose
of analysing timber supply. Similarly, much of the bull trout layer was constructed using
datasets that were built to measure the impacts of harvesting on riparian areas and watershed
quality. There were very few datasets available that were produced solely for conservation or
92

ecological research sake. In this way, the selection of conservation features was somewhat
biased to those elements valued by the natural resource sector and less a reflection of the
most efficient surrogates for abiotic/biotic diversity.
An additional limitation of the SCP process involved translating the goals for
conservation into quantifiable targets. The goal of this study was to prioritize lands that, if
protected, would complement existing PAs in promoting biodiversity, maintaining natural
disturbance regimes, and building climate change resilience within the WHSA. The criteria
that make up this goal are based on large-scale biophysical patterns and processes.
Attempting to quantify how much area was needed to account for climatological variables,
stochastic disturbance events, geomorphological processes, and interactions between
biological communities proved to be a challenging task. There remains the potential that,
even if all lands prioritized by conservation targets were protected, the resultant PA network
would still fall short of achieving all components of the study goal.
Another limitation was the extent to which the large woodland caribou targets masked
the full capabilities of MARXAN-ILP. Meaning, the spatial footprint of 90% representation
for each of the woodland caribou herds resulted in many of the other targets being met within
that footprint through complementarity. In addition, the woodland caribou conservation
feature layer had a comparatively low edge-to-area ratio and a tendency to occur outside of
resource development — both characteristics that lessened the influence of cost surfaces.
This made for relatively uniform results across all scenarios when compared to some of the
trial runs conducted in the early stages of this research, at lower representation targets, with
the same user-defined constraints (Appendix 3). This result was not ideal given the resources
expended on scenario development, but was welcomed because it highlighted the
93

effectiveness of woodland caribou as a surrogate for other conservation features in the
WHSA.
There is potential that I have been overly generous in describing the distribution of
select conservation features across the WHSA. Given the scale of this analysis, I drew upon
methodologies that could be applied across the entire planning region when constructing
conservation feature layers. This resulted in some layers being built using rudimentary
selection criteria for what was considered suitable habitat (e.g., bull trout and grizzly bear).
These layers might have resulted in an over-estimation of the amount of area needed to
ensure the persistence of select conservation features. The inclusion of surplus area could
have implications if these portfolios are used to delineate administrative boundaries for PAs.
These implications would come in the form of unnecessary opportunity and acquisition cost
that occurs when area is used inefficiently to achieve conservation.
7.2 Limitations of the MARXAN-ILP Model
One of the major limitations involved in the construction of the MARXAN-ILP
model was the WHSA boundary itself, specifically the northern and southern extents. In
hindsight, Northern Rocky Mountains and Kakwa Provincial Parks should have been
excluded from the boundary as they are not representative of the WHSA. Northern Rocky
Mountains Provincial Park is part of the Muskwa-Kechika Management Area and Kakwa
Provincial Park is more closely associated with the Central Rocky Mountains PA complex
than with the WHSA. The rationale for their initial inclusion was to assess the extent to
which selected planning units interacted with borders of these PAs and whether any
connectivity was achieved naturally. However, including these large PAs created an
inaccurate representation of how much area in the WHSA is actually protected. In reality, the
94

core sections of the WHSA are poorly represented with only 4.8% falling within PAs once
Northern Rocky Mountains Provincial Park and Kakwa Provincial Park are removed.
The inclusion of these large PAs also raised issues when completing step 3 in the SCP
process (review existing PAs in the WHSA). These PAs captured large portions of many
conservation feature’s total area, thus reducing the amount of outstanding area needed to
achieve the various conservation targets for each feature respectively. Essentially, those large
parks created the impression that some features were well-represented by PAs when, in fact,
the actual representation of these features was poor within the core portions of the WHSA.
Accordingly, MARXAN-ILP would not have selected additional area to accommodate those
features thus leaving them vulnerable to resource development.
Another limitation that arose when running scenarios in MARXAN-ILP was the
similarity in portfolios produced by scenarios using medium and low-sensitivity cost
surfaces. This was likely due to the large planning unit size of 1 km2. The planning units
were too coarse a scale for MARXAN-ILP to make the distinction between temporary and
permanent development as these forms of land use were commonly in close proximity to one
another. Meaning, the same planning units that were avoided due to cutblocks in the
medium-sensitivity scenarios would also have been avoided in the low-sensitivity scenarios
because of the road used for access. Smaller planning unit size would result in differentiation
between cutblocks and permanent landscape conversion and produce solution spaces whose
perimeters exhibit greater sensitivity to these different levels of land use.

95

8.0 RECOMMENDATIONS
This research helped to identify coarse-scale priority areas for conservation action in
the WHSA. The finalized conservation portfolio can be used to direct the establishment of
PA boundaries. It can also be used for the assessment of risk to conservation value, and the
completion of a fine scale connectivity analysis.
As noted above, the MARXAN-ILP model is a live interface that can be manipulated
to achieve alternative solution spaces. With this in mind, a way to improve the current model
would be to include a conservation feature representing a mid-trophic level carnivore such as
a furbearing mammal (e.g., fisher, American marten). Small, carnivorous mammals are
commonly included in conservation planning process as they are often heavily harvested and
are representative of late-seral stage forests (Apps, 2013).
The working model would also benefit from research conducted in partnership with
First Nations whose traditional territories overlap the WHSA. The incorporation of
traditional ecological knowledge would complement the empirically-based data used in this
study and would be invaluable for addressing some of the gaps present in those datasets.
Aside from the ecological contribution, First Nations involvement would add a cultural
element to the conservation efforts taking place in the WHSA. Additional conservation
features could be developed that encapsulate areas being used by First Nations within
contemporary contexts for exercising their constitutional rights of hunting, fishing, and
gathering. Such research would ensure that inclusivity is built into any portfolio meant to
inform decision-making processes regarding PA creation in the WHSA.

96

Another avenue for future research would be to explore the extent to which shifting
climate regimes will interact with lands prioritized for conservation under current state
conditions. Many of the values targeted in this study have the potential to display latitudinal
migration in response to changes in temperature and precipitation (e.g., forest pattern and
process features). Research aimed at predicting these movements, and where climate velocity
might be most pronounced in the WHSA, would lend insights into which lands should be
proactively selected to accommodate novel assemblages of species — regardless of whether
those areas contain high conservation value under current state conditions (Brito-Morales et
al., 2018). Similarly, this research would also help to identify areas unlikely to retain their
conservation value.

97

9.0 CONCLUSION
At the local scale, the WHSA supports a wide range of ecological diversity and
represents some of the last remaining intact landscapes in northeastern British Columbia. At
the continental scale, the WHSA serves as a vital corridor connecting networks of PAs in the
south to large PAs in the north. To lose landscape function in the WHSA would effectively
sever the band of contiguous habitats spanning the length of the North American Cordillera
and create two isolated PA complexes. This has implications for those biological
communities that migrate along latitudinal gradients and could result in the isolation of
populations. The WHSA is currently underrepresented by existing PAs and competing
resource interests from the surrounding Peace River Region threaten to encroach upon the
ecological values present there. This research was aimed at producing a suite of possible
options for conservation in the WHSA that could be used to inform decision-making
processes concerning future PA creation in the area.
The SCP approach was effective for conducting a prioritization exercise at the scale
of the WHSA. The MARXAN-ILP software allowed me to identify areas of high
conservation value and explore the extent to which varying intensities of resource use
influenced prioritization. The software performed large computations at high speed and
consistently achieved the optimal solution at the lowest cost. The results of this research
represent a wide range of options for conservation in the WHSA and present additional
research questions.
The final conservation portfolio shows priority lands (i.e., Portfolio A) that are best
suited for immediate protection measures. These areas met targets for all conservation
features and best displayed principles of good PA design. Complementing priority areas are
98

integrated lands (i.e., Portfolio B, C, and D) that retain high conservation value despite
containing varying levels of resource development. The integrated lands serve as a
transitionary buffer between heavily-developed landscapes and the values present within
priority lands. These areas offer an opportunity to prescribe restorative efforts or special
management strategies to achieve conservation alongside other forms of land-use in the
WHSA.
The results of this research are not meant to serve as legal boundaries for future PAs.
More research is required to further refine the selected areas and explore solutions that take
into account First Nations interests or different climate scenarios. What this research does
provide is a decision-support portfolio that is repeatable, transparent, and scientificallydefensible. Land planners can be sure that the areas prioritized in this research represent the
foundational and mechanistic elements that comprise the ecology of the WHSA on the
whole. The final portfolio is meant to inform the discussion surrounding protected area
creation in the WHSA and contribute to the preservation of one of British Columbia’s last
true wildernesses.

99

10.0 REFERENCES

Airamé, S., Dugan, J.E., Lafferty, K.D., Leslie, H., McArdle, D.A., and Warner, R.R. (2003).
Applying ecological criteria to marine reserve design: a case study from the
California Channel Islands. Ecological Applications, 13: 170–184.
Apps, C. (2013). Assessing cumulative impacts to wide-ranging species across the Peace
Break region of northeastern British Columbia. Version 3.0 Yellowstone to Yukon
Conservation Initiative, Canmore, AB: Yellowstone to Yukon Conservation Initiative.
Ball, I.R., Possingham, H.P., and Watts, M.E. [computer software] (2009). Marxan and
relatives: software for spatial conservation prioritization. Oxford, UK: Oxford
University Press.
Banks, S.A., Skilleter, G.A., and Possingham, H.P. (2005). Intertidal habitat conservation:
identifying conservation targets in the absence of detailed biological information.
Aquatic Conservation: Marine and Freshwater Ecosystems, 15: 271–288.
Bannerman, S., and Province of British Columbia. (1998). Biodiversity and interior habitats:
the need to minimize edge-effects. Victoria, BC: Ministry of Forests, Research
Program.
BEACONS. (2006). Creating a scientific foundation for conservation planning in boreal
Canada. Edmonton, AB: University of Alberta. Retrieved November 19, 2016, from:
https://sites.ualberta.ca/~fschmieg/Beacons/PDFs/BEACONs.pdf
Beier, P., and Brost, B. (2010). Use of land facets to plan for climate change: conserving the
arenas not the actors. Conservation Biology, 24: 701–710.
Berkes, F. (1993). Traditional ecological knowledge in perspective. Ottawa, ON:
International Program on Traditional Ecological Knowledge and International
Development Research Center.
Beyer, H.L., Dujardin, Y., Watts, M.E., and Possingham, H.P. (2016). Solving conservation
planning problems with integer linear programming. Ecological Modelling, 328: 14–
22.
Bonn, A., Rodrigues, A.S.L., and Gaston, K.J. (2002). Threatened and endemic species: are
they good indicators of patterns of biodiversity on a national scale? Ecology
Letters, 5: 733–741.
Brito-Morales, I., Garc�, M. J., Schoeman, D. S., Burrows, M. T., Poloczanska, E. S.,
Brown, C. J., Ferrier, S., and Richardson, A. J. (2018). Climate Velocity Can Inform
Conservation in a Warming World. Trends in Ecology & Evolution, 33: 441–457.
Canadian Parks and Wilderness Society. (2016). Protecting Canada: is it in our nature?:
how Canada can achieve its international commitment to protect our land and
freshwater. Ottawa, ON: Canadian Parks and Wilderness Society.

100

Carroll, C., Noss, R.F., Paquet, P.C., and World Wildlife Fund (Canada). (2001). Carnivores
as focal species for conservation planning in the Rocky Mountain region. Toronto,
ON: World Wildlife Fund Canada.
Carroll, C., Roberts, D.R., and Michalak, et al., (2017). Scale-dependent complementarity of
climatic velocity and environmental diversity for identifying priority areas for
conservation under climate change. Retrieved September 15, 2017, from:
https://adaptwest.databasin.org/pages/environmental-diversity-north-america
Chadwick, D.H., Gehman, R., and National Geographic Society (U.S.). (2000). Yellowstone
to Yukon. Washington, D.C.: National Geographic Society.
Ciarniello, L.M. (2006). Demography and habitat selection by grizzly bears (Ursus arctos L.)
in central British Columbia. Edmonton. AB: University of Alberta.
Crutzen, P., and Stoermer, E. (2000). The 'Anthropocene'. IGBP Global Change Newsletter
41: 17–18.
Dearden, P., and Rollins, R. (1993). Parks and protected areas in Canada: Planning and
management. Toronto, ON: Oxford University Press.
DeLong, S.C., and Province British Columbia. (2011). Land units and benchmarks for
developing natural-disturbance based forest management guidance for Northeastern
British Columbia. Victoria, B.C.: Ministry of Forests and Range, Forest Science
Program.
Demarchi, D. (2011). The British Columbia ecoregion classification. Victoria, B.C.:
Ecosystem Information Section Ministry of Environment.
Dirzo, R., Young, H.S., Galetti, M., Ceballos, G., Isaac, N.J., and Collen, B. (2014).
Defaunation in the Anthropocene. Science, 345: 401–406.
Doyle, J., and Province of British Columbia. (2010). Conservation of ecological integrity in
B.C. parks and protected areas. Victoria, B.C: Office of the Auditor General of
British Columbia.
Dudley, N. (2008). Guidelines for Applying Protected Area Management Categories. Gland,
CH: IUCN.
Dyer S.J., O’Neill, J.P., Wasel, S.M., and Boutin, S. (2001). Avoidance of industrial
development by woodland caribou. Journal of Wildlife Management, 65: 531–542.
Eagles, P.F.J., McCool, S.F., and Haynes, C.D. (2002). Sustainable tourism in protected
areas: guidelines for planning and management. IUCN Gland, SH and Cambridge,
UK: IUCN.
Environment and Climate Change Canada. (2016). Canadian environmental sustainability
indicators: Canada's protected areas. Retrieved February 13, 2017, from:
www.ec.gc.ca/indicateurs-indicators/default.asp?lang=en&n=478A1D3D-1.
Environmental Systems Research Institute [computer software] (2011). ArcGIS Desktop:
Release 10.3 Redlands, CA: Environmental Systems Research Institute.
101

Evans, S. (2003). Selecting candidate marine reserves: A test of the efficacy of current
approaches. Master’s Thesis, University of Guelph, Guelph, ON.
Geselbracht, L., and Torres, R. (2005). Florida marine assessment: prioritization of
marine/estuarine sites and problems adversely affecting marine/estuarine habitats
and associated species of greatest conservation need (interim report). Gainesville,
FL: The Nature Conservancy. Retrieved November 17, 2016, from:
http://myfwc.com/wildlifelegacy/PDF/marineestuarinethreats.pdf
Government of Canada. (1972). National parks system planning manual. Ottawa, ON:
National & Historic Parks Branch.
Groves, C.R., Jensen, D.B., Valutis, L.L., Redford, K.H., Shaffer, M.L., Scott, J.M., and
Anderson, M. G. (2002). planning for biodiversity conservation: putting conservation
science into practice a seven-step framework for developing regional plans to
conserve biological diversity, based upon principles of conservation biology and
ecology, is being used extensively by the nature conservancy to identify priority areas
for conservation. BioScience, 52: 499–512.
Grantham, H.S., Pressey, R.L., Wells, J.A., and Beattie, A.J. (2010). Effectiveness of
biodiversity surrogates for conservation planning: different measures of effectiveness
generate a kaleidoscope of variation. Plos ONE, 5: e11430.
doi:10.1371/journal.pone.0011430.
Gustine, D.D., and Parker, K.L. (2008). Variation in the seasonal selection of resources by
woodland caribou in northern British Columbia. Canadian Journal of Zoology, 86:
812–825.
Hagen, J., and Decker, A.S. (2011). The status of bull trout in British Columbia: a synthesis
of available distribution, abundance, tend, and threat information. Victoria, B.C.:
Province of British Columbia
Heinemeyer, K., Tingey, R., Pollock, J., Ciruna, K., Lind, T., Butterfield, B., Rumsey, C.,
Griggs, J., Iachetti, P., Bode, C., Olenicki, T., Parkinson, E., and Sizemore, D. (2004).
Conservation area design for the Muskwa-Kechika Management Area. Prepared for
the BC Ministry of Sustainable Resource Management.
Holdgate, M. (1999). The green web – a union for world conservation. Environmental Policy
and Planning, 2: 261–262.
Horn, H. (2011). Strategic conservation planning for terrestrial animal species in the Central
Interior of British Columbia. BC Journal of Environment and Management, 12: 36–
53.
Hummel, M. (1989). Endangered spaces: The future for Canada’s wilderness. Toronto, ON:
Key Porter Books.
Jones, E.S. (2008). Seasonal habitat use and selection by woodland caribou herds in the
South Peace region, central British Columbia. Victoria, B.C.: Ministry of
Environment.
Johnson, C.J., Seip, D.R., and Boyce, M.S. (2004). A quantitative approach to conservation
planning: using resource selection functions to map the distribution of mountain
caribou at multiple spatial scales. Journal of Applied Ecology, 41: 238–251.
102

Kavanagh, K., Rowe, J.S., Iacobelli, T., and World Wildlife Fund (Canada). (1995). A
protected areas gap analysis methodology: planning for the conservation of
biodiversity. Toronto, ON: World Wildlife Fund Canada.
Lambeck, R.J. (1997). Focal species: a multi-species umbrella for nature conservation.
Conservation Biology, 11: 849–856.
Leslie, H., Ruckelshaus, M., Ball, I.R., Andelman, S., and Possingham, H.P. (2003). Using
siting algorithms in the design of marine reserve networks. Ecological Applications,
13: 185–198.
Lindenmayer, D.B., Meffe G.K., Manning, A.D., Smith, P.L., Possingham, H.P., Fischer, J.,
Oliver, I., and McCarthy, M.A. (2002). The focal-species approach and landscape
restoration: a critique. Landscape and Urban Planning, 62: 149–158.
Locke, H. (2014). Nature needs half: a necessary and hopeful new agenda for protected areas
in North America and around the World. George Wright Forum, 31: 359–371.
Loos, S.A., and Canessa, R.R. (2006). Exploration of MARXAN for utility in marine
protected area zoning. Masters Thesis, University of Victoria, Victoria, B.C.
Lothian, W. F., and Parks Canada. (1976). A history of Canada's national parks. Ottawa,
ON: Parks Canada.
MacKinnon D., Lemieux, C.J., Beazley, K., Woodley, S., Helie, R., Perron, J., Elliot, J.,
Haas, C., Langlois, J., and Lazaruk, H. (2016). Canada and Aichi Biodiversity Target
11: understanding ‘other effective area-based conservation measures’ in the context
of the broader target. Biodiversity Conservation, 24: 3559–3581.
Margules, C.R., and Pressey, R.L. (2000). Systematic conservation planning. Nature, 405:
243–253.
Margules, C.R., Nicholls A.O., and Pressey, R.L. (1988). Selecting networks of reserves to
maximize biological diversity. Biological Conservation, 43: 63–76.
Michalak, J., Carroll, C., Lawler, J.J., Neilson, S., Roberts, D.R., and Hamann, A. (2015)
Land facet data for North America at 100m resolution. Retrieved September 15,
2017, from: https://adaptwest.databasin.org/pages/adaptwest-landfacets
Ministry of Environment (2013). Implementation plan for the ongoing management of South
Peace Northern Caribou (Rangifer tarandus caribou pop. 15) in British Columbia.
Victoria, B.C.: Province of British Columbia.
Ministry of Sustainable Resource Management (2005). Background information and
supporting documentation for the process involved in developing the recommended
biodiversity objectives in the PG TSA. Prince George, B.C.: Province of British
Columbia.
Molles, M.C. (1999). Ecology: concepts and applications. Boston, MA: WCB/McGraw-Hill.
Morecroft, M. D., Crick, H. Q. P., Duffield, S. J., and Macgregor, N. A. (2012). Resilience to
climate change: translating principles into practice. Journal of Applied Ecology, 49:
547–551.
103

Morrell, N., Schuster, R., and Arcese, P. (2015). Marxan tutorial for the Coastal Douglas-Fir
and Conservation Partnership Study Area. Vancouver, B.C.: University of British
Columbia.
Noss, R.F., and Cooperrider (1994). Saving nature’s legacy: protecting restoring
biodiversity. Washington, DC: Island Press.
Noss, R.F., World Wildlife Fund (Canada), and World Wildlife Fund (U.S.).
(1995). Maintaining ecological integrity in representative reserve networks. Toronto:
World Wildlife Fund, Canada.
Noss, R.F. (1996). Protected areas: how much is enough? Cambridge, MA: Blackwell
Science.
Oakleaf, J.R., Kennedy, C.M., Baruch-Mordo, S., West, P.C., Gerber, J.S., Jarvis, L.,
Kiesecker, J., and Kumar, L. (2015). A world at risk: aggregating development trends
to forecast global habitat conversion. Plos One, 10: e0138334.
doi:10.1371/journal.pone.0138334.
Parks Canada (2016). State of Canada's natural and cultural heritage places. Ottawa, ON:
Parks Canada.
Poiani, K.A, Richter, B.D., Anderson, M.G., and Richter, H.E. (2000). Biodiversity
conservation at multiple scales: functional sites, landscapes, and networks.
Bioscience, 50: 133–146.
Polfus, J.L., Hebblewhite, M., and Heinemeyer, K. (2011). Identifying indirect habitat loss
and avoidance of human infrastructure by northern mountain woodland caribou.
Biological Conservation, 144: 2637–2646.
Pressey, R.L. (1999). Applications of irreplaceability analysis to planning and management
problems. Parks, 9: 42–51.
Primack, R.B. (2010). Essentials of conservation biology. Sunderland, Mass: Sinauer
Associates Inc.
Province of British Columbia (2015a). Tree Farm Licence 48: rationale for allowable annual
cut determination. Prince George, B.C.: Ministry of Forest, Lands, Natural Resource
Operations and Rural Development.
Province of British Columbia (2015b). Cumulative effects framework grizzly bear values
summary. Victoria, B.C.: Provincial Grizzly Bear Technical Working Group.
Province of British Columbia, and Taku River Tlinget First Nation. (2011). Wooshtin
Wudidaa Atlin Taku land use plan. Smithers, B.C: Province of British Columbia.
Province of British Columbia (1995). Biodiversity guidebook. Victoria, B.C.: Ministry of
Forests, Lands, Natural Resource Operations and Rural Development.
Province of British Columbia. (1993). A protected areas strategy for British Columbia.
Victoria, BC: Province of British Columbia.
104

Rea, R.V., Hodder, D.P., and Child, K.N. (2004). Considerations for natural mineral licks
used by moose in land use planning and development. Alces, 40: 161–167.
Rumsey, C., Ardron, J., Ciruna, K., Curtis, T., Doyle, F., Ferdana, Z., Hamilton, T.,
Heinemeyer, K., Jachette, P., Jeo, R., Kaiser, G., Narver, D., Noss, R., Sizemore, D.,
Tautz, A., Tingey, R., and Vance-Borland, K. (2003). An ecosystem spatial analysis
for Haida Gwaii, Central Coast and North Coast, British Columbia. Victoria, B.C.:
Coast Information Team. Retrieved September 15, 2016, from:
http://www.citbc.org/c-esa-fin-04may04.pdf
Saetersdal, M., Line, J.M., and Birks, H.J.B. (1993). How to maximize biological diversity in
nature reserve selection: vascular plants and breeding birds in deciduous woodlands.
Western Norway Biological Conservation, 66: 131–138.
Schmiegelow, F.K.A., Cumming, S.G., Harrison, S., Leroux, S., Lisgo, K., Noss, R., and
Olsen, B. (2006) Conservation beyond crisis management: a conservation-matrix
model. A Discussion Paper for the Canadian BEACONs Project. Retrieved December
04, 2017, from: http://natureneedshalf.org/wp-content/uploads/2012/02/BorealBeacons_at-least-half-matrix_conservation.pdf
Schneider, R.R., Hauer, G., Farr, D., Adamowicz, W.L., and Boutin, S. (2011). Achieving
conservation when opportunity costs are high: optimizing reserve design in Alberta’s
oil sands region. PLoS ONE, 6: e23254.
Schuster, R., Crombie, M., Morrell, N., and Arcese P.A. (2016). Prioritization tool for the
conservation of coastal douglas-fir forest and savannah habitats of the Georgia Basin
and Puget Sound Lowlands. Vancouver, B.C.: University of British Columbia.
Secretariat of the Convention on Biological Diversity, and United Nations Environment
Programme (2001). The convention on biological diversity. Geneva, CH: United
Nations Environment Programme.
Simberloff, D. (1998). Flagships, umbrellas, and keystones: is single-species management
passé in the landscape era? Biological Conservation, 83: 247–257.
Soulé, M.E., and Sanjayan, M.A. (1998). Conservation targets: do they help? Science, 279:
2060–2061.
Stevens, V., and Darling, L. (2004). Conservation of species and ecosystems at risk: BC
parks and protected areas challenges. Paper presented at Species at Risk 2004
Pathways to Recovery Conference, Victoria, BC.
Stewart, R.R., Noyce, T., and Possingham, H.P. (2003). Opportunity cost of ad hoc marine
reserve design decisions: an example from South Australia. Marine Ecology Progress
Series, 253: 25–38.
Svancara, L.K, Brannon, R., Scott, J.M., Groves, C.R., Noss, R.F., and Pressey, R.L. (2005).
Policy-driven versus evidence-based conservation: a review of political targets and
biological needs. BioScience, 55: 989–995.
105

Tingley, M.W., Darling, E.S., and Wilcove, D.S. (2014). Fine and coarse-filter conservation
strategies in a time of climate change. The Year in Ecology and Conservation
Biology, 1322: 92–109.
Venter, O., Fuller, R.A., Segan, D.B., Carwardine, J., Brooks, T., and Butchart, S.H.M.
(2014) Targeting global protected area expansion for imperiled biodiversity. PLoS
Biol 12: e1001891. doi:10.1371/journal.pbio.1001891
Warman, L.D. (2001). Identifying priority conservation areas using systematic reserve
selection and GIS at a fine spatial scale: a test case using threatened vertebrate
species in the Okanagan, British Columbia. Masters Thesis, University of British
Columbia, Vancouver, B.C.
Waters, C.N., Zalasiewicz, J., Summerhayes, C., Barnosky, A.D., Poirier, C., Galuszka, A.,
Cearreta, A., Edgeworth, M., Ellis, E.C., Ellis, M., Jeandel, C., Leinfelder, R.,
McNeill, J.R., Richter, D., Steffen, W., Syvitski, J., Vidas, D., Wagreich, M.,
Williams, M., Zhisheng, A., Grinevald, J., Odada, E., Oreskes, N., and Wolfe, A.P.
(2016). The Anthropocene is functionally and stratigraphically distinct from the
Holocence. Science, 351: http://dx.doi.org/10.1126/ science.aad2622
Werring, J. (2015). Investigation by the David Suzuki Foundation into issues of potential
environmental concern related to oil and gas development in the Montney Shale Play
in northeastern British Columbia, August 14 – 28, 2015. Vancouver, B.C.: David
Suzuki Foundation.
Wiersma, Y.F., and Sleep, D.J.H. (2018). The effect of target setting on conservation in
Canada’s boreal: what is the right amount of area to protect? Biodiversity
Conservation, 27: 733–748.
Wiersma, Y.F., and Nudds, T.D. (2009). Efficiency and effectiveness in representative
reserve design in Canada: The contribution of existing protected areas. Biological
Conservation, 142: 1639–1646.
Wiersma, Y.F., and Canadian Council on Ecological Areas. (2006). Protected areas in
northern Canada: Designing for ecological integrity, phase 1 report. Gatineau, QC:
Canadian Council on Ecological Areas.
Wilson, S.F. (2016). Managing zone-of-influence effects of oil and gas activities on
terrestrial wildlife and habitats in British Columbia. Journal of Ecosystems and
Management, 16: 1–14.
World Wide Fund for Nature. (2016). Living planet report 2016 summary: risk and resilience
in a new era. Retrieved January 08, 2017, from:
http://www.footprintnetwork.org/content/documents/2016_Living_Planet_Report_Su
mmary_Lo.pdf
Wright, P. (2016). Protection and Persistence in the Canadian Protected Areas System: A
Review of Conservation Science Research and Approaches. Paper presented at 2016
Canadian Parks Summit: Canmore, AB. Retrieved January 06, 2017, from:

106

http://lin.ca/sites/default/files/attachments/conservation-science-Final-commissioned2016.pdf
Yellowstone to Yukon Conservation Initiative (2012). Muskwa-Kechika Management Area
biodiversity conservation and climate change assessment. Canmore, AB:
Yellowstone to Yukon Conservation Initiative

107

APPENDIX 1: Spatial information used for MARXAN-ILP model
LAYER
Land Facet
Diversity and
Rarity
Forest Pattern and
Process

Woodland Caribou

Grizzly Bear

Bull Trout

Special Features

LAYER NAME
Land Facets Incorporating
Latitude-adjusted Elevation

SOURCE
Adaptwest

Vegetation Resource Inventory
Biogeoclimatic Zones
Natural Disturbance Zones

Data BC
Data BC
MFLNRORD - Geospatial Services
Team Omineca Region
MFLNRORD - Ecosystems Branch
Prince George
parker@unbc.ca

Core Habitat Mapping South
Peace
Resource Selection Function
Northern Mountain
Resource Selection Function
Southern Mountain
Ungulate Winter Range Proposed
and Approved
Legal Herd Boundaries
Biogeoclimatic Zones
Broad Ecosystem Inventory
Wildlife Habitat Areas Proposed
and Approved
Third-order and Greater
Watersheds (50,000)
Fisheries Sensitive Watersheds
Proposed and Approved
Critical Rearing Habitat
Wildlife Habitat Areas Proposed
and Approved
Reconnaissance Karst Potential
Mapping
Freshwater Atlas Wetlands
Mineral Licks
Baseline Thematic Mapping
Urban
TANTALIS - Residential Leases
TRIM Enhanced Base Map
Urban
Digital Road Atlas
Forest Tenure Road Sections
TRIM Transportation Lines
108

johnsoch@unbc.ca
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
MFLNRORD - Ecosystems Branch
Prince George
Data BC
DataBC
Data BC
MFLNRORD - Ecosystems Branch
Prince George
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC

Cost Surfaces

Petroleum Development Roads
Petroleum Access Road
Crown Tenures Pipelines and
Processing
Oil and Gas Right-of-way
Oil and Gas Facility
Oil and Gas Ancillary
Oil and Gas Well Sites
Oil and Gas Surface Hole
TRIM Well Sites
TRIM Pipelines
TRIM Seismic Lines
OGC 2003-2012, 2002-2006,
1996-2004 Seismic Lines
TANTALIS Quarrying Licence
TANTALIS Quarrying Lease
TANTALIS Right-of-way
Mineral Production
Baseline Thematic Mapping
Mining
TANTALIS Industrial Licence
TANTALIS Industrial Lease
TANTALIS Industrial Right-ofway
Baseline Thematic Mapping
Agriculture
Agricultural Land Reserve
Consolidated Cutblocks

109

Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC
Data BC

APPENDIX 2: Cost Surface Geoprocessing Steps
Step 1.
Step 2.
Step 3.
Step 4.
Step 5.
Step 6.
Step 7.
Step 8.
Step 9.
Step 10.
Step 11.
Step 12.
Option A
Option B

Merge all datasets depicting roads described in Appendix 1.
Buffer road polylines based on surface type: paved = 15 m from centerline;
FSR = 10 m from centerline; secondary unpaved road = 7.5 m from centerline;
trail = 3.5 m from centerline
Dissolve buffered roads on a single attribute field to reduce the amount of data
in the layer
Merge all datasets depicting seismic lines described in Appendix 1
Buffer seismic polylines 6 m from centerline
Dissolve buffered seismic lines on a single attribute field to reduce the amount
of data in the layer
Merge all datasets depicting oil and gas development described in Appendix 1
and dissolve on single attribute field
Merge all datasets depicting industrial development described in Appendix 1
and dissolve on single attribute field
Merge all datasets depicting quarrying/mineral/mining development described
in Appendix 1 and dissolve on single attribute field
Merge all datasets depicting urban/residential development described in
Appendix 1 and dissolve on single attribute field
Merge all datasets depicting agricultural development described in Appendix 1
and dissolve on single attribute field
Merge all dissolved layers into a single polygon and dissolve on single
attribute field to produce low-sensitivity cost surface
Prior to completing step 12, add consolidated cutblocks <20 years old to
produce medium-sensitivity cost surface
Prior to completing step 12, buffer forms of development based on avoidance
distances described in Table 4 to produce high-sensitivity cost surface

110

APPENDIX 3: Solution spaces produced in trial scenarios prioritizing areas for conservation
in the Wild Harts Study Area.

Figure 20. Trial scenarios at 50% target for all conservation features; E. no cost surface used; F. low-sensitivity
cost surface used; G. medium-sensitivity cost surface used; H. high-sensitivity cost surface used.

111