Content uploaded by William L. Gaines
Author content
All content in this area was uploaded by William L. Gaines on Jan 12, 2018
Content may be subject to copyright.
Canada Lynx Carrying Capacity in Washington
Final Report
December 2016
PREPARED BY:
WASHINGTON CONSERVATION SCIENCE INSTITUTE
Andrea L. Lyons, MS; William L. Gaines, PhD; James Begley, MS
USDA FOREST SERVICE, PACIFIC NORTHWEST RESEARCH STATION
Peter H. Singleton, PhD
WASHINGTON DEPARTMENT OF FISH AND WILDLIFE
Jeffrey C. Lewis, PhD; Benjamin T. Maletzke, PhD
Submitted to the Washington Department of Fish and Wildlife
Olympia, Washington
Contract #16-06588
2
Acknowledgements
This project was funded by the Washington Department of Fish and Wildlife, the National Fish and Wildlife
Foundation and the Wenatchee Forestry Sciences Lab. Special thanks go to Chris Loggers, Scott Fisher, John
Rohrer, Scott Fitkin and Jen Watkins for contributing their time and expertise to get this project off the ground.
We are very grateful to all those who helped make this project possible. The views and conclusions contained
in this document are those of the authors and should not be interpreted as representing the opinions or policies
of the U.S. Government or the National Fish and Wildlife Foundation and its funding sources. Mention of trade
names or commercial products does not constitute their endorsement by the U.S. Government, or the National
Fish and Wildlife Foundation or its funding sources.
Table of Contents
Page
List of Tables and Figures
3
Introduction & Objective
4
Methods
5
Results and Discussion
13
Literature Cited
27
Appendix S1.
30
3
List of Tables and Figures
Table 1. Variables used in resource selection function development.
Table 2. Annual female lynx survival values for all combinations of age classes and resource quality classes used in
population model. Values were determined for each life stage in the high habitat quality class as the highest value from our
literature review less 5%, in the moderate habitat quality class as the mean value from the literature, and in the low habitat
quality class as 25% less than the lowest value in the literature.
Table 3. Annual female lynx fecundity values for all combinations of age classes and resource quality classes used in
population model. Values were determined for each life stage in the high habitat quality class as the highest value from our
literature review less 5%, in the moderate habitat quality class as the mean value from the literature, and in the low habitat
quality class as 25% less than the lowest value in the literature.
Table 4. Description of model scenarios developed to estimate carrying capacity for lynx in the OLMZ. The number in the
Scenario name refers to the home range size used in the model. All models used the same initial resource layer as indicated
by pre-fire or post-fire.
Table 5. Rapid Assessment of Vegetation Condition after Wildfire (RAVG) classes and associated loss of basal area
vegetation resulting from wildfire.
Table 6. Parameters and associated coefficients in the Okanogan LMZ habitat model.
Table 7. Simulation-duration mean number of female individuals for the total, group and floater populations in the
Okanogan and Kettle LMZs for six scenarios. The change in population carrying capacity as a result of changes in habitat
due to wildfire and other disturbances, as well as succession, was calculated as the percent change in total population size
between scenarios (Post-Fire – Pre-Fire). Group members were female lynx in the total population with established home
ranges and floaters were dispersing female lynx in the total population without home ranges.
Figure 1. Lynx management zones (LMZs) in Washington indicate the general areas historically occupied by lynx in
northcentral and northeastern Washington (from Lewis 2016).
Figure 2. Okanogan and Kettle Lynx Management Zones, Lynx Analysis Units and wildfire activity from 2000 through
2015 within Washington State, US.
Figure 3. HexSim flowchart – HexSim is an individual-based population modeling framework that represents population
function based on a series of annual life history events. Events in the lynx model included: 1) Survival, 2) Reproduction, 3)
Movement, 4) Resource Acquisition / Home Range Establishment, and 5) Population Census. The process repeats as
individuals age by a year.
Figure 4. Annual mixed effect resource selection function reclassified to relative habitat quality classes mapped within the
Okanogan Lynx Management Zone. 4a) depicts the 2000 pre-fire condition, prior to substantial wildfire activity from 2000-
2013, while 4b) depicts the 2013 post-fire condition.
Figure 5. Okanogan Lynx Management Zone 2013 post-fire resource map with fire activity polygons from 2000-2015. The
majority of fire activity after 2013 occurred outside of lynx habitat and burn intensity data was unavailable at the time of
this assessment so was not included here.
Figure 6. Annual mixed effect resource selection function reclassified to relative habitat quality classes mapped within the
Kettle Lynx Management Zone. 6a) depicts the 2000 pre-fire condition prior to substantial wildfire activity from 2000 -
2015, while 6b) depicts the 2015 post-fire condition. Red polygons depict fire boundaries for wildfire activity from 2000
through 2015.
Figure 7. Spatial distribution of mean annual territorial female lynx density (# per 100km2) by LAU in the Okanogan LMZ.
Differences between scenarios were a result of resource changes due to disturbance, primarily wildfire, with three different
home range sizes (39km2, 55km2, and 72 km2). Color scheme and range of values were held constant within each home
range to show the influence of disturbance on modeled density outcomes.
Figure 8. Spatial distribution of mean annual territorial female lynx density (# per 100km2) by LAU in the Kettle LMZ.
Differences between scenarios were a result of resource changes due to disturbance, primarily wildfire, with three different
home range sizes (39km2, 55km2, and 72 km2). Color scheme and range of values were held constant within each home
range to show the influence of disturbance on modeled density outcomes.
4
INTRODUCTION AND OBJECTIVE
Canada lynx (Lynx canadensis) is a species native to boreal forests of north-central and north-eastern
Washington State. Once found in large numbers, particularly in the northern portion of the state, populations
have declined due to habitat loss and hunting (Lewis 2016). The Canada lynx was listed as a Washington State
threatened species in 1993, resulting in development of a recovery plan in 2001 (Stinson 2001). Lynx were listed
as a Threatened species under the Federal Endangered Species Act in April 2000. Key threats identified in the
Washington recovery plan included: forest management, fire and fire suppression, insect epidemics, and
management of lynx harvest and habitats in southern British Columbia. Fifteen years later, these threats are still
identified as important issues while recent research suggests climate change is likely to impact lynx, potentially
exacerbating habitat loss through increased wildfire, leading to even smaller and more isolated populations in
the future with decreased habitat suitability and genetic diversity (Hoving et al. 2005, Gonzalez et al. 2007, Yan
et al. 2013).
Historically lynx were believed to occupy six Lynx Management Zones (LMZ), but Washington’s lynx
population is now largely restricted to the Okanogan LMZ (Fig. 1) (Lewis 2016), which falls under multiple
land-management jurisdictions, including the Okanogan-Wenatchee National Forest, North Cascades National
Park, Mount Baker-Snoqualmie National Forest, and Washington Department of Natural Resources. Occasional
detections of lynx in the Kettle LMZ suggest lynx are present but likely not part of a resident population (Lewis
2016). The Kettle LMZ is predominantly managed by the Colville National Forest and the Confederated Colville
Tribes.
Over the past 35 years understanding of lynx habitat use and population ecology, and methods to estimate the
potential carrying capacity of wildlife populations within ecosystems have advanced tremendously. Our goal
was to synthesize these advances and integrate spatial habitat data and demographic parameter estimates using
a spatially explicit, individual-based population modeling approach. We used this model to address two
questions: 1) What is the potential carrying capacity for Canada lynx in the Okanogan and Kettle LMZs, and 2)
How have changes in habitat influenced carrying capacity over time within those LMZs?
Figure 1. Lynx management zones (LMZs) in Washington indicate the general areas historically occupied by lynx in
northcentral and northeastern Washington (WDNR 2006).
5
ANALYSIS AREA
Our analysis area included the portion of the Okanogan LMZ north of Lake Chelan and the entire Kettle LMZ.
The Okanogan LMZ is approximately 9,200 km2 (3,552 mi2) and the Kettle LMZ is approximately 3,300 km2
(1,274 mi2) (Fig 2). Both LMZs include a range of land uses from designated wilderness to multiple use resource
lands to heavily populated urban areas.
The Okanogan LMZ varies from extensive lush subalpine forests and alpine meadows along the central spine of
the North Cascades Mountains, transitioning rapidly to dry forests and dry, lowland valleys on the eastern portion
of the ecosystem. Elevation ranges from 242 m in the eastern valleys, to peaks reaching 2755 m. Road densities
vary across the landscape with a large expanse of predominantly roadless area in the western and northwestern
portion of the LMZ. Similarly, the Kettle LMZ varies from subalpine forests and alpine meadows along the
central spine of the Kettle Range, then transitions rapidly to dry forests and dry, lowland valleys along all edges
of the LMZ. Elevation ranges from 390 m around the edges of the LMZ, to peaks exceeding 2,100 m along the
center of the LMZ. Road densities vary across the landscape with minimal expanses of roadless areas.
Both LMZs share a northern border with British Columbia. The LMZs are divided into Lynx Analysis Units
(LAUs) to identify assessment units for monitoring and evaluation of cumulative effects (Gaines et al. 2003,
ILBT 2013). These analysis units approximate a female lynx home range and are large enough to allow the
assessment of seasonal habitats and the cumulative effects of human activities on these habitats. There are 50
LAUs in the Okanogan LMZ and 14 LAUs in the Kettle LMZ (Fig 2).
Both LMZs are also located in fire-prone landscapes with varying fire return intervals and risk for large fires
(Hessburg et al. 2005, Perry et al. 2011). Over the past 15 years both of these areas have experienced an increase
in substantial wildfire activity. Wildfires have impacted over 2000 km2 within the Okanogan LMZ study area
and 360km2on the Kettle LMZ.
METHODS
To estimate carrying capacity we developed a suite of spatially-explicit, individual-based population models
using HexSim software (version 4.0.3.0, Schumaker 2016) that integrated information on habitat selection, and
population dynamics and changes in resource availability. HexSim software provides a framework for
implementing population simulation models that has been used to investigate potential population outcomes
based on empirical information regarding habitat associations and demographic rates (Heinrichs et al. 2010,
Spencer et al. 2011, Huber at al. 2014). We developed Canada lynx population models that provided the
appropriate information and flexibility to address two key questions: 1) What is the potential carrying capacity
for lynx in the Okanogan and Kettle LMZs? and 2) How have wildfire and habitat changes influenced carrying
capacity?
Application of HexSim required information on resource selection, home range size, dispersal, survival, and
fecundity. We used data primarily from the Okanogan LMZ and expert knowledge from biologists familiar with
lynx in Washington to populate these parameters. When site-specific data were unavailable we used information
from the literature, primarily from ecosystems that resembled the southern periphery nature of the Okanogan
and Kettle LMZs. We then extrapolated this model to the Kettle LMZ.
6
Figure 2. Okanogan and Kettle Lynx Management Zones, Lynx Analysis Units and wildfire activity from 2000 through
2015 within Washington State, US.
Development of Resource Layers – Lynx Habitat Modeling
HexSim requires that each hexagon within the model be assigned a habitat resource value based on the quality
of habitat within the hexagon. To estimate carrying capacity for the two different LMZs at two time periods, we
built several different resource layers based on available spatial data. This resulted in resource layers that
represented:
1) OLMZ 2013 post-fire
2) OLMZ 2000 pre-fire
3) KLMZ 2015 post-fire
4) KLMZ 2000 pre-fire
To include spatial data that could show changes on the landscape we included canopy cover, forest structure and
greenness in our model selection process. We anticipated that these variables would change between the pre and
7
post-fire conditions in response to the effect of not only wildfire but also effects of harvest, insect and disease,
and succession.
The initial resource values and habitat quality classifications were calculated using a resource selection function
(RSF) (Manley et al. 2002, Proctor et al. 2015). We used GPS data from radio-collared lynx within the Okanogan
LMZ to assess habitat selection at the scale of the Lynx Management Zone for both female and male lynx to
develop a general annual model for the region. Developing an RSF model for the Okanogan LMZ based on
telemetry data provided a model of “current” habitat selection. We acquired GPS data from Washington
Department of Fish and Wildlife (WDFW) for 16 lynx over the time period: February 2008 through June 2013.
We selected a random sample of 80% of available lynx locations for model development and withheld the
remaining 20% for model evaluation.
We estimated a 100% minimum convex polygon range of all of the lynx telemetry locations to delineate our
study area for model development. We only used locations within the United States, due to spatial data layer
constraints. We used GPS radio-collar locations and an equal number of random locations from within the 100%
MCP home range of all lynx to develop the RSF. We estimated model parameters with mixed effects logistic
regression with individual lynx as a random effect (n=14) and applied a square root transformation to transform
and normalize the positive skewing of exponential RSF values using R software (version 3.2.2, R Development
Core Team, Vienna, Austria) and ArcGIS (version 10.4, ESRI, Inc.). Only three of the collared lynx were female
so we pooled male and female lynx telemetry data to provide a more robust analysis of lynx habitat selection.
We acknowledge that male selection may differ from females but the pooled information likely provides an
adequate general picture of habitat selection. We tested all covariates for pairwise correlations (Spearman) and
when correlations were found (r>= 0.7) we did not use those pairings in the same model. All continuous variables
were standardized to examine relative influence on RSFs.
We developed a set of 30 a priori models and usedAkaike’sInformationCriterion(AIC,BurnhamandAnderson
2002) to select the model that best fit the structure of our data. We calculated a bootstrapped (n=100) mean and
confidence interval for resultant coefficient estimates. Although we recognize that this set of models may be
conservative, our objective was to provide a descriptive and biologically meaningful multivariate model of lynx
resource selection that could be generalized to the population and provide the foundation for subsequent
population carrying capacity modeling.
We examined selection ratios (use/availability) for individual lynx between the model evaluation dataset and
model development dataset to assess how well the RSF predicted use. We mapped the resource selection function
across the Okanogan LMZ and initially classified the transformed RSF values into 10 equal interval bins where
habitat use equaled the proportion of withheld GPS locations within each bin (relative to total locations) and
availability equaled the proportion of area within each bin (relative to total area). Selection ratios also determined
the subsequent break points for final resource layer classes.
To develop the initial resource map and to classify habitat for HexSim we classified the RSF scores into three
categories based on habitat selection where 1 = habitat selected less than available (low quality habitat); 2 =
selection approximately equal to availability (moderate quality habitat); and 3 = habitat selection greater than
available (high quality habitat). We removed non-habitat types of ice, rock, and water bodies larger than 10km2.
This initial resource map functioned as our post-fire scenario. We also mapped RSF values with the pre-fire data
layers to create the “pre-fire”scenarioresourcemap.
Data Layers
Occupancy, reproduction and habitat selection have been documented during several studies that occurred during
the 1980-2012 time period in the Okanogan LMZ (Brittell et al. 1989, Koehler 1990, McKelvey et al. 2000, von
Keinast 2003, Maletzke 2004, Koehler et al. 2008, Vanbianchi 2015). These studies and work in other
8
ecosystems (Squires et al. 2010, Squires et al. 2013) provided a group of variables we used in our suite of models
to describe habitat. Our final set of variables included those that provided a direct measurement (i.e. canopy
cover) as well as those that could be considered as an index or surrogate (i.e. greenness and solar radiation). Our
static terrain variables included slope, elevation, aspect, surface ruggedness, topographic wetness index and solar
radiation derived with a Digital Elevation Model (smoothed to remove banding. See Copeland et al. 2007) in
ARCGIS. We linearized aspect (McKelvey et al. 2000) to provide values that ranged from 0 on the coldest,
wettest slopes (due northeast) to 180 on the warmest, driest slopes (due southwest).
We also considered dynamic variables that could potentially capture changes in habitat over time, such as
increases due to succession, or decreases due to disturbances such as wildfire, insects and disease or harvest.
Our dynamic variables included greenness, vegetative cover type (forest structure type) and canopy cover.
Greenness is an index of leafy green productivity calculated with a tasseled cap transformation (Baig et al. 2014).
We derived greenness from Landsat 8 imagery during the summer of 2000 and 2013. These two time periods
provided the most recent comparable imagery of good quality for the Okanogan and Kettle LMZs that coincided
with vegetative data for the post-fire time period as well as a comparable pre-fire snapshot. Vegetative
classification was based on GNN data from 2000 and 2012 (Ohmann et al. 2011). We reclassified the original
GNN forest type into four categories: mesic-forest (i.e. Engelmann spruce, lodgepole, ABLA dominated), dry
forest (i.e. PSME & PIPO dominated), Other forest and Non-forest (i.e. agricultural lands, remnant forest with
<10% canopy, grasslands, wet and mesic shrublands and shrub-steppe) (Table 1). Canopy cover data was
obtained from the GNN dataset. We included quadratic forms for canopy cover and elevation in some models.
For the sake of simplicity the resource maps developed for the post-fireanalysisarereferredtoas“2013post-
fire”.
Table 1. Variables used in resource selection function development.
Category
Variable
Units
Vegetation
Canopy Cover
continuous
Dry Forest
categorical
Mesic Forest
categorical
Other forest
categorical
Non-forest
categorical
Ecological
Greenness
continuous
Solar radiation
kj/m2
Elevation
meters
Slope
degrees
Aspect
continuous
Surface Roughness
ratio
Wildfire and other changes to habitat
By using Landsat and GNN data from 2000 as compared to 2013 we hoped to capture changes in vegetation due
to fires over a 13 year period. This would capture effects from approximately 2000 km2 of wildfire on the OLMZ.
Most of the wildfire activity on the OLMZ after 2013 occurred outside of lynx habitat and burn intensity data
was unavailable at the time of this assessment so was not included here (Fig. 5).
In contrast, substantial fire activity within lynx habitat on the Kettle LMZ occurred in 2015. As such we adjusted
the 2013 resource layer for the Kettle to account for those changes (see below under Kettle LMZ Resource Layer
development). Wildfire impacted approximately 360 km2 on the Kettle LMZ from 2000 through 2015. We
attempted to use the best available data consistently across the LMZs to describe the current habitat condition
but did not want to disregard the significant recent fire activity on the KLMZ.
9
Carrying Capacity Model – HexSim Input
HexSim is an individual-based population modeling framework that represents population function based on a
series of annual life history events. The Canada lynx model incorporated survival, reproduction, movement,
resource acquisition, home range establishment and a population census, all of which are influenced by habitat
conditions. We used a female-only, single-sex model structure because: 1) lynx are polygynous, 2) reproductive
output is limited by the number of females of reproductive age, 3) female survival influences population trend
more than male survival (Aubry et al. 2000, Anderson and Lovallo 2003) and 4) to reduce the complexity of the
model. Model parameters were based on local empirical information, estimates from the literature or professional
opinion depending on availability of information.
Figure 3. HexSim flowchart – HexSim is an individual-based population modeling framework that represents population
function based on a series of annual life history events. Events in the lynx model included: 1) Survival, 2) Reproduction, 3)
Movement, 4) Resource Acquisition / Home Range Establishment, and 5) Population Census. The process repeats as
individuals age by a year.
Habitat Resources
HexSim uses a hexagonal grid to represent habitat conditions that influence individual movement, survival, and
reproduction. The Okanogan and Kettle landscapes were represented as a grid of 16.2 ha (500m diameter)
hexagons. We chose this hexagon size because it captured relatively fine-scale landscape patterns that we expect
would influence lynx habitat selection without becoming computationally limiting. Each hexagon was assigned
a resource score based on underlying habitat values. We calculated a focal sum of Habitat Classes 1 (poor
quality) through 3 (high quality) at a 250m radius across the study area. We attributed hexagons with the focal
sum value at the center of the hexagon. Simulated individuals were assigned to a resource quality class based on
the total resource scores of all hexagons within their home ranges (see Resource Acquisition / Home Range
section below).
Survival
Survival rates of females were incorporated into the model relative to age class and resource quality. Modeled
individuals were assigned to four age classes: kitten (<1 year), yearling (age 1 year), sub-adult (age 2 years) and
adult (age >2 years). Survival values for each age class were estimated based on data available from other lynx
populations. Although there were extensive data available in the literature relative to survival estimates for the
four age classes, (kitten, yearling, subadult and adult), no quantifiable information on the relationship between
10
survivorship and habitat quality was available. As such we estimated female survival for kittens, yearlings,
subadults, and adults in low, moderate and high quality habitat based on general published values. We
determined the values for each life stage in the high habitat quality class as the highest values from our literature
review less 5%, in the moderate habitat quality class as the mean value from the literature, and in the low habitat
quality class as 25% less than the lowest value in the literature (Table 2). The resource quality class refers to
lynx whose home range meets the home range requirements as defined in HexSim. A home range in the high
resource quality class had a minimum of 60% of the home range in the high quality category. A home range in
the Moderate resource quality class had 40 to 59% of the home range in the high quality category. Home ranges
that did not meet the high or moderate classes defaulted to the low resource quality class.
Table 2. Annual female lynx survival values for all combinations of age classes and resource quality classes used in
population model. Values were determined for each life stage in the high habitat quality class as the highest value from our
literature review less 5%, in the moderate habitat quality class as the mean value from the literature, and in the low habitat
quality class as 25% less than the lowest value in the literature.
Resource Quality Class
Age Class
Low
Moderate
High
Kitten
0.09
0.45
0.74
Yearlings*
0.39
0.60
0.65
Sub-adult
0.39
0.60
0.65
Adult
0.56
0.85
0.88
*data specific to yearlings was unavailable in the literature so were set equal to sub-adults.
Reproduction
Like many aspects of lynx population dynamics, lynx reproduction is closely tied to hare populations and will
fluctuate according to hare density (Aubry et al. 2000). Lynx have a moderate reproductive rate, resulting
primarily from the early age of first reproduction (as early as one year old), a litter size that generally ranges
from 1-4 kittens and a short interval between litters (at most annually, but interval may increase dependent on
prey densities) (Brittell et al. 1989, Koehler 1990, Brainerd 1985, Squires 2016). Fecundity in lynx is defined as
the average number of young per adult female per year. Fecundity values were estimated based on data available
from other lynx populations. In our model only yearling, subadult, and adult females with home ranges that met
the moderate or high habitat quality class as defined in HexSim were allowed to reproduce. Similar to the
survival estimates, we determined fecundity rates in the high habitat quality class as the highest value from our
literature review less 5%, in the moderate habitat quality class as the mean value from the literature, and zero in
the low habitat quality class (Table 3). The age of first reproduction was set at one year.
Table 3. Annual female lynx fecundity values for all combinations of age classes and resource quality classes used in
population model. Values were determined for each life stage in the high habitat quality class as the highest value from our
literature review less 5%, in the moderate habitat quality class as the mean value from the literature, and in the low habitat
quality class as 25% less than the lowest value in the literature.
Resource Quality Class
Age Class
Low
Moderate
High
Kitten
0
0
0
Yearling*
0
0.15
0.29
Sub-adult
0
0.15
0.29
Adult
0
0.83
1.20
*data specific to yearlings was unavailable in the literature so were set equal to sub-adults.
11
Movement
Movement parameters for dispersing individuals were based on information from other lynx populations and
data from the OLMZ. Published information on female lynx dispersal is limited in southern boreal forests (Aubry
et al. 2000) and available information in other ecosystems suggests a wide range of possibilities (Poole 1997,
Mowat et al. 2000). Although information has been recorded on long-distance movements, female lynx may not
disperse long distances, and may establish home ranges that are near or overlap their natal home range (Aubry
et al. 2000, Mowat et al. 2000). We calculated mean home range diameters for female lynx in the OLMZ of
approximately 13km. As such we set the dispersal value as 11km to allow lynx to disperse but still allow for
overlap. Only individuals that failed to acquire adequate resources to establish a home range dispersed. Generally
we assumed that lynx were dispersal habitat generalists and were not strongly influenced by habitat suitability
in their dispersal movements. Marcot et al. (2015) found that HexSim population estimates had relatively low
sensitivity to dispersal movement parameters compared to other model parameters they investigated.
Resource Acquisition and Home Range
Home Range
To determine the home range sizes for HexSim scenarios we calculated 95% minimum convex polygon home
ranges (Calenge 2006) for female lynx in the OLMZ and also used values from lynx work completed earlier in
the analysis area (Koehler 1990). As such the home-range sizes used in the carrying capacity models were 39km2
(Koehler 1990), 55km2 (mid-range value) and 72km2 (the mean OLMZ female home range estimate). In our
model, individual lynx were classified as group members (female lynx with established home ranges), or floaters
(dispersing female lynx without home ranges).
Territoriality
Although lynx may have home ranges that overlap, the degree of overlap, or territoriality, often depends on the
sex of the individuals. Related females and opposite sex tend to be more tolerant of overlap (Poole 1995, Mowat
et al. 2000). We incorporated territoriality by requiring lynx to defend a proportion of their home range, thus
preventing other lynx from using those resources. Quantitative data on territoriality is limited so we analyzed
the spatial and temporal overlap of female lynx in the OLMZ with a straightforward method. We estimated a
60% fixed kernel core home range and then examined the degree of overlap. We found female lynx (n=2) that
exhibited spatial and temporal overlap had 79 – 87% overlap of the core home range. We also examined model
sensitivity to territoriality values and found population size increased to an optimal value when territoriality was
equal to 30%. As territoriality increased the population size and variability decreased. For our scenarios we set
territoriality at 30%, recognizing that this would optimize population densities and values will vary depending
on actual defended territory.
Carrying Capacity Model –Scenarios
Because data on lynx demographics and habitat use can vary considerably, we created several different model
scenarios to examine carrying capacity of the Okanogan and Kettle LMZs and the influence of disturbance. We
developed multiple scenarios to assure key model variables were included and to address the uncertainty
associated with modeling a potential population. A complete description of all model input is provided in
Appendix S1.
Our preliminary analysis resulted in a suite of three different model scenarios that we believed were most
plausible and likely bound the actual carrying capacity of the OLMZ (Table 4). Each model was run for a total
of 175 years, including a 75 year “burn-in” period followed by a 100 year simulation period. Models were
initiated with1000individualsrandomlyplacedacrossthelandscape.The“burn-in”periodallowedpopulations
to approach equilibrium in the landscape and develop a representative distribution of age classes prior to the
12
simulation period (Singleton 2013). We ran the three scenarios on the 2013 post-fire resource map and on the
2000 pre-fire resource map to examine effects of habitat changes on population outcomes. However, it should
be noted that the two different time steps are represented by snapshots of habitat quality at defined intervals. The
model outputs are best interpreted as indices of habitat carrying capacity under landscape conditions at two
specific times, given model uncertainty and assumptions.
Table 4. Description of model scenarios developed to estimate carrying capacity for lynx in the OLMZ. The number in the
Scenario name refers to the home range size used in the model. All models used the same initial resource layer as indicated
by pre-fire or post-fire.
Scenario
Description
39_pre-fire
39 km2 home range size. Resource layer to
describe pre-fire habitat conditions in 2000.
39_post-fire
39 km2 home range size. Resource layer to
describe post-fire habitat conditions in 2013.
55_pre-fire
55 km2 home range size. Resource layer to
describe pre-fire habitat conditions in 2000.
55_post-fire
55 km2 home range size. Resource layer to
describe post-fire habitat conditions in 2013.
72_pre-fire
72 km2 home range size. Resource layer to
describe pre-fire habitat conditions in 2000.
72_post-fire
72 km2 home range size. Resource layer to
describe post-fire habitat conditions in 2013.
We ran 25 population simulation replicates per scenario. Preliminary analysis indicated that 25 replicates were
adequate to capture the variability in annual population size and distribution estimates produced by repeated
simulations. We used simulation-duration mean number of individuals to represent the carrying capacity metric.
We summarized patterns of spatial distribution of the modeled populations across the LMZ by calculating the
annual mean number of female lynx with home ranges by LAU. All model output compilation, statistical analysis
and mapping were conducted using R software (version 3.2.2, R Development Core Team, Vienna, Austria) and
ArcGIS (version 10.4, ESRI, Inc.).
To calibrate our model results we compared our population outcomes with previously calculated density
estimates for the OLMZ and other similar ecosystems. We calculated a population size for each LMZ where:
population size = density estimate from literature (#lynx /100km2) * LAU area (100km2). Although these other
ecosystems may not be at carrying capacity, a comparison of density estimates provided a plausibility test of
model outcomes.
Resource Layer and Carrying Capacity Model – Kettle LMZ
Because site specific information on lynx in the Kettle LMZ is unavailable, we applied the Okanogan LMZ
model to the Kettle LMZ landscape. We developed the resource layer using the same spatial data sources,
specific to the Kettle region. To account for the 2015 fire activity and to provide a more accurate representation
of the current situation, we adjusted the initial RSF output with recent fire activity data. We overlayed Rapid
Assessment of Vegetation Condition after Wildfire (RAVG) data from the US Forest Service to discount RSF
values within fire boundaries in 2015 (http://www.fs.fed.us/postfirevegcondition/whatis.shtml).
RAVG provided a seven-class basal area loss layer (Table 5) that was used to adjust the Kettle 2015 Resource
Map. For the purposes of this exercise, areas within RAVG Classes 1 and 2 did not change the resource map
13
habitat class. Areas within RAVG Class 3 decreased the resource map habitat class by 1 (from high quality to
moderate quality habitat or from moderate quality habitat to low quality habitat). Areas within RAVG Classes
4-7 decreased the resource map habitat class to Class 1-low quality habitat.
Table 5. Rapid Assessment of Vegetation Condition after Wildfire (RAVG) classes and associated loss of basal area
vegetation resulting from wildfire.
RAVG Class
% basal area loss
1
0
2
0 - < 10
3
10 - < 25
4
50 - < 75
5
25 - < 50
6
75 - < 90
7
90 or greater
RESULTS and DISCUSSION
Lynx Habitat Modeling – Okanogan and Kettle LMZs
The most parsimonious RSF habitat model contained greenness, mesic forest, non-forest, surface ruggedness,
aspect, elevation and canopy cover and the quadratics for elevation and canopy cover (Table 6.). This RSF model
had considerably greater empirical support as compared to the remaining models (second “best” model ∆AICc
=30). Habitat selection results indicated a positive relationship with greenness, mesic forest, non-forest,
elevation and canopy cover, a negative relationship with aspect, surface ruggedness and a quadratic effect for
elevation and canopy cover. Correlations between surface ruggedness and slope and between aspect and solar
radiation eliminated pairing of these variables in subsequent models.
Table 6. Parameters and associated coefficients in the Okanogan LMZ habitat model.
Fixed Effects
95% Confidence Interval
Estimate
Standard Error
Lower
Upper
greenness
0.39
0.01
0.29
0.47
mesic forest
0.44
0.03
0.29
0.59
non-forest
2.18
0.08
1.58
2.79
surface ruggedness
-0.6
0.02
-1.04
-0.17
aspect
-0.18
0.01
-0.26
-0.1
elevation
15.54
0.28
11.83
20.26
elevation2
-13.29
0.24
-17.57
-10.08
canopy cover
3.06
0.10
1.96
4.19
canopy cover2
-2.18
0.08
-3.08
-1.15
Intercept
-4.3
0.13
-5.14
-3.36
Selection ratios from model development and evaluation datasets indicated that the threshold for habitat selection
occurred when transformed RSF scores were ≥0.4. We looked at selection of individual lynx to determine
classification and categorized the resource layer map as follows:
Class 1 = RSF score <0.4. Selection less than available (poor quality lynx habitat)
Class 2 = RSF score ~0.4-0.5. Selection equal to or slightly greater than available (moderate quality
lynx habitat).
Class 3 = RSF score >0.5. Selection greater than available (high quality lynx habitat).
14
Eighty-seven percent ofwithheldlynxlocationshadRSFvalues≥0.4(moderatetohighqualityhabitat),whereas
only36%ofthemodeldevelopmentareahadvalues≥0.4.Theresultingresourcemapsdepict relative quality
of habitat across the LMZ in the pre-fire (Fig 4a) and post-fire (Fig. 4b) time periods. Approximately 260 km2
(3%) of lynx habitat in the OLMZ decreased from high quality habitat (Class 3) to low quality habitat (Class 1),
primarily as a result of wildfire.
Our resource maps provided a reasonable and consistent general description of lynx habitat selection in the
Okanogan analysis area. Previous studies in the Okanogan area found lynx select for Engelmann spruce and
subalpine fir forest, moderate canopy cover, flat to moderate slopes, and relatively high elevations; and select
against Douglas-fir and ponderosa pine forests, forest openings, recent burns, sparse canopy and understory, and
relatively steep slopes (Koehler et al. 2008, Maletzke et al. 2008).
Other studies have shown quality foraging habitat for lynx occurs where forest structure provides habitat for
snowshoe hares (Koehler 1990, Agee 2000, Hodges 2000) in the form of dense, multi-layered understory
(Hodges 2000, Lewis et al. 2011) that maximizes cover and browse at varying snow depths throughout the
winter. Such habitat structure is common in early-seral stages but may also occur in coniferous forests with
mature but relatively open overstories (Hodges 2000, Lewis et al. 2011). Another important component of lynx
habitat is areas that are used for denning (Koehler 1990, Moen et al. 2008, Squires et al. 2008) which generally
consists of large woody debris, in the form of either down logs or root wads (Koehler 1990, Mowat et al. 2000,
Slough 1999, Squires et al. 2008). These structures are often associated with late-successional forests and may
be located within older regenerating stands (>20 years since disturbance) or in mature conifer or mixed conifer-
deciduous (typically spruce/fir or spruce/birch) forests (Koehler 1990, Slough 1999, Squires et al. 2008). Lynx
habitat selection was strongly associated with elevation as lynx are highly adapted to environments that receive
considerable winter snowpack (Koehler and Aubry 1994, Aubry et al. 2000, von Keinast 2003, Maletzke 2004).
Recent research in the Okanogan area indicated that lynx avoid recently burned areas, particularly areas that
burned with higher intensity, but may use unburned stands within fire boundaries (Vanbianchi 2015). These
“skips” may provide connectivity across large burned areas. It should be noted that lynx survivorship,
productivity and population dynamics are closely related to snowshoe hare density, although potentially to a
lesser degree in the southern boreal forests (Aubry et al. 2000, Mowat et al. 2000), but subsequent model
complexity and a lack of data prohibited including prey density in our model.
Kettle LMZ Resource Layers
Application of the RSF model to the Kettle LMZ suggested lynx habitat is primarily located along the center of
the LMZ where elevations and habitat types fall into those preferred by lynx. (Fig. 6a and 6b). Approximately
95 km2 (3%) of lynx habitat in the KLMZ decreased from high and moderate quality habitat (Class 3 and 2) to
low quality habitat (Class 1), primarily as a result of wildfire.
Figure 4. Annual mixed effect resource selection function reclassified to relative habitat quality classes mapped within the Okanogan Lynx Management Zone. 4a)
depicts the 2000 pre-fire condition, prior to substantial wildfire activity from 2000-2013, while 4b) depicts the 2013 post-fire condition.
4a) 2000 pre-fire 4b) 2013 post-fire
16
Figure 5. Okanogan Lynx Management Zone post-fire resource map with fire activity polygons from 2000-2015. The majority of fire activity after 2013 occurred outside
of lynx habitat and burn intensity data was unavailable at the time of this assessment so was not included here.
17
Figure 6. Annual mixed effect resource selection function reclassified to relative habitat quality classes mapped within the Kettle Lynx Management Zone. 6a) depicts
the 2000 pre-fire condition prior to substantial wildfire activity from 2000-2015, while 6b) depicts the 2015 post-fire condition. Red polygons depict fire boundaries for
wildfire activity from 2000 through 2015.
6a) 2000 pre-fire 6b) 2015 post-fire
Carrying Capacity Estimates – Okanogan and Kettle LMZs
The range of model outcomes for the pre-fire time period indicated the OLMZ was capable of supporting a lynx
population that ranged from a low of four females to a high of 70 females (Table 7). Results varied greatly depending
on the home range size and, as expected, larger home ranges resulted in smaller carrying capacity estimates. The HexSim
modeling framework also demonstrated the negative impact that wildfire has on carrying capacity for lynx. Habitat
changes due to wildfire resulted in a reduction in total female population estimates ranging from 36-68% (Table 7) as
compared to the pre-fire scenarios. Several simulations reached a population size of zero before the completion of the
run. This suggests that the LMZ may not be capable of sustaining a lynx population in isolation and may be dependent
on immigration, particularly given larger home range size assumptions.
The Kettle LMZ displayed similar results. The range of model outcomes for the pre-fire time period indicated the KLMZ
was capable of supporting a lynx population that ranged from a low of three females to a high of 24 females (Table 7).
Results varied greatly depending on the home range size and, as expected, larger home ranges resulted in smaller carrying
capacity estimates. Habitat changes due to wildfire resulted in a reduction in total female population estimates ranging
from 30-52% (Table 7) as compared to the pre-fire scenarios. Only the 39_prefire scenario replicates reached simulation
completion each time, suggesting the KLMZ may be even more limited than the OLMZ with regard to sustaining a lynx
population in isolation.
Model Calibration: Are these estimates plausible?
Our simulation results provided a range of potential lynx carrying capacity values for the Okanogan and Kettle LMZs.
To examine if these estimates were plausible we compared our results to density estimates from a variety of other
ecosystems. Density estimates ranged considerably depending on the ecosystem (i.e. northern boreal vs. southern boreal)
and snowshoe hare density. As such our population estimate comparisons may be conservative. For this exercise we
compared our estimates to those of southern boreal forests such as the Okanogan (Brittell et al. 1989, Koehler 1990), or
more northerly ecosystems during low hare density years such as the Yukon (Slough and Mowat 1996) and the NW
Territories (Poole 1994), which provided a range of densities from 2 – 3 lynx /100km2.
Based on those densities reflected in the literature we estimated approximately 60-91 females within the Okanogan LMZ.
Our post-fire simulations resulted in population estimates that ranged from 1-45 females in the OLMZ, which was
slightly lower than the range estimated from other Washington studies or ecosystems. Based on those densities reflected
in the literature we estimated approximately 10-15 females within the Kettle LMZ. Our post-fire simulations resulted in
population estimates that ranged from 1-16 females in the KLMZ, which was similar to the range estimated from other
Washington studies or ecosystems.
Spatial patterns of lynx occupancy within both LMZs were generally consistent across the model variants (Fig. 7 and 8).
Lynx were predicted to occur throughout the LAUs within the LMZs in the pre-fire scenarios, while LAUs in the current
scenarios occasionally equaled zero. Predicted lynx abundance generally followed the pattern of the resource map with
higher densities occurring in areas of contiguous higher quality habitat, and then shifted with the post-fire resource layer
and correlated with the location of large wildfires that occurred from 2000-2015.
Okanogan LMZ
LAUs on the east side of the ecosystem (Loomis Central/North/South) and in the Pasayten Wilderness generally had the
highest density of territorial females across pre-fire scenarios (Fig. 7). Including the influence of habitat changes
decreased overall densities throughout the LMZ. Some LAUs with substantial fire activity reached densities of zero.
These patterns were relatively consistent across scenarios.
19
Kettle LMZ
Lambert, Indian and U.S. LAUs generally had the highest density of territorial females across pre-fire scenarios (Fig. 8).
This seemed reasonable given the high quality habitat mapped by the RSF model along the central portion of the LMZ.
However, including the influence of habitat changes decreased overall densities throughout the LMZ. Lambert and North
Sherman LAUs, centrally located in the LMZ, had the highest post-fire densities. The lowest density LAUs were on the
north end of the LMZ where the majority of the wildfire impacts occurred. These patterns were relatively consistent
across scenarios.
The spatial distribution estimates along the international border may be somewhat inaccurate because our analysis area
created a false barrier along the northern edge where hypothetical lynx could not disperse and habitat values diminished.
This was an artifact of our model framework that could be addressed by expanding the spatial extent of the model into
Canada. This approach should be considered as our simulations indicated these populations are likely dependent on
immigration from BC for persistence. Because habitat is limited, lynx populations in the US are small relative to the
larger populations in Canada. As such US lynx populations may depend on immigration from populations in Canada to
ensure genetic diversity and population persistence (Ruggiero et al. 2000). Koehler et al. (2008) discussed how trapping,
wildfire and timber harvest contributed to decreased lynx populations in the Okanogan and Kettle regions over 10 years
ago. The challenges associated with these LMZs, such as isolated habitat, increased wildfire, and potential dispersal
obstacles (fencing, major roadways), are still significant conservation challenges.
Conclusion
Through modeled simulations we have estimated the carrying capacity of lynx in the Okanogan and Kettle Lynx
Management Zones, which can inform efforts to manage lynx in these ecosystems. Lynx populations in Washington
have experienced a decline over the past 20 years that can be partially attributed to the loss of quality habitat to wildfire.
Our modeling approach involved estimating carrying capacities for two landscapes that received no immigration from
outside population sources. Using this approach, the small carrying capacities we estimated may correspond to low
probabilities of persistence until habitat conditions in these LMZs could support larger populations. However, the lynx
population in Washington is not isolated and because it is on the margin of their range, the connection with the larger
population to the north in Canada is likely sustaining the Washington population. On-going habitat loss and
fragmentation warrants further consideration relative to population persistence. We explored carrying capacity with a
single sex model, acknowledging these models have limitations for representing small population processes (including
Allee effects and demographic stochasticity) that can contribute to small population extinction and meta-population
instability. Additionally, the population we modeled is based on demographic characteristics from resident populations,
however the characteristics of the lynx population currently residing in the Okanogan LMZ may differ substantially from
a typical resident population. The population’srelatively small size, its position at the margin of the range, the possible
limitations of demographic support (i.e. immigration) from BC, and the fragmented configuration of habitat within the
LMZ, may significantly influence the sex ratio, age structure and reproductive potential of this population, and
ultimately, its probability of persistence. As such, creating a two-sex model to simulate population viability would be a
logical next step to further assess the stability, viability and probability of persistence of the Washington population to
inform recovery objectives and strategies.
20
Table 7. Simulation-duration annual mean number of female individuals for the total, group and floater populations in the Okanogan and Kettle LMZs for six scenarios.
The change in population carrying capacity as a result of changes in habitat due to wildfire and other disturbances, as well as succession, was calculated as the percent
change in total population size between scenarios (Post-Fire – Pre-Fire). Group members were female lynx in the total population with established home ranges and
floaters were dispersing female lynx in the total population without home ranges.
LMZ
Scenario
Total
Population
(#)
90%
quantile
range
SE
Group
Members
(#)
90%
quantile
range
SE
Floaters
(#)
90%
quantile
range
SE
Decrease
in
Population
Size (%)
Number of
simulations
that reached
100 years
(out of 25
simulations)
Mean
persistence
(years)
OKANOGAN
39_pre-fire
70
49-89
0.5
40
29-51
0.3
30
19-40
0.3
25
100
39_post-fire
45
29-63
0.4
25
16-34
0.2
19
11-29
0.2
36
25
100
55_pre-fire
39
24-54
0.4
22
14-31
0.2
16
9-24
0.2
25
100
55_post-fire
21
0-36
0.4
12
0-19
0.2
9
0-17
0.2
46
4
90
72_pre-fire
4
0-14
0.1
2
0-7
0.1
2
0-7
0.1
3
44
72_post-fire
1
0-5
0.1
1
0-3
0.03
1
0-2
0.03
68
0
16
KETTLE
39_pre-fire
24
12-35
0.2
14
7-20
0.1
10
4-16
0.1
25
100
39_post-fire
16
0-28
0.2
9
0-16
0.1
7
0-13
0.1
33
20
87
55_pre-fire
12
0-22
0.2
7
0-13
0.1
5
0-10
0.1
19
87
55_post-fire
8
0-19
0.2
5
0-11
0.1
4
0-9
0.1
30
10
72
72_pre-fire
3
0-9
0.1
1
0-4
0.04
1
0-4
0.04
4
30
72_post-fire
1
0-5
0.1
1
0-3
0.03
1
0-2
0.04
52
0
4
21
Figure 7. Spatial distribution of mean annual territorial female lynx density (# per 100km2) by LAU in the Okanogan LMZ. Differences between scenarios were a result
of resource changes due to disturbance, primarily wildfire, with three different home range sizes (39km2, 55km2, and 72 km2). Color scheme and range of values were
held constant within each home range to show the influence of disturbance on modeled density outcomes.
Home Range: 39km2
Scenario: 39_pre-fire Scenario: 39_post-fire
Annual Female Lynx Density by LAU
(# per 100km2)
0.0
0.1 - 0.5
0.6 - 1.0
1.1 - 1.5
1.6 - 2.0
2.1 - 2.5
OKA_Pop_1kmbuff
22
Figure 7. continued
Home Range: 55km2
Scenario: 55_pre-fire Scenario: 55_post-fire
Annual Female Lynx Density by LAU
(# per 100km2)
0.0
0.1 - 0.3
0.3 - 0.5
0.6 - 0.8
0.8 - 1.0
1.1 - 1.3
1.4 - 1.5
OKA_Pop_1kmbuff
23
Figure 7. continued
Home Range: 72 km2
Scenario: 72_pre-fire Scenario: 72_post-fire
Annual Female Lynx Density by LAU
(# per 100km2)
0.00
0.01 - 0.03
0.04 - 0.06
0.07 - 0.09
0.10 - 0.12
0.13 - 0.14
OKA_Pop_1kmbuff
24
Figure 8. Spatial distribution of mean annual territorial female lynx density (# per 100km2) by LAU in the Kettle LMZ. Differences between scenarios were a result of
resource changes due to disturbance, primarily wildfire, with three different home range sizes (39km2, 55km2, and 72 km2). Color scheme and range of values were held
constant within each home range to show the influence of disturbance on modeled density outcomes.
Home Range: 39km2
Scenario: 39_pre-fire Scenario: 39_post-fire
Annual Female Lynx Density by LAU
(# per 100km2)
0.01 - 1.0
1.1 - 1.5
1.6 - 2.0
2.1 - 2.5
2.6 - 3.0
Kettle LMZ Study Area Boundary
25
Figure 8. continued
Home Range: 55km2
Scenario: 55_pre-fire Scenario: 55_post-fire
Annual Female Lynx Density by LAU
(# per 100km2)
0.01 - 0.5
0.6 - 0.8
0.8 - 1.0
1.1 - 1.3
1.4 - 1.5
Kettle LMZ Study Area Boundary
26
Figure 8. continued
Home Range: 72 km2
Scenario: 72_pre-fire Scenario: 72_post-fire
Annual Female Lynx Density by LAU
(# per 100km2)
0.00
0.01 - 0.05
0.06 - 0.10
0.11 - 0.15
0.16 - 0.20
0.21 - 0.25
Kettle LMZ Study Area Boundary
27
LITERATURE CITED
Agee, J.K. 2000. Disturbance ecology of North American boreal forests and associated northern/mixed subalpine forests.
Pages 39-82 in L.F. Ruggiero, K.B. Aubry, S.W. Buskirk, G.M. Koehler, C.J. Krebs, K.S. McKelvey, and J.R. Squires,
editors. Ecology and conservation of lynx in the United States. University of Colorado Press, Boulder, CO.
Anderson, E. and M. Lovallo. 2003. Bobcat and lynx. Pages 758–786 in G.A. Feldhamer, B.C. Thompson, and J.A.
Chapman, editors. Wild Mammals of North America: Biology, Management, and Economics. Johns Hopkins Press,
Baltimore, Maryland.
Aubry, K.B., G.M. Koehler, and J.R. Squires. 2000. Ecology of Canada lynx in southern boreal forests. Pages 373–396
in L.F. Ruggiero, K.B. Aubry, S.W. Buskirk, G.M. Koehler, C.J. Krebs, K.S. McKelvey, and J.R. Squires, editors.
Ecology and conservation of lynx in the United States. University of Colorado Press, Boulder, CO.
Baig M.H.A., L. Zhang, T. Shuai and Q. Tong. 2014. Derivation of a tasselled cap transformation based on Landsat 8
at-satellite reflectance, Remote Sensing Letters, 5:5, 423-431, DOI: 10.1080/2150704X.2014.915434
Brainerd, S.M. 1985. Reproductive ecology of bobcats and lynx in western Montana. M.S. Thesis. University of
Montana, Missoula.
Brittell, J.D., R.J. Poelker, S.J. Sweeney, and G.M. Koehler. 1989. Native cats of Washington. Unpublished. Olympia,
WA, Washington Department of Wildlife.
Burnham, K.P., and D.R. Anderson. 2002. Model selection and inference: a practical information-theoretic approach.
Springer-Verlag, New York, NY.
Calenge, C. 2006. The package adehabitat for the R software: a tool for the analysis of space and habitat use by animals.
Ecological Modelling, 197: 516-519.
Copeland, J.P., J.M. Peek, C.R. Groves, W.E. Melquist, K.S. McKelvey, G.W. McDaniel, C.D. Long, and C.E. Harris.
2007. Seasonal habitat associations of the wolverine in central Idaho. Journal of Wildlife Management, 71(7):2201–
2212.
Devineau, O., T.M. Shenk, G.C. White, P.F. Doherty Jr., P.M. Lukacs, and R.H. Kahn. 2010. Evaluating the Canada
lynx reintroduction programme in Colorado: patterns in mortality. Journal of Applied Ecology, 47(2010):524–531.
Gaines, W.L., P.H. Singleton, and R.C. Ross. 2003. Assessing the Cumulative Effects of Linear Recreation Routes on
Wildlife Habitats on the Okanogan and Wenatchee National Forests. General Technical Report PNW-GTR-586.
November. 89 pages.
Gonzalez, P., R.P. Neilson, K.S. McKelvey, J.M. Lenihan, and R.J. Drapek. 2007. Potential Impacts of Climate Change
on Habitat and Conservation Priority Areas for Lynx canadensis (Canada Lynx). Report to: Watershed, Fish, Wildlife,
Air, and Rare Plants Staff; National Forest System; Forest Service; U.S. Department of Agriculture; Washington, DC
and NatureServe, Arlington, VA. 19pp.
Hessburg, P.F., J.K. Agee, and J.F. Franklin. 2005. Dry forest and wildland fires of the inland Northwest USA:
Contrasting the landscape ecology of the pre-settlement and modern eras. Forest Ecology and Management,
211(2005):117-139.
Hodges, K.E. 2000. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163-206 in L.F. Ruggiero,
K.B. Aubry, S.W. Buskirk, G.M. Koehler, C.J. Krebs, K.S. McKelvey, and J.R. Squires, editors. Ecology and
conservation of lynx in the United States. University of Colorado Press, Boulder, CO.
Hoving, C.L., D.J. Harrison, W.B. Krohn, J.A. Ronald, M. O'Brien. 2005. Broad-scale predictors of Canada lynx
occurrence in eastern North America. Journal of Wildlife Management, 69(2):739–751.
Huber, P.R., S.E. Greco, N.H. Schumaker, J. Hobbs. 2014. A priori assessment of reintroduction strategies for a native
ungulate: Using HexSim to guide release site selection. Landscape Ecology, 29(4), pp.689–701.
28
Interagency Lynx Biology Team (ILBT). 2013. Canada lynx conservation assessment and strategy. 3rd edition. USDA
Forest Service, USDI Fish and Wildlife Service, USDI Bureau of Land Management, and USDI National Park Service.
Forest Service Publication R1-13-19, Missoula, MT. 128 pp.
Koehler, G.M. 1990. Population and habitat characteristics of lynx and snowshoe hares in north central Washington.
Canadian Journal of Zoology 68(5):845-851.
Koehler, G.M., B.T. Maletzke, J.A. von Kienast, K.B. Aubry, R.B. Wielgus, and R.H. Naney. 2008. Habitat
fragmentation and the persistence of lynx populations in Washington State. Journal of Wildlife Management,
72(7):1518-1524.
Lewis, C.W., K.E. Hodges, G.M. Koehler, and L.S. Mills. 2011. Influence of stand and landscape features on snowshoe
hare abundance in fragmented forests. Journal of Mammalogy, 92:561–567.
Lewis, J. C. 2016. Periodic status review for the Lynx in Washington. Washington Department of Fish and Wildlife,
Olympia, Washington. 17 + iii pp.
Maletzke, B.T. 2004. Winter habitat selection of lynx (Lynx canadensis) in northern Washington. Masters of Science,
Natural Resource Sciences. Washington State University. December 2004. 39pp.
Maletzke, B.T., G.M. Koehler, R.B. Wielgus, K.B. Aubry, and M.A. Evans. 2008. Habitat conditions associated with
lynx hunting behavior during winter in northern Washington. Journal of Wildlife Management, 72(7):1473–1478.
Manly, B.F.J., L.L. McDonald, D.L. Thomas, T.L. McDonald and W.P. Erickson. 2002. Resource selection by animals:
Statistical design and analysis for field studies. 2nd Edition. Kluwer Academic Publishers, Nederlands. 221 pp.
Marcot, B.G., P.H. Singleton, and N.H. Schumaker. 2015. Analysis of sensitivity and uncertainty in an individual-based
model of a threatened wildlife species. Natural Resource Modeling, 28(1):37-58.
McKelvey, K.S., Y.K. Ortega, G.M. Koehler, K.B. Aubry, and D. Brittell. 2000. Canada lynx habitat and topographic
use patterns in north central Washington: a re-analysis. Pages 307-336 in L.F. Ruggiero, K.B. Aubry, S.W. Buskirk,
G.M. Koehler, C.J. Krebs, K.S. McKelvey, and J.R. Squires, editors. Ecology and conservation of lynx in the United
States. University of Colorado Press, Boulder, CO.
Moen, R., C.L. Burdett, and G.J. Niemi. 2008. Movement and habitat use of Canada lynx during denning in Minnesota.
Journal of Wildlife Management, 72:1507-1513.
Mowat,G.,K.G.Poole,andM.O’Donoghue.2000.EcologyoflynxinnorthernCanadaandAlaska. Pages 265-306 in
L.F. Ruggiero, K.B. Aubry, S.W. Buskirk, G.M. Koehler, C.J. Krebs, K.S. McKelvey, and J.R. Squires editors. Ecology
and conservation of lynx in the United States. University of Colorado Press, Boulder, CO.
Ohmann, J.L., M.J. Gregory, E.B. Henderson, and H.M. Roberts. 2011. Mapping gradients of community composition
with nearest-neighbor imputation: extending plot data for landscape analysis. Journal of Vegetation Science 22(4):660-
676.
Perry, D.A., P.F. Hessburg, C.N. Skinner, T.A. Spies, S.L. Stephens, A.H. Taylor, J.F. Franklin, B McComb, and G.
Riegel. 2011. The ecology of mixed severity fire regimes in Washington, Oregon, and Northern California. Forest
Ecology and Management, 262(2011):703-717.
Poole, K. G. 1994. Characteristics of an unharvested lynx population during a snowshoe hare decline. Journal of Wildlife
Management, 58:608-618.
Poole, K.G. 1995. Spatial organization of a lynx population. Canadian Journal of Zoology, 73:632-641.
Poole, K.G. 1997. Dispersal patterns of lynx in the Northwest Territories. Journal of Wildlife Management, 61:497–505.
Proctor, M.F., S.E. Nielsen, W.F. Kasworm, C. Servheen, T.G. Radandt, A.G. Machutchon, and M.S. Boyce. 2015.
Grizzly bear connectivity mapping in the Canada-United States trans-border region. Journal of Wildlife Management,
79(4):544-558.
Ruggiero L.F., K.B. Aubry, S.W. Buskirk, G.M. Koehler, C.J. Krebs, K.S. McKelvey, J.R. Squires. 2000. The scientific
basis for lynx conservation: qualified insights. Pages 443-454 in L.F. Ruggiero, K.B. Aubry, S.W. Buskirk, G.M.
29
Koehler, C.J. Krebs, K.S. McKelvey, and J.R. Squires, editors. Ecology and conservation of lynx in the United States.
University Press of Colorado. Boulder, Colorado, USA.
Schumaker, N.H. 2016. HexSim (Version 4.0.3.0). U.S. Environmental Protection Agency, Environmental Research
Laboratory, Corvallis, Oregon, USA. www.hexsim.net
Singleton, P.H. 2013. Barred owls and northern spotted owls in the eastern Cascade Range, Washington. PhD
Dissertation. University of Washington, Seattle, WA. 197pp.
Slough, B.G. 1999. Characteristics of Canada lynx, Lynx canadensis, maternal dens and denning habitat. Canadian Field
Naturalist, 113:605-608.
Slough, B.G. and G. Mowat. 1996. Lynx population dynamics in an untrapped refugium. Journal of Wildlife
Management, 60:946-961.
Spencer, W., H. Rustigian-Romsos, J. Strittholt, R. Scheller, W. Zielinkski, and R. Truex. 2011. Using occupancy and
population models to assess habitat conservation opportunities for an isolated carnivore population. Biological
Conservation 144:788-803.
Squires, J.R., N.J. Decesare, J.A. Kolbe, L.F. Ruggiero. 2008. Hierarchical den selection of Canada lynx in western
Montana. Journal of Wildlife Management, 72(7):1497-1506.
Squires, J.R., N.J. Decesare, J.A. Kolbe and L.F. Ruggiero. 2010. Seasonal resource selection of Canada lynx in managed
forests of the northern Rocky Mountains. Journal of Wildlife Management, 74(8):1648-1660.
Squires, J.R., N.J. Decesare, L.E. Olson, J.A. Kolbe, M. Hebblewhite and S.A. Parks. 2013. Combining resource
selection and movement behavior to predict corridors for Canada lynx at their southern range periphery. Biological
Conservation, 157:187–195.
Squires, J. 2016. Current Distribution, status, and threats to Canada lynx in Montana and Wyoming. Pages 20-21 in H.
Bell, K. Broderdorp, J. Cummings, B. Holt, M. McCollough, M. Parkin, T. Smith, and J. Zelenak, Canada Lynx Expert
Elicitation Workshop October 13-15, 2015. Final Report.
Stinson, D. W. 2001. Washington state recovery plan for the lynx. Washington Department of Fish and Wildlife,
Olympia, Washington. 78 pp. + 5 maps.
Vanbianchi, C. M. 2015. Habitat use and connectivity for Canada lynx in the North Cascade Mountains, Washington.
Masters of Science, University of British Columbia. Retrieved from
https://open.library.ubc.ca/cIRcle/collections/24/items/1.0166325 (Original work published 2015). 271pp.
Vashon, J., S. McLellan, S. Crowley, A. Meehan, and K. Laustsen. 2012. Canada lynx assessment. Maine Department
of Inland Fish and Wildlife, Research and Assessment Section, Bangor, ME.
von Kienast, J.A. 2003. Winter habitat selection and food habits of lynx on the Okanogan Plateau, Washington. Thesis,
University of Washington, Seattle, USA.
Washington Department of Natural Resources [WDNR]. 2006. Lynx habitat management plan for all DNR managed
lands. Washington Department of Natural Resources. Olympia, WA. 159 pp.
Yan, C., N. Chr. Stenseth, C.J. Krebs, and Z. Zhang. 2013. Linking climate change to population cycles of hares and
lynx. Global Change Biology, 19: 3263–3271. doi:10.1111/gcb.12321.
30
Appendix S1. Literature sources and associated data values for demographic parameters used in the development of the Okanogan
and Kettle Lynx Management Zones Canada lynx carrying capacity models.
Table S1. Summary of lynx GPS data and calculated homerange area based on 95% minimum convex polygon (calculated in R,
adehabitatHR, Calenge 2006). Range and mean home range area for female and male lynx within the OLMZ was used to determine
habitat composition.
ID
Sex
Dates
# locations
95% MCP (km2)
Mean (km2)
LF1
F
3/12-12/12
1243
106.5
LF2
F
3/10-11/10
863
42.1
LF3
F
3/12-8/12
745
67.3
Females: 72.0
LM1
M
2/08-12/09
1787
76.1
LM2
M
2/12-1/13
1733
73.2
LM3
M
2/08-1/09
1574
36.5
LM4
M
3/12-6/13
1604
125.2
LM5
M
3/09-10/10
1900
689.7
LM6
M
3/11-10/11
838
231.4
LM7
M
4/11-10/11
543
19
LM8
M
2/09-10/09
941
118.9
LM9
M
4/08-3/09
631
24.9
LM10
M
2/10-11/10,
3/11-4/12
2998
106
LM11
M
3/11-10/11
740
102.7
Males: 92.6
LM12
M
2/07-3/07
193*
Na
LM13
M
1/07-2/07
126*
Na
*Did not use in model. Too few locations.
Table S2. Sources used to determine survival estimates for NCE lynx carrying capacity models. Data was unavailable specific to
yearlings so we used subadult values for yearlings in all models.
Survival Rate by Age Class
Location
Kittens
SubAdult
Adult
Source
Washington
0.12
Koehler 1990
Maine
0.78
Vashon et al 2012
Seely Lake, MT
0.52
0.75
Squires 2016
Purcells, MT
0.68
0.85
Squires 2016
Washington
0.83
Koehler 1990
Washington
0.89
Brittel et al 1989
Colorado
0.93
Devineau et al. 2010
Colorado
0.82
Devineau et al. 2010
Yukon
0.75
Slough and Mowatt 1996,
O'Donoghue et al. 1997
Yukon
0.90
Slough and Mowatt 1996,
O'Donoghue et al. 1997
NW territories
0.90
Poole 1994
31
Table S3. Sources used to determine fecundity estimates for NCE lynx carrying capacity models.
Litter Size
Study Area
Sample Size
Years
Source
2.24
Seeley Lake, MT
33
1999-2007
Squires 2016
2.95
Purcell Mountains, MT
22
2003 - 2007
Squires 2016
2.0
Okanogan Co, WA
4 litters, 4 litters
1980 -1987
Brittel et al. 1989, Koehler 1990
3.22
MN
9
2004-2006
Moen et al. 2008
2.25 (low hare years)
ME
NA
2006-2010
Vashon et al. 2012
2.74 (high hare years)
ME
NA
1999-2005
Vashon et al. 2012
1.75 (1-3)
MT
18
1985
Brainerd 1985
3.25 (1-5)
MT
18
1985
Brainerd 1985
3.2
NS
154
1977-1980
Parker et al. 1983
3.6
NS
154
1977-1980
Parker et al. 1983
Table S4. Lynx population density estimates from other ecosystems used in comparison with carrying capacity estimates for
Okanogan and Kettle LMZs.
Location
Date of estimate
Density (lynx/100 km2)
Source
Southern Boreal Forests
Okanogan National Forest,
Washington
1985-1987
2.3
Koehler 1990
Okanogan National Forest,
Washington
1989
2
Brittell et al. 1989
Yukon
1987
2.7
Slough and Mowat 1996
NW Territories
1989-1993, population low
after decline in hare numbers
3
Poole 1994
