FORESTRY, FIRE AND FUR: FACTORS DRIVING THE DECLINE OF
FISHERS (PEKANIA PENNANTI) IN CENTRAL INTERIOR BRITISH
COLUMBIA
by
Rory Fogarty
BSc, Simon Fraser University, 2011

A Thesis Submitted in Partial Fulfillment
of the Requirements for the Degree of
Master of Science
in
Environmental Science

Thompson Rivers University
Kamloops, British Columbia, Canada
November 14, 2023

Thesis Examining Committee:

Karl Larsen (PhD), Thesis Supervisor
Professor, Department of Natural Resource Sciences, Thompson Rivers University

Dexter Hodder, Committee Member,
Director of Research and Education, John Prince Research Forest, Adjunct Faculty,
Department of Natural Resource Science, Thompson Rivers University
David Hill (PhD), Committee, Member,
Associate Professor, Department of Geography, Thompson Rivers University
Dr. Cole Burton, External Examiner,
Canada Research Chair (Tier 2) in Terrestrial Mammal Conservation, Department of
Forest Resources Management, University of British Columbia

ABSTRACT
Understanding the factors driving the population dynamics of endangered species is
critical to effective conservation. Demographic information on how environmental and
anthropogenic influences affect a species distribution and abundance provides valuable
information to wildlife managers looking to accurately assign conservation status, deploy limited
resources efficiently, and set sustainable harvest levels. Unfortunately, obtaining detailed
population data for rare and cryptic species that naturally occur at low densities is challenging,
and in many cases not enough is known to develop appropriate conservation measures. One such
species is the fisher (Pekania pennanti), a mid-sized mustelid considered a species-at-risk over
much of its North American range.
Using a combination of DNA-based mark-recapture surveys and population viability
analysis (PVA), I evaluated the factors driving the declines being seen in the geographically
isolated and endangered Columbian population of fishers (Pekania pennanti) in central interior
British Columbia (BC), Canada. I estimated the density and abundance of fishers in two spatially
distinct ecosystems in central interior BC where current data were lacking, and assessed which
ecological and anthropogenic factors best explained the distribution of fishers in each ecosystem.
Additionally, I constructed population viability models to assess the specific impacts that
trapping mortality would have on the persistence of the Columbian population of fishers under
different trapping scenarios. I found the density of fishers in my two study areas varied
substantially, primarily due to the level of trapping mortality occurring during each survey
season. The density of fishers in the Chilcotin study area west of Williams Lake was among the
highest ever reported for the province (~21 fishers/1000 km2), whereas density in the Enterprise
study area south and east of Williams Lake was among the lowest (~9 fishers/1000 km2). Habitat
covariates that best explained the density of fishers were similar between both study areas;
forested stands with wet soil moisture regimes composed of older deciduous and spruce trees
were most strongly related to higher densities in both the Chilcotin and Enterprise study areas.
My population modelling suggests that if current mortality rates continue, including deaths from
trapping, the Columbian population of fishers will likely be extirpated from central interior BC
within two decades. In the absence of additional fur harvest mortality, the Columbian population
ii

appeared unlikely to persist beyond 37 years without additional measures being taken to increase
female survival and reproductive output.
This study provides valuable information to wildlife managers looking to allocate limited
resources to aid in the recovery of an isolated and endangered species in central interior BC.
Conservation priorities for the Columbian population of fishers should focus on eliminating
mortality from trapping, and protecting the high-value habitats the species requires to survive
and reproduce.

Keywords: Fisher, Pekania pennanti, density, spatially explicit capture-recapture, trapping
mortality, population modelling, British Columbia

Photo credit: Helen Davis and Rich Weir

iii

ACKNOWLEDGEMENTS
This research would not have been possible without the support and collaboration of
many people, to whom I would like to wish my sincere gratitude and thanks. Firstly, I would like
to thank my supervisor Karl Larsen, for taking a chance on someone who didn’t necessarily fit
the mold of a typical graduate student. I am forever grateful that you could see past a less than
stellar transcript, and will never forget the time we spent musing on the fate of the world over
Black & Tans during a pandemic.
I would like to thank Larry Davis, for mentoring me through thick and thin and
introducing me to the wonderful world of weasels – I would not be where I am today without
your guidance and support. A special thank you to Rich Weir, who also took a chance on me and
showed me how passion and determination for wildlife conservation can be translated into real
world action. I want to thank the dedicated members of the BC Fisher Habitat Working Group
who supported this work from beginning to end: Rich Weir, Larry Davis, Scott Yaeger, Eric
Lofroth, Inge-Jean Hansen, Benita Kaytor, Joanna Burgar, and Sandra Frey. I would like to
recognize and thank the Tsideldel and Esk’etemc First Nations for providing field support and
the use of their traditional territories for this research. Another special thank you to Kent Watson
and Mike Tomlinson, for seeing the bigger picture and realizing I would eventually make it back
where I belong.
Finally, I would like to thank my family. To my parents Reg and Lea Fogarty, words
cannot express how grateful I am to you for instilling in me the values of hard work and respect
for others. I want to thank my brother Sean and his partner Kylie for providing accommodations
during my field work free of charge, and apologies again for the unwelcome ermine that got into
the house during my stay! A very special thank you to my brother for providing the motivation
needed to complete this work.
All field sampling procedures used in this work were approved by the Thompson Rivers
University Animal Care Committee under file #102619. Financial support for this project was
generously provided by the BC Ministry of Environment and Climate Change Strategy and the
Ministry of Forests, Lands, Natural Resource Operations and Rural Development.

iv

TABLE OF CONTENTS
ABSTRACT.................................................................................................................................... ii
ACKNOWLEDGEMENTS ........................................................................................................... iv
TABLE OF CONTENTS ................................................................................................................ v
LIST OF FIGURES ...................................................................................................................... vii
LIST OF TABLES ......................................................................................................................... ix
CHAPTER 1. INTRODUCTION ................................................................................................... 1
The Ecological Importance of Mesocarnivores .......................................................................... 1
Humans and Furbearing Mesocarnivores .................................................................................... 2
Fishers as a North American Furbearing Mesocarnivore............................................................ 4
Objectives and Thesis Organization ............................................................................................ 7
Study Area ................................................................................................................................... 8
Literature Cited ......................................................................................................................... 16
CHAPTER 2. FACTORS DRIVING POPULATION DENSITIES OF ENDANGERED
FISHERS (PEKANIA PENNANTI) IN TWO GEOGRAPHICALLY DISTINCT REGIONS OF
BRITISH COLUMBIA, CANADA.............................................................................................. 21
Introduction ............................................................................................................................... 21
Materials and Methods .............................................................................................................. 25
Study area .............................................................................................................................. 25
Fisher DNA sampling ............................................................................................................ 26
Spatial capture-recapture analysis ......................................................................................... 28
Detection covariates .............................................................................................................. 29
Density covariates.................................................................................................................. 30
Results ....................................................................................................................................... 33
Chilcotin study area ............................................................................................................... 33
Enterprise study area ............................................................................................................. 43
Discussion ................................................................................................................................. 47
Conservation implications ..................................................................................................... 51
Literature Cited ......................................................................................................................... 53
CHAPTER 3. TRAPPING MORTALITY ACCELERATES THE DECLINE OF THE FISHER
(PEKANIA PENNANTI), AN ENDANGERED MESOCARNIVORE IN BRITISH COLUMBIA,
CANADA.† ................................................................................................................................... 60
Introduction ............................................................................................................................... 60
Materials & Methods ................................................................................................................. 62
v

Study area .............................................................................................................................. 62
Population viability analysis software ................................................................................... 62
Input parameters .................................................................................................................... 63
Model manipulation and sensitivity testing ........................................................................... 65
Results ....................................................................................................................................... 68
No Trapping Scenario (Scenario 0 – no annual fur harvest mortality) ................................. 68
Alternative scenarios ............................................................................................................. 69
Sensitivity testing .................................................................................................................. 71
Discussion ................................................................................................................................. 74
Conclusions ............................................................................................................................... 77
Literature Cited ......................................................................................................................... 79
CHAPTER 4 .CONCLUSION...................................................................................................... 85
Summary of Thesis.................................................................................................................... 85
Management Recommendations ............................................................................................... 87
Population Management ........................................................................................................ 87
Habitat Management.............................................................................................................. 88
Harvest Management ............................................................................................................. 89
Limitations and Future Research............................................................................................... 90
Conclusion................................................................................................................................. 92
Literature Cited ......................................................................................................................... 93
APPENDIX A. DESCRIPTION OF VARIABLES USED IN DENSITY MODELS. ................ 95
APPENDIX B. SET OF CANDIDATE MODELS CONSIDERED FOR PREDICTING
DENSITY OF FISHERS IN THE CHILCOTIN AND ENTERPRISE STUDY AREAS. .......... 97
APPENDIX C. 95% CONFIDENCE SET OF CANDIDATE MODELS FOR PREDICTING
THE DENSITY OF FISHERS IN THE CHILCOTIN AND ENTERPRISE STUDY AREAS. . 99

vi

LIST OF FIGURES
Figure 1.1. Fishers (Pekania pennanti) are an elusive member of the Mustelid family (Photo
credit by author). ................................................................................................................. 5
Figure 1.2. Current range of the two fisher (Pekania pennanti) populations (Boreal and
Columbian, respectively) in British Columbia, Canada (BC CDC 2020). ......................... 9
Figure 1.3. Map of the survey area for the 2018-19 Chilcotin Fisher Survey. IDF = Interior
Douglas-fir BEC zone; MS = Montane Spruce BEC zone; SBPS = Sub-Boreal PineSpruce BEC zone; ESSF = Engelmann-Spruce Subalpine fir BEC Zone. ....................... 10
Figure 1.4. Map of the survey area for the 2020-21 Enterprise Fisher Survey. IDFdk3 = Interior
Douglas-fir dry cool 3 BEC subzone variant; IDFxm = Interior Douglas-fir very dry mild
BEC subzone variant; SBPSmk = Sub-Boreal Pine-Spruce moist cool BEC subzone
variant; SBSdw2 = Sub-Boreal Spruce dry warm 2 BEC subzone variant. ..................... 11
Figure 1.5. Forests in the Chilcotin study area consisted of a mosaic of lodgepole pine stands of
various ages with older spruce stands concentrated around streams and wetlands. ......... 13
Figure 1.6. Typical spruce and pine stand with hair-snagging device set up in the Chilcotin study
area. ................................................................................................................................... 13
Figure 1.7. Forests in the Enterprise study area consisted of multi-layered Douglas-fir stands
with spruce and aspen localized around riparian areas. .................................................... 14
Figure 1.8. Example of a multi-layered Douglas-fir stand in the Enterprise study area. .............. 14
Figure 1.9. Approximately 30% of the Chilcotin study area was burned during the 2017 Plateau
wildfire. ............................................................................................................................. 15
Figure 2.1. Typical set up of triangular wooden hair-snagging ‘cubby’ fastened vertically to a
tree. Inset picture shows the inside of the device with bait and adhesive glue strips used to
collect fisher hair............................................................................................................... 27
Figure 2.2. Relationship between density of fishers and variables included in the top-ranked
models for the Chilcotin study area. Panel A shows the top-ranked model pooled across
sexes, Panel B is for females only, and Panel C is for males only; dashed lines represent
the upper and lower 95% confidence intervals. ................................................................ 39
Figure 2.3. Predicted density of fishers (fishers/1000 km2), pooled across sexes, in Sub-Boreal
Pine-Spruce forests in central BC estimated from spatial capture-recapture analysis of
genetically identified fishers sampled during the winter of 2018-19 as predicted by the
proportion of surrounding 30 km2 containing primary branch resting stands. SBPSxc =
Sub-Boreal Pine-Spruce very dry cold BEC subzone variant. ......................................... 40
Figure 2.4. Predicted density of male fishers (fishers/1000 km2) in the Sub-Boreal Pine-Spruce
forests in central BC estimated from spatial capture-recapture analysis of genetically
identified fishers sampled during the winter of 2018-19 as predicted by the proportion of

vii

surrounding 30 km2 containing primary branch resting stands. SBPSxc = Sub-Boreal
Pine-Spruce very dry cold BEC subzone variant. ............................................................. 41
Figure 2.5. Predicted density of female (fishers/1000 km2) in the Sub-Boreal Pine-Spruce forests
in central BC estimated from spatial capture-recapture analysis of genetically identified
fishers sampled during the winter of 2018-19 as predicted by the proportion of
surrounding 30 km2 containing primary branch resting stands. SBPSxc = Sub-Boreal
Pine-Spruce very dry cold BEC subzone variant. ............................................................. 42
Figure 2.6. Modelled fisher density (pooled across sexes) in stands with wet soil moisture
regimes (i.e., wetter than mesic) within Interior Douglas-fir forest types of the Enterprise
study site in central interior BC. ....................................................................................... 45
Figure 2.7. Predicted density of fishers (fishers/1000 km2) in the Interior Douglas-fir forests of
central interior BC estimated from spatial capture-recapture analysis of genetically
identified fishers sampled during the winter of 2020-21 as predicted by the proportion of
surrounding 30 km2 containing stands with soil moisture regimes wetter than mesic.
IDFdk3 = Interior Douglas-fir dry cool 3 BEC subzone variant. ..................................... 46
Figure 3.1. Influence of alternative levels of mortality from trapping bycatch on the probability
of extinction over 100 years for the Columbian population of fishers (Pekania pennanti)
in British Columbia, Canada, using parameterizations listed in Table 3.1 and Trapping
Scenarios in Table 3.3. Modelled using Vortex software (version 10.5.5.0). .................. 70
Figure 3.2. The effects of varying input parameters for breeding and mortality rates and their
associated values of Environmental Variation (EV) by ±10% on the stochastic population
growth rate for the Columbian population of fishers (Pekania pennanti) in British
Columbia, Canada under Scenario 0 – no trapping mortality. The black dashed line is set
at the stochastic population growth rate value for Scenario 0. Red indicates the parameter
value was decreased by 10% and black indicates the parameter value was increased by
10%. Each parameter was varied while holding all other parameters constant. Modelled
using Vortex software (version 10.5.5.0).......................................................................... 73
Figure 3.3. The effects of decreasing natural female mortality in increments of 2% on the
probability of extinction over 100 years for the Columbian population of fishers (Pekania
pennanti) in British Columbia, Canada. The red dashed line at 0 represents natural female
mortality rates used for initial population projections, detailed in Table 1, that were
derived from three radio-telemetry studies on fishers in the central interior of BC between
1995 – 2009. These initial rates were subsequently decreased in increments of 2% to
evaluate the effect on probability of extinction, in the absence of any mortality from
trapping. Modelled using Vortex software (version 10.5.5.0). ......................................... 74

viii

LIST OF TABLES
Table 2.1. Density estimates for fishers from elsewhere in North America, with 95% confidence
intervals (95% CI) included if reported. Density estimates from British Columbia are in
bold. .................................................................................................................................. 23
Table 2.2. Variables used in candidate models known from previous research to influence the
density of fishers within the Chilcotin and Enterprise study areas in the central interior of
British Columbia, Canada. I report the variable name, variable description, and whether it
was used in the candidate model set for each study area (C = Chilcotin, E = Enterprise).
The area, length, or density of each variable was calculated for both the 4.84 km2 (core
use scale) and the 30 km2 (home range scale) buffer areas around each habitat mask
point. A full description of variables with their applicable reference can be found in
Appendix A. ...................................................................................................................... 32
Table 2.3. Model selection table to evaluate detection for fishers in the Chilcotin and Enterprise
study areas in the central interior of British Columbia, Canada. I estimated 3 parameters:
density (D), detection probabiity (g0), and a spatial parameter (sigma). No density
covariates were included in this analysis. The table shows the number of model
parameters (K), model log likelhood (logLik), Akaike's Information Criterion corrected
for small sample sizes (AICc), the difference in AICc values (∆AICc), and relative model
weight (wi). Models ≤2.0 AICc units of the top-ranked models are displayed here, with
the null model assuming g0 and sigma are fixed shown for reference. ............................ 35
Table 2.4. Model selection table showing the top-ranked density models for fishers in the
Chilcotin study area in the central interior of British Columbia, Canada. I estimated 3
parameters: density (D), detection probabiity (g0), and a spatial parameter (sigma) at the
female fisher core use area scale (4.84 km2) and the home range scale (30 km2). The table
shows the number of model parameters (K), model log likelhood (logLik), Akaike's
Information Criterion corrected for small sample sizes (AICc), the difference in AICc
values (∆AICc), and relative model weight (wi). Models ≤2.0 AICc units of the topranked models are displayed here, along with the null model that assumes all parameters
are constant. ...................................................................................................................... 36
Table 2.5. Beta coefficients (β) and associated 95% confidence intervals (CI) for the top-ranked
density models for fishers at both the core use area scale (4.84 km2) and the home range
scale (30 km2) in the Chilcotin and Enterprise study areas in the central interior of British
Columbia, Canada. ............................................................................................................ 38
Table 2.6. Model selection table showing the top-ranked density models for fishers in the
Enterprise study area in the central interior of British Columbia, Canada. I estimated 3
parameters: density (D), detection probabiity (g0), and a spatial parameter (sigma) at the
female fisher core use area scale (4.84 km2) and the home range scale (30 km2). The table
shows the number of model parameters (K), model log likelhood (logLik), Akaike's
Information Criterion corrected for small sample sizes (AICc), the difference in AICc
values (∆AICc), and relative model weight (wi). Models ≤2.0 AICc units of the topranked models are displayed here, along with the null model that assumes all parameters
are constant. ...................................................................................................................... 44
ix

Table 3.1. Input parameters and associated data sources for population viability analysis of the
Columbian population of fishers (Pekania pennanti) in British Columbia (BC), Canada.
Most parameters were derived from three radio-telemetry studies on fishers in central BC
and one small, well-studied population in California. Environmental Variation (EV)
refers to the annual variation in reproduction and survival due to random changes in the
environment and is modelled using the standard deviation for each applicable parameter
(Lacy et al. 2020). ............................................................................................................. 64
Table 3.2. The proportion of fishers (Pekania pennanti) harvested from the Columbian
population in British Columbia, Canada by trap class as calculated from the provincial fur
harvest database. ............................................................................................................... 66
Table 3.3. Fur harvest scenarios considered in the projection of population outcomes for fishers
(Pekania pennanti) in the Columbian population, British Columbia, Canada. The
percentage of the total population removed under each scenario was based upon the
annual average reported harvest of fishers in the Columbian population between 2009 and
2017 (shown in brackets) and a projected initial population size of 571 fishers. We
applied a constant trapping mortality rate using this percentage of the initial population
harvested annually to account for the decline in harvest we expected to see each year as
the overall population declines (i.e., as the population declines over time so does the
number of fishers harvested). ............................................................................................ 68
Table 3.4. Stochastic growth rate, probability of extinction, and mean time to extinction (±
standard error) for the Columbian population of fishers (Pekania pennanti) in British
Columbia (BC), Canada under different levels of trapping mortality. Simulations were
run 1000 times over a period of 100 years using Vortex software (version 10.5.5.0). Both
males and females were available for trapping in these scenarios with an age distribution
used from fisher carcass data from BC (Weir 2003). Extinction was considered when
only one sex remained in the population. ......................................................................... 69
Table 3.5. Effect of using different age structures on the predicted stochastic population growth
rate, probability of extinction, and mean time to extinction (± standard error) for the
Columbian population of fishers (Pekania pennanti) in British Columbia, Canada, over
100 years. Extinction was considered when only one sex remained in the population.
Modelled using Vortex software (version 10.5.5.0). ........................................................ 72
Table 3.6. Stochastic population growth rate, probability of extinction, and mean time to
extinction (± standard error) for the Columbian population of fishers (Pekania pennanti)
in British Columbia, Canada over 100 years with different initial population sizes.
Extinction was considered when only one sex remained in the population. Modelled
using Vortex software (version 10.5.5.0).......................................................................... 72

x

CHAPTER 1
INTRODUCTION
The Ecological Importance of Mesocarnivores
Global biodiversity is in decline, primarily due to habitat loss and human-caused
mortality (Maxwell et al. 2016). Where these threats to wildlife populations co-occur, they can
produce strong complementary effects that are larger than the sum of their parts (Brook et al.
2008). However, these threats do not affect all species equally, and rare species with restricted
geographical ranges (Pimm 1991, Gaston 1994), low fecundity (MacArthur and Wilson 1967),
poor dispersal abilities (de Vries et al. 1996), and large home range sizes (Woodroffe and
Ginsberg 1998) often are the most susceptible to environmental and anthropogenic disturbances.
Many mammalian carnivores (principally Order Carnivora) possess some or all of these
characteristics and as such are particularly vulnerable to habitat loss and human-caused
mortality; unsurprisingly, this means species within the Carnivora continue to be among the most
threatened across the globe (Wolf and Ripple 2017).
The ecological importance of large carnivores is well-established. Even a few individuals
can strongly influence community structure directly by limiting prey populations, causing
cascading effects on vegetation, and providing competition between larger and smaller predators
(Ripple and Beschta 2004, Ripple et al. 2014, Sivy et al. 2017). Additionally, carnivores are ideal
indicators of ecosystem health because their large home ranges and low population densities
make them sensitive to environmental change (Buskirk and Zielinski 2003). As such,
conservation planning often focuses on carnivores with the largest home range requirements
based on the premise that protecting the habitat necessary to support viable populations also
likely protects the habitats of other species with smaller area needs (Noss et al. 1996). To this
end, it is usually the large carnivores such as grizzly bears (Ursus arctos) and wolves (Canis
lupus) that are used as ‘umbrella species’ due to their extensive spatial needs as well as their
charismatic appeal (Buskirk et al. 2003).
However, scientists increasingly are understanding that smaller carnivores can also play
this role due to their similarly large home range requirements (e.g., 405 - 1366 km2 for
wolverines [Gulo gulo] - Lofroth 2001, 50 – 500 km2 for badgers [Taxidea taxus] - Weir and
1

Hoodicoff 2002), and because in many regions they have replaced larger carnivores as the apex
predator (Roemer et al. 2009). Small- and mid-sized carnivores (typically <20 kg), collectively
referred to as “mesocarnivores”, represent an ecologically diverse array of species whose
influence on ecosystem structure and function may have previously been underestimated
(Roemer et al. 2009). With the decline of large carnivores, many mesocarnivores now play the
role of top predator (Laliberte and Ripple 2004), or act as important drivers of ecosystems due to
their impact on prey populations or the surrounding environment (Eldridge and Whitford 2009).
For example, where wolves have disappeared from certain areas of California, coyotes (Canis
latrans) have become the top predator, and by consuming smaller predators that typically predate
on eggs from songbird nests they have a positive effect on songbird survival rates (Crooks and
Soule 1999). Badgers have a substantial influence on rodent populations in grassland ecosystems
and the burrows they dig affect water infiltration, soil aeration, organic decomposition, and plant
diversity (Eldridge 2004). Ocelots (Leopardus palus) in Panama are the main predators of
Central American agoutis (Dasyprocta punctata; Moreno et al. 2006), and because agoutis are
primary consumers of large-bodied seeds, ocelot predation indirectly affects the way forests are
structured (Roemer et al. 2009). Taken together, the wide range of roles that mesocarnivores
fulfill in a variety of habitats suggest these species may have exceptionally important effects on
ecosystem function, structure, or dynamics (Roemer et al. 2009).
Given the importance of mesocarnivores to ecosystems, the best approach to large-scale
conservation may be to target multiple focal species, including those species that specialize on
rare habitats that have high ecological value (Carroll et al. 2001). Moreover, recent research
suggests that the presence of an intact community of specialist species can be used as a reliable
indicator of ecosystem health and functionality and the absence of these species suggests an
ecosystem out of balance (Clavel et al. 2011). Thus, a high degree of habitat specialization
coupled with the considerable spatial requirements of some mesocarnivores may make them
more suitable and compelling species on which to focus conservation efforts (Buskirk and
Zielinski 2003).

Humans and Furbearing Mesocarnivores
Despite their importance to ecosystems, humans have been responsible for the direct and
indirect mortality of carnivores for millennia. In North America, Indigenous peoples hunted and
2

trapped carnivores for food, clothing, cultural reasons, and to protect their families from
predators (Proulx 1999). Furbearing species, the majority of which would be considered
mesocarnivores, were harvested at low levels by Indigenous people who used fur for clothing,
bedding, and shelter (Wright 1987). Prior to European colonization, Indigenous people used
primitive trapping techniques such as deadfalls and snares, and population effects were
negligible as focus was placed on acquiring the meat and hides of larger animals (Banci and
Proulx 1999). Colonists brought with them new trapping technology in the form of steel traps
and economic incentives, and the large-scale harvesting of furbearers soon became a keystone
enterprise (Gerstell 1985).
The evolution of the fur trade into a commercial entity changed how humans used
furbearers as a resource and the harvest and trade of furs resulted in significant declines in many
species (Banci and Proulx 1999). In these early Colonial days, the natural resources of North
America seemed inexhaustible and unregulated fur harvest quickly resulted in the substantial
reduction, extirpation, or extinction of many once-common furbearers like the American beaver
(Castor canadensis), the sea mink (Neovison macrodon), and the sea otter (Enhydra lutis)(Ray
1987). In the case of the sea otter, extirpation from the North Pacific resulted in a cascade of
effects including excessive growth of their primary prey, sea urchin, and the subsequent
decimation of kelp forests upon which a multitude of other species depend (Estes et al. 1998).
While fur harvest does not appear to have had a long-term effect on ermine (Mustela erminea)
populations, American marten (Martes americana), Pacific marten (Martes caurina), wolverines,
and North American river otters (Lontra canadensis) had all been extirpated from portions of
their former ranges by the early twentieth century (Harrington et al. 2017). This excessive use of
resources spawned the first regulations restricting the harvest of various wildlife in North
America although more systematic and widespread conservation efforts took much longer to
develop (Organ et al. 2012).
By the mid-1900s, the conversion of agricultural lands back to forest, trapping closures,
stricter fur harvest regulations, and translocation efforts allowed many furbearing
mesocarnivores to recolonize portions of their historical ranges (White et al. 2021). Although
increased habitat protections and reduced human-caused mortality has been successful in
allowing harvests of several species to once again become sustainable (White et al. 2021), most
North American mesocarnivores continue to be killed both directly and indirectly by humans to
3

this day. Forest-dependent mesocarnivores such as the American marten, for example, continue
to be impacted by the negative effects of timber harvest removing large areas of late-successional
forests, and in most jurisdictions still are trapped for their furs (White et al. 2015). While some
species appear to be resilient to high levels of fur harvest (e.g., coyotes; Malhotra 2022), most
furbearers fall somewhere between low and intermediate resiliency levels with substantial
variation depending on habitat conditions at the regional level (Banci and Proulx 1999).
Unfortunately, trapping systems currently approved for use worldwide still are not truly speciesspecific and it is nearly impossible to ensure that only those species with high resiliency to
trapping are harvested (Virgos et al. 2016, Proulx 2022).

Fishers as a North American Furbearing Mesocarnivore
Fishers (Pekania pennanti; Figure 1.1) are one example of a low-density and wideranging mesocarnivore species that has experienced significant range contractions over the past
two centuries due to human-related activities (Powell and Zielinski 1994, Lofroth et al. 2010).
Fishers are housecat-sized members of the Mustelid family once widespread and abundant across
North America (Gibilisco 1994); following the settlement of Europeans their distribution and
abundance declined primarily due to overexploitation in the fur trade and habitat loss (Lofroth et
al. 2010). This large-scale habitat loss, along with extremely high pelt prices and a lack of fur
harvest regulations in the early 1900s led to the extirpation of fishers from many regions across
the southern extent of their range (Lewis et al. 2012). Increased protective measures, trapping
restrictions, reintroduction efforts, and habitat recovery programs have allowed fishers to
recolonize much of their historic range in eastern North America; however, west of the Rocky
Mountains they remain uncommon, existing as fragmented remnant populations in California,
Oregon, Idaho, Montana, and British Columbia (BC)(Lofroth et al. 2010).
Throughout North America, fishers inhabit low elevation boreal and temperate coniferous
and mixed-wood forests (Lofroth et al. 2010). They are considered a forest-obligate species and
typically are associated with increasing amounts of tree canopy cover (Powell 1994, Weir 2003).
Like many other forest-dwelling species, fishers use cavities or chambers in live and dead trees
as daily refugia for resting, and for reproduction (Zielinski et al. 2004, Lofroth et al. 2010).
Resting and denning structures typically are found in the largest diameter standing live trees,
snags, or logs (Raley et al. 2012), but other structural features such as platforms, branches or
4

Figure 1.1. Fishers (Pekania pennanti) are an elusive member of the Mustelid family (Photo credit by author).

5

mistletoe brooms also are important for resting (Davis 2009, Aubry et al. 2013, Green 2017).
Although these structures typically occur in higher densities in older forests, fishers are not truly
an old-growth dependent species but rather require a mosaic of different stand types within their
home range (Lofroth et al. 2010, Raley et al. 2012, Sauder and Rachlow 2015).
Specific habitat requirements for denning and resting make fishers in BC forest obligates,
relying on these distinct structural components of forested ecosystems to fulfill their life
requisites (Lofroth et al. 2010). Female fishers in BC give birth and den in the cavities of large
diameter trees where some form of damage (e.g., frost crack, branch scar, fire-scar) has allowed
decay organisms to enter and form internal cavities over time (Weir et al. 2012). When not
actively travelling or hunting fishers use specific habitat features to conserve energy, avoid
predators, thermoregulate, and to consume prey safely (Raley et al. 2012). These resting
structures also are rare and include heart rot cavities, large branches, coarse woody debris,
brooms in white spruce (Picea glauca) formed by spruce broom rust (Chrysomyxa
arctostaphyli), among others (Weir 2003, Davis 2009).
The disease and decay processes that create these essential habitat features for fishers
take a long time to develop (>100 years) making the animal highly vulnerable to disturbances
that remove these structures faster than forests can redevelop them (Weir et al. 2012). In BC, the
permanent loss of forested habitats due to land conversion and hydroelectric developments is a
threat in some areas of the province (Weir 2003), but forest harvest has the greatest potential to
impact fisher habitat negatively in BC due to the prevalence of clear-cut harvesting practices
(Weir and Corbould 2010, BC CDC 2020). Over the past two decades, the low-elevation forests
that fishers rely on in the province have been significantly impacted by large-scale insect
infestations, unprecedented wildfire seasons, and associated salvage logging. Concern over the
loss of fisher habitat in our province is nothing new (Weir 2003, Lofroth et al. 2010), yet the
current rate and magnitude of forest harvest following these large-scale disturbances has the
potential to be highly detrimental to this already low-density species.
Recent genetic work in BC indicates there are two distinct fisher populations within the
province, the Boreal and Columbian population units (BC CDC 2020). During its most recent
status review of fishers in March of 2020, the BC Conservation Data Centre (BC CDC) assessed
these two fisher populations separately and the Columbian population was placed on the
province’s ‘Red-list’ as Endangered (BC CDC 2020). This up-listing was due to a low
6

population estimate of 299 – 517 adult individuals, genetic and geographic isolation, and shortterm population trends showing declines of 30 – 50% due to loss of forest habitat (BC CDC
2020). Current conditions also suggest this population is declining further as a result of
increasing levels of forest harvest causing further habitat loss and degradation, and trapping
where the habitat of fishers has already been compromised (BC CDC 2020). In August 2021, the
commercial trapping season for fishers was closed within the range of the Columbian population
one year after the animals were reclassified as Endangered (Province of British Columbia 2021).
However, fishers are regularly taken as bycatch in traps designed and set for other furbearing
species (e.g., traps designed for American marten account for 52% of annual fisher bycatch;
Province of British Columbia unpubl. data: Compulsory Reporting 2013–2018 [accessed 15 July
2021]) therefore the closure of the trapping season is likely of limited benefit for reducing the
number of fishers harvested annually.
Although recent trends indicate that the Columbian population is declining, fishers
historically have been poorly monitored across the province and the true extent of any decline
remains unknown (R. Weir pers. comm.). As a result, there still are substantial gaps in
knowledge about the current distribution and abundance of fishers in several regions of the
province, including areas in the central interior where large-scale habitat losses already have
occurred (BC CDC 2020). In addition, in areas where the habitat of fishers has already been
significantly impacted little is known about the influence that trapping mortality may be having
on this small and isolated population.

Objectives and Thesis Organization
The overarching goal of my research was to investigate the key environmental,
demographic, and anthropogenic factors driving the dynamics of the Columbian population of
fishers. In conducting this research, I sought much-needed information on the current distribution
and abundance of the species in central BC while investigating the effect that continuing trapping
mortality will have on its sustainability over the longer term. In Chapter 2, I used information
from non-invasive DNA-based surveys in two spatially distinct ecosystems within the range of
the Columbian population of fishers; this provided estimates of density and abundance where
current data do not exist. Additionally, I investigated which anthropogenic and environmental
factors best predicted density across these regions, thereby gaining valuable insight into current
7

habitat associations specific to these ecosystems. In Chapter 3, I used survival and reproductive
rates from fishers in central BC and Population Viability Analysis (PVA) methodologies to
evaluate the effects that continuing mortality from trapping bycatch may be having on the
sustainability of the Columbian population of fishers. Finally, in Chapter 4, I briefly summarize
my main findings and detail the management implications of this research for the Columbian
population of fishers. In conclusion, I provide recommendations for recovery planning efforts
and identify areas future research should focus on with respect to fishers in the central interior of
BC.
In the remainder of this chapter, I provide a broad description of the range of the
Columbian population of fishers in BC, and a more detailed description of the two study areas
where DNA-based surveys were conducted.

Study Area
In the central interior of BC, the Columbian population of fishers range throughout
forested habitats at low to moderate elevations extending north to the Rocky Mountains and
south to Lytton and Lillooet (Figure 1.2; Weir 2003). In this province, ecosystems are classified
into ‘biogeoclimatic’, or ‘BEC’, zones based on climate, elevation, and vegetation
characteristics. Northern and central portions of the range of the Columbian population consist of
the Sub-Boreal Spruce BEC zone (Meidinger et al. 1991), whereas the southerly portion of the
range consists of drier ecological zones including the Sub-Boreal Pine-Spruce (Steen &
Demarchi 1997), Montane Spruce (Hope et al. 1991a), and Interior Douglas-fir (Hope et al.
1991b) BEC zones.
For my research, I conducted DNA-based hair-snagging surveys in two study areas
within the range of the Columbian population of fishers; one on the Chilcotin Plateau west of
Williams Lake during the winter of 2018-19 (hereafter named the Chilcotin study area; Figure
1.3), and one on the interior Fraser Plateau south and east of Williams Lake during the winter of
2020-21 (hereafter named the Enterprise study area; Figure 1.4). The 2,440 km2 Chilcotin study
area extended from Chantslar Lake in the west to Alexis Lakes in the east, and Highway 20 in
the south to Satah Mountain in the north. The 2,580 km2 Enterprise study area extended from the
grasslands above the Fraser River in the west to Highway 97 in the east, and Williams Lake in
the north to Meadow Lake Road in the south. Both study areas consisted of level to gently rolling
8

Figure 1.2. Current range of the two fisher (Pekania pennanti) populations (Boreal and Columbian,
respectively) in British Columbia, Canada (BC CDC 2020).

9

Figure 1.3. Map of the survey area for the 2018-19 Chilcotin Fisher Survey. IDF = Interior Douglas-fir BEC zone; MS = Montane Spruce BEC
zone; SBPS = Sub-Boreal Pine-Spruce BEC zone; ESSF = Engelmann-Spruce Subalpine fir BEC Zone.

10

Figure 1.4. Map of the survey area for the 2020-21 Enterprise Fisher Survey. IDFdk3 = Interior Douglasfir dry cool 3 BEC subzone variant; IDFxm = Interior Douglas-fir very dry mild BEC subzone variant;
SBPSmk = Sub-Boreal Pine-Spruce moist cool BEC subzone variant; SBSdw2 = Sub-Boreal Spruce dry
warm 2 BEC subzone variant.

11

topography and surveys were conducted in non-mountainous areas at low to moderate elevations
(750 – 1700 m).
The Chilcotin study area was within the driest ecological zone where fishers are found in
BC having forests dominated by lodgepole pine (Pinus contorta) with smaller components of
Douglas-fir (Pseudotsuga menziessi), trembling aspen (Populus tremuloides), and hybrid spruce
(Picea glauca x engelmannii) concentrated around streams and wetlands (Figures 1.5 and 1.6Sub-Boreal Pine-Spruce very dry cold [SBPSxc], Montane Spruce very dry very cold [MSxv],
and Interior Douglas-fir dry cool 4 [IDFdk4] BEC zones, respectively; Meidinger and Pojar
1991). The Enterprise study area consisted of forests dominated by either multi-aged Douglas-fir
or uniform-aged lodgepole pine stands, with small trembling aspen and hybrid spruce stands
occurring locally (Figures 1.7 and 1.8 - Interior Douglas-fir dry cool 3 [IDFdk3] BEC zone;
Steen and Coupé 1997).
The Chilcotin study area was located on the leeward side of the Coast Range Mountains
where the rain shadow effect of the mountains is most pronounced with relatively little
precipitation falling (mean annual precipitation 355 – 389 mm) resulting in severely limited
vegetation production and soil development in this area (Meidinger and Pojar 1991). The climate
in the Chilcotin study area consisted of warm, dry summers (mean warmest month = 12.3 ℃;
Steen and Coupé 1997) and cold, snowy winters (mean coldest month = -11.8 ℃; Steen and
Coupé 1997). Mean annual precipitation (433 mm) and mean annual temperature (3.3 ℃) were
both slightly higher in the Enterprise study area, and the annual average snowfall ranged from
179 cm in the SBPSxc to 231 cm in the IDFdk3 (Steen and Coupé 1997). Approximately 30% of
the Chilcotin study area was burnt to varying degrees by the Plateau Fire (Figure 1.9) and ~3%
of the Enterprise study area was burnt by the Gustafsen Lake Fire during the 2017 wildfire
season, which at the time was the most severe wildfire season on record for BC (Province of BC
2017).

12

Figure 1.5. Forests in the Chilcotin study area consisted of a mosaic of lodgepole pine stands of various
ages with older spruce stands concentrated around streams and wetlands.

Figure 1.6. Typical spruce and pine stand with hair-snagging device set up in the Chilcotin study area.

13

Figure 1.7. Forests in the Enterprise study area consisted of multi-layered Douglas-fir stands with spruce
and aspen localized around riparian areas.

Figure 1.8. Example of a multi-layered Douglas-fir stand in the Enterprise study area.

14

Figure 1.9. Approximately 30% of the Chilcotin study area was burned during the 2017 Plateau wildfire.

15

Literature Cited
Banci V, Proulx G (1999) Resiliency of furbearers to trapping in Canada. In: Proulx G (ed)
Mammal Trapping. Alpha Wildlife Research & Management Ltd., Sherwood Park,
Alberta
Brook BW, Sodhi NS, Bradshaw CJA (2008) Synergies among extinction drivers under global
change. Trends in Ecology and Evolution 23:453-460
Buskirk SW, Zielinski WJ (2003) Small and mid-sized carnivores. In: Zabel CJ, Anthony RF
(eds) Mammal Community Dynamics: Management and Conservation in the Coniferous
Forests of North America. Cambridge University Press, Cambridge, UK
Carroll C, Noss RF, Paquet PC (2001) Carnivores as focal species for conservation planning in
the Rocky Mountain region. Ecological Applications 11:961-980
BC Conservation Data Centre (2020) Conservation Status Report: Pekania pennanti pop. 5. In.
BC Ministry of Environment
Clavel J, Julliard R, Devictor V (2011) Worldwide decline of specialist species: toward a global
functional homogenization? Frontiers in Ecology and the Environment 9:222-228
Crooks KR, Soule ME (1999) Mesopredator release and avifaunal extinctions in a fragmented
system. Nature 400:536-566
Davis RL (2009) Denning ecology and habitat use by fisher (Martes pennanti) in pine dominated
ecosystems of the Chilcotin Plateau. Master of Science Research, Simon Fraser
University, Burnaby, BC, Canada
de Vries HH, den Boer PJ, van Dijk TS (1996) Ground beetle species in heathland fragments in
relation to survival, dispersal, and habitat preference. Oecologia 107:332-342
Eldridge DJ (2004) Mounds of the American badger (Taxidea taxus): Significant features of
North American shrub-steppe ecosystems. Journal of Mammalogy 85:1060-1067
Eldridge DJ, Whitford WG (2009) Badger (Taxidea taxus) disturbances increase soil
heterogeneity in a degraded shrub-steppe ecosystem. Journal of Arid Environments
73:66-73
Eng M, Fall A, Hughes J, Shore T, Riel B, Hall P, Walton A (2005) Provincial level projection
of the current mountain pine beetle outbreak: An overview of the model and results of
Year 3 of the project. . In. Natural Resources Canada. Mountain Pine Beetle Initiative
Working Paper
16

Estes JA, Tinker MT, Williams TM, Doak DF (1998) Killer whale predation on sea otters linking
oceanic and nearshore ecosystems. Science 282:473-476
Gaston KJ (1994) Rarity, Vol. Chapman & Hall, London, UK
Gerstell R (1985) The steel trap in North America, Vol. Stackpole Books, Mechanicsburg,
Pennsylvania
Gibilisco CJ (1994) Distributional dynamics of modern Martes in North America. In: Buskirk
SW, Harestad AH, Raphael MG, Powell RA (eds) Martens, sables and fishers: biology
and conservation. Cornell University Press, Ithaca, NY
Hope GD, Mitchell WR, Lloyd DA, Erickson WR, Harper WL, Wikeem BM (1991a) Interior
Douglas-fir Zone. In: Meidinger DV, Pojar J (eds) Ecosystems of British Columbia.
British Columbia Ministry of Forests Special Report Series 6, Victoria, Canada
Hope GD, Mitchell WR, Lloyd DA, Harper WL, Wikeem BM (1991b) Montane Spruce Zone.
In: Meidinger DV, Pojar J (eds) Ecosystems of British Columbia. British Columbia
Ministry of Forests Special Report Series 6, Victoria, Canada
Klafki RW (2014) Road ecology of a northern population of badgers (Taxidea taxus) in British
Columbia, Canada. Master of Science, Thompson Rivers University, Kamloops, BC
Laliberte AS, Ripple WJ (2004) Range contractions of the North American carnivores and
ungulates. Bioscience 54:123-138
Lewis JC, Powell RA, Zielinski WJ (2012) Carnivore translocations and conservation: insights
from population models and field data for fishers (Martes pennanti). PLoS One 7:1-15
Lofroth EC, Raley CM, Higley JM, Truex RL, Yaeger JS, Lewis JC, Happe PJ, Finley LL (2010)
Conservation of Fishers (Martes pennanti) in south-central British Columbia, Western
Washington, Western Oregon, and California. Volume I: Conservation Assessment, Vol
I. USDI Bureau of Land Management, Denver, Colorado, USA
MacArthur RH, Wilson EO (1967) The equilibrium theory of island biogeography, Vol.
Princeton University Press, Princeton, NJ
Maxwell SL, Fuller RA, Brooks TM, E.M. WJ (2016) Biodiveristy: The ravages of guns, nets
and bulldozers. Nature 536:143-145
Meidinger DV, Pojar J (1991) Ecosystems of British Columbia, Vol. British Columbia Ministry
of Forests Special Report Series 6, Victoria, BC, Canada

17

Meidinger DV, Pojar J, Harper WL (1991) Sub-Boreal Spruce Zone. In: Meidinger DV, Pojar J
(eds) Ecosystems of British Columbia. British Columbia Ministry of Forests Special
Report Series 6, Victoria, Canada
Moreno RS, Kays RW, Samudio RJ (2006) Competitive release in diets of ocelot (Leopardus
pardalis) and puma (Puma concolor) after jaguar (Panthera onca) decline. Journal of
Mammalogy 87:808-816
Noss RF, Quigley HB, Hornocker MG, Merrill T, Paquet PC (1996) Conservation Biology and
Carnivore Conservation in the Rocky Mountains. Conservation Biology 10:949-963
Organ JF, Geist V, Mahoney P, Williams S, Krausman PR, Batcheller GR, Decker T,
Carmichael R, Nanjappa P, Regan R (2012) The North American Model of Wildlife
Conservation. The Wildlife Society Technical Review 12-04. In, Bethesda, Maryland,
USA
Pimm SL (1991) The balance of nature?, Vol. University of Chicago Press, Chicago, Illinois
Powell RA, Zielinski WJ (1994) Fisher. In: Ruggiero LF, Aubry KB, Buskirk SW, Lyon LJ,
Zielinski WJ (eds) The Scientific basis for conserving forest carnivores: American
marten, fisher, lynx and wolverine in the western United States
Proulx G (1999) Review of current mammal trap technology in North America. In: Mammal
trapping, Alpha Wildlife Publications, Sherwood Park, Alberta, Canada
Proulx G (2021) How human-small carnivore conflicts can repeat themselves. Canadian Wildlife
Biology and Management 10:1-24
Province of British Columbia (2017) Wildfire Season Summary.
https://www2.gov.bc.ca/gov/content/safety/wildfire-status/about-bcws/wildfirehistory/wildfire-season-summary
Province of British Columbia (2020) Hunting and trapping regulations synopsis 2020 - 2022.
Accessed 15 August 2021. https://www2.gov.bc.ca/gov/content/sportsculture/recreation/fishing-hunting/hunting/regulations-synopsis
Province of British Columbia (2021) Hunting and trapping regulations synopsis corrections.
Accessed 15 August 2021. https://www2.gov.bc.ca/gov/content/sportsculture/recreation/fishing- hunting/hunting/regulations-synopsis/regulation-synopsiscorrections-and-updates

18

Ray AJ (1987) The fur trade in North America: an overview from a historical geographical
perspective. In: Novak M, Baker JA, Obbard ME, Malloch B (eds) Wild Furbearer
Management in North America. Ontario Trappers Association, North Bay, Ontario
Ripple WJ, Beschta RL (2004) Wolves, elk, willows, and trophic cascades in the upper Gallatin
Range of Southwestern Montana, USA. Forest Ecology and Management 200:161-181
Ripple WJ, Estes JA, Beschta RL, Wilmers CC, Ritchie EG, Hebblewhite M, Berger J,
Elmhagen B, Letnic M, Nelson MP, Schmitz OJ, Smith DW, Wallach AD, Wirsing AJ
(2014) Status and ecological effects of the world's largest carnivores. Science
343:1241484
Roemer GW, Gompper ME, Van Valkenburgh B (2009) The Ecological Role of the Mammalian
Mesocarnivore. BioScience 59:165-173
Sivy KJ, Pozzanghera CB, Grace JB, Prugh LR (2017) Fatal Attraction? Intraguild Facilitation
and Suppression among Predators. Am Nat 190:663-679
Steen O, Demarchi DA (1991) Sub-Boreal Pine-Spruce Zone. In: Meidinger DV, Pojar J (eds)
Ecosystems of British Columbia. British Columbia Ministry of Forests Special Report
Series 6, Victoria, Canada
Steen OA, Coupe RA (1997) A field guide to forest site identification and interpretation for the
Cariboo Forest Region, Vol. BC Ministry of Forests, Land Management Handbook No.
39, Victoria, BC, Canada
Weir RD (2003) Status of the fisher in British Columbia. BC Ministry of Sustainable Resource
Management, Conservation Data Centre, and Ministry of Water, Land and Air Protection,
Biodiversity Branch, Victoria, BC No. B-105:38
Weir RD, Corbould FB (2010) Factors Affecting Landscape Occupancy by Fishers in NorthCentral British Columbia. Journal of Wildlife Management 74:405-410
Weir RD, Hoodicoff C (2002) Development of conservation strategies for badgers in the
Thompson & Okanagan regions. 2001-02 annual report. In, Armstrong, BC
Weir RD, Phinney M, Lofroth EC (2012) Big, sick, and rotting: Why tree size, damage, and
decay are important to fisher reproductive habitat. Forest Ecology and Management
265:230-240
White HB, Batcheller GR, Boggess EK, Brown CL, Butfiloski JW, Decker TA, Erb JD, Fall
MW, Hamilton DA, Hiller TL, Hubert GF, Lovallo MJ, Olson JF, Roberts NM (2021)
19

Best Management Practices for Trapping Furbearers in the United States. Wildlife
Monographs 207:3-59
White HB, Decker T, O'Brien MJ, Organ JF, Roberts NM (2015) Trapping and furbearer
management in North American wildlife conservation. Int J Environ Stud 72:756-769
Wolf C, Ripple WJ (2017) Range contractions of the world's large carnivores. R Soc Open Sci
4:170052
Woodroffe R, Ginsberg JR (1998) Edge effects and the extinction of populations inside protected
areas. Science 280:2126-2128
Wright JV (1987) Archaeological evidence for the use of furbearers in North America. In: Novak
M, Baker JA, Obbard ME, Malloch B (eds) Wild Furbearer Management in North
America. Ontario Trappers Association, North Bay, Ontario

20

CHAPTER 2
FACTORS DRIVING POPULATION DENSITIES OF ENDANGERED FISHERS
(PEKANIA PENNANTI) IN TWO GEOGRAPHICALLY DISTINCT REGIONS OF
BRITISH COLUMBIA, CANADA.
Introduction

Accurately and precisely estimating population density is critical for effective
conservation and management of wildlife. Population data are crucial for proper evaluation of
the conservation status of a species (Jimenez et al. 2017), making decisions about deploying
limited resources to protect threatened species (Luo et al. 2020), and setting sustainable harvest
levels (Fuller et al. 2016). However, estimating the density of some carnivores can be
particularly difficult because they often range widely and tend to occur naturally at low densities
(Long et al. 2010, Obbard et al 2010, Krohner 2021). Rare carnivores also are often highly
sensitive to disturbance and at greater risk of extinction, thus being frequently prioritized for
conservation (Travaini et al. 1997, Thompson 2013). Unfortunately, for many threatened and
rare species, detailed population data are lacking and not enough is known to develop
appropriate conservation measures (Duggan et al. 2015, White et al. 2020).
Fishers (Pekania pennanti) are a rare and elusive member of Family Mustelidae
distributed across the boreal and temperate coniferous and mixed-wood forests of North America
(Proulx et al. 2004). In the late 1800s, the range of fishers in North America underwent
significant contractions stemming from habitat loss and fragmentation, overexploitation in the
fur trade, and mortality from predator-control programs (Powell 1993, Krohn et al. 2012, Lewis
et al. 2012). The consequence of this large-scale habitat loss coupled with unregulated fur
harvest was the disappearance of fishers from many regions across the southern extent of their
range by the early 1900s (Lofroth et al. 2010, Lapoint et al. 2015).
Throughout their range, fishers are linked to forests with relatively high overhead cover
(Powell 1993, Weir 2003); although not solely dependent on old-growth coniferous forests, the
abundance of complex vertical and horizontal structure (e.g., large live trees, snags, logs, and
moderate-to-dense-canopy cover) usually found in older temperate forests appears to be an
accurate predictor of fisher habitat use and occupancy (Raley et al. 2012). Like other populations
west of the Rocky Mountains, fishers in British Columbia (BC) depend on structural components
21

typically found in higher densities in late-successional forests to fulfill their life requisites (Raley
et al. 2012, Weir et al. 2012). Structures used for resting and denning (e.g., cavities, platforms,
and other microstructures) are critical habitat features that can be rare across the landscape (Weir
and Harestad 2003, Aubry and Raley 2006, Purcell et al. 2009). In fact, female fishers in BC
appear to have the most stringent habitat requirements for reproduction, with the dens where they
birth and raise their kits being found almost exclusively in the internal cavities of large diameter
trees (Lofroth et al. 2010). Suitable cavities have highly specific dimensions and must be large
enough to accommodate adult female fishers and up to three kits (Weir and Corbould 2008)
while having an opening small enough to exclude predators, including adult male fishers (Raley
et al. 2012). In early April, females use multiple different den trees continuously for the threemonth period following kit parturition (Weir and Corbould 2008); this makes having an adequate
supply of den trees critical for successful reproduction (Green et al. 2019, Berg et al. 2020). The
disease and decay processes that form these vital denning and resting structures can take up to a
century to develop, making fishers highly susceptible to habitat disturbances, such as rotational
forestry, that remove these structures at rates faster than forests can redevelop them (Weir et al.
2012).
Since the mid-1900s, fishers east of the Rocky Mountains recolonized many portions of
their historic range thanks to enhanced protective measures (e.g., the closing of trapping
seasons), translocation efforts, and natural forest succession or reforestation (Bowman et al.
2006, Hapeman et al. 2011, Greenhorn et al. 2018). However, despite some fur harvest bans and
five translocation programs, fishers west of the Rocky Mountains remain geographically
restricted and occur as smaller and more isolated populations in British Columbia (BC), Canada,
and Washington, Oregon, California, Idaho, and Montana (Lewis et al. 2012, Lapoint et al. 2015,
Lewis et al. 2016).
The densities of fishers in parts of BC are among the lowest recorded from anywhere
within their range (e.g., 8.8 fishers/1000 km2; Weir and Corbould 2006), with the highest density
for the province being reported in the northeast (16.7 fishers/1000 km2; Weir et al. 2011). The
density of fishers at the southwestern extent of the species range in the province recently was
estimated at 13.1 fishers/1000 km2 (Davis and Weir 2021). These estimates, however, are much
lower than those documented in eastern North America (49 - 385 fishers/1000 km2; as cited in
Weir et al. 2003) and California (140 fishers/1000 km2; Matthews et al. 2011). The reasons for
22

these differences in the density of fishers in BC remain unclear but could be related to
contemporary habitat loss (Weir 2003), lower prey densities (Weir et al. 2009), an increased cost
of locomotion resulting from deeper snow conditions at higher latitudes (Raine 1983), as well as
other factors such as community predator dynamics of which much is still unknown. Table 2.1
shows density estimates for fishers from other areas of North America.

Table 2.1. Density estimates for fishers from elsewhere in North America, with 95% confidence
intervals (95% CI) included if reported. Density estimates from British Columbia are in bold.

Study Area

Estimation
Method

Density
(Fishers/1000
km2)

Southcentral Maine, USA

Mark-resight

50 - 120

Southern Quebec, Canada
Northcentral Massachusetts,
USA
Northcentral British
Columbia, Canada
Southern Sierra Nevada,
California, USA
Hoopa Valley Indian
Reservation, California, USA
Hoopa Valley Indian
Reservation, California, USA
Northeastern British
Columbia, Canada

Mark-resight

300

Mark-resight
Minimum
number alive

250

267 - 333
Not
reported

8.8

7.7 - 9.9

Fuller et al. (2001)
Weir and Corbould
(2006)

Mark-resight

100

67 - 144

Jordan (2007)

Mark-resight

520

430 - 640

Matthews et al. (2011)

Mark-resight
Minimum
number alive
Spatial capturerecapture
Spatial capturerecapture

140

130 - 160

Matthews et al. (2011)

16.3

11.6 - 21

Weir et al. (2011)

45

20 - 101

Linden et al. (2017)

13.1

6.3 - 27.4

Davis and Weir (2021)

Western New York, USA
Southcentral British
Columbia, Canada

95% CI
Not
reported

Source
Arthur et al. (1989)
Garant and Crete (1997)

Recent research has elucidated there are two distinct populations of fishers in BC - Boreal
and Columbian population, respectively - separated by a high snowpack zone delineated by the
Rocky Mountain divide and the mountainous region south of the Spatsizi Plateau (see Figure 1.2
in Chapter 1). The most recent estimate for the Boreal population is 896 – 1519 adult individuals
(BC CDC 2020a) with this population presumably still being relatively contiguous with other
populations in the boreal forests of Alberta. The most recent estimate for the Columbian
population, however, is exceedingly low at 299 – 517 adult individuals (BC CDC 2020b); given
that fishers have effectively been extirpated from most of southern BC this new research
suggests the Columbian population of fishers is geographically isolated from other populations in
23

North America with demographic rescue through immigration unlikely. In recent decades, the
low-elevation forests these fishers rely on in the central interior of BC have been subject to
extensive disturbance from industrial development (e.g., hydroelectric, oil and gas), ongoing
timber harvesting activities, large-scale insect infestations, extraordinary wildfire seasons, and
the accelerated salvage harvest of insect- and fire-damaged trees (Eng et al. 2005, Province of
BC 2017, Province of BC 2018). Isolation, habitat loss, and the accompanying decline in
population size resulted in the BC government recently revising the conservation status of the
Columbian population to endangered (BC CDC 2020b).
Although recent trends indicate the Columbian population is declining, the true extent of
any declines remains unknown due to inadequate monitoring of fishers at the provincial scale. In
BC, ecosystems are classified based on climate, elevation, and vegetation characteristics and thus
far surveys for fishers have occurred only in the relatively higher productivity ecosystems in the
northcentral (Weir and Corbould 2006) and northeastern portions of the province (Weir et al.
2011), and one smaller scale study at the extreme southwestern periphery of the species
provincial range (Davis and Weir 2021). Additionally, naïve density estimates were generated
for two 400 km2 pilot study areas in moderate to high capability fisher habitat using DNA-based
methods; however, these did not include any spatial information and were based on small sample
sizes (2 and 5 fisher individuals identified; Davis 2004). The range of the Columbian population
in BC encompasses a diversity of ecosystems and because the density of fishers can vary widely
among regions depending on habitat quality (Davis and Weir 2021), detailed inventory
information from different parts of the province is needed to improve the reliability of the
population estimate and identify key factors affecting densities in these areas.
As a result of the patchy nature of these monitoring efforts there still are considerable
gaps in knowledge about the current distribution and abundance of fishers over expansive areas
of BC’s central interior, including where large-scale habitat losses have already occurred.
Without representative estimates of the density of fishers across the range of the Columbian
population, it will be challenging for wildlife managers to accurately assess population trends
and the true extent of the apparent declines being seen in central BC. Therefore, the main
objective of this study was to use non-invasive DNA-based surveys in two representative
ecosystems within the range of the Columbian population of fishers coupled with spatially
explicit capture-recapture methodologies to estimate density and abundance where current data
24

do not exist. Additionally, I investigated habitat factors known from previous research to be
important for predicting fisher density across these regions, thus gaining valuable insight into
current habitat associations specific to these ecosystems. I predicted that the density of fishers
would be positively associated with stands containing more denning and resting structures,
overhead cover, mature forest, riparian features, and higher soil moisture levels, and negatively
associated with open areas (e.g., wetlands, recent burns, young cut blocks). The information
gained from this study can be used by land managers to refine the current population estimate
and identify key habitats where fishers are enduring in the central interior of BC, allowing these
areas to be prioritized for increased protection and other fisher management strategies.

Materials and Methods
Study area
I conducted DNA-based hair-snagging surveys in two study areas within the range of the
Columbian population of fishers. The first was a 2,440 km2 region on the Chilcotin Plateau west
of Williams Lake, where data were collected during winter 2018-19 (hereafter named the
Chilcotin study area; see Figure 1.3 in Chapter 1); the second was located southeast of Williams
Lake (2,580 km²) where data were collected during winter 2020-21 (hereafter named the
Enterprise study area; see Figure 1.4 in Chapter 1). Both study areas consisted of level to gently
rolling topography and surveys were conducted in non-mountainous areas at low to moderate
elevations (750 – 1700 m).
The Chilcotin study area was within the driest ecological zone where fishers are found in
BC with forests dominated by lodgepole pine (Pinus contorta) and a small component of
Douglas-fir (Pseudotsuga menziessi), and trembling aspen (Populus tremuloides), and hybrid
spruce (Picea glauca x engelmannii) concentrated around streams and wetlands (Meidinger and
Pojar 1991). The Enterprise study area also was within the drier central interior of the province
where forests are dominated by either multi-aged Douglas-fir or uniform-aged lodgepole pine
stands, with small trembling aspen and hybrid spruce stands occurring locally (Steen and Coupé
1997). For additional details on my study areas see Chapter 1.
Both study areas have extensive histories of logging and mining, with large numbers of
pine trees also killed by mountain pine beetle (Dendroctonus ponderosae) and subsequently
salvage harvested at accelerated rates during the past two decades (Eng et al. 2005). These
25

industries built and continue to build pervasive road networks in this region, although the mining
industry is less active than it was historically. Western portions of the Enterprise study area were
previously logged using a partial cutting silvicultural system to maintain mule deer (Odocoileus
hemionus) winter range habitat and were selectively logged in attempts to suppress several
patchily-distributed Douglas-fir beetle (Dendoctronus pseudotsugae) infestations.
Approximately 30% of the Chilcotin study area was burnt to varying degrees by the Plateau Fire
and ~3% of the Enterprise study area was burnt by the Gustafsen Lake Fire during 2017, which
at the time was the most severe wildfire season on record for BC (Province of BC 2017).
Registered traplines overlapped both study areas with a legal trapping season for fishers
open from November 1st to February 15th during both survey winters (Province of BC 2020).
Since the year 2000, a total of 142 fishers were trapped from the 9 registered traplines that
overlapped with the Enterprise study area (Province of British Columbia unpubl. data: BC Wild
Fur Harvest Database 2000 - 2019 [accessed 15 July 2022]). Within the same time period, a total
of 30 fishers were trapped from the 6 traplines overlapping the Chilcotin study area (Province of
British Columbia unpubl. data: BC Wild Fur Harvest Database 2000 - 2019 [accessed 15 July
2022]).

Fisher DNA sampling
I conducted DNA-based surveys by remotely collecting hair samples following consistent
sampling methodology in the two study areas. I divided each study area into 20 km2 cells to
approximate the smallest expected size of a female home range in the central interior (Weir et al.
2009). Field crews established sampling sites in each cell using detectors that snagged and
collected hair and follicle tissue but did not restrain animals. Crews deployed detectors following
the design of Foran et al. (1997), with two pieces of wood (2 x 19 x 60 cm) screwed together to
form a triangular ‘cubby’ and both ends left open (Figure 2.1). Field personnel baited cubbies
with a piece of chicken attached via tie wire to the middle of the inside of the cubby, and
smeared a commercial beaver castor (Sharpe’s Beaver Lure)and fisher lure (Sharpe’s Fisher
Lure) mixture on the chicken and a woolen jute string hung on branches next to the cubby to act
as attractants. Cubbies had four pieces (approximately 1.5 x 5 cm) of adhesive-based mouse-trap
paper fastened to the inside (2 at each end) to collect hair from individual fishers when they tried
26

to access the bait inside the cubby. Crews affixed cubbies vertically to the bole of a tree using 4 –
7.5 cm long screws and placed a 19 x 30 cm roof board above the unit to prevent rain or snow
from degrading any hair samples present.
Field crews accessed sampling sites by foot, pick up truck, snowmobile, and helicopter,
and deployed cubbies at sites in the best available habitat (Mowat and Paetkau 2002). Crews
revisited sites at 21-day intervals beginning in mid-December and ending in mid-March for N =
4 sampling sessions each winter. Upon each revisit crews assessed the glue pads and cubby
edges for hair and follicle tissues. Glue pads that had collected hair were covered with plastic
paper and placed in a paper envelope and stored in a dry environment to be sent for processing at
the end of the survey. Field personnel moved sampling sites at least once within a cell (>800 m
from the previous site) during the winter to reduce the possibility of habituation, maximize the
number of spatial recaptures of the same individuals at different locations, and evenly distribute
sampling efforts across the study areas.

Figure 2.1. Typical set up of triangular wooden hair-snagging ‘cubby’ fastened vertically to a tree. Inset
picture shows the inside of the device with bait and adhesive glue strips used to collect fisher hair.

27

I sent all hair samples to Wildlife Genetics International (WGI) in Nelson, BC for
microsatellite genotyping. An additional 7 fisher hair and tissue samples collected from a trapper
operating in the Enterprise study area at the time of the survey were analyzed to determine if
those animals were also detected at our sampling sites. Lab technicians selected the best
available samples for analysis, ideally using clipped roots of 10 guard hairs where possible,
and/or up to 30 whole under-fur hairs if needed to supplement guard hair. Technicians used
QIAGEN DNeasy Blood and Tissue kits (Qiagen, Toronto, ON, Canada) to extract DNA. The
lab identified hair samples to species by partially sequencing the mitochondrial 16S rRNA gene
(Johnson and O’Brien 1997). For samples identified as fishers, technicians determined the
individual identity of each animal using 7 microsatellite markers previously identified in an
earlier study on fishers in BC (Lut604, MP0055, Ma-1, MP0247, Mvis072, MP0144, MP0182;
Weir et al. 2013). For each individual identified by its multi-locus genotype, the lab was able to
determine sex using the ZFX/ZFY/SRY gender marker (Davis and Weir 2021). Error checking
followed Paetkau (2003), proven with other species to consistently produce low error rates
(Kendall et al. 2009).

Spatial capture-recapture analysis
I employed spatially explicit capture-recapture (SECR) methods that used a maximum
likelihood framework (Borchers and Efford 2008, Efford et al. 2009) to estimate the density of
fishers in each study area. This approach uses the detection history of identifiable animals (i.e.,
location and timing of captures) to typically estimate three parameters: density (D), detection
probability (g0), and a spatial parameter σ (Efford et al. 2009). Like many other methods used
for estimating animal abundance, SECR combines a state model to describe variation in the data
due to an ecological process (i.e., how animals are distributed on the landscape), and an
observation model to describe imperfections in the observation of the process (i.e., just because a
fisher is present does not mean it will consistently be detected). In the state model, the
distribution of home range centres in a population initially is treated as a homogeneous Poisson
point process where density (= intensity) is the sole parameter of the process (Borchers and
Efford 2008). An inhomogeneous Poisson process also may be fitted to evaluate the effects that
28

different covariates have on the predicted density (Efford 2021). For the observation model, a
function is used to describe the decline in detection probability (g0) with distance from an
animal’s home range center (Efford 2021). Therefore, g0 is technically defined as the
detectability of an individual at a certain detector if the individual’s activity center was at that
exact location. The spatial scale parameter σ is related to the range size during sampling
occasions and is used along with animal capture histories to estimate range centers of individuals
in the sampling area and the associated area of integration (Borchers and Efford 2008).
Estimation of density, g0, and σ is achieved by numerically maximizing the likelihood with
respect to the parameters of each of the sub-models (Efford et al. 2009).
I used package secr (Efford 2021) in Program R (R Version 4.0.2; R Core Team 2020) to
estimate the density of fishers in the two study areas. I first constructed a ‘habitat mask’, which
is a grid of points used to facilitate computation, to delineate the bounds of the area of integration
(i.e., the total area over which density was estimated), and to store habitat covariates for spatial
models of density. Following Efford (2021), I estimated the root pooled spatial variance to
generate a preliminary and biased estimate of σ and buffered the trap array for each study area by
four times this preliminary σ (10,000 m and 18,000 m buffer widths for the Chilcotin and
Enterprise study areas, respectively). I then evaluated each of these buffer sizes to ensure that the
masks were large enough that animals outside the habitat mask would have a negligible
probability of being detected within the sampling areas so as not to potentially affect density
estimates (Efford 2021). I chose a mask point spacing of 2200 m that fell within the range of
recommended sizes (<1*σ; Efford 2021) and yielded a computationally reasonable number of
mask points for both study areas (991 for the Chilcotin, 1220 for the Enterprise). The Interior
Douglas-fir very dry-mild (IDFxm) BEC zone is not considered viable fisher habitat in BC (Weir
and Almuedo 2010) therefore I excluded any mask points falling in this zone, and those within
large bodies of water, from the final habitat masks.

Detection covariates
The SECR framework allows for both standard and user-defined covariates to be
assigned to all three modelling parameters (Efford 2021), and I first evaluated the effects of
ecologically and behaviourally relevant covariates on g0 and σ to determine the candidate
29

observation model most supported by the data to be included in the second step of the analysis
(i.e., density estimation). For the detection models, I used a half-normal function to model the
shape of decline in detection probability with distance from home range center.
Previous research has indicated that mustelids may become ‘trap-happy’ or ‘trap-shy’
(Royle et al. 2011, Mowat et al. 2019) therefore I included automated behavioural response
covariates (Efford 2021) where parameters may depend on detection at the preceding occasion
(B), there is a step change after first detection (b), site effectiveness changes once any animal is
caught (k), and site effectiveness changes if there was a detection on the preceding occasion (K).
I expected that detection parameters might change over the course of the winter, so I evaluated
this potential effect by including a time covariate with one level for each occasion (t) and a time
trend factor where there is a linear trend over occasions on the link scale (T). Male fisher home
ranges are substantially larger than females in BC (Weir et al. 2009) and both sexes use space
differently: I also therefore included a sex covariate (h2). For the Chilcotin study area, I expected
that burn severity intensity also may influence detection parameters and included it as separate
user-defined covariates (i.e., whether a detector was in an unburned, low burn intensity, medium
burn intensity, or high burn intensity polygon).

Density covariates
I evaluated habitat factors that could affect fisher density by creating several candidate
models containing covariates that previous studies have shown to be related to fisher use (Table
2.2). I used spatial data from the BC Data Catalogue (Province of BC 2021) and the BC Fisher
Habitat Forestry Web module (BC Fisher Habitat Forestry Web Module 2021) to assess the
effect of habitat covariates on fisher density in our study areas (see Appendix A for a complete
list of references and variable descriptions). Because fishers have been shown to select habitat at
multiple scales (Weir and Harestad 2003), I used a moving window analysis to re-scale
covariates using two different buffer sizes around mask points representing the average size of a
female fisher core use area (4.84 km2; Weir 1995) and an average home range size for female
fishers (30 km2, Davis 2009: Weir et al. 2009) in central interior BC. I based these buffer sizes
on the female portion of the population because they appear to have the most specific habitat
requirements in BC (Weir and Corbould 2008, Davis 2009, Lofroth et al. 2010). I calculated the
30

proportion, length, or density of covariates within these buffer areas and these data were
appended to the habitat mask prior to fitting candidate models. I tested for correlation between
habitat covariates using the corrplot function in the R package corrplot (Wei 2021) and did not
include highly correlated covariates in the same model (Spearman’s r Rank >0.6); Dormann et al.
2013).
For both detection and density models, I used Akaike’s Information Criterion corrected
for small samples sizes (AICc) to evaluate the support for each model in a candidate set (see
Appendix B for a full list of model candidate sets) to identify the model that was best supported
by the data (Burnham and Anderson 2002). I first compared detection models from each
candidate set to determine the detection variables best supported by the data. I chose detection
models based on support from the data and knowledge of fisher biology while also considering
model parsimony. I then included this detection model in the second step of the analysis
identifying which density model was best supported by the data. Models within two AICc of each
other were considered equivalent and those >4 AICc units away from the top model were
considered to have negligible support from the data (Burnham and Anderson 2002). I quantified
the strength of evidence for top models using Akaike weights (wi; Burnham and Anderson 2002)
and identified the 95% confidence set of best models where ∑wi ≥0.95.

31

Table 2.2. Variables used in candidate models known from previous research to influence the density of
fishers within the Chilcotin and Enterprise study areas in the central interior of British Columbia, Canada.
I report the variable name, variable description, and whether it was used in the candidate model set for
each study area (C = Chilcotin, E = Enterprise). The area, length, or density of each variable was
calculated for both the 4.84 km2 (core use scale) and the 30 km2 (home range scale) buffer areas around
each habitat mask point. A full description of variables with their applicable reference can be found in
Appendix A.
Variable
Primary branch resting

Description
Area of spruce-leading, secondary, or tertiary stands aged
≥83 years old with crown closure ≥40%

Study Area
C+E

Primary denning

Area of aspen- or cottonwood-leading stands aged ≥135
years old or Douglas-fir-leading stands aged ≥207 years
old with crown closure ≥20%
Area of stands with spruce or aspen content ≥25% and
aged ≥100 years old
Area of stands with total cover ≥50% (≥30% shrub cover
and ≥20% tree cover)
Area of stands with trees dead, needles, twigs, and
understory consumed
High intensity description or, stands with trees dead,
scorched needles remain on trees, understory burned

C+E

Density of streams in km/km2
Total length of all streams in km
Estimated total relative density of digestible energy based
on kcal/g provided by snowshoe hares and squirrels

C+E

Area of stands with subhygric, hygric, or subhydric soil
moisture regimes. Located primarily along wetlands and
streams, contain spruce, and a relatively well-developed
shrub layer
Area of spruce-or aspen-leading stands within 100 m of a
wetland
Area of stands with ≥30% crown closure within 100 m of
wetland
Area of recently logged or burned stands (<12 years old),
wetlands, and agricultural fields
Area of stands >100 years old within 100 m of wetland or
stream

C+E

Primary coarse wood resting
Primary movement
High intensity burn
Medium or high intensity burn
Stream density
Stream length
Prey density
Wet soil moisture regime

Spruce-aspen riparian forest
Riparian closed canopy forest
Open areas
Old riparian forest
Cut blocks harvested <20 years
ago
At-leading stands
Sx-dominated riparian forest
Road density
Active trap line
Mature and old stands
Old Douglas-fir forest

C+E
C+E
C
C

C+E
C+E

C+E
C+E
C+E
C+E
C+E

Area of cut blocks harvested <20 years ago
Area of aspen- or cottonwood-leading stands
Area of spruce-leading stands within 100 m of a stream
or wetland
Density of roads in km/km2
Area of one active registered trap line during the
Enterprise survey
Area of stands >80 years old
Area of Douglas-fir-leading or secondary stands ≥100
years old

32

E
E
C+E
E
E
E

As with other capture-recapture modelling, the SECR framework uses model formulae
that define variation in each parameter as a function of covariates that is linear on a ‘link’ scale
(Efford 2021). Fitting a model provides estimates of its ‘beta coefficients’ which can then be
used to make predictions (Efford 2021). For each model parameter (i.e., D, g0, and σ) I
calculated beta coefficients and their associated 95% confidence intervals (CI) and if the 95% CI
spanned zero we treated those variables as potentially informative but not reliable predictors. I
used the top-ranked density models (which included the top-ranked detection sub-model) and the
region.N function in secr (Efford 2021) to estimate population size of fishers across habitat mask
areas. I used the predictDsurface function included in secr (Efford 2021) which takes the beta
coefficients from the top-ranked density models on the default log-link scale to predict the
density of fishers across habitat mask areas. I then created another habitat mask encompassing
the larger ecologically similar areas immediately adjacent to both study areas and used the same
top-ranked model coefficients to predict the density of fishers at each mask point. I created a
raster layer of the predicted density at each mask point using the raster function in secr (Efford
2021) and developed final maps for a visual representation of the density of fishers across the
larger surrounding ecologically similar areas as predicted by the top models.
I initially ran density models using combined data for both sexes as I had relatively small
sample sizes for the Enterprise study area and combined sex models often produce nearly
identical density estimates (Efford and Mowat 2014, Mowat et al. 2019) For the Chilcotin study
area, I also ran separate models for each sex.

Results
Chilcotin study area
In the Chilcotin study area I identified 48 individual fishers (15 males, 33 females)
through 127 detections across 256 sampling sites over the course of the winter. A total of three
fishers were harvested from a trapline overlapping part of the Chilcotin study area in the winter
of 2018-19, according to the BC Provincial Wild Fur Harvest database (Province of British
Columbia unpubl. data: BC Wild Fur Harvest Database 2018 - 2019 [accessed 15 July 2022]),
however, the individual identity of these animals went unconfirmed as hair samples were not
shared with us by the trapper. The top-ranked detection model (pooled across sexes) included a
33

behavioural effect, where detection probability depended on whether a fisher was detected at a
site on the previous occasion. This best detection model was >16 AICc units lower than the null
model (Table 2.3). This learned effect also was included in the three top-ranked detection models
for females only that were well supported by the data (∆AICc <2.0), and the top model ranked
>19 AICc units lower than the null female-only model (Table 2.3). The top-ranked male-only
detection model included a site learned response where site effectiveness changes once any
animal is caught, however, the model with detection probability and σ fixed also performed well
(∆AICc <1.0; Table 2.3). Although multiple detection models were well-supported by the data
for the combined sex and female-only models, I retained the top-ranked detection models for
future fitting as I felt they were reasonable based on the ecology of the species. I retained the
male-only detection model with detection probability and σ fixed since this minimized the
overall number of model parameters.
For the top-ranked detection model pooling both sexes, detection probability at the home
range center was 0.17 ± 0.039 SE for naïve fishers detected at the trap for the first time, and 0.42
± 0.060 SE if there had been any previous fisher detection at the site. This model also predicted a
ratio of 31 ± 0.067 SE ♂♂ to 69 ± 0.067 ♀♀. Detection probability for the female-only model
followed a similar pattern with the value at the home range center being 0.12 ± 0.045 SE for
naïve fishers and 0.47 ± 0.079 SE if there was a detection on the previous occasion. Detection
probability at the home range center for the male-only model (i.e., null model) was 0.28 ± 0.067
SE. σ values were similar between all three model categories: for the model combining both
sexes σ was 2.75 km ± 0.16 SE, was 2.68 km ± 0.19 SE for the female-only model, and was 2.83
km ± 0.29 SE for the male-only model.
After selecting the detection models, the top-ranked models involving habitat variables
showed consistent patterns at both spatial scales (i.e., core use area and home range), producing
nearly identical density and abundance estimates. Several of the 95% confidence set of candidate
models contained nearly all of the candidate models I considered (Appendix C). Density for the
combined sex model was most strongly related to area of primary branch resting stands at both
the core use area (4.84 km²) and home range area (30 km²) scales (Table 2.4). The top-ranked
combined sex model at the core use scale was 16 times more likely and scored 5.5 AICc units
lower than the null model that predicted no effect of habitat on density, and the top-ranked
combined sex model at the home range scale was 5 times more likely and scored 3.1 AICc units
34

lower than the null model. Several other combined sex models that included other habitat
covariates in addition to area of primary branch resting stands were well-supported by the data,
however, the confidence intervals of the beta coefficients for the second variable in each of these
models spanned zero, suggesting they were potentially informative but no longer a reliable
predictor.

Table 2.3. Model selection table to evaluate detection for fishers in the Chilcotin and Enterprise study
areas in the central interior of British Columbia, Canada. I estimated 3 parameters: density (D), detection
probabiity (g0), and a spatial parameter (sigma). No density covariates were included in this analysis. The
table shows the number of model parameters (K), model log likelhood (logLik), Akaike's Information
Criterion corrected for small sample sizes (AICc), the difference in AICc values (∆AICc), and relative
model weight (wi). Models ≤2.0 AICc units of the top-ranked models are displayed here, with the null
model assuming g0 and sigma are fixed shown for reference.
Study Area

Sex

Chilcotin

Both

Female

Male

Enterprise

Model
g0~B sigma~1
g0~B + T sigma~1
g0~1 sigma~1 (null)

K
5
6
4

logLik
-445.1
-444.8
-454.4

AICc
901.59
903.68
917.79

∆AICc
0.00
2.09
16.21

wi
0.37
0.13
0.00

g0~B sigma~1
g0~B sigma~B
g0~B + T sigma~1
g0~1 sigma~1 (null)

4
5
5
3

-285.5
-284.3
-284.6
-297.2

580.41
580.79
581.44
601.24

0.00
0.38
1.04
20.83

0.40
0.33
0.24
0.00

g0~k sigma~1
g0~bk sigma~1
g0~1 sigma~1 (null)
g0~Low sigma~1
g0~Bk sigma~1

4
4
3
4
4

-154.8
-155.1
-157.1
-155.4
-155.8

321.55
322.23
322.43
322.84
323.52

0.00
0.68
0.88
1.29
1.96

0.22
0.16
0.14
0.12
0.08

Both

g0~1 sigma~B
5
-149.6
313.27
0.00
0.37
g0~1 sigma~1 (null)
4
-152.3
315.08
1.81
0.15
*Detection covariates included whether there was a detection on the previous occasion (B), a trap-specific
behaviour (bk), a trend over trapping occasion (T), site effectiveness changes once any animal is caught
(k), and low intensity burn (Low). A ‘1’ indicates g0 or sigma remained constant.

35

Table 2.4. Model selection table showing the top-ranked density models for fishers in the Chilcotin study
area in the central interior of British Columbia, Canada. I estimated 3 parameters: density (D), detection
probabiity (g0), and a spatial parameter (sigma) at the female fisher core use area scale (4.84 km2) and the
home range scale (30 km2). The table shows the number of model parameters (K), model log likelhood
(logLik), Akaike's Information Criterion corrected for small sample sizes (AICc), the difference in AICc
values (∆AICc), and relative model weight (wi). Models ≤2.0 AICc units of the top-ranked models are
displayed here, along with the null model that assumes all parameters are constant.
Scale

Model

K

logLik

AICc

∆AICc

wi

Core

D ~ Primary branch resting, g0 ~ B, sigma ~1

6

-441.0

896.09

0.00

0.26

use

D ~ Primary branch resting + stream density, g0 ~ B, sigma ~1

7

-440.5

897.83

1.75

0.11

Null

5

-445.1

901.59

5.50

0.02

Home

D ~ Primary branch resting, g0 ~ B, sigma ~1

6

-442.2

898.49

0.00

0.17

range

D ~ Primary denning, g0 ~ B, sigma ~1

6

-442.9

899.91

1.41

0.08

D ~ Primary denning + high intensity burn, g0 ~ B, sigma ~1

7

-441.7

900.22

1.72

0.07

D ~ wet soil moisture, g0 ~ B, sigma ~1

6

-443.2

900.40

1.90

0.07

Null

5

-445.1

901.59

3.09

0.04

D ~ wet soil moisture, g0 ~ B sigma ~ 1

5

-282.7

577.56

0.00

0.24

Sex

Both

Core
use
Female
Home
range

D ~ Primary denning, g0 ~ B sigma ~ 1

5

-283.2

578.70

1.14

0.13

D ~ Sx- or At-leading stands near wetland, g0 ~ B sigma ~ 1

5

-283.4

578.95

1.39

0.12

Null

4

-285.5

580.41

2.85

0.06

D ~ wet soil moisture, g0 ~ B sigma ~ 1

5

-283.2

578.72

0.00

0.22

D ~ Primary denning, g0 ~ B sigma ~ 1

5

-283.9

580.06

1.34

0.11

D ~ Sx- or At-leading stands near wetland, g0 ~ B sigma ~ 1

5

-284.0

580.21

1.49

0.10

Null

4

-285.5

580.41

1.69

0.09

Core

D ~ Primary branch resting, g0 ~ 1 sigma ~ 1

4

-152.8

317.60

0.00

0.44

use

D ~ Primary branch resting + stream density, g0 ~ 1 sigma ~ 1

5

-151.1

318.89

1.29

0.23

Null

3

-157.1

322.43

4.83

0.04

Home

D ~ Primary branch resting, g0 ~ 1 sigma ~ 1

4

-153.6

319.21

0.00

0.42

range

D ~ Primary branch resting + stream density, g0 ~ 1 sigma ~ 1

5

-152.7

322.10

2.89

0.10

Null

3

-157.1

322.43

3.23

0.08

Male

covariates in addition to area of primary branch resting stands were well-supported by the data,
however, the confidence intervals of the beta coefficients for the second variable in each of these
models spanned zero, suggesting they were potentially informative but no longer a reliable
predictor.
Density in the top male-only model also was most strongly related to area of primary
branch resting stands at both the core use (4.84 km²) and home range (30 km²) scales (Table 2.4).
36

The top-ranked male-only model at the core use scale was 11 times more likely and scored 4.8
AICc units lower than the null model that predicted no effect of habitat on density, and the topranked male-only model at the home range scale was 5 times more likely and scored 3.3 AICc
units lower than the null model. The second-ranked male-only model that included area of
primary branch resting stands and length of streams within both the 4.84 km2 and 30 km² buffer
areas also had good support from the data (i.e., within 1.75 AICc units of top model), but the beta
coefficient of the stream covariate had a 95% confidence interval that spanned zero, making it no
longer a reliable predictor.
The top-ranked model from the candidate set for females predicted that density was
related to the area of stands with wet soil moisture regimes (i.e., subhygric, hygric, or subhydric
soils) within both the 4.84 and 30 km² buffer areas, with the models including area of primary
denning stands and spruce- or aspen-leading stands within 100 m of a wetland also being wellsupported (Table 2.4). Correlation analysis indicated these three habitat variables were also
weakly correlated (Spearman’s r Rank <0.4) which aligned with the modelling results. The topranked female-only model at the core use scale was 4 times more likely and scored 2.8 AICc
units lower than the null model that predicted no effect of habitat on density. The top-ranked
female-only model at the home range scale was twice as likely and scored 1.7 AICc units lower
than the null model that predicted no effect of habitat on density.
The top-ranked model for all fishers estimated an average density of 21.3 ± 3.8 SE
fishers/1000 km2, or an abundance of 101.9 ± 18.0 SE fishers, across the entire 4,790 km2
Chilcotin habitat mask area. The top-ranked male-only model estimated an average density of
5.2 ± 1.4 SE male fishers/1000 km2, or an abundance of 25.1 ± 6.7 SE male fishers, whereas the
female-only model estimated an average density of 15.3 ± 4.3 SE female fishers/1000 km2, or an
abundance of 73.2 ± 20.6 SE female fishers. Density in the top-ranked combined sex model was
positively related to the area of branch resting stands within both the 4.84 km2 (Table 2.5; β =
10.9; 95% CI: 5.7 – 16.1) and the 30 km² (Figure 2.2a, Table 2.5; β = 13.8; 95% CI: 4.8 – 22.9)
buffer areas. The top-ranked male-only model also was positively related to the area of branch
resting stands within both the 4.84 km2 (Table 2.5; β = 15.2; 95% CI: 5.7 – 16.1) and the 30 km²
buffer areas (Figure 2.2b, Table 2.5; β = 21.6; 95% CI: 10.2 – 33.0). Density in the female-only
model was positively related to the area of stands with wet soil moisture regimes within both the

37

4.84 km2 (Table 2.5; β = 4.9; 95% CI: 0.9 – 8.8) and the 30 km² buffer areas (Figure 2.2c, Table
2.5; β = 4.9; 95% CI: 0.1 – 9.7).
I used the top-ranked models at the home range scale to predict the density of fishers over
a larger area that was ecologically similar to the survey grid. The combined sex and male-only
models predicted a relatively uniform density over the majority of this zone, with localized areas
of higher densities of fishers overlapping our study area, in the northwest and northeast corners,
and a handful of small areas further south (Figures 2.3 and 2.4). The top-ranked female-only
model also predicted high densities in the study area and in the northwest corner, but also with
higher density pockets more well-distributed across the remainder of the area (Figure 2.5).
Table 2.5. Beta coefficients (β) and associated 95% confidence intervals (CI) for the top-ranked density
models for fishers at both the core use area scale (4.84 km2) and the home range scale (30 km2) in the
Chilcotin and Enterprise study areas in the central interior of British Columbia, Canada.
95% CI
Study
Area

Sex

Scale

Model

β

Lower

Upper

Core Use

D ~ Primary branch resting, g0 ~ B, sigma ~1

10.9

5.7

16.1

Home
Range

D ~ Primary branch resting, g0 ~ B, sigma ~1

13.8

4.8

22.9

Core Use

D ~ wet soil moisture, g0 ~ B sigma ~ 1

4.9

0.9

8.8

Home
Range

D ~ wet soil moisture, g0 ~ B sigma ~ 1

4.9

0.1

9.7

Core Use

D ~ Primary branch resting, g0 ~ k sigma ~ 1

15.1

5.7

16.1

Home
range

D ~ Primary branch resting, g0 ~ k sigma ~ 1

21.6

10.2

33.0

Core Use

D ~ wet soil moisture, g0 ~ 1 sigma ~ 1

15.8

6.8

24.7

Home
Range

D ~ Crown closure >50%, g0 ~ 1 sigma ~ 1

5.1

1.5

8.8

Both

Chilcotin

Female

Male

Enterprise

Both

38

Figure 2.2. Relationship between density of fishers and variables included in the top-ranked models for
the Chilcotin study area. Panel A shows the top-ranked model pooled across sexes, Panel B is for females
only, and Panel C is for males only; dashed lines represent the upper and lower 95% confidence intervals.

39

Figure 2.3. Predicted density of fishers (fishers/1000 km2), pooled across sexes, in Sub-Boreal Pine-Spruce forests in central BC estimated from
spatial capture-recapture analysis of genetically identified fishers sampled during the winter of 2018-19 as predicted by the proportion of
surrounding 30 km2 containing primary branch resting stands. SBPSxc = Sub-Boreal Pine-Spruce very dry cold BEC subzone variant.

40

Figure 2.4. Predicted density of male fishers (fishers/1000 km2) in the Sub-Boreal Pine-Spruce forests in central BC estimated from spatial
capture-recapture analysis of genetically identified fishers sampled during the winter of 2018-19 as predicted by the proportion of surrounding 30
km2 containing primary branch resting stands. SBPSxc = Sub-Boreal Pine-Spruce very dry cold BEC subzone variant.

41

Figure 2.5. Predicted density of female (fishers/1000 km2) in the Sub-Boreal Pine-Spruce forests in central BC estimated from spatial capturerecapture analysis of genetically identified fishers sampled during the winter of 2018-19 as predicted by the proportion of surrounding 30 km2
containing primary branch resting stands. SBPSxc = Sub-Boreal Pine-Spruce very dry cold BEC subzone variant.

42

Enterprise study area

I sampled fishers between December 2020 and March 2021 in the Enterprise study area at
304 unique sampling sites and identified 22 individual fishers (6 males, 16 females) that were
detected 31 times over the course of the winter. One female fisher was confirmed by genetic
analysis to have been killed by a trapper operating in the study area. Five fishers from the study
area were killed by trappers before my survey began, with a total of 9 unique fisher individuals
killed by trappers operating within the study area during the 2020-21 fur harvest season. I was
unable to assess whether 4 other trapped fishers were detected in our sampling because I did not
have samples for genetic identification. Because I could not determine the identity of all
individuals nor the timing and location of the fishers killed by trappers and because they were no
longer part of the population, I did not include these 13 animals in my density calculations.
I documented 21 individuals recaptured at different locations only nine times and
therefore did not attempt to model each sex separately and report here results from models
pooled across sexes. The top-ranked detection model included a behavioural effect where the
spatial scale parameter σ depended on whether a fisher was detected on the preceding occasion
(Table 2.3). The null model which assumed both detection probability and σ remained constant
also performed well (∆AICc <2.0), and I selected this model to retain for future fitting since it
reduced the number of overall model parameters. Detection probability at the home range center
was very low for the Enterprise study area at 0.039 ± 0.0 SE and the spatial parameter σ was
large at 4.67 km ± 0.0 SE. This model also estimated that the population had a male-to-female
ratio of 29 ± 9.9 SE males to 71 ± 9.9 SE females.
I evaluated a similar suite of variables as the Chilcotin study area thought to potentially
influence the density of fishers, with no variation in detection probability and σ (i.e., I included
the null detection model in the second step of the analysis). At both spatial scales I investigated
(i.e., female core use area and home range, 4.84 km2 and 30 km2, respectively) the top four
models contained the same covariates and were well-supported (∆AICc <2.0) but at the home
range scale their ranking order was slightly different than at the core use area scale (Table 2.6).
The top-ranked model at the core use area scale predicted that density was most strongly related
to the area of stands with wetter than mesic soil moisture regimes (i.e., subhygric, hygric, or
subhydric soils - Table 2.6; β = 10.9; 95% CI: 5.7 – 16.1). At the home range scale this model
43

was also well-supported, but the top-ranked model predicted that the density of fishers was
related to the area of stands with crown closure greater than 50%. At the core use scale, the topranked model was 15 times more likely and scored 5.4 AICc units lower than the null model with
no predicted effect of habitat on density. At the home range scale, the top-ranked model was 4
times more likely and scored 2.6 AICc units lower than the null model with no predicted effect of
habitat on density.

Table 2.6. Model selection table showing the top-ranked density models for fishers in the Enterprise
study area in the central interior of British Columbia, Canada. I estimated 3 parameters: density (D),
detection probabiity (g0), and a spatial parameter (sigma) at the female fisher core use area scale (4.84
km2) and the home range scale (30 km2). The table shows the number of model parameters (K), model log
likelhood (logLik), Akaike's Information Criterion corrected for small sample sizes (AICc), the difference
in AICc values (∆AICc), and relative model weight (wi). Models ≤2.0 AICc units of the top-ranked
models are displayed here, along with the null model that assumes all parameters are constant.
Sex

Scale

Model

K

logLik

AICc

∆AICc

wi

Core

D ~ wet soil moisture, g0 ~ 1 sigma ~ 1

5

-147.8

309.64

0.00

0.50

use

Null

4

-152.3

315.08

5.44

0.03

Home

D ~ Crown closure >50%, g0 ~ 1 sigma ~ 1

5

-149.2

312.45

0.00

0.22

range

D ~ Open areas analysis, g0 ~ 1 sigma ~ 1

5

-149.3

312.52

0.06

0.21

D ~ wet soil moisture, g0 ~ 1 sigma ~ 1

5

-149.8

313.60

1.15

0.12

D ~ Primary movement, g0 ~ 1 sigma ~ 1

5

-150.0

314.02

1.57

0.10

Null

4

-152.3

315.08

2.63

0.06

Both

I used the model including stands with wet soil moisture regimes to estimate the density
and abundance of fishers in the Enterprise study area and to predict the density of fishers over
the larger ecologically similar surrounding areas as it had strong support from the data at both
spatial scales. The density estimates generated from this model were largely consistent between
the two scales with an average of 8.9 ± 2.7 SE fishers/1000 km2 estimated at the core use scale
and 8.5 ± 2.7 SE fishers/1000 km2 estimated at the home range scale. These density estimates
translated into abundance estimates of 52.8 ± 16.1 SE fishers and 50.5 ± 15.8 SE fishers at each
spatial scale, respectively, for the Enterprise study area. The density of fishers was positively
associated with the area of stands with wet soil moisture regimes (Figure 2.6, Table 2.5 - β =
13.7; 95% CI: 4.1 – 23.3). This model predicted a relatively low density (<10 fishers/1000 km²)
over most of the extrapolated area, with a few localized areas of higher densities in the center
and along the southern boundary (Figure 2.7).
44

Figure 2.6. Modelled fisher density (pooled across sexes) in stands with wet soil moisture regimes (i.e.,
wetter than mesic) within Interior Douglas-fir forest types of the Enterprise study site in central interior
BC.

45

Figure 2.7. Predicted density of fishers (fishers/1000 km2) in the Interior Douglas-fir forests of central
interior BC estimated from spatial capture-recapture analysis of genetically identified fishers sampled
during the winter of 2020-21 as predicted by the proportion of surrounding 30 km2 containing stands with
soil moisture regimes wetter than mesic. IDFdk3 = Interior Douglas-fir dry cool 3 BEC subzone variant.

46

Discussion
These results affirm the notion that the density of fishers varies markedly within the
Pacific Northwest, but they also suggest populations in this study region are at levels worthy of
concern. For example, the density estimates for the Chilcotin study area were among the highest
reported across the province (cf. 16.3 fishers/1000km2; Weir et al. 2011), whereas the estimates
from the Enterprise study area were among the lowest (cf. 8.8 fishers/1000 km2 - Weir and
Corbould 2006). Yet, these density estimates are substantially lower than that reported from
eastern North America (50 to 327 fishers/1000 km2; Powell and Zielinski 1994) and California
(140 fishers/1000 km2; Matthews et al. 2011). The reason behind these differences in the density
of fishers within and beyond BC is not clear, but have been hypothesized to relate to
mesocarnivore community dynamics (D. Hodder pers. comm.), differences in prey availability
(Weir et al. 2011), and snow conditions (Raine 1983) which differ regionally within BC, as well
as across the eastern and southern portions of the species’ entire range.
Differences in biological, abiotic, and anthropogenic factors may help explain such
notable differences in the density of fishers between the two study areas. Both of these two study
areas encompass zones predicted by Lofroth (2004) to support moderate to high capabilities of
fishers based on climate, vegetation, and prey communities. Recently, however, forests in both
study areas were highly affected by the Mountain Pine Beetle Epidemic which resulted in
massive swaths of pine trees being killed and salvage harvested at unprecedented rates (Eng et al.
2005). Interestingly, nearly one-third of the Chilcotin study area was also burned by the Plateau
Wildfire in 2017 which at the time was the largest wildfire on record for BC (Province of BC
2017) yet fishers were detected making forays into the more heavily burned areas as well as
along the fringes of the fire boundary. I would have expected that having such a large area
recently disturbed by wildfire would have negatively affected the density of fishers in the
Chilcotin study area, however, the number of animals detected was much higher there than in the
less disturbed Enterprise study area.
Perhaps the most direct explanation behind the difference in the density of fishers
estimated for the two study areas involves the relative levels of human-caused mortality. The
only trapline overlapping the Chilcotin study area that documented fisher harvest during the
2018-19 season reported 3 animals being trapped (Province of British Columbia unpubl. data:
BC Wild Fur Harvest Database 2000 - 2019 [accessed 15 July 2022]) but I was unable to
47

confirm whether I had also detected these animals during the survey. At the time of these surveys
in BC, trappers were required to report any fisher trapped in an open season to the government
within 15 days of the season closing date of February 15 (Province of BC 2020) and aside from
the 3 individuals mentioned above, no other fishers were reported from the Chilcotin study area,
suggesting overall trapping pressure was very low at the time of our survey. This aligns with the
findings of Davis (2009) who stated that fishers in the Chilcotin generally experience low levels
of trapping pressure, and is further supported by the low numbers of fishers trapped on traplines
overlapping the Chilcotin study area since 2000 (Province of British Columbia unpubl. data: BC
Wild Fur Harvest Database 2000 - 2019 [accessed 15 July 2022]). Conversely, in the Enterprise
study area I was able to confirm via genetic analysis or timing of capture that 9 unique fisher
individuals were killed by trappers during the winter of my survey. Four other fishers were
reported as harvested by trappers from within the Enterprise study area, however, I was unable to
confirm these animals as unique individuals without hair or tissue samples. All told, this suggests
between 30 and 38% of the fishers that occurred in the Enterprise population prior to the trapping
season were removed via fur harvest, substantially affecting density and abundance estimates for
this study area.
Overall, I found that the density of fishers in the Chilcotin study area was positively
associated with area of primary branch resting habitat. In fact, when the proportion of primary
branch resting stands in the surrounding 30 km2 increased from 0 to 0.05, the density of fishers
was predicted to double from 16 fishers/1000 km2 to 32 fishers/1000 km2. When not actively
hunting or traveling, fishers use protected resting sites to conserve energy, avoid predation,
thermoregulate, and to safely consume prey (Lofroth et al. 2010, Raley et al. 2012). Primary
branch resting habitat in the drier ecological regions of BC consists of older (>83 years), largediameter spruce stands with relatively high amounts (>40 %) of crown closure (Weir and
Almuedo 2010). These results are supported by numerous other studies in BC and western North
America that show fishers use large-diameter spruce trees relatively often for resting (Weir and
Harestad 2003, Weir et al. 2004, Davis 2009). Davis (2009) found the most frequently used
arboreal rest sites in their West Chilcotin study area were in hybrid spruce (Picea glauca x
engelmannii) trees, with fishers primarily resting on platforms created by spruce broom rust
(Chrysomyxa arctostaphyli), on squirrel nests, or on branches. Spruce stands in the Chilcotin
study area typically were concentrated around streams and wetlands (Meidinger and Pojar 1991)
48

that may contain older trees found within this landscape due to historical fire patterns and
logging practices that preserved riparian vegetation. The fact that stream density was directly
related to fisher density in the Chilcotin also may be a reflection of this past timber harvesting
practice. Protecting these older spruce stands from future forest harvest should help ensure
fishers have an adequate amount of high-value resting habitat available on the landscape, which
will in turn help the species continue to persist at higher densities.
The male-only model for the Chilcotin also predicted that density was most strongly
related to primary branch resting habitat, but the top-ranked female-only model indicated the
density of female fishers was best predicted by stands with wet soil moisture regimes (i.e.,
subhygric, hygric, or subhydric soils). Density models including primary denning habitat and
spruce- or aspen-leading stands within 100 m of a wetland also were well-supported by our
female detection data, and the correlation analysis indicated some weak association between
these three variables. Stands with wet soil moisture regimes in the Chilcotin are small and
primarily occur at the fringe of wetlands and along stream channels and are differentiated from
drier stands due to the presence of hybrid spruce, a more well-developed shrub layer, and a lesser
abundance of lichens (Meidinger and Pojar 1991), and these distinctions help explain the
correlation between the variables included in the top three female-only density models. The
Chilcotin plateau is one of the driest regions where fishers are found in BC, and these locally
wetter stand types have been shown to provide higher densities of both denning and resting
structures (Davis 2009) while also supporting higher amounts of important prey species (Davis
2003, Davis 2004). Across the range of fishers, deciduous trees provide important denning
structures critical to reproduction (Powell 1993, Weir and Harestad 2003, Weir 2003) and in the
Chilcotin large diameter trembling aspen are most abundant in these wetter, more productive
stands around riparian features (Davis 2009). Spruce and aspen stands are important predictors of
fisher denning and resting habitats in BC (Weir and Harestad 2003, Davis 2009, Weir et al.
2012), and my modelling results support these other studies by showing the density of female
fishers also is positively associated with these types of habitats. Protecting and reserving these
older, large diameter spruce and aspen stands adjacent to riparian features should help ensure
there are more of the crucial denning and resting structures that fishers require for both survival
and reproduction available on the landscape, thereby increasing the chances the species will
continue to persist at higher densities in the Chilcotin region of BC.
49

These results demonstrate that throughout the drier ecological zones of interior BC,
wetter stand types are associated with higher densities of fishers. Similar to the Chilcotin results,
my top-ranked density models for the Enterprise study area predicted the density of fishers was
strongly related to stands with wet soil moisture regimes. When the proportion of stands with wet
soil moisture regimes in the surrounding 30 km2 area increased from 0.1 to 0.3, the predicted
density of fishers increased from 1 fisher/1000 km2 to 21 fishers/1000 km2. Although the
Enterprise study area is not quite as dry as the Chilcotin, it is also considered one of the drier
ecosystems where fishers are found in BC (BC Fisher Habitat Forestry Web Module 2021).
Stands with these higher soil moisture levels in the Enterprise study area are generally found
near the base of north-facing slopes, and adjacent to streams and wetlands (Meidinger and Pojar
1991). These stands typically have closed forest canopies and primarily consist of a mixture of
hybrid white spruce and Douglas-fir, with greater shrub cover than other drier stands (Hope et al.
1991). Although no specific data exists for the Enterprise study area, given their characteristics
these stands likely provide higher densities of denning and resting structures and prey similar to
their counterparts in other drier ecological zones of interior BC (e.g., the Chilcotin Plateau),
making the positive association I found with higher density of fishers consistent with other areas
of the province. Protecting these high-value riparian stands from disturbance likely is important
to ensure the structures fishers need to fulfill their life requisites remain available in sufficient
quantities on the landscape.
While my model selection results generally supported habitat associations suggested by
previous research, one interesting departure from this trend was the link between fisher densities
and the amount of ‘open’ areas (e.g., wetlands, recent burns, young cutblocks) in one of my two
study areas. Throughout their range, fishers have consistently been tied to areas with overhead
cover, being reported to avoid more open habitats (Powell and Zielinski 1994, Weir 2003),
Lofroth et al. 2010, Raley et al. 2012). In north-central BC, Weir and Corbould (2010) found a
strongly inverse relationship between the probability of a fisher home range being occupied and
increasing amounts of ‘open’ areas within that home range, and an independent study from Idaho
found nearly the exact same relationship (Sauder and Rachlow 2014). When I included ‘open’
areas as a covariate in our density models the relationship between the density of fishers was
only slightly positive in the Enterprise study area, but this model did rank within the top four
from the candidate set. In the Chilcotin study area the ‘open’ areas model ranked last and areas
50

that were burned at moderate or higher intensities also did not appear to be strongly related to
density, although the exact influence these covariates may be having on the density of fishers is
somewhat unclear (95% confidence intervals of beta coefficients spanning zero). The observed
lack of response to recently burned areas may be related to how soon after the fire I surveyed
(i.e., within one year), and habitat-related effects from this disturbance may be lagging behind
and will be more apparent over the longer term in the Chilcotin study area. I hypothesize that in
the drier ecological zones of BC the Weir and Corbould (2010) model is primarily confounded
because the density of fishers here appears to be strongly related to forested stands that are often
found along the fringes of wetlands, and these stands likely also represent the highest value
fisher habitat within these regions.
Extrapolating my density models to areas larger than my study areas but similar
ecologically was informative but also comes with important limitations, and I caution against
using these models to estimate the total abundance of fishers for these larger areas. My trapping
arrays only covered ~20% and ~30% of these larger ecologically similar areas and estimates are
based on only one winter of survey data. Even within the areas immediately adjacent to my
survey grids there is a high amount of ecosystem diversity, and assuming models based on one
year of detection data have obvious limitations, especially for the Enterprise study area where a
limited number of recaptures resulted in less precise density estimates. Alternatively, these
density models built on and supported previous research regarding the habitat associations of
fishers in central interior BC and could be used to identify priority areas for habitat protections,
areas to direct future population monitoring efforts, and to help identify other areas of the
province where fishers may be enduring at higher densities.

Conservation implications

This study provides land managers with important information regarding the distribution
and abundance of fishers within the endangered Columbian population in the central interior of
British Columbia (BC), Canada, and the methods I used can be applied to other small and
isolated populations facing the same cumulative threats of habitat loss, fragmentation, and
degradation, and human-caused mortality. Population density estimates reported for these two
study areas are among the highest and lowest ever documented in BC, which has important
51

implications for the future conservation and management of this population. The difference in
density estimates between the two study areas can largely be attributed to the different levels of
trapping mortality in these areas. Based on my findings, the negative effect of trapping on the
density of fishers is quite clear and unless trapping regulations and methods are adjusted to
substantially reduce this source of mortality it is likely that the density of fishers in the Enterprise
study area will continue to decline and ultimately may never recover. In chapter 3, I explore in
further detail the effects that trapping mortality may have on the persistence of the Columbian
population of fishers. The Chilcotin region of central BC appears to have one of the highest
densities of fishers in the province and may be an important source population should
translocation programs become necessary; therefore, protective measures aimed at ensuring this
population cluster persists into the future should be a priority.
Density estimates from these two study areas were substantially lower than elsewhere in
North America and given the current status of the Columbian population the importance of
maintaining viable clusters of fishers where they continue to persist cannot be understated. The
density of fishers had strong positive associations to similar habitat types in both my study areas
which are identifiable using existing spatial data layers and information from the interior of BC. I
suggest that land managers make use of this existing information and the management levers at
their fingertips to identify and protect these high-value habitats from further industrial
disturbances, improving the chances that fishers will continue to persist in this region.
This study has provided a snapshot look at the state of two fisher population clusters in
representative ecosystems from central BC, but long term and continuous monitoring of these
population clusters and others should be built into the management of the Columbian population
of fishers going forward. Without being able to accurately assess population trends across years
from rigorous and repeatable monitoring at regular intervals, it will be challenging to determine
whether conservation measures being implemented are improving the long-term sustainability of
this endangered population.

52

Literature Cited
Arthur S. M., Krohn W. B., Gilbert J.R. 1989. Home range characteristics of adult fishers.
Journal of Wildlife Management. 53:674-679.
Aubry K. B., Raley C. M. 2006. Ecological characteristics of fishers (Martes pennanti) in the
southern oregon cascade range. Pp. 1-31, USDA Forest Service, Olympia, WA.
Berg S. S., Erb J. D., Spaid V. T., Dewey D. L., Coy P. L., Sampson B. A., Fieberg J. R., Arnold
T. W., Forester J. D. 2020. The role of local cavity tree density in the selection of den
sites by female fishers (Pekania pennanti) in northern minnesota. Canadian Journal of
Forest Research 50:742-750.
Borchers D. L.,Efford M. G. 2008. Spatially explicit maximum likelihood methods for capturerecapture studies. Biometrics 64:377-385.
Bowman J., Donovan D., Rosatte R. C. 2006. Numerical response of fishers to synchronous prey
dynamics. Journal of Mammalogy 87:480-484.
Burnham K.,Anderson D. 2002. Model selection and multimodel inference - a practical
information-theoretic approach, 2nd edition. Springer-Verlag Inc., New York, NY, USA.
Centre B. C. D. 2020a. Conservation status report: Pekania pennanti pop. 4.
https://a100.gov.bc.ca/pub/eswp/. Accessed 2021/06/02 2021.
Centre B. C. D. 2020b. Conservation status report: Pekania pennanti pop. 5. .
https://a100.gov.bc.ca/pub/eswp/. Accessed 2020/11/06 2020.
Davis R. L. 2003. Stand level habitat use by furbearer species in the Anahim Lake area of British
Columbia Pp. 1-47, Yun Ka Whu'ten Holdings Ltd., Williams Lake, BC, Canada.
Davis R. L. 2004. Yun ka whu’ten furbearer track transect project - 2003/2004 summary report.
Pp. 1-51, Yun Ka Whu'ten Holdings Ltd., Williams Lake, BC, Canada.
Davis R. L. 2009. Denning ecology and habitat use by fisher (Martes pennanti) in pine
dominated ecosystems of the Chilcotin Plateau. Master of Science Research, Simon
Fraser University, Burnaby, BC, Canada.
Davis R. L.,Weir R. D. 2021. Density of fishers (Pekania pennanti) at the southwestern edge of
the species' range in British Columbia. Northwestern Naturalist 102:232-238.
Dormann C. F., et al. 2013. Collinearity: A review of methods to deal with it and a simulation
study evaluating their performance. Ecography 36:27-46.
53

Duggan J. M., Eichelberger B. A., Ma S., Lawler J. J., Ziv G. 2017. Informing management of
rare species with an approach combining scenario modeling and spatially explicit risk
assessment. Ecosystem Health and Sustainability 1:1-18.
Efford M. G. 2021. Secr 4.4 - spatially explicit capture-recapture in R.
https://www.otago.ac.nz/density/pdfs/secr-overview.pdf. Accessed July 5, 2021 2021.
Efford M. G., Borchers D. L., Byrom A. E. 2009. Density estimation by spatially explicit
capture-recapture: Likelihood-based methods. In: Thompson DLea, editors. Modeling
demographic processes in marked populations, environmental and ecological statistics.
Springer Science+Business Media, LLC2009.
Efford M. G.,Mowat G. 2014. Compensatory heterogeneity in spatially explicit capture-recapture
data. Ecology 95:1341-1348.
Eng M., Fall A., Hughes J., Shore T., Riel B., Hall P., Walton A. 2005. Provincial level
projection of the current mountain pine beetle outbreak: An overview of the model and
results of year 3 of the project. Natural Resources Canada. Mountain Pine Beetle
Initiative Working Paper.
Foran D. R., Minta S. C., Heinemayer K. S. 1997. DNA-based analysis of hair to identify species
and individuals for population research and monitoring. Wildlife Society Bulletin 25:840847.
Fuller A. K., Linden D. W., Royle J. A. 2016. Management decision making for fisher
populations informed by occupancy modeling. The Journal of Wildlife Management
80:794-802.
Garant Y., Crete M. 1997. Fisher, Martes pennanti, home range characteristics in a high density
untrapped population in southern Quebec. The Canadian Field-Naturalist 111:359-364.
Green R. E., Purcell K. L., Thompson C. M., Kelt D. A., Wittmer H. U. 2019. Microsites and
structures used by fishers (Pekania pennanti) in the southern Sierra Nevada: A
comparison of forest elements used for daily resting relative to reproduction. Forest
Ecology and Management 440:131-146.
Greenhorn J. E., Bowman J., Wilson P. J. 2018. Genetic monitoring suggests increasing structure
following recolonization by fishers. The Journal of Wildlife Management 82:1403-1416.
Group B. F. H. W. 2021. Bc fisher habitat forestry web module. https://www.bcfisherhabitat.ca/.

54

Hapeman P., Latch E. K., Fike J. A., Rhodes O. E., Kilpatrick C. W. 2011. Landscape genetics
of fishers (Martes pennanti) in the northeast: Dispersal barriers and historical influences.
Journal of Heredity 102:1-9.
Jimenez J., Nunez-Arjona J. C., Rueda C., Gonzalez L. M., Garcia-Dominguez F., MunozIgualada J., Lopez-Bao J. V. 2017. Estimating carnivore community structures. Scientific
Reports 7:41036.
Johnson W. E.,O'Brien S. J. 1997. Phylogenetic reconstruction of the Felidae using 16s rRNA
and NADH-5 mitochondrial genes. Journal of Molecular Evolution 44:S98-S116.
Kendall K. C., Stetz J. B., Boulanger J., Macleod A. C., Paetkau D., White G. C. 2009.
Demography of a recovering grizzly bear population. Journal of Wildlife Management
73:3-16.
Koen E. L., Bowman J., Findlay C. S., Zheng L. 2007. Home range and population density of
fishers in eastern Ontario. Journal of Wildlife Management 71:1484-1493.
Krohn W. B. 2012. Distributional changes of American martens and fishers in eastern north
america, 1699-2001. In: Aubry KB, Zielinski WJ, Raphael MG, Proulx G, Buskirk SW,
editors. Biology and conservation of martens, sables, and fishers: A new synthesis.
Cornell University Press, Ithaca, New York, USA.
LaPoint S. D., Belant J. L., Kays R. W. 2015. Mesopredator release facilitates range expansion in
fisher. Animal Conservation 18:50-61.
Lewis J. C., Jenkins K. J., Happe P. J., Manson D. J., McCalmon M. 2016. Landscape-scale
habitat selection by fishers translocated to the Olympic Peninsula of Washington. Forest
Ecology and Management 369:170-183.
Lewis J. C., Powell R. A., Zielinski W. J. 2012. Carnivore translocations and conservation:
Insights from population models and field data for fishers (Martes pennanti). PLoS One
7:1-15.
Lofroth E. C. 2004. Fisher (martes pennanti) british columbia population science review. p. 22.
Lofroth E. C., Higley J. M., Naney R. H., Raley C. M., Yaeger J. S., Livingston S. A., Truex R.
L. 2011. Conservation of fishers (Martes pennanti) in south-central British Columbia,
western Washington, western Oregon, and California. Volume ii: Key findings from
fisher habitat studies in British Columbia, Montana, Idaho, Oregon, and California. USDI
Bureau of Land Management, Denver, Colorado, USA.
55

Lofroth E. C., Raley C. M., Higley J. M., Truex R. L., Yaeger J. S., Lewis J. C., Happe P. J.,
Finley L. L. 2010. Conservation of fishers (Martes pennanti) in south-central British
Columbia, western Washington, western Oregon, and California. Volume i: Conservation
Assessment. USDI Bureau of Land Management, Denver, Colorado, USA.
Long R. A., Donovan T. M., MacKay P., Zielinski W. J., Buzas J. S. 2010. Predicting carnivore
occurrence with noninvasive surveys and occupancy modeling. Landscape Ecology
26:327-340.
Luo G., Wei W., Dai Q., Ran J. 2020. Density estimation of unmarked populations using camera
traps in heterogeneous space. Wildlife Society Bulletin 44:173-181.
Matthews S. M., Mark Higley J., Scott Yaeger J., Fuller T. K. 2011. Density of fishers and the
efficacy of relative abundance indices and small-scale occupancy estimation to detect a
population decline on the Hoopa Valley Indian Reservation, California. Wildlife Society
Bulletin 35:69-75.
Meidinger D. V.,Pojar J. 1991. Ecosystems of British Columbia. British Columbia Ministry of
Forests Special Report Series 6, Victoria, BC, Canada.
Mowat G., Clevenger A. P., Kortello A. D., Hausleitner D., Barrueto M., Smit L., Lamb C.,
DorsEy B., Ott P. K. 2019. The sustainability of wolverine trapping mortality in southern
Canada. The Journal of Wildlife Management 84:213-226.
Mowat G.,Paetkau D. 2002. Estimating marten (Martes americana) population size using hair
capture and genetic tagging. Wildlife Biology 8:201-209.
Obbard M. E., Howe E. J., Kyle C. J. 2010. Empirical comparison of density estimators for large
carnivores. Journal of Applied Ecology 47:76-84.
Paetkau D. 2003. An empirical exploration of data quality in DNA-based population inventories.
Molecular Ecology 12:1375-1387.
Powell R. A. 1993. The fisher - life history, ecology & behavior, 2nd ed. University of
Minnesota Press, Minneapolis, Minnesota, USA.
Powell R. A.,Zielinski W. J. 1994. Fisher. In: Ruggiero LF, Aubry KB, Buskirk SW, Lyon LJ,
Zielinski WJ, editors. The scientific basis for conserving forest carnivores: American
marten, fisher, lynx and wolverine in the western United States.
Proulx G., Aubry K., Birks J., Buskirk S., Fortin C., Frost H., Krohn W., Mayo L., Monakhov
V., Payer D. 2004. World distribution and status of the genus Martes in 2000. In:
56

Harrison DJ, Fuller AK, Proulx G, editors. Martens and fishers (Martes) in human-altered
environments. Springer, New York, New York, USA.
Province of British Columbia. 2017. Wildfire season summary.
https://www2.gov.bc.ca/gov/content/safety/wildfire-status/about-bcws/wildfirehistory/wildfire-season-summary.
Province of British Columbia. 2018. Wildfire season summary 2018.
https://www2.gov.bc.ca/gov/content/safety/wildfire-status/about-bcws/wildfirehistory/wildfire-season-summary.
Province of British Columbia. 2020. Hunting and trapping regulations synopsis 2020 - 2022.
https://www2.gov.bc.ca/gov/content/sports-culture/recreation/fishinghunting/hunting/regulations-synopsis Accessed 15 August 2021.
Province of British Columbia. 2021a. BC Data Catalogue. https://catalogue.data.gov.bc.ca/.
Purcell K. L., Mazzoni A. K., Mori S. R., Boroski B. B. 2009. Resting structures and resting
habitat of fishers in the southern Sierra Nevada, California. Forest Ecology and
Management 258:2696-2706.
Raine R. M. 1983. Winter habitat use and responses to snow cover of fisher (Martes pennanti)
and marten (Martes americana) in southeastern Manitoba. Canadian Journal of Zoology
61:25-34.
Raley C. M., Lofroth E. C., Truex R. L., Yaeger J. S., Higley J. M. 2012. Habitat ecology of
fishers in North America. In: Aubry KB, Zielinski WJ, Raphael MG, Proulx G, Buskirk
SW, editors. Biology and conservation of martens, sables and fishers: A new synthesis.
Cornell University Press.
Royle J. A., Chandler R. B., Gazenski K. D., Graves T. A. 2013. Spatial capture-recapture
models for jointly estimating population density and landscape connectivity. Ecology
94:287-294.
Royle J. A., Magoun A. J., Gardner B., Valkenburg P., Lowell R. E. 2011. Density estimation in
a wolverine population using spatial capture–recapture models. The Journal of Wildlife
Management 75:604-611.
Sauder J.D., Rachlow, J.L. 2014. Forest heterogeneity influences habitat selection by fishers
(Pekania pennanti) within home ranges. Forest Ecology and Management 347:49-56.

57

Steen O. A.,Coupe R. A. 1997. A field guide to forest site identification and interpretation for the
Cariboo Forest Region. BC Ministry of Forests, Land Management Handbook No. 39,
Victoria, BC, Canada.
Thompson W. 2013. Sampling rare or elusive species: Concepts, designs, and techniques for
estimating population parameters. Island Press, Washington, D.C., USA.
Travaini A., Delibes M., Ferrerras P., Palomares F. 1997. Diversity, abundance or rare species as
a target for the conservation of mammalian carnivores: A case study in southern Spain.
Biodiversity and Conservation 6:529-535.
Wei T., Simko V., Levy M., Xie Y., Jin Y., Zemla J. 2017. Package 'corrplot'.
https://peerj.com/articles/9945/Supplemental_Data_S10.pdf. Accessed July 5, 2021 2021.
Weir R. D. 2003. Status of the fisher in British Columbia. BC Ministry of Sustainable Resource
Management, Conservation Data Centre, and Ministry of Water, Land and Air Protection,
Biodiversity Branch, Victoria, BC No. B-105:38.
Weir R. D., Almuedo P. L. 2010. British Columbia's interior: Fisher wildlife decision aid. BC
Journal of Ecosystems & Management 10:35-41.
Weir R. D., Corbould F. B. 2006. Density of fishers in the Sub-boreal Spruce biogeoclimatic
zone of British Columbia. Northwestern Naturalist 87:118-127.
Weir R. D., Corbould F. B. 2008. Ecology of fishers in the sub-boreal forests of north-central
British Columbia, final report. in Peace/Williston Fish and Wildlife Compensation
Program p. 1-178
Weir R. D., Harestad A. H. 2003. Scale-dependent habitat selectivity by fishers in south-central
British Columbia. Journal of Wildlife Management 67:73-82.
Weir R. D., Harestad A. H., Corbould F. B. 2009. Home ranges and spatial organization of
fishers, Martes pennanti, in central British Columbia. Canadian Field-Naturalist 123:126132.
Weir R. D., Lofroth E. C., Phinney M. 2011. Density of fishers in boreal mixed wood forests of
northeastern British Columbia. Northwestern Naturalist 92:65-68.
Weir R. D., Lofroth E. C., Phinney M., Harris L. R. 2013. Spatial and genetic relationships of
fishers in boreal mixed-wood forests of northeastern British Columbia. Northwest
Science 87:114-125.

58

Weir R. D., Phinney M., Lofroth E. C. 2012. Big, sick, and rotting: Why tree size, damage, and
decay are important to fisher reproductive habitat. Forest Ecology and Management
265:230-240.
White N. E., Guzik M. T., Austin A. D., Moore G. I., Humphreys W. F., Alexander J., Bunce M.
2020. Detection of the rare australian endemic blind cave eel (Ophisternon candidum)
with environmental DNA: Implications for threatened species management in
subterranean environments. Hydrobiologia 847:3201-3211.
Yaeger J. S. 2005. Habitat at fisher resting sites in the Klamath province of northern California.
Master of Science, Humboldt State University, Arcata, CA.
Zielinski W. J., Truex R. L., Schmidt G. A., Schlexer F. V., Schmidt K. N., Barrett R. H. 2004.
Resting habitat selection by fishers in California. Journal of Wildlife Management
68:475-492.

59

CHAPTER 3
TRAPPING MORTALITY ACCELERATES THE DECLINE OF THE FISHER
(PEKANIA PENNANTI), AN ENDANGERED MESOCARNIVORE IN BRITISH
COLUMBIA, CANADA.†

Introduction
Carnivores of all sizes play influential roles in regulating ecosystems through trophic
interactions influencing prey abundance, the subsequent effects on vegetation dynamics, and
intraguild competition between larger and smaller predators (Prugh et al. 2009, Ritchie et al.
2012). However, mammalian carnivores worldwide face a multitude of threats including habitat
loss and fragmentation, predator control programs, and unsustainable levels of hunting and
trapping (Prugh et al. 2009, Wolf & Ripple 2017). In North America, several carnivore species
have contracted their ranges over the past two centuries following Euro-American colonization
of the continent and the ensuing development and demand for resources necessary to support an
expanding human population (Laliberte & Ripple 2004). Although some species have reestablished in portions of their historical ranges, many have not and continue to decline in
abundance (Gittleman et al. 2001, Wolf & Ripple 2017). Understanding the key environmental,
demographic, and anthropogenic factors driving the dynamics of endangered carnivore
populations is crucial to ensuring the persistence of these important species over time (Yackulic
et al. 2011, Wolf & Ripple 2017).
Fishers (Pekania pennanti) are a medium-sized member of Family Mustelidae inhabiting
the boreal and temperate forests of North America (Powell 1993) and, like many other
carnivores, the species has experienced significant range contractions since the late 1800s
(Laliberte & Ripple 2004, Lofroth et al. 2010). Following European settlement, fisher range
contracted northward primarily due to habitat loss via commercial timber harvest and agricultural
land clearing, and overexploitation in the fur trade (Lofroth et al. 2010, Lewis et al. 2012,
Lapoint et al. 2015).

†

This chapter was previously published in the journal Endangered Species Research 49 (2022).
60

By the early 1900s, large-scale habitat loss, coupled with high pelt prices and minimal fur
harvest regulations, led to the decimation of many fisher populations across the southern extent
of their range (Lewis et al. 2012). Subsequent protective measures, translocation efforts, and the
reversion of agricultural lands back to forest have enabled fishers to recolonize some of their
historical range in eastern North America, yet fishers continue to exist west of the Rocky
Mountains as smaller and more isolated populations in British Columbia (BC), Canada, and a
handful of western states in the United States of America (Lofroth et al. 2010).
As elsewhere within their range, fishers in BC depend on unique structural elements
typically found in late-successional forests for both survival and reproduction (Raley et al. 2012,
Weir et al. 2012). Structures used for denning and resting are essential habitat features primarily
associated with deformed and decaying live trees and include cavities, platforms, and other
microstructures (Weir & Harestad 2003, Aubry & Raley 2006, Purcell et al. 2009). Because
these structures are rare and can take up to a century to develop, fishers are highly susceptible to
habitat alterations that remove these structures at rates faster than forests can redevelop them
(Weir et al. 2012). Over the past three decades, the low elevation forests of central BC have
undergone widespread habitat change through ongoing forest harvest, large-scale insect
infestations, unprecedented wildfire seasons, and the accelerated rates of salvage logging that
followed these disturbances (Eng et al. 2005, Province of BC 2017, Province of BC 2018b).
Fisher densities in BC are some of the lowest documented from anywhere within their range
(e.g., 8.8 fishers/1000 km2; Weir & Corbould 2006), prompting concerns among wildlife
managers, trappers, stakeholders, First Nations, and conservation groups over the decline of
these animals and their habitat. Moreover, recent research indicates there are two distinct fisher
populations in BC (Boreal population and Columbian population, respectively) separated by a
high snowpack zone encompassing the Rocky Mountain divide and the mountainous region
south of the Spatsizi Plateau (see Figure 1.2 in Chapter 1). Given that fishers have already been
extirpated from most of southern BC, this work suggests that the Columbian fisher population
effectively has been isolated from any other population on the continent.
As a result of patterns in habitat loss and the accompanying decline in population size,
the status of the Columbian fisher population was revised to endangered by the BC government
(Province of BC 2020). This was accompanied by the cessation of commercial fisher trapping
61

within the range of the Columbian fisher population (Province of BC 2021b). However, fishers
continue to be taken as bycatch in traps designed and set for other furbearing species (e.g.,
American marten (Martes americana), 52% of annual fisher mortality; Province of BC 2018a).
Therefore, the objective of this study was to specifically evaluate the effects this mortality from
trapping bycatch may be having on the sustainability of the Columbian fisher population. I used
field-collected data on fisher reproduction and survival from central BC along with data from the
provincial fur harvest database in population viability models to project the population response
to different levels of trapping mortality. I predicted that low reproductive output and high rates of
natural female mortality in Columbian fishers would affect this population’s ability to endure
additional, additive mortality from trapping. The knowledge gained from this study will allow
wildlife managers to craft more effective population management measures to help ensure that
this population persists over the long term.

Materials & Methods
Study area
The range of fishers (Pekania pennanti) within the Columbian population encompasses
forested habitats at low to moderate elevations in central BC (Weir 2003). Northern and central
portions of this range consist of the Sub-Boreal Spruce biogeoclimatic (BEC) zone (Meidinger et
al. 1991), whereas the southerly portion of the range primarily consists of drier ecological zones
including the Sub-Boreal Pine-Spruce (Steen & Demarchi 1997), Montane Spruce (Hope et al.
1991a), and Interior Douglas-fir (Hope et al. 1991b) BEC zones.

Population viability analysis software
I built population models using Vortex version 10.5.5.0 (Lacy & Pollak 2020), a software
program shown to be suitable for populations with low reproductive rates (Lacy 1993, Kim et al.
2016, Winton et al. 2020). Vortex is an individual-based modelling process that simulates
population outcomes by sequentially stepping through a series of events describing the annual
cycle of a typical sexually reproducing organism (Lacy et al. 2020). It incorporates demographic,
environmental, and genetic stochasticity as it follows the fate of individuals from birth to death
62

based on probabilities and user-defined parameters (Lacy et al. 2020). Model outputs included
probability of extinction, mean stochastic population growth rate (r), and mean time to
extinction.

Input parameters
I parameterized the initial population model primarily using the survival and reproduction
information from Lofroth et al. (2022) who analyzed radio-telemetry and other data from the
studies of 60 free-ranging fishers in central BC by Weir (1995), Weir & Corbould (2008), and
Davis (2009), along with published literature from a well-studied fisher population in California
(Table 3.1; Sweitzer et al. 2015). I ran 1000 simulations of each model to predict population
outcomes over a period of 100 years using a one-year time step and did not include density
dependence, catastrophes, inbreeding, or dispersal effects. I set the quasi-extinction threshold as
the point when only one sex remained and considered this to be the point at which the extirpation
of Columbian fishers occurred in the Vortex programming language; hereafter ‘extinction’ refers
to reaching this threshold. I did not assign any correlation between reproduction and survival due
to a lack of empirical data needed to inform this relationship. The current estimate for the
Columbian fisher population is 299 – 517 adult individuals (adult fishers are those animals >2
years old; BC CDC 2020). Therefore, I used an initial population size scaled upwards from the
midpoint of this adult population estimate to include animals in all age cohorts for each
simulation (i.e., initial population size of 571 individuals). I used a proportional age distribution
for the initial population based upon the most recent data available from central interior BC
(Weir & Corbould 2008). I modelled three different age cohorts for each sex: Adults aged >2
years old, subadults aged 1 – 2 years old, and kits aged 0 – 1 years old.
Vortex defines mortality rates in the language of matrix life-table analysis as the
percentage of animals alive at age x that die before reaching age x + 1 (Lacy et al. 2020).
Survival and subsequent mortality rates were calculated following methods detailed in Lofroth et
al. (in press) using monitoring data collected from the three radio-telemetry studies from central
BC. Because these survival calculations included mortalities from trapping, I reran their analyses
excluding fishers that died in traps using the survival package in the R programming language (R
Version 4.0.2; R Core Team 2020). This ensured that only non-trapping sources of mortality
63

Table 3.1. Input parameters and associated data sources for population viability analysis of the
Columbian population of fishers (Pekania pennanti) in British Columbia (BC), Canada. Most parameters
were derived from three radio-telemetry studies on fishers in central BC and one small, well-studied
population in California. Environmental Variation (EV) refers to the annual variation in reproduction and
survival due to random changes in the environment and is modelled using the standard deviation for each
applicable parameter (Lacy et al. 2020).
Parameter

Value (± EV)

Source

Location

Population parameters
Initial Population Size
Carrying Capacity

571
10000

BC CDC (2020)

British Columbia, CA

Reproductive parameters

2
8

Lofroth et al. (2010), Smith et
al. (2020)
Weir (2003), Lofroth et al.
(2010)
Weir (2003), Lofroth et al.
(2010)
Lofroth et al. (2010)

4

Lofroth et al. (2010)

Western North America

1.7 (± 0.69)

British Columbia, CA

50

Lofroth et al. (in press)
Frost & Krohn (1997),
Matthews et al. (2019)

54 (± 41)

Lofroth et al. (in press)

British Columbia, CA

Sweitzer et al. (2015)
Sweitzer et al. (2015)
Lofroth et al. (in press), this
study
Lofroth et al. (2022), this
study
Lofroth et al. (2022), this
study
Lofroth et al. (2022), this
study

California, USA
California, USA

Breeding system

Polygynous

Age at first litter (female)
Age at sexual maturity
(male)
Maximum age of breeding
Maximum number of
progeny per brood
Mean number of kits per
litter
Sex ratio at birth (in %
males)
% Females breeding
annually

2

% Males breeding annually
Mortality rates (%)
Kits (age 0 - 1; female)
Kits (age 0 - 1; male)

100

Subadult (age 1 - 2; female)

40 (± 22)

Subadult (age 1 - 2; male)

14 (± 13)

Adult (age >2 years; female)

21 (± 9)

Adult (age >2 years; male)

10 (± 10)

43 (± 19)
43 (± 19)

Western North America
British Columbia, CA
British Columbia, CA
Western North America

North America

British Columbia, CA
British Columbia, CA
British Columbia, CA
British Columbia, CA

were included in the initial population progression since the effect of annual fur harvest mortality
was simulated in Vortex in a separate step. I used kit mortality rates from birth to the age of 1
from an untrapped, well-studied population in California, as it was the most proximate published
study that included data on fisher kit survival (Sweitzer et al. 2015).

64

I used a polygynous breeding system for our models as both female and male fishers have
been documented breeding with multiple partners in the same year (Smith et al. 2020). I set the
age at first breeding to be 2 years old for both males and females. Although female fishers can
breed at 1 year of age, because they exhibit delayed implantation they will not give birth to kits
until they are at least 2 years old (Lofroth et al. 2010). Male fishers 1-year-old in age can
produce sperm, but it is believed their baculum is not developed enough to cause females to
ovulate until they are two years old (Douglas & Strickland 1987, Frost et al. 1997). I used a sex
ratio of 50:50 at birth (Frost et al. 1997, Matthews et al. 2019). I set the maximum number of
litters per year to be one and maximum litter size to be 4 kits per litter (Paragi et al. 1994, Aubry
& Raley 2006). Not all female fishers successfully breed every year (Powell 1993, Lofroth et al.
2010), and I used the proportion of females breeding annually and average litter size as
calculated from the three radio-telemetry studies from central BC (Lofroth et al. in press). I
assumed that 100% of extant males of reproductive age would be in the breeding pool each year.
I incorporated environmental variation (EV), or the annual variation in reproduction and
survival due to random changes in the environment, by incorporating the standard deviation or
error observed from the empirical data with each demographic parameter in our modelling
inputs. For each annual iteration, fluctuations in the annual probabilities of mortality and
reproduction were modelled as binomial processes based on the mean and standard deviation
specified for each parameter (Lacy et al. 2020).

Model manipulation and sensitivity testing
I initially tested the effect of fur harvest levels on the probability of future extinction
using the annual average number of fishers harvested from the Columbian fisher population
between 2009 and 2017 (mean = 169 ± 37 fishers, or 29.6% of the initial population; Province of
BC 2021a). More recent data were unavailable since there was a time lag between harvest
reporting and entry in the BC Wild Fur Harvest database. Although the commercial fisher
trapping season was recently closed, killing traps certified for trapping other furbearing species
consistently caught a substantial number of fishers every year; therefore, I used provincial fur
harvest data to calculate the proportion of fishers caught in different trap types to inform our
alternative modelling scenarios (Table 3.2; Province of BC 2018a).
65

Table 3.2. The proportion of fishers (Pekania pennanti) harvested from the Columbian population in
British Columbia, Canada by trap class as calculated from the provincial fur harvest database.
Trap
Class
120a
160
220
280
330
Foothold
Snare

Species typically set for:

Overall percentage of fisher harvest

American marten Martes
americana, fisherb Pekania
pennanti

52%

American marten, fisher
Fisher, lynx Lynx canadensis,
bobcat Lynx rufus,
Lynx, bobcat
Lynx, bobcat, wolverine Gulo
gulo
Lynx, bobcat, coyote Canis
latrans, wolf Canis lupus
Lynx, bobcat, coyote, wolf

7%
8%
6%
16%
3%
8%

ª 120-class killing traps are currently the only trap type that can be modified to
specifically exclude fishers.
b

The commercial fisher trapping season was closed in August of 2021.

Both before and after the trapping season closure, most trappers in BC did not specifically set
traps to target fishers but typically set traps for the equally valuable and more abundant
American marten (Martes americana), as either species could be harvested using the same trap
set. As such, 120-class traps (e.g., Belisle SUPER X 120, LDL B120 Magnum), which are
primarily used, and certified, for trapping American martens and occasionally American mink
(Neovison vison), accounted for a substantial proportion of fisher harvest during each trapping
season (52% of fisher harvest; Province of BC 2018a). Fishers also were killed in slightly larger
traps set specifically for them, but which also captured American martens (e.g., 160- and 220class traps; 15% of fisher harvest), as well as 280- and 330-class traps set primarily for Canada
lynx (Lynx canadensis), bobcats (Lynx rufus), and wolverines (Gulo gulo)(22% of fisher
harvest). Leghold traps and snares commonly used for trapping Canada lynx, bobcats, coyotes
(Canis latrans), and grey wolves (Canis lupus) accounted for 11% of fisher harvest each year.
I simulated the effects of varying levels of trapping mortality on population viability
based upon the mortality attributed to these different classes of traps. I considered 4 different
scenarios and modelled the effect of mortality under these scenarios on population persistence
(Table 3.3). In Trapping Scenario 1, the harvest of fishers in any trap type continued unabated
which, because fishers are killed in all trap types certified and set for other furbearing species in
66

BC, I expected to remain very similar to the current annual average harvest rate of 169 fishers
per year (29.6% of the initial population). In trapping Scenario 2, I considered the possibility that
mortality from 120-class traps could be eliminated since this is currently the only trap class that
can be modified to specifically exclude fishers (I-J Hansen pers. comm.) but kills from other
traps continued. Trapping Scenario 2 simulated the harvest of 81 fishers (14.2% of the initial
population) from the population annually and was equal to the annual average number of fishers
trapped in non-120-class traps. In Trapping Scenario 3, I considered the possibility that mortality
from traps previously certified for use on fishers could be removed from the harvest but kills
from other traps continued. Trapping Scenario 3 simulated the harvest of 56 fishers (9.8% of the
initial population) which is the annual average number of fishers caught in traps not certified for
use on fishers (i.e., 280- and 330-class kill traps, foothold, and snare traps). I assumed the
number of fishers harvested each year under these scenarios would decline in concert with
overall population declines, therefore I applied a constant trapping mortality rate using the
percentage of the initial population harvested annually for each of the three trapping scenarios
(e.g., 29.6% of the population was harvested annually for Trapping Scenario 1). The percentage
of animals harvested from each age-sex class was based on the proportional distribution of
carcasses submitted by trappers across BC (Weir 2003). I also considered a No Trapping
Scenario (Scenario 0) where no fishers were removed from the population through fur harvest.
I used sensitivity testing to assess the relative influence of each individual input
parameter on the predicted model outcomes. I simulated the effects of altering model parameters
one at a time while holding all other values constant to better understand the relative sensitivity
of each input parameter on the projected stochastic population growth rate. Given the uncertainty
associated with the current estimate for the Columbian fisher population, I varied the initial
population size in Scenario 0 from 300 to 2000 fishers to evaluate how this factor would change
the predicted fate of the population. The age structures of mustelid populations are thought to be
inherently unstable (Powell 1994), therefore I used different starting age distributions from
published sources (Weir 1995, Weir & Corbould 2008, Buskirk et al. 2012, Lacy & Pollak 2020)
to determine if this would change the projected outcomes. Following similar methodologies used
in other studies on fisher demographics in the western United States (Lewis et al. 2012, Sweitzer
et al. 2015), I then manipulated additional model parameters by ± 10% around initial values to
evaluate changes to stochastic population growth rate. These parameters included male and
67

female mortality rates for each age-sex class, litter size, and the proportion of females breeding
each year, as well as each parameter’s associated value of environmental variation.

Table 3.3. Fur harvest scenarios considered in the projection of population outcomes for fishers (Pekania
pennanti) in the Columbian population, British Columbia, Canada. The percentage of the total population
removed under each scenario was based upon the annual average reported harvest of fishers in the
Columbian population between 2009 and 2017 (shown in brackets) and a projected initial population size
of 571 fishers. We applied a constant trapping mortality rate using this percentage of the initial population
harvested annually to account for the decline in harvest we expected to see each year as the overall
population declines (i.e., as the population declines over time so does the number of fishers harvested).
Trap types that
harvest fishers

Typical target species

Percentage of population
harvestedª

0: No fishers harvested in
any trap type

None

N/A

0% (0 fishers)

1: Fishers harvested in all
trap types (status quo)

All trap types

See list below

29.6% (169 fishers)

2: Fishers harvested in nonmarten sets only

160-, 220-, 280-,
330- class killing,
foothold, and snare
traps

Fisher Pekania pennanti,
lynx Lynx canadensis,
bobcat Lynx rufus,
wolverine Gulo Gulo,
coyote Canis latrans,
wolf Canis lupus

14.2% (81 fishers)

3: Fishers harvested in nonfisher-certified sets only

280-, 330- class
killing, foothold,
and snare traps

lynx Lynx canadensis,
bobcat Lynx rufus,
wolverine Gulo Gulo,
coyote Canis latrans,
wolf Canis lupus

9.8% (56 fishers)

Trapping Scenario

ª Based on an initial population size of 571 fishers.
b

120-class killing traps are currently the only trap type that can be modified to specifically exclude fishers.

Results

No Trapping Scenario (Scenario 0 – no annual fur harvest mortality)
Scenario 0, which used known rates of natural mortality and reproduction from freeranging fishers and included no mortality from fur harvest, predicted that the Columbian fisher
population in the central interior of BC would become extirpated in 36.9 years (SE = 0.43), on
average (Table 3.4). The results of this modelling scenario showed a steadily declining
68

Columbian fisher population with a mean population growth rate of -0.1261 (SE = 0.0015)
across all years, and a probability of extinction of 100% in 100 years.

Table 3.4. Stochastic growth rate, probability of extinction, and mean time to extinction (± standard
error) for the Columbian population of fishers (Pekania pennanti) in British Columbia (BC), Canada
under different levels of trapping mortality. Simulations were run 1000 times over a period of 100 years
using Vortex software (version 10.5.5.0). Both males and females were available for trapping in these
scenarios with an age distribution used from fisher carcass data from BC (Weir 2003). Extinction was
considered when only one sex remained in the population.

Trapping
Scenario

Percentage of
population
trapped annually
(%)

0

0

1

29.6

2

3

Stochastic
population growth
rate (r)

Probability
of
extinction
(%)

Mean time to
extinction (yrs)

-0.13 ± 0.0015

100

36.9 ± 0.43

-0.27 ± 0.0023

100

10.7 ± 0.03

14.2

Average number of
fishers trapped
annually between
2009-2017 in non120 class trapsª

-0.23 ± 0.0023

100

16.5 ± 0.12

9.8

Average number of
fishers trapped
annually between
2009-2017 in nonfisher-certified traps

-0.21 ± 0.0021

100

20.6 ± 0.19

Rationale
No fishers trapped in
any trap types
Average number of
fishers trapped
annually between
2009-2017

ª 120 class traps typically used for American Marten Martes americana can be modified to specifically exclude
fishers - other trap classes currently cannot.

Alternative scenarios
I modelled a suite of alternative population scenarios based on status quo annual fur
harvest rates to evaluate and predict the probability of extinction for the Columbian fisher
population. When fur harvest removed any animals from the population, the probability of
extinction was 100% over 100 years and extinction predicted within 21 years for all three
scenarios that we considered (Figure 3.1; Table 3.4). Generally, the predicted population growth
rate for the Columbian population decreased substantially with increasing levels of fur harvest.
69

Figure 3.1. Influence of alternative levels of mortality from trapping bycatch on the probability of
extinction over 100 years for the Columbian population of fishers (Pekania pennanti) in British
Columbia, Canada, using parameterizations listed in Table 3.1 and Trapping Scenarios in Table 3.3.
Modelled using Vortex software (version 10.5.5.0).

Specifically, under the status quo scenario (Scenario 1), which assumed the average
annual fur harvest rate of 29.6% of the population continued unabated, the population was
predicted to become extinct within 10.7 years (SE = 0.03). Trapping Scenario 2, which assumed
that fishers were excluded and not killed in 120-class American marten (Martes americana) traps
but continued to be caught in all other trap types (48% of annual trapping mortality; Province of
BC 2018a), predicted a population growth rate of -0.2272 (SE = 0.0023) and a mean time to
extinction of 16.5 years (SE = 0.12). Trapping Scenario 3, which assumed that mortality from
traps certified for use on fishers could be removed from the harvest but kills from other traps
70

continued, predicted a population growth rate of -0.2050 (SE = 0.0021) and a mean time to
extinction of 20.6 years (SE = 0.19).

Sensitivity testing
I used sensitivity testing to explore the effects that altering other model parameters had
on the stochastic population growth rate. Age structures of mustelid populations are thought to
be inherently unstable (Powell 1994), however, changing the age distribution used for the initial
population had negligible effects on probability of extinction and mean time to extinction, and a
moderate effect on the population growth rate (range: -0.1513 to -0.1261; Table 3.5). Altering
the initial population size from 300 to 2000 fishers had very little effect on the predicted
population growth rate (range: -0.1336 to -0.1189) or probability of extinction (>99.8% in all
scenarios), however it did result in differences in mean time to extinction (range: 34.2 – 44.3
years; Table 3.6). Population models were most sensitive to adult female mortality followed by
the percentage of females breeding annually, subadult female mortality, and female kit mortality,
in order of decreasing significance (Figure 3.2). Varying male mortality rates and changing the
environmental variation associated with each parameter had little effect on the stochastic
population growth rate. Therefore, although there was some uncertainty associated with the
values used for the environmental variation in each demographic rate, refining these values
would not change the conclusions drawn about the viability of the population.
To further evaluate the effects that varying female mortality would have on the predicted
probability of extinction for the Columbian fisher population, I decreased mortality rates for
female fishers in increments of 2% (i.e., mortality rate decreased from 20% to 18%, 18% to 16%,
etc.) from the initial values used in Scenario 0, with no additional mortality from trapping
(Figure 3.3). Decreasing female mortality rates by 10% resulted in less than half of the 1000
populations simulated going extinct within 100 years. Decreasing female mortality rates by 12%
resulted in a slightly positive stochastic population growth rate and a probability of extinction of
less than 25% within 100 years. When female mortality rates were decreased by 10% and the
percentage of females successfully breeding was increased by 10%, the population growth rate
became positive at 0.0227 (SE = 0.0008) with a probability of extinction less than 6% within 100
years.
71

Table 3.5. Effect of using different age structures on the predicted stochastic population growth rate,
probability of extinction, and mean time to extinction (± standard error) for the Columbian population of
fishers (Pekania pennanti) in British Columbia, Canada, over 100 years. Extinction was considered when
only one sex remained in the population. Modelled using Vortex software (version 10.5.5.0).

Age Distribution

Source

Proportional based on radiotelemetry
research in BC
Proportional based on fisher carcass
data from 1988-1993 in BC
Proportional based on population
model
Stable age distribution (as calculated
by Vortex 10.5.0.0)

Weir & Corbould
(2008)

Stochastic
population
growth rate (r)

Probability
of
extinction
(%)

Mean time to
extinction
(yrs)

-0.13 ± 0.0015

100

36.9 ± 0.43

-0.13 ± 0.0015

100

36.6 ± 0.40

-0.14 ± 0.0015

100

34.4 ± 0.39

-0.15 ± 0.0016

100

31.5 ± 0.39

Weir (1995)
Buskirk et al.
(2012)
Lacy & Pollak
(2020)

Table 3.6. Stochastic population growth rate, probability of extinction, and mean time to extinction (±
standard error) for the Columbian population of fishers (Pekania pennanti) in British Columbia, Canada
over 100 years with different initial population sizes. Extinction was considered when only one sex
remained in the population. Modelled using Vortex software (version 10.5.5.0).

Initial population size
300
400
571ª
700
1000
2000

Stochastic population
growth rate (r)
-0.12 ± 0.0015
-0.12 ± 0.0014
-0.13 ± 0.0015

Probability of
extinction
(%)
100
100
100

Mean time to
extinction (yrs)
34.2 ± 0.41
35.3 ± 0.39
36.9 ± 0.43

-0.13 ± 0.0015
-0.13 ± 0.0014
-0.13 ± 0.0013

99.9
99.9
99.8

38.4 ± 0.42
40.1 ± 0.43
44.3 ± 0.45

ª Approximate midpoint of the current estimate for the Columbian fisher population, British
Columbia, Canada. This was the value used for initial population size in all Trapping
Scenarios.

72

Figure 3.2. The effects of varying input parameters for breeding and mortality rates and their associated
values of Environmental Variation (EV) by ±10% on the stochastic population growth rate for the
Columbian population of fishers (Pekania pennanti) in British Columbia, Canada, under Scenario 0 – no
trapping mortality. The black dashed line is set at the stochastic population growth rate value for Scenario
0. Red indicates the parameter value was decreased by 10% and black indicates the parameter value was
increased by 10%. Each parameter was varied while holding all other parameters constant. Modelled
using Vortex software (version 10.5.5.0).

73

Figure 3.3. The effects of decreasing natural female mortality in increments of 2% on the probability of
extinction over 100 years for the Columbian population of fishers (Pekania pennanti) in British
Columbia, Canada. The red dashed line at 0 represents natural female mortality rates used for initial
population projections, detailed in Table 1, that were derived from three radio-telemetry studies on fishers
in the central interior of BC between 1995 – 2009. These initial rates were subsequently decreased in
increments of 2% to evaluate the effect on probability of extinction, in the absence of any mortality from
trapping. Modelled using Vortex software (version 10.5.5.0).

Discussion
Population simulations indicated that, under current reproductive and natural mortality
rates, the Columbian population of fishers appears unable to sustain any additional mortality in
the form of fur harvest. Because fishers are easily trapped (Powell 1979, Powell 1994), small
populations can be put at further risk by the trapping of other furbearers in these same areas
(Douglas & Strickland 1987, Powell & Zielinski 1994). The commercial trapping season was
discontinued in 2021 within the range of the Columbian population (Province of BC 2021b),
however fishers in central BC continue to be regularly killed through the legal trapping of other
furbearing species every year (Province of BC 2021a). During the two years that the fisher
trapping season was previously closed in BC in the early 1990’s, the number of fishers
74

incidentally caught exceeded the number legally caught the preceding year (Powell & Zielinski
1994), which suggests the closure of the fisher trapping season alone will have minimal benefit
for this population. Unless trapping regulations are changed in central BC to substantially reduce
the number of fishers trapped as bycatch each year, my modelling indicates the Columbian
population will become extirpated from the region in just over a decade.
Closing the commercial trapping season or eliminating the use of body-gripping traps
(i.e., killing traps) for other terrestrial furbearers within Columbian fisher range are likely the
most effective ways to eliminate the significant threat that bycatch mortality represents to the
persistence of this fisher population. In central BC most fishers are caught in 120-class killing
traps designed for American martens (Martes americana)(52%; Province of BC 2018a). In 2019
- 2020, the BC Fisher Habitat Working Group initiated a pilot project to design and construct a
“fisher exclusion box” which can be used to modify 120-class traps to specifically exclude
fishers (I-J Hansen pers. comm.). Legally requiring these “fisher exclusion boxes” to be used by
trappers would eliminate most of the risk of fishers being caught in this class of traps and
decrease considerably the fisher bycatch in central BC while still allowing some level of marten
trapping. However, steadily increasing lumber prices makes the cost of producing these
“exclusion boxes” not insignificant and is currently limiting the voluntary adoption of their use
by trappers in the province (I-J Hansen pers. comm.). Trap classes designed for other larger
furbearing animals, leghold traps, and snares also account for a substantial proportion of the
fisher harvest in central BC (48%; Province of BC 2018a), and at this time these other trap
classes and types do not have modifications designed to exclude fishers. When I simulated a
scenario where only those fishers caught in non-120 class traps were harvested from the
population, the mean time to extinction for the Columbian population increased from 10.7 (status
quo) to 16.5 years. These results further demonstrate that without eliminating the bycatch of
fishers in traps set for other furbearers through increased trapping regulations, it is highly
unlikely that the Columbian fisher population will persist in central BC.
Even when fur harvest mortality was removed, projections still indicated a steadily
declining Columbian fisher population. While mortality from trapping bycatch may be the
proximate threat towards the fisher population in the interior of BC, habitat loss, fragmentation,
and degradation are reported to be the primary drivers behind the long-term population declines
being seen in the province (Weir 2003, Lofroth et al. 2010, BC CDC 2020).
75

Throughout their range fishers depend on structures associated with late-successional
forests for both survival and reproduction (Lofroth et al. 2010), and reproductive output has been
linked to habitat quality (Raley et al. 2012). The dens where fishers birth and raise their kits in
BC are exclusively found in the cavities of large diameter trees (Lofroth et al. 2010, Weir et al.
2012), which are atypical and uncommon on the landscape. Fishers use protected resting sites
when not actively hunting or traveling to conserve energy, avoid predation, for thermoregulatory
purposes, and to consume prey safely (Lofroth et al. 2010, Raley et al. 2012). In BC, fishers rest
in cavities in large diameter trees, on platforms formed in spruce and subalpine fir trees from
abnormal growths caused by spruce broom rust (Chrysomyxa arctostaphyli) or fir broom rust
(Melampsorella caryophyllacearum), and on large branches (Weir 2003, Davis 2009). Forest
management has the greatest potential to negatively impact fisher habitat in BC due to the
prevalence of clear-cut harvesting with short cutting rotations which removes these important
structures at rates faster than they develop under current forest harvesting and secondary
successional timelines (Weir & Corbould 2010, BC CDC 2020). Limiting further disturbance to
high-value fisher denning and resting habitats in central interior BC will also likely be necessary
to promote the recovery of the species in the region.
My population models were highly sensitive to the loss of reproductive-aged females and
the percentage of females successfully producing offspring each year. These results align with
other studies from a small, untrapped, and isolated fisher population in the southern Sierra
Nevada mountains of California (Lamberson 2000, Sweitzer et al. 2015) as well as published
literature on other mammalian carnivores (Hebblewhite et al. 2003, Mills et al. 2018, Hooker et
al. 2020), which all found female demographic response important to their study populations.
Simulations also were sensitive to subadult female and female kit mortality, indicative of an
overall vulnerability of the population to all forms of female mortality. Furthermore, these
modelling results demonstrate that in the absence of any additional mortality from trapping,
moderate decreases of 10% in female mortality and increases of 10% in breeding success would
have considerably positive impacts on the persistence of the Columbian fisher population. Taken
together, the observations that populations are sensitive to female survival and reproductive
success and that the loss of fisher denning and resting habitats is likely detrimental to both are
concerning, and these results suggests that without identifying and protecting areas where fishers

76

are actively reproducing it will be challenging to recover the Columbian fisher population to
sustainable levels over the longer term.
As with all modelling approaches, there are assumptions and limitations to these
predictions. I assumed that reproductive and mortality rates remained constant through time,
which may not be the case. I did not include dispersal effects in our models and our analysis was
aspatial and applied to the entirety of Columbian fisher range. Suitable fisher habitat is highly
fragmented in BC and there are likely negative effects on the population due to increased
dispersal distances for both kits leaving their mothers to establish their own territories, and males
and females seeking breeding opportunities. I did not include any effects due to inbreeding
depression, which would also likely exert detrimental effects on the population if it continued to
decline. Catastrophes such as the large-scale wildfires that are becoming more common were
also not included, and I expect the destruction of such extensive areas of the forested habitats
that fishers require may have further adverse effects on the population. The potential effects that
prey availability may have on fisher populations also went unmodelled. Importantly, it is likely
the factors not included in the modelling (e.g., dispersal effects) would have additional impacts
and not improve the outlook for this fisher population. I provide a more detailed discussion on
the limitations of this model in Chapter 4.

Conclusions
Like many other low-density and wide-ranging carnivores, fishers (Pekania pennanti) in
the Columbian population in central BC face numerous threats, most notably habitat loss and
fragmentation, and human-caused mortality. As a result, this population is declining and without
intervention by management authorities is at risk of extirpation from the region in the very near
future. My modelling strongly suggests that the annual level of mortality from trapping within
the Columbian population of fishers is currently unsustainable and accelerating the declines
being seen in this population. My analysis suggests that in addition to the recent closure of the
trapping season for fishers in central BC, it is likely necessary to modify the trapping seasons
and regulations (including restricting the use of kill traps) for other furbearers found within the
range of the Columbian population to reduce mortality from bycatch to give fishers in this region
a chance to persist long enough to allow for active habitat recruitment. Additionally, identifying
77

areas where fishers are currently successfully breeding and protecting these habitats from further
disturbances will be needed to help increase survival and reproductive rates to levels high
enough to reverse population declines over the longer term. Future research should focus on
investigating other management tools aside from regulating trapping and habitat protection that
also may be effective in helping to increase survival rates and reproductive outputs within the
Columbian population of fishers in BC.

78

Literature Cited
Aubry KB, Raley CM (2006) Ecological Characteristics of Fishers (Martes pennanti) in the
southern Oregon Cascade Range. USDA Forest Service, Olympia, WA
Berg SS, Erb JD, Spaid VT, Dewey DL, Coy PL, Sampson BA, Fieberg JR, Arnold TW,
Forester JD (2020) The role of local cavity tree density in the selection of den sites by
female fishers (Pekania pennanti) in northern Minnesota. Canadian Journal of Forest
Research 50:742-750
Boyce M (1992) Population Viability Analysis. Annual Review of Ecology and Systematics
23:481-506
Buskirk SW, Bowman J, Gilbert JH (2012) Population biology and matrix demographic
modelling of American martens and fishers. In: Aubry KB, Zielinski WJ, Raphael MG,
Proulx G, Buskirk SW (eds) Biology and Conservation of Martens, Sables, and Fishers:
A New Synthesis
BC Conservation Data Centre (2020) Conservation Status Report: Pekania pennanti pop. 5. BC
Ministry of Environment
Davis RL (2009) Denning ecology and habitat use by fisher (Martes pennanti) in pine dominated
ecosystems of the Chilcotin Plateau. Master of Science Research, Simon Fraser
University, Burnaby, BC, Canada
Douglas CW, Strickland MA (1987) Fisher. In: Novak M, Baker JA, Obbard ME, Malloch B
(eds) Wild Furbearer Conservation and Management in North America. Ontario Ministry
of Natural Resources and Ontario Trappers Association, Toronto, ON
Eberhart L, Siniff D (1977) Population dynamics and marine mammals management policies. 3.
Fisheries Res Board Canada 34:183-190
Eng M, Fall A, Hughes J, Shore T, Riel B, Hall P, Walton A (2005) Provincial level projection
of the current mountain pine beetle outbreak: An overview of the model and results of
Year 3 of the project. Natural Resources Canada. Mountain Pine Beetle Initiative
Working Paper
Frost HC, Krohn WB, Wallace CR (1997) Age-specific reproductive characteristics in fishers.
Journal of Mammalogy 78:598-612
Gabriel MW, Woods LW, Wengert GM, Stephenson N, Higley JM, Thompson C, Matthews SM,
Sweitzer RA, Purcell K, Barrett RH, Keller SM, Gaffney P, Jones M, Poppenga R, Foley
79

JE, Brown RN, Clifford DL, Sacks BN (2015) Patterns of Natural and Human-Caused
Mortality Factors of a Rare Forest Carnivore, the Fisher (Pekania pennanti) in California.
PLoS One 10:1-19
Gittleman JL, Funk SM, Macdonald D, Wayne RK (2001) Carnivore Conservation, Cambridge
University Press, Cambridge
Hebblewhite M, Percy M, Serrouya R (2003) Black bear (Ursus americanus) survival and
demography in the Bow Valley of Banff National Park, Alberta. Biological Conservation
112:415-425
Hooker MJ, Chandler RB, Bond BT, Chamberlain MJ (2020) Assessing Population Viability of
Black Bears using Spatial Capture‐Recapture Models. The Journal of Wildlife
Management 84:1100-1113
Hope GD, Mitchell WR, Lloyd DA, Erickson WR, Harper WL, Wikeem BM (1991a) Interior
Douglas-fir Zone. In: Meidinger DV, Pojar J (eds) Ecosystems of British Columbia.
British Columbia Ministry of Forests Special Report Series 6, Victoria, Canada
Hope GD, Mitchell WR, Lloyd DA, Harper WL, Wikeem BM (1991b) Montane Spruce Zone.
In: Meidinger DV, Pojar J (eds) Ecosystems of British Columbia. British Columbia
Ministry of Forests Special Report Series 6, Victoria, Canada
Kim B-J, Lee B-K, Lee H, Jang G-S (2016) Considering threats to population viability of the
endangered Korean long-tailed goral (Naemorhedus caudatus) using VORTEX. Animal
Cells and Systems 20:52-59
Krebs J, Lofroth EC, Copeland JP, Banci V, Cooley D, Golden H, Magoun A, Mulders R, Shults
B (2004) Synthesis of survival rates and causes of mortality in North American
wolverines. The Journal of Wildlife Management 68:493-502
Lacy R (1993) VORTEX: a computer simulation model for population viability analysis.
Wildlife Research 20:45-65
Lacy RC, Miller PS, Traylor-Holzer K (2020) Vortex 10 User's Manual. 1 April update. IUCN
SSC Conservation Planning Specialist Group, and Chicago Zoological Society, Apple
Valley, Minnesota, USA
Lacy RC, Pollak JP (2020) Vortex: A Stochastic Simulation of the Extinction Process. Version
10.5.5.0. Chicago Zoological Society, Brookfield, Illinois, USA

80

Laliberte AS, Ripple WJ (2004) Range contractions of the North American carnivores and
ungulates. Bioscience 54:123-138
Lamberson RH, Truex RL, Zielinski WJ, McFarlane D (2000) Preliminary analysis of fisher
populaiton viability in the S. Sierra NV. USDA Forest Service, Arcata, CA
LaPoint SD, Belant JL, Kays RW (2015) Mesopredator release facilitates range expansion in
fisher. Animal Conservation 18:50-61
Lewis JC, Powell RA, Zielinski WJ (2012) Carnivore translocations and conservation: insights
from population models and field data for fishers (Martes pennanti). PLoS One 7:1-15
Lofroth EC, Raley CM, Higley JM, Truex RL, Yaeger JS, Lewis JC, Happe PJ, Finley LL (2010)
Conservation of Fishers (Martes pennanti) in south-central British Columbia, Western
Washington, Western Oregon, and California-Volume I: Conservation Assessment. In:
Lofroth EC, Raley CM (eds), Book I. USDI Bureau of Land Management, Denver, CO,
USA
Matthews SM, Green DS, Higley JM, Rennie KM, Kelsey CM, Green RE (2019) Reproductive
den selection and its consequences for fisher neonates, a cavity-obligate mustelid. Journal
of Mammalogy 100:1305-1316
Meidinger DV, Pojar J, Harper WL (1991) Sub-Boreal Spruce Zone. In: Meidinger DV, Pojar J
(eds) Ecosystems of British Columbia. British Columbia Ministry of Forests Special
Report Series 6, Victoria, Canada
Mills LS, Hebblewhite M, Eacker DR (2018) Bayesian population viability analysis for lynx and
wolverine in Scandinavia. Book Report 6793. Swedish Environmental Protection Agency
Paragi TF, Krohn WB, Arthur SM (1994) Using estimates of fisher recruitment and survival to
evaluate population trends. Northeast Wildlife 51:1-11
Pollock KH, Winterstein SR, Bunck CM, Curtis PD (1989) Survival analysis in telemtry studies:
The staggered entry design. Journal of Wildlife Management 53:7-15
Powell RA (1993) The fisher - life history, ecology & behavior, University of Minnesota Press
Powell RA (1994) Structure and spacing of Martes populations. In: Buskirk SW, Harestad AS,
Raphael MG, Powell RA (eds) Martens, Sables, and Fishers: Biology and Conservation.
Cornell University Press, Ithaca, NY

81

Province of British Columbia (2017) Wildfire Season Summary.
https://www2.gov.bc.ca/gov/content/safety/wildfire-status/about-bcws/wildfirehistory/wildfire-season-summary
Province of British Columbia (2018a) Compulsory Reporting 2013 -2018. Accessed 15 July
2021.
Province of British Columbia (2018b) Wildfire Season Summary 2018.
https://www2.gov.bc.ca/gov/content/safety/wildfire-status/about-bcws/wildfirehistory/wildfire-season-summary
Province of British Columbia (2020) Hunting and trapping regulations synopsis 2020 - 2022.
Accessed 15 August 2021. https://www2.gov.bc.ca/gov/content/sportsculture/recreation/fishing-hunting/hunting/regulations-synopsis
Province of British Columbia (2021b) Hunting and trapping regulations synopsis corrections.
Accessed 15 August 2021. https://www2.gov.bc.ca/gov/content/sportsculture/recreation/fishing-hunting/hunting/regulations-synopsis/regulation-synopsiscorrections-and-updates
Prugh LR, Stoner CJ, Epps CW, Bean WT, Ripple WJ, Laliberte AS, Brashares JS (2009) The
Rise of the Mesopredator. BioScience 59:779-791
Purcell KL, Mazzoni AK, Mori SR, Boroski BB (2009) Resting structures and resting habitat of
fishers in the southern Sierra Nevada, California. Forest Ecology and Management
258:2696-2706
R Core Team (2020) R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria
Raley CM, Lofroth EC, Truex RL, Yaeger JS, Higley JM (2012) Habitat Ecology of Fishers in
North America. In: Aubry KB, Zielinski WJ, Raphael MG, Proulx G, Buskirk SW (eds)
Biology and Conservation of Martens, Sables and Fishers: A New Synthesis. Cornell
University Press
Ritchie EG, Elmhagen B, Glen AS, Letnic M, Ludwig G, McDonald RA (2012) Ecosystem
restoration with teeth: what role for predators? Trends Ecol Evol 27:265-271
Sauder JD, Rachlow JL (2014) Both forest composition and configuration influence landscapescale habitat selection by fishers (Pekania pennanti) in mixed coniferous forests of the
Northern Rocky Mountains. Forest Ecology and Management 314:75-84
82

Shaffer ML (1981) Minimum population sizes for species conservation. Bioscience 31:131-134
Smith KP, Facka AN, Powell RA (2020) Using reintroduction to elucidate breeding system,
breeding phenology, and infanticide for a solitary carnivore (fisher, Pekania pennanti).
Canadian Journal of Zoology 98:433-447
Steen O, Demarchi DA (1991) Sub-Boreal Pine-Spruce Zone. In: Meidinger DV, Pojar J (eds)
Ecosystems of British Columbia. British Columbia Ministry of Forests Special Report
Series 6, Victoria, Canada
Sweitzer RA, Popescu VD, Barrett RH, Purcell KL, Thompson CM (2015) Reproduction,
abundance, and population growth for a fisher (Pekania pennanti) population in the
Sierra National Forest, California. Journal of Mammalogy 96:772-790
Sweitzer RA, Popescu VD, Thompson CM, Purcell KL, Barrett RH, Wengert GM, Gabriel MW,
Woods LW (2016) Mortality risks and limits to population growth of fishers. Journal of
Wildlife Management 80:438-451
Weir RD (1995) Diet, Spatial organization and habitat relationships of fishers in south-central
British Columbia. Master of Science, Simon Fraser University, Burnaby, BC
Weir RD (2003) Status of the fisher in British Columbia. BC Ministry of Sustainable Resource
Management, Conservation Data Centre, and Ministry of Water, Land and Air Protection,
Biodiversity Branch, Victoria, BC No. B-105:38
Weir RD, Corbould FB (2006) Density of fishers in the Sub-Boreal Spruce biogeoclimatic zone
of British Columbia. Northwestern Naturalist 87:118-127
Weir RD, Corbould FB (2008) Ecology of fishers in the Sub-boreal forests of north-central
British Columbia, Final Report. Peace/Williston Fish and Wildlife Compensation
Program, Vol. 315
Weir RD, Corbould FB (2010) Factors Affecting Landscape Occupancy by Fishers in NorthCentral British Columbia. Journal of Wildlife Management 74:405-410
Weir RD, Corbould FB, Harestad AH (2005) Effect of ambient temperatures on the selection of
rest structures by fishers. In: Martens and Fishers (Martes) in human-altered
environments. Springer, Boston, MA
Weir RD, Harestad AH (2003) Scale-dependent habitat selectivity by fishers in south-central
British Columbia. Journal of Wildlife Management 67:73-82

83

Weir RD, Harestad AH, Corbould FB (2009) Home ranges and spatial organization of fishers,
Martes pennanti, in central British Columbia. Canadian Field-Naturalist 123:126-132
Weir RD, Phinney M, Lofroth EC (2012) Big, sick, and rotting: Why tree size, damage, and
decay are important to fisher reproductive habitat. Forest Ecology and Management
265:230-240
Winton SA, Bishop CA, Larsen KW (2020) When protected areas are not enough: low-traffic
roads projected to cause a decline in a northern viper population. Endangered Species
Research 41:131-139
Wolf C, Ripple WJ (2017) Range contractions of the world's large carnivores. R Soc Open Sci
4:170052
Yackulic CB, Sanderson EW, Uriarte M (2011) Anthropogenic and environmental drivers of
modern range loss in large mammals. Proc Natl Acad Sci U S A 108:4024-4029
Zielinski WJ, Dunk JR, Yaeger JS, LaPlante DW (2010) Developing and testing a landscapescale habitat suitability model for fisher (Martes pennanti) in forests of interior northern
California. Forest Ecology and Management 260:1579-1591

84

CHAPTER 4
CONCLUSION
Summary of Thesis
The overall objective of this thesis was to investigate the environmental, demographic,
and anthropogenic factors influencing the decline of an isolated and endangered carnivore
population to provide science-based information for conservation and management. More
specifically, I intended to fill the considerable knowledge gaps regarding the current distribution
and abundance of the Columbian population of fishers in central BC, and assess the effect that
continuing trapping mortality will have on its sustainability over the long term. To achieve this, I
(1) used information from non-invasive DNA-based surveys at two study sites within the range
of the Columbian population of fishers to estimate density and abundance where current data are
lacking, (2) investigated which anthropogenic and environmental factors best predicted fisher
density across these areas to gain insight into current habitat associations within these
ecosystems, and (3) conducted a Population Viability Analysis (PVA) to assess the impacts that
trapping mortality would have on the persistence of the Columbian population of fishers under
alternative management scenarios.

The primary findings from my thesis were:
•

The density of fishers in the Chilcotin and the Enterprise study areas differed markedly.
The former was among the highest reported for BC at ~21 fishers/1000 km2, whereas the
estimate for the latter was among the lowest at ~9 fishers/1000 km2. Both of these density
estimates are substantially lower than those reported for fishers in other areas of North
America.

•

Trappers harvested 13 fishers from the Enterprise study area during the time of the
survey. Although I was unable to confirm the individual identity of 4 of these animals,
this catch suggested 30-38% of the fishers present within the Enterprise study area were
removed via fur harvest having a significant impact on the density and abundance
estimates of the animals in this study area.
85

•

Covariates that best predicted the density of fishers were similar between the two study
areas. Older, large-diameter spruce stands with high amounts of crown closure best
predicted the density of fishers in the Chilcotin study area, and stands with wet soil
moisture regimes best predicted the density of fishers in the Enterprise study area. Both
stand types are typically found adjacent to streams and wetlands, and these stands are
known to contain higher densities of the critical habitat features fishers require for
denning and resting, as well as higher numbers of available prey species.

•

Population modelling suggested that if current fur harvest mortality rates continue the
Columbian population of fishers may become extirpated within 11 years. When even
greatly reduced levels of fur harvest mortality were included in the models, extirpation of
this population still was predicted within 21 years.

•

When fur harvest mortality was removed from my modelling, the Columbian population
appeared unlikely to persist beyond 37 years. This suggests that other factors such as
habitat loss, fragmentation, and degradation also are contributing to the negative
population growth rate predicted by the models.

•

Population growth rates were most sensitive to the loss of reproductive-aged females and
the percentage of females successfully producing offspring each year. In the absence of
additional fur harvest mortality, modest increases of 10% in both female survival and
breeding success resulted in a positive population growth rate, suggesting that identifying
and protecting critical denning and resting habitat also may be worth considering to help
reverse population declines.

Overall, these findings demonstrate the density of fishers in the central interior of BC is
highly variable among regions but continues to be among the lowest from anywhere within the
species’ range. Given the current status of the Columbian population, the importance of
identifying and protecting existing clusters of fishers that are actively reproducing should not be
understated. These results also support the idea that fishers in central BC are associated with
older spruce stands and increasing amounts of overhead cover, and riparian forests adjacent to
streams and wetlands. Protecting these stand types from further disturbance will likely be
important to ensure the structures fishers require to fulfill their life requisites remain available in
sufficient numbers on the landscape. As with other small and isolated populations, my analysis
86

showed that the Columbian population of fishers is highly susceptible to the removal of even a
few individuals, particularly reproductive-aged females. Not only did fur harvest mortality result
in substantial differences in density estimates between my two study areas, but population
modelling showed that if fur harvest rates of fishers continue at current levels this population
will likely disappear from the region in just over a decade. This study was based on the most
current and accurate population data for fishers in central BC, and as such these results have
significant implications for the conservation and persistence of the species in the region.

Management Recommendations
My study provides important information regarding the conservation of fishers in central
interior BC, and the methods I used can be applied to other small and isolated populations facing
the same cumulative threats of habitat loss, fragmentation, and degradation, and human-caused
mortality. During its most recent status assessment, the BC Conservation Data Centre identified
habitat loss due to forest harvesting as the primary threat to the Columbian population of fishers
in central BC (BC CDC 2020). Secondary threats include loss or degradation of habitat from
agriculture, transportation, and wildfire, and trapping mortality where habitats have been
compromised (BC CDC 2020). Although the International Union for the Conservation of Nature
(IUCN) lists the status of fishers as “Least Concern” globally (Helgen and Reid 2018), this
organization still recommends as the primary conservation measure for the species ‘to prevent
excessive harvest’ (Helgen and Reid 2018). Here I present management recommendations based
on the findings of this thesis, along with historical information, to provide decision-makers with
ways to reduce or eliminate these threats to the Columbian population of fishers in central BC to
help prevent extirpation of the species from the region in the very near future.

Population Management
Until very recently, the monitoring of fisher populations in British Columbia has been
sporadic at best. Fur harvest records have been used the longest as a surrogate for rigorous
population monitoring, however, trends in harvest rarely reflect actual population trends and
estimates of population size because of differences in trapper effort and pelt prices (Banci 1989,
Banci and Proulx 1999). Population estimates for fishers in BC have largely been based on two
87

radio-telemetry studies, one in the Williston region of north-central BC (8.8 fishers/1000 km2;
Weir and Corbould 2006) and one northeast of the Rocky Mountains (16.3 fishers/1000 km2;
Weir et al. 2011). Estimates of fisher abundance for the province were derived using a
combination of these estimates, along with habitat capability mapping (Lofroth 2004). This study
provides much needed current density and abundance estimates for fishers in two distinct
ecosystems in central BC, however, the lack of long-term monitoring data over the entire range
of the Columbian population makes it challenging to accurately determine the true magnitude of
fisher population declines in central BC. To address this issue going forward, I recommend:
•

Updating the current population estimate for the Columbian population using the density
and abundance estimates from this study.

•

Repeating the survey work for both the Chilcotin and Enterprise study areas within the
next 5 years to begin to build the dataset necessary to accurately determine trends in
fisher densities in these two regions.

•

Developing rigorous and repeatable monitoring plans for fishers throughout the province
and implementing them at regular intervals (e.g., every 5 years) across different regions.
These may include DNA-based surveys using methods similar to this study, camera
trapping surveys, or further radio-telemetry work to help focus management efforts on
those population clusters in most need of immediate action.

Habitat Management
Fishers have been described as one of the most habitat-specialized mammals in North
America (Buskirk and Powell 1994), relying on unique and uncommon structures in forests for
both survival and reproduction (Raley et al. 2012). The processes that form these structures can
take centuries to develop them, therefore fishers are highly susceptible to forest disturbances
such as rotational logging and wildfire (Lofroth et al. 2010, BC CDC 2020). The results of this
study showed that the density of fishers was most strongly related to stand types containing
higher densities of the critical denning and resting structures required to fulfill their life
requisites. These stand types have been spatially mapped across the province and are freely
available to any individual via the BC Data Catalogue (Province of BC 2022). The BC Fisher
Habitat Forestry Web Module also provides a free GIS-based tool that allows users to input
proposed forest harvest unit shapes which it uses to calculates the density of denning and resting
88

stands that remain in the surrounding landscape. Based on this information, users are given the
number of denning and resting features to retain within their cut block, as well as the number of
coarse woody debris piles and logs to create. When the surrounding landscape is depauperate in
these crucial denning and resting stands, the user is given a warning to avoid the harvest of
certain stands. Currently this guidance is only legally required to be used by forest licensees in
the Cariboo Region, but with flexibility to be implemented such that a harvest warning does not
preclude that polygon from still being harvested. In other words, currently forest licensees and
others proposing timber harvest still can harvest stands with high numbers of denning and resting
structures regardless of how many might remain in the surrounding areas. To ensure that these
high-value denning and resting stands remain available in sufficient quantities on the landscape, I
recommend:
•

Forest licensees and others proposing timber harvesting activities in all regions of the
province where fishers are found should be legally required to use the GIS-based tool
prior to harvest and adhere to the guidelines for habitat retention as specified.

•

When proposed cut blocks contain areas with a harvest warning these polygons should
be ground-truthed to confirm they contain denning and resting structures that fishers
require for survival and reproduction. If these polygons have the appropriate
characteristics, they should be removed from the cut block and reserved from harvest
until the surrounding forests in the landscape can provide the number of denning and
resting stands necessary for a fisher to occupy that area.

Harvest Management
Furbearing carnivores throughout Canada and the United States, including British Columbia,
continue to be killed every year through targeted and/or incidental trapping. While some species
appear to be resilient to higher levels of fur harvest mortality (e.g., coyotes, red fox; Banci and
Proulx 1999), others are not and the effect of trapping on these species has the potential to cause
population declines and extirpations in a short period of time (Harrington et al. 2017). In his
extensive review of mammal trapping, Proulx (2021) outlines 3 conditions that should always be
met to allow fur trapping to occur: 1) when it employs species-specific trapping systems, 2)
when it will not impact species at risk, and 3) when it uses only trapping systems that meet the
highest standards of animal welfare. In the case of fishers and other furbearing species in BC,
89

none of these three conditions are met. Traps certified for use in this province routinely catch
multiple different species (see Chapter 3), and not one trapping system is truly species-specific.
Given that there are multiple species at risk in BC that are also furbearers (e.g., wolverine,
fisher), current circumstances in this province simply do not justify fur trapping until systems
can be improved to specifically target individual species. California has completely banned
commercial and recreational fur trapping for animal pelts aside from rodents (Newsome 2019),
and Washington no longer allows the use of foot-hold traps, body-gripping traps (including all
conibear types) and snares, except by special permit (Washington Department of Fish and Game
2022). It would not be unreasonable for the government of BC to implement similar restrictions
until trapping systems can be employed that specifically exclude those species with low
resiliency to trapping. With this in mind, I recommend:
•

The trapping season be closed for all terrestrial furbearing species (aside from rodents)
within the range of the Columbian population of fishers until the population has
recovered to levels where a harvest season does not result in population declines.

•

Notwithstanding the above, 120-class traps could continue to be used to trap mink and
marten if they have been retrofitted to specifically exclude fishers as per the
specifications and designs provided by the BC Fisher Habitat Working Group.

•

The trapping season should remain closed within the range of the Columbian population
until trapping systems that are truly species-specific have been designed and approved
for use such that only species with a high resiliency to trapping mortality may be
captured.

Limitations and Future Research
My study provided much needed information on the current density and abundance of
fishers in two geographically distinct ecosystems in central interior BC, and identified which
habitat types best predicted the density of fishers in those areas. However, given the variability
within these ecosystems and that estimates are based on one year of surveys for each study site, I
caution against extrapolating the density models over larger areas until more detection data can
be collected. Additionally, the low recapture rates in the Enterprise study site prevented me from
determining whether there were sex-based differences affecting male and female fisher densities,
which is certainly a possibility. Future research should focus on not only re-surveying these same
90

study areas but also surveying additional areas within the same ecosystem types to help refine
the predictive density models to make them applicable to a larger area.
Population Viability Analysis (PVA) is a powerful tool and, when used appropriately, can
provide critical information to inform the management of endangered species. My PVA for the
Columbian population of fishers utilized survival and reproduction data from studies within
central interior BC as well as a small and isolated population of fishers in California, and
therefore was informed by the best available empirical information. However, this work also
highlights the fact that much of what is known about fishers in BC is based on a few localized
studies with limited sample sizes. For example, the litter sizes I used in this model were based on
one study area along with data from an artificial den box study (Davis unpubl. data), which may
not be representative for the entire Columbian population. This work has shown there are
substantial differences in the density of fishers in different regions of BC, and future research
should focus on determining whether there are specific differences in the survival and
reproductive output of fishers on a regional basis.
The effect that trapping mortality is having on this population appears to be quite
significant, however, there were assumptions that needed to be made due to a lack of data about
certain aspects of the population including dispersal rates and inbreeding depression. I also did
not include a spatial component to my analysis and assumed that demographic rates were
consistent throughout the range of the Columbian population. An updated version of this model
that includes a spatial component is currently being worked on (Burgar unpubl. data), and my
PVA provided the framework for this work. The density estimates I derived for both the
Enterprise and Chilcotin study areas also are being incorporated into this new model, and
continuing to collect detection data from other areas of the province will help improve this new
model going forward. Implementing rigorous and repeatable surveys for fishers on a region-byregion basis would allow changes in population trends to be tracked more effectively and could
then be used to test the predictions from the PVA model. Future work looking at whether there
are regional differences in survival and reproductive rates for fishers in different areas of the
province (e.g., Chilcotin versus Enterprise study areas) would also be highly beneficial. This
would then allow for a more detailed metapopulation analysis to determine if fishers in different
regions are at greater risk of extirpation so that management efforts could be focused most
effectively.
91

Conclusion
Management strategies that attempt to conserve populations of endangered carnivores
usually consist of protecting and/or restoring critical habitats to promote increased recruitment,
and reducing the risk of mortality to increase survival rates. These ideas are not new, but they
continue to be implemented for a multitude of species quite simply for one reason – they
typically are the most successful. In the case of the fisher, these strategies already have been
proven to work, with populations in eastern Canada and the United States having recolonized
much of their historic range. This study provides further evidence these strategies would be
successful in helping to recover the endangered Columbian population of fishers in central BC,
and the population monitoring and modelling methods used here can be applied to identify where
wildlife managers could focus efforts to protect and recover other at-risk carnivore populations.
The extirpation of carnivores by humans continues to repeat itself either directly through
shooting, trapping, and poisoning, or indirectly through impacts on prey species or habitat loss.
Without revisiting the conservation measures proven to effectively manage these species we may
once again lose the important ecosystem services they provide.

92

Literature Cited
Banci V (1989) Fisher management strategy for British Columbia. In: Wildlife Bulletin, Book B63. BC Ministry of Environment, Victoria, BC
Banci V, Proulx G (1999) Resiliency of furbearers to trapping in Canada. In: Proulx G (ed)
Mammal Trapping. Alpha Wildlife Research & Management Ltd., Sherwood Park,
Alberta
Buskirk SW, Powell RA (1994) Habitat ecology of fisher and American martens. In: Buskirk
SW, Harestad AH, Raphael MG, Powell RA (eds) Martens, sables, and fishers: biology
and conservation. Comstock Publishing Associates, Cornell University Press, Ithaca,
New York
BC Conservation Data Centre (2020) Conservation Status Report: Pekania pennanti pop. 5. In.
BC Ministry of Environment
Harrington LA, Marino J, King CM (2017) People and wild musteloids. In: Macdonald DW,
Newman C, Harrington LA (eds) Biology and Conservation of Musteloids. Oxford
University Press, Oxford, UK
Helgen K, Reid F (2018) Martes pennanti (amended version of 2016 assessment). The IUCN Red
List of Threatened Species. Accessed 04 January 2018.
https://www.iucnredlist.org/species/41651/125236220
Lofroth EC, Raley CM, Higley JM, Truex RL, Yaeger JS, Lewis JC, Happe PJ, Finley LL (2010)
Conservation of Fishers (Martes pennanti) in south-central British Columbia, Western
Washington, Western Oregon, and California. Volume I: Conservation Assessment, Vol
I. USDI Bureau of Land Management, Denver, Colorado, USA
Newsome G (2019) Wildlife Protection Act of 2019. In: Fish and Game Code, Book 216,
California
Proulx G (2022) The five Ws of mammal trapping. In: Proulx G (ed) Mammal Trapping Wildlife Management, Animal Welfare & International Standards. Alpha Wildlife
Publications, Sherwood Park, Alberta
Province of British Columbia (2021) BC Data Catalogue. https://catalogue.data.gov.bc.ca/
Raley CM, Lofroth EC, Truex RL, Yaeger JS, Higley JM (2012) Habitat Ecology of Fishers in
North America. In: Aubry KB, Zielinski WJ, Raphael MG, Proulx G, Buskirk SW (eds)

93

Biology and Conservation of Martens, Sables and Fishers: A New Synthesis. Cornell
University Press
Weir RD, Corbould FB (2006) Density of fishers in the Sub-Boreal Spruce biogeoclimatic zone
of British Columbia. Northwestern Naturalist 87:118-127
Weir RD, Lofroth EC, Phinney M (2011) Density of fishers in boreal mixed wood forests of
northeastern British Columbia. Northwestern Naturalist 92:65-68
Washington Department of Fish and Wildlife (2022) Furbearer Trapping Regulations. Accessed
6 Jan 2023. https://wdfw.wa.gov/sites/default/files/publications/02333/wdfw02333.pdf

94

APPENDIX A. DESCRIPTION OF VARIABLES USED IN DENSITY MODELS.
Table A.1. Variables used in candidate models known from previous research to influence the density of
fishers within the Chilcotin and Enterprise study areas in the central interior of British Columbia, Canada,
along with their associated references. A + symbol indicates a fisher would typically select for this
attribute and a – symbol indicates a fisher would typically select against this variable.
Select (+)
or Avoid
(-)
+

Variable Name
Primary branch
resting stands

Description
Area of spruce-leading, secondary, or tertiary stands aged ≥83
years old with crown closure ≥40% calculated as a proportion
of the 4.84 km2 and 30 km2 buffer areas.

Primary
denning stands

Area of aspen- or cottonwood-leading stands aged ≥135 years
old or Douglas-fir-leading stands aged ≥207 years old with
crown closure ≥20% calculated as a proportion of the 4.84 km2
and 30 km2 buffer areas.
Area of stands with spruce or aspen content ≥25% and aged
≥100 years old calculated as a proportion of the 4.84 km2 and
30 km2 buffer areas.
Area of stands with total cover ≥50% (≥30% shrub cover and
≥20% tree cover) calculated as a proportion of the 4.84 km2
and 30 km2 buffer areas.
Area of stands burned at a high intensity with trees dead,
needles, twigs, and understory consumed, calculated as a
proportion of the 4.84 km2 and 30 km2 buffer areas.
Area of stands burned at either a high or medium intensity with
trees dead, scorched needles remain on trees, understory
burned, calculated as a proportion of the 4.84 km2 and 30 km2
buffer areas.
Density of streams in km/km2 within the 4.84 km2 and 30 km2
buffer areas.

+

Weir and Almuedo
2010

+

Weir and Almuedo
2010

+

Weir and Almuedo
2010

-

Hope et al. 2015

-

Hope et al. 2015

+

Davis 2009

Total length of all streams in km within the 4.84 km2 and 30
km2 buffer areas.
Estimated total relative density of digestible energy based on
kcal/g provided by snowshoe hares and squirrels within the
4.84 km2 and 30 km2 buffer areas.
Area of stands with subhygric, hygric, or subhydric soil
moisture regimes calculated as a proportion of the 4.84 km2
and 30 km2 buffer areas. Located primarily along wetlands and
streams, contain spruce, and a relatively well-developed shrub
layer
Area of spruce-or aspen-leading stands within 100 m of a
wetland calculated as a proportion of the 4.84 km2 and 30 km2
buffer areas.
Area of stands with ≥30% crown closure within 100 m of
wetland calculated as a proportion of the 4.84 km2 and 30 km2
buffer areas.

+

Davis 2009

+

Weir unpubl. data

+

Weir and Corbould
2010

+

Davis 2009

+

Weir and Corbould
2010

Primary coarse
wood resting
stands
Primary
movement
stands
High intensity
burn
Medium or high
intensity burn

Stream density
Stream length
Prey density

Wet soil
moisture regime

Spruce-aspen
riparian forest
Riparian closed
canopy forest

Table continued on next page…

95

Reference
Weir and Almuedo
2010

Variable
Name
Open areas

Description
Recently logged or burned areas (<12 years old), wetlands,
and agricultural fields calculated as a proportion of the 4.84
km2 and 30 km2 buffer areas.

Select
(+) or
Avoid (-)
-

Old riparian
forest

Area of stands >100 years old within 100 m of wetland or
stream calculated as a proportion of the 4.84 km2 and 30 km2
buffer areas.

Cut blocks
harvested <20
years ago

Area of cut blocks harvested <20 years ago calculated as a
proportion of the 4.84 km2 and 30 km2 buffer areas.

At-leading
stands
Sx-dominated
riparian forest

Area of aspen- or cottonwood-leading stands calculated as a
proportion of the 4.84 km2 and 30 km2 buffer areas.
Are of spruce-leading stands within 100 m of a stream or
wetland calculated as a proportion of the 4.84 km2 and 30
km2 buffer areas.

+

Davis 2009

+

Davis 2009

Road density

Density of roads in km/km2 within the 4.84 km2 and 30 km2
buffer areas.

-

Kordosky et al. 2021

Active trap line

Area of one active registered trap line during the Enterprise
survey where 8 fishers were trapped, calculated as a
proportion of the 4.84 km2 and 30 km2 buffer areas.

-

Province of BC 2020

Old Douglas-fir
forest

Area of Douglas-fir-leading or secondary stands ≥100 years
old calculated as a proportion of the 4.84 km2 and 30 km2
buffer areas.

+

Weir and Almuedo
2010

96

+

Reference
Weir and Corbould
2010
Weir and Corbould
2010
Weir and Corbould
2010

APPENDIX B. SET OF CANDIDATE MODELS CONSIDERED FOR PREDICTING
DENSITY OF FISHERS IN THE CHILCOTIN AND ENTERPRISE STUDY AREAS.
Table B.1. Set of candidate models considered for predicting the density of fishers in the Chilcotin study
area in the central interior of British Columbia, Canada, during the winter of 2018-19. All models also
included a detection sub-model (not shown here) depending on whether I predicted density for both sexes
combined, or males and females separately, and candidate sets were considered at both the core use (4.84
km2) and home range (30km2) scales.

Model Name
C_1

Model
D ~ Primary branch resting stands

C_2

D ~ Primary denning stands

C_3

D ~ Coarse woody debris resting stands

C_4

D ~ Primary branch resting stands + stands burned at high intensity

C_5

D ~ Primary denning stands + stands burned at high intensity

C_6

D ~ Primary branch resting stands + stands burned at either medium or high intensity

C_7

D ~ Primary denning stands + stands burned at either medium or high intensity

C_8

D ~ Primary denning stands + primary branch resting stands + primary movement stands

C_9

D ~ Primary denning stands + primary branch resting stands

C_10

D ~ Primary branch resting stands + primary movement stands

C_11

D ~ Primary denning stands + stream density

C_12

D ~ Primary denning stands + total length of streams

C_13

D ~ Primary branch resting stands + stream density

C_14

D ~ Primary denning stands + prey density

C_15

D ~ Stands with wet soil moisture regimes

C_16
C_17

D ~ Sx or At-leading stands within 100 m of wetland
D ~ Sx or At-leading stands within 100 m of wetland + stands with >30% crown closure
within 100 m of wetland

C_18

D ~ Open areas

C_19

D ~ Primary movement stands

C_20

D ~ Stands >100 years old within 100 m of wetland

C_21

D ~ Cut blocks harvested <20 years ago

Null

D ~ 1 (Density remained uniform across the study area)

97

Table B.2. Set of candidate models considered for predicting the density of fishers in the Enterprise study
area in the central interior of British Columbia, Canada, during the winter of 2020-21. All models also
included a detection sub-model with no variation in g0 or σ and were considered at both the core use
(4.84 km2) and home range (30km2) scales.

Model Name

Model

E_1

D ~ Primary branch resting stands

E_2

D ~ Primary denning stands

E_3

D ~ Coarse woody debris resting stands

E_4

D ~ Primary branch resting stands + primary movement stands

E_5

D ~ Primary denning stands + primary branch resting stands

E_6

D ~ Primary branch resting stands + aspen-leading stands

E_7

D ~ Primary denning stands + Stands with wet soil moisture regimes

E_8

D ~ Primary denning stands + stream density

E_9

D ~ Primary denning stands + total length of streams

E_10

D ~ Sx-dominated riparian forest

E_11

D ~ Open areas

E_12

D ~ Density of roads

E_13

D ~ Active trapline

E_14

D ~ Cut blocks harvested <20 years ago

E_15

D ~ Primary movement stands

E_16

D ~ Crown closure >50%

E_17

D ~ Old Douglas-fir forest

E_18

D ~ Mature and old stands

E_19

D ~ Stands >100 years old within 100 m of wetland

E_20

D ~ Stands >100 years old within 100 m of a stream

Null

D ~ 1 (Density remained uniform across the study area)

98

APPENDIX C. 95% CONFIDENCE SET OF CANDIDATE MODELS FOR
PREDICTING THE DENSITY OF FISHERS IN THE CHILCOTIN AND ENTERPRISE
STUDY AREAS.

Table C.1. Comparing model fit between the 95% confidence set of candidate density models for fishers
(both sexes) from the Chilcotin study area in the central interior of British Columbia, Canada, during the
winter of 2018-19. I estimated 3 parameters: density (D), detection probability (g0), and a spatial
parameter (sigma) at the home range scale (30 km2). All models included a behavioural effect where g0
depends on detection at the preceding occasion and no variation in sigma. I report here the number of
model parameters (K), model log likelhood (logLik), Akaike's Information Criterion corrected for small
sample sizes (AICc), the difference in AICc values (∆AICc), and relative model weight (wi).

Model

∆AICc

K

logLik

AICc

Primary branch resting stands

6

-442.22277

898.494

0

0.1699

Primary denning stands

6

-442.92899

899.907

1.413

0.0838

Primary denning stands + high intensity burn

7

-441.70911

900.218

1.724

0.0717

Stands with wet soil moisture regimes
Primary branch resting stands + high intensity burned
stands

6

-443.17355

900.396

1.902

0.0656

7

-441.86828

900.537

2.043

0.0612

Primary branch resting stands + primary denning stands

7

-441.91671

900.633

2.139

0.0583

Sx- or At-leading stands within 100 m of wetland

6

-443.37728

900.803

2.309

0.0535

Primary branch resting stands + stream density
Primary denning stands + medium or high intensity
burned stands
Primary branch resting stands + primary movement
stands
Primary branch resting stands + medium or high
intensity burn stands

7

-442.01885

900.838

2.344

0.0526

7

-442.0941

900.988

2.494

0.0488

7

-442.17371

901.147

2.653

0.0451

7

-442.21686

901.234

2.74

0.0432

Null

5

-445.07928

901.587

3.093

0.0362

Sx or At-leading stands within 100 m of wetland +
stands with >30% crown closure within 100 m of
wetland

7

-442.45834

901.717

3.223

0.0339

Stands >100 years old within 100 m of wetland

6

-443.83987

901.729

3.235

0.0337

Primary denning + prey density

7

-442.70218

902.204

3.71

0.0266

Cut blocks harvested <20 years ago

6

-444.29887

902.647

4.153

0.0213

Primary denning + stream density

7

-442.92854

902.657

4.163

0.0212

Primary denning + total length of streams

7

-442.92854

902.657

4.163

0.0212

Primary coarse wood resting habitat

6

-444.6061

903.261

4.767

0.0157

99

wi

Table C.2. Comparing model fit between the 95% confidence set of candidate density models for fishers
(both sexes) from the Chilcotin study area in the central interior of British Columbia, Canada, during the
winter of 2018-19. I estimated 3 parameters: density (D), detection probability (g0), and a spatial
parameter (sigma) at the core use scale (4 km2). All models included a behavioural effect where g0
depends on detection at the preceding occasion and no variation in sigma. I report here the number of
model parameters (K), model log likelhood (logLik), Akaike's Information Criterion corrected for small
sample sizes (AICc), the difference in AICc values (∆AICc), and relative model weight (wi).

Model

∆AICc

K

logLik

AICc

Primary branch resting stands

6

-441.01821

896.085

0

0.2569

Primary branch resting stands + stream density

7

-440.51565

897.831

1.746

0.1073

Primary branch resting stands + Primary denning stands
Primary branch resting stands + stands burned at a high
intensity
Primary branch resting stands + primary movement
stands
Primary branch resting stands + stands burned at a high
or medium intensity

7

-440.73139

898.263

2.178

0.0865

7

-440.77092

898.342

2.257

0.0831

7

-440.99884

898.798

2.713

0.0662

7

-441.01742

898.835

2.75

0.065

Sx- or At-leading stands within 100 m of wetland

6

-442.62004

899.289

3.204

0.0518

Stands with wet soil moisture regimes

6

-442.89244

899.834

3.749

0.0394

Primary denning stands

6

-442.90056

899.85

3.765

0.0391

Stands >100 years old within 100 m of wetland

6

-442.98999

900.029

3.944

0.0358

7

-441.62669

900.053

3.968

0.0353

8

-440.66814

901.029

4.944

0.0217

7

-442.14425

901.088

5.003

0.0211

7

-442.22734

901.255

5.17

0.0194

5

-445.07928

901.587

5.502

0.0164

Sx or At-leading stands within 100 m of wetland +
stands with >30% crown closure within 100 m of
wetland
Primary branch resting stands + primary denning stands
+ primary movement stands
Primary density stands + stands burned at a high or
medium intensity
Primary denning stands + stands burned at a high
intensity
Null

100

wi

Table C.3. Comparing model fit between the 95% confidence set of candidate density models for fishers
(females only) from the Chilcotin study area in the central interior of British Columbia, Canada, during
the winter of 2018-19. I estimated 3 parameters: density (D), detection probability (g0), and a spatial
parameter (sigma) at the home range scale (30 km2). All models included a behavioural effect where g0
depends on detection at the preceding occasion and no variation in sigma. I report here the number of
model parameters (K), model log likelhood (logLik), Akaike's Information Criterion corrected for small
sample sizes (AICc), the difference in AICc values (∆AICc), and relative model weight (wi).

Model

∆AICc

K

logLik

AICc

wi

Stands with wet soil moisture regimes

5

-283.24974

578.722

0

0.2197

Primary denning stands

5

-283.9211

580.064

1.342

0.1123

Sx- or At-leading stands within 100 m of wetland

5

-283.99545

580.213

1.491

0.1042

Null

4

-285.48957

580.408

1.686

0.0945

Stands >100 years old within 100 m of wetland

5

-284.37415

580.971

2.249

0.0713

Primary denning stands + prey density

6

-283.1026

581.436

2.714

0.0565

Primary denning stands + high intensity burn

6

-283.41174

582.054

3.332

0.0415

Primary branch resting stands

5

-285.05834

582.339

3.617

0.036

Sx or At-leading stands within 100 m of wetland +
stands with >30% crown closure within 100 m of
wetland

6

-283.61939

582.47

3.748

0.0337

Cut blocks harvested <20 years ago
Primary denning stands + medium or high intensity
burn

5

-285.14261

582.507

3.785

0.0331

6

-283.80833

582.847

4.125

0.0279

Primary denning stands + stream density

6

-283.83379

582.898

4.176

0.0272

Primary coarse wood resting stands

5

-285.35081

582.924

4.202

0.0269

Open areas

5

-285.39535

583.013

4.291

0.0257

Primary branch resting stands + primary denning

6

-283.91947

583.07

4.348

0.025

Primary movement stands

5

-285.48286

583.188

4.466

0.0235

101

Table C.4. Comparing model fit between the 95% confidence set of candidate density models for fishers
(females only) from the Chilcotin study area in the central interior of British Columbia, Canada, during
the winter of 2018-19. I estimated 3 parameters: density (D), detection probability (g0), and a spatial
parameter (sigma) at the core use area scale (4.84 km2). All models included a behavioural effect where
g0 depends on detection at the preceding occasion and no variation in sigma. I report here the number of
model parameters (K), model log likelhood (logLik), Akaike's Information Criterion corrected for small
sample sizes (AICc), the difference in AICc values (∆AICc), and relative model weight (wi).

Model

∆AICc

K

logLik

AICc

Stands with wet soil moisture regimes

5

-282.66874

577.56

0

0.2353

Primary denning
Sx- or At-leading stands within 100 m of
wetland
Stands >100 years old within 100 m of
wetland

5

-283.23751

578.697

1.137

0.1333

5

-283.36233

578.947

1.387

0.1176

5

-283.76019

579.743

2.183

0.079

Primary denning + prey density score

6

-282.51393

580.259

2.699

0.061

Null

4

-285.48957

580.408

2.848

0.0566

Primary denning + high intensity burn
Sx or At-leading stands within 100 m of
wetland + stands with >30% crown closure
within 100 m of wetland

6

-282.94761

581.126

3.566

0.0396

6

-282.97175

581.174

3.614

0.0386

Primary denning + stream density
Primary denning + medium or high intensity
burn

6

-283.11155

581.454

3.894

0.0336

6

-283.12796

581.487

3.927

0.033

Primary branch resting + primary denning

6

-283.14642

581.524

3.964

0.0324

Primary branch resting

5

-284.71097

581.644

4.084

0.0305

Primary coarse wood resting

5

-285.00629

582.235

4.675

0.0227

Cut blocks harvested <20 years ago

5

-285.19081

582.604

5.044

0.0189

Open areas

5

-285.3736

582.969

5.409

0.0157

Primary movement

5

-285.48869

583.2

5.64

0.014

102

wi

Table C.5. Comparing model fit between the 95% confidence set of candidate density models for fishers
(males only) from the Chilcotin study area in the central interior of British Columbia, Canada, during the
winter of 2018-19. I estimated 3 parameters: density (D), detection probability (g0), and a spatial
parameter (sigma) at the home range scale (30 km2). All models included no variation in g0 or sigma. I
report here the number of model parameters (K), model log likelhood (logLik), Akaike's Information
Criterion corrected for small sample sizes (AICc), the difference in AICc values (∆AICc), and relative
model weight (wi).
∆AICc

Model

K

logLik

AICc

Primary branch resting stands

4

-153.60432

319.209

0

Primary branch resting stands + stream density

5

-152.71768

322.102

2.893

0.099

Null
Primary branch resting stands + primary denning
stands

3

-157.12617

322.434

3.225

0.0838

5

-153.18098

323.029

3.82

0.0623

Primary branch resting stands + high intensity burn
Primary branch resting stands + stands burned at
high or medium intensity

5

-153.41982

323.506

4.297

0.0491

5

-153.42511

323.517

4.308

0.0488

Primary branch resting + primary movement stands

5

-153.5893

323.845

4.636

0.0414

Open areas

4

-156.11872

324.237

5.028

0.034

Primary denning stands

4

-156.44751

324.895

5.686

0.0245

Primary movement stands

4

-156.46024

324.92

5.711

0.0242

Primary coarse wood resting stands

4

-156.60483

325.21

6.001

0.0209

Cut blocks harvested <20 years ago

4

-156.62557

325.251

6.042

0.0205

Sx- or At-leading stands within 100 m of wetland

4

-156.82312

325.646

6.437

0.0168

Stands >100 years old within 100 m of wetland

4

-156.90104

325.802

6.593

0.0156

103

wi
0.4205

Table C.6. Comparing model fit between the 95% confidence set of candidate density models for fishers
(males only) from the Chilcotin study area in the central interior of British Columbia, Canada, during the
winter of 2018-19. I estimated 3 parameters: density (D), detection probability (g0), and a spatial
parameter (sigma) at the core area use scale (4.84 km2). All models included no variation in g0 or sigma. I
report here the number of model parameters (K), model log likelhood (logLik), Akaike's Information
Criterion corrected for small sample sizes (AICc), the difference in AICc values (∆AICc), and relative
model weight (wi).
∆AICc

Model

K

logLik

AICc

Primary branch resting stands
Primary branch resting stands + stream density
stands
Primary branch resting stands + primary
denning stands
Primary branch resting stands + primary
movement stands
Primary branch resting stands + medium or high
intensity burn
Primary branch resting stands + high intensity
burn

4

-152.7998

317.6

0

0.4402

5

-151.11257

318.892

1.292

0.2307

5

-152.48264

321.632

4.032

0.0586

5

-152.61673

321.9

4.3

0.0513

5

-152.61774

321.902

4.302

0.0512

5

-152.74792

322.163

4.563

0.045

Null

3

-157.12617

322.434

4.834

0.0393

Cut blocks harvested <20 years ago

4

-156.37194

324.744

7.144

0.0124

Primary movement stands

4

-156.41179

324.824

7.224

0.0119

Open areas

4

-156.41889

324.838

7.238

0.0118

104

wi

Table C.7. Comparing model fit between the 95% confidence set of candidate density models for fishers
(both sexes) from the Enterprise study area in the central interior of British Columbia, Canada, during the
winter of 2020-21. I estimated 3 parameters: density (D), detection probability (g0), and a spatial
parameter (sigma) at the home range scale (30 km2). All models included no variation in g0 or sigma. I
report here the number of model parameters (K), model log likelhood (logLik), Akaike's Information
Criterion corrected for small sample sizes (AICc), the difference in AICc values (∆AICc), and relative
model weight (wi).

Model

∆AICc

K

logLik

AICc

Crown closure >50%

5

-149.22637

312.453

0

0.2195

Open areas

5

-149.259

312.518

0.065

0.2125

Stands with wet soil moisture regimes

5

-149.79926

313.599

1.146

0.1238

Primary movement stands
Sx-leading stands >100 years old within 100 m
of a wetland

5

-150.01018

314.02

1.567

0.1003

5

-150.48161

314.963

2.51

0.0626

Null

4

-152.29054

315.081

2.628

0.059

Road density

5

-150.76034

315.521

3.068

0.0473

Cut blocks harvested <20 years ago

5

-151.12867

316.257

3.804

0.0328

Primary coarse wood resting stands

5

-151.41791

316.836

4.383

0.0245

Mature and old structural stages

5

-151.42813

316.856

4.403

0.0243

Active trap line

5

-151.49095

316.982

4.529

0.0228

Primary denning stands

5

-151.74062

317.481

5.028

0.0178

Fd-leading stands >100 years old

5

-151.84659

317.693

5.24

0.016

105

wi

Table C.8. Comparing model fit between the 95% confidence set of candidate density models for fishers
(both sexes) from the Enterprise study area in the central interior of British Columbia, Canada, during the
winter of 2020-21. I estimated 3 parameters: density (D), detection probability (g0), and a spatial
parameter (sigma) at the fisher core use area scale (4.84 km2). All models included no variation in g0 or
sigma. I report here the number of model parameters (K), model log likelhood (logLik), Akaike's
Information Criterion corrected for small sample sizes (AICc), the difference in AICc values (∆AICc),
and relative model weight (wi).
Model

∆AICc

K

logLik

AICc

Stands with wet soil moisture regimes

5

-147.819535

309.639

0

0.5005

Crown closure >50%

5

-148.994511

311.989

2.35

0.1546

Open areas

5

-149.513937

313.028

3.389

0.0919

Primary movement stands

5

-149.772111

313.544

3.905

0.071

Null

4

-152.290541

315.081

5.442

0.0329

Road density
Sx-leading stands >100 years old within 100
m of a wetland

5

-150.621095

315.242

5.603

0.0304

5

-150.876009

315.752

6.113

0.0235

Cut blocks harvested <20 years ago

5

-151.312198

316.624

6.985

0.0152

Mature and old structural stages

5

-151.338915

316.678

7.039

0.0148

Active trap line

5

-151.488459

316.977

7.338

0.0128

Primary denning stands

5

-151.571582

317.143

7.504

0.0117

106

wi

107