SPATIAL ECOLOGY OF GRAY WOLVES (CANIS LUPUS) IN THE WESTERN GREAT 
LAKES REGION IN RESPONSE TO ANTHROPOGENIC DISTURBANCE 

By 

Merijn van den Bosch 

A DISSERTATION 

Submitted to 
Michigan State University 
in partial fulfillment of the requirements 
for the degree of 

Fisheries and Wildlife –– Doctor of Philosophy 

2023 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Large carnivores tend to avoid humans and their activities spatially and temporally, yet 

often experience high rates of anthropogenic mortality and their conservation frequently depends 

ABSTRACT 

on coexistence with humans. I investigated how human landscape disturbances influences 

landscape use by gray wolves (Canis lupus) in the western Great Lakes region of the United 

States. Specifically, I investigated whether landscape use by wolves has changed temporally, 

whether it differs among individuals, and how human disturbances may affect the future 

distribution of gray wolves. I characterized wolf recolonization of the western Great Lakes 

region during 1989–2020 by investigating whether the relationship between wolf habitat 

suitability and two indices of human disturbance, human population density and proportions of 

agricultural land cover, has changed during this period. I also modeled habitat selection for the 

western Great Lakes region, distinguishing between territorial and non-territorial wolves and 

conducted a subsequent connectivity analysis. Finally, I assessed wolf recolonization potential 

throughout their former range in the eastern USA by modeling habitat suitability and 

connectivity potential. During 1989–2020, wolves occupied areas with progressively higher 

levels of human disturbance, resulting in an estimated 35% increase in wolf habitat across 

Minnesota, Wisconsin, and Michigan, though range expansion appeared to slow over time. 

Habitat selection of territorial and non-territorial wolves was similar and connectivity analysis 

suggested limited habitat connectivity due to human landscape disturbance, constrained 

recolonization of suitable areas within the western Great Lakes region. I estimated 65.3% of 

former range in the eastern USA is currently unsuitable for wolves, but identified five 

unoccupied areas in the eastern USA that could maintain viable wolf populations. However, 

connectivity between current and potential wolf range appears limited due to high human 

disturbance and the presence of the Great Lakes. This work demonstrates an apparent range 

stabilization of wolves in the Great Lakes region, where little unoccupied habitat remains 

available, but there appears limited potential for further natural recolonization of unoccupied 

habitat throughout the eastern United States. This can inform policy discussions such as the 

desirability of human-assisted recolonization of additional historical range, and the 

accomplishment of targets set when wolves received federal protection. 

 
 
 
 
ACKNOWLEDGEMENTS 

I am deeply indebted to my supervisor, Dr. Jerry Belant, for granting me an opportunity 

beyond my wildest imaginations, for his extensive support in the form of invaluable advice and 

insights, and moreover, for his unwavering belief in my abilities to successfully complete the 

trajectory resulting in this dissertation. I would also like to extend my sincere thanks to those 

serving as my committee members: Dr. Mariela Gantchoff, Dr. Brent Patterson, Dr. Thomas 

Getty, and last but not least, Dr. Ken Kellner, whose abilities to provide analytical support in an 

accessible and pedagogic manner are unmatched. 

This research was made possible by cooperation among the Minnesota, Wisconsin, and 

Michigan Departments of Natural Resources, the USDA, and various tribal agencies. I will not 

attempt individual recognitions but want to thank everyone involved with this gargantuan effort, 

from the wildlife professionals I have had the pleasure of working with directly to the many 

members of the public volunteering their time to collect data upon which most of my analyses 

relied. Funding was provided by the U.S. Fish and Wildlife Service through the Great Lakes Fish 

and Wildlife Restoration Act and the Boone and Crockett Program at Michigan State University.  

I gratefully acknowledge the support I received from my peers, both past and present 

members of our lab, as well as other friends I made along the way.  

I would also like to thank my friends back in Veusseleir and Antwaarpe, who have been 

important to me for many years already and provided me with remote friendship as well as 

revitalizing good times upon my visits home. I will be eternally grateful to my family, 

particularly my parents, for their unconditional love and support, despite my having placed the 

Atlantic Ocean between us. I therefore dedicate this dissertation to them. Finally, though he will 

never know, I thank my very own “Tennessee stud”, Tucker, for coming into my life and being 

my pride and joy ever since he did. 

iii 

 
 
 
 
 
 
 
 
 
 
CHAPTER 1: INTRODUCTION ................................................................................................1                          

TABLE OF CONTENTS 

CHAPTER 2: HUMAN DISTURBANCE LIMITS GRAY WOLF RANGE EXPANSION IN 
THE GREAT LAKES REGION, USA ........................................................................................3 

CHAPTER 3: HABITAT SELECTION OF RESIDENT AND NON-RESIDENT GRAY 
WOLVES: IMPLICATIONS FOR HABITAT CONNECTIVITY ............................................ 17 

CHAPTER 4: IDENTIFYING POTENTIAL GRAY WOLF HABITAT AND CONNECTIVITY 
IN THE EASTERN USA .......................................................................................................... 34 

LITERATURE CITED ............................................................................................................. 58 

APPENDIX .............................................................................................................................. 71 

iv 

 
 
 
 
 
 
 
CHAPTER 1: INTRODUCTION  

The overarching goal of this research is to contribute to knowledge on the spatial ecology 

of gray wolves (Canis lupus) in the western Great Lakes region of the United States. The 

recurring theme is how human landscape disturbances influence landscape use by wolves. Large 

carnivores tend to avoid humans and their activities spatially (Laundre et al., 2010) and 

temporally (Gaynor et al., 2018), yet experience high rates of anthropogenic mortality (Ripple et 

al., 2014; Hill et al., 2022) and their conservation often depends on coexistence with humans 

(Oriol-Cotterill et al., 2015).    

While expanding wolf populations may initially occupy high-quality territories, 

saturation of these areas can lead to occupation of lower quality territories, potentially with more 

human disturbance (Guisan and Theurillat, 2000). This variation in occupied space use should be 

accounted for to accurately forecast distributions of recolonizing carnivores (Guisan and 

Thuiller, 2005). In Chapter 1, I used long-term winter survey data on gray wolves in Minnesota, 

Wisconsin, and Michigan to study changes in the relationship between wolf presence and human 

disturbance during recolonization. The prediction of a negative relationship between gray wolf 

habitat suitability and two indices of human disturbance, human population density and 

proportions of agricultural land cover, was supported. The prediction of wolves expanding their 

range to include areas with higher human disturbance was also supported. I estimated a 35% 

increase in occupied habitat from 1989–1994 to 2016–2020. However, wolf range expansion into 

areas with higher human disturbance appeared to decline over time. Further wolf recolonization 

of the western Great Lakes region appears largely constrained by human disturbances, resulting 

in a stabilization or near stabilization of current range. 

Assessing the spatial ecology of wolves requires understanding movements of dispersing 

individuals, an important biological stage in the life of wolves (Mech and Boitani, 2003). Studies 

generally consider habitat selection across individuals as being equivalent, which can affect the 

accuracy of habitat selection and connectivity analyses. In Chapter 2, I modeled habitat selection 

for the western Great Lakes distinct population segment of gray wolves and assessed habitat 

connectivity between current range and potential areas for range expansion. Using GPS data 

facilitates distinction between wolves established in a territory from those that are dispersing 

(Bunnefeld et al., 2011), and I used a step-selection function (Thurfjell et al., 2014) to test for 

differences between habitat selection of resident and dispersing wolves. I based a connectivity 

1 

 
 
 
analysis on the latter if differences were found. The prediction wolves would select against areas 

of greater human disturbance was supported, but the prediction that selection against disturbance 

would be greater for resident wolves (Rio-Maior et al., 2019) was not supported. I therefore 

conducted connectivity analysis on combined resident and disperser movements and concluded 

limited landscape permeability due to wolf avoidance of human landscape disturbance has 

constrained recolonization of suitable areas within current range and bordering states. 

Increased legal protection and shifting human attitudes has allowed wolves to partially 

reoccupy former range in the USA (Gompper et al., 2015) and Europe (Chapron et al., 2014), yet 

it is unknown to what extent recolonization could continue. As wolf persistence and 

recolonization depends on human tolerance, estimating the potential for further recolonization of 

former range can facilitate the development of management and policy strategies before further 

recolonization occurs (Treves et al., 2004). In Chapter 3, I assessed recolonization potential 

throughout former wolf range in the eastern USA (Novak, 1995), by modeling habitat suitability 

and connectivity potential. I used recent winter survey data to develop an ensemble species 

distribution model for current wolf range in the western Great Lakes region and projected this 

model throughout former wolf range in the eastern USA and current wolf range in southern 

Canada. I estimated 65.3% of former range in the eastern USA is currently unsuitable for wolves 

but identified 6 core areas in the eastern USA that could maintain viable wolf populations, of 

which only the Great Lakes core area is currently occupied. Large areas of wolf habitat remain 

available throughout the eastern USA, but connectivity between current and potential wolf range 

appears limited due to extensive agriculture, high human populations, and presence of the Great 

Lakes.  

Overall, I have demonstrated how recolonizing gray wolves have adapted to human 

disturbance and may further adapt in space and time. There is an apparent range stabilization of 

wolves in the Great Lakes region, where little unoccupied habitat remains available, but there 

appears limited potential for further natural recolonization of unoccupied habitat throughout the 

eastern United States. This can inform policy discussions, such as the desirability of human-

assisted recolonization of additional historical range, and the accomplishment of targets set when 

wolves received federal protection. 

2 

 
 
 
 
CHAPTER 2: HUMAN DISTURBANCE LIMITS GRAY WOLF RANGE 

EXPANSION IN THE GREAT LAKES REGION, USA  

2.1 ABSTRACT 

Species distribution models can facilitate conservation planning and action but presume 

species-environment relationships are stable, which is not the case for invasive or recolonizing 

species only partially occupying their potential distributions. This complicates our understanding 

of colonization and recolonization processes and their effects on species distributions. We 

grouped snow tracking data collected during gray wolf (Canis lupus) recolonization of the 

western Great Lakes region into six periods spanning 1989–2020 and used a species distribution 

model to assess temporal variation in wolf distribution in response to two indices of human 

landscape disturbance: human population density and proportion of agricultural land cover. We 

found consistent negative relationships between these covariates and wolf habitat suitability, 

whereby the strength of these negative relationships decreased over time. Increases in habitat 

area decreased over time and our model estimated about 148 500 km2 of wolf habitat in our 

study area in the first (1989–1994) period, increasing 35% to about 201 000 km2 by the last 

(2016–2020) period. Wolves increasingly occupied areas with higher levels of human 

disturbance during recolonization, demonstrating temporal dynamics in the relationship between 

wolves and indices of human disturbance. The western Great Lakes wolf population may now 

occupy most areas currently suitable and is limited by human landscape disturbances, resulting in 

apparent stabilization of regional wolf range.  

2.2 INTRODUCTION 

The 21st century is characterized by an ongoing anthropogenic mass extinction event, 

with vertebrate extinctions having accelerated up to one hundredfold (Ceballos et al., 2015). 

Species distribution models facilitate conservation planning and action (Guisan and Thuiller, 

2005) by predicting how species ranges shift in response to anthropogenic factors including 

habitat loss and fragmentation (Fahrig et al., 2003) or climate change (Thomas et al., 2004). 

Range shift studies are also crucial to assess the spread of recolonizing, reintroduced, or invasive 

species, whether these promote biodiversity (Ripple and Beschta, 2012) or degrade it (Clavero 

and García-Berthou, 2005; Lockwood et al., 2013). Accurate forecasts of species colonization 

processes are therefore prerequisite to mitigate or effectuate range shifts (Mladenoff and Sickley, 

3 

 
 
 
1998; Jiménez-Valverde et al., 2011).  

Unraveling mechanisms influencing colonization or recolonization includes 

understanding the species’ realized niche through modeling relationships between species 

occurrences and environmental variables (Guisan et al., 2017). However, these models presume 

species-environment relationships are stable (Guisan and Zimmermann, 2000; Elith and 

Leathwick, 2009), which rarely occurs for invasive (Václavík and Meentemeyer, 2012) or 

recolonizing species (Svenning and Skov, 2004) that often only partially occupy their potential 

distribution and are not in spatial equilibrium with their environment (Guisan and Thuiller, 

2005). Using species distribution models developed from presence data that do not represent the 

full range of suitable conditions in an area can reduce model accuracy and usefulness (Václavík 

and Meentemeyer, 2012). 

 Large carnivores are recolonizing parts of their historical range in North America 

(Gompper et al., 2015) and Europe (Chapron et al., 2014). Partial recolonizations of historical 

carnivore ranges resulted from increased legal protection of carnivores and their habitat, and 

changing public perception (Chapron et al., 2014; Gompper et al., 2015). However, human 

disturbance remains the most limiting factor driving large carnivore distributions in human-

dominated landscapes, such as Eurasian lynx (Lynx lynx, Ripari et al., 2022), brown bears (Ursus 

arctos; Lamb et al., 2018), and mountain lions (Puma concolor, Suraci et al., 2019). Gray wolves 

(Canis lupus) occupied most of the conterminous United States but were largely extirpated from 

this area by 1970 (Boitani, 2003). Following federal protection in 1974 through the Endangered 

Species Act (ESA), wolves recolonized large portions of the western Great Lakes region of the 

USA (Ruid et al., 2009). The western Great Lakes population has recently stabilized at around 

4,200 wolves, more recently for Michigan (2011) and Wisconsin (2017) than for Minnesota, 

where stability has been suggested since 2005 or earlier (USFWS, 2020a).   

Wolf recolonization may be spatially constrained by anthropogenic disturbance and 

mortality risk (Mech, 2017; Hill et al., 2022). Earlier research found consistently negative 

relationships between wolf habitat suitability and indices of human disturbance (Mladenoff et al., 

1995; Carroll et al., 2003; Martínez-Meyer et al., 2021), but may have underestimated limits of 

wolf tolerance for human disturbance (Mladenoff et al., 1995; Mladenoff and Sickley, 1998, 

Carroll et al., 2003) by not accounting for wolves occupying an incomplete realized niche 

(Mladenoff et al., 2009; O’Neil et al., 2019). For example, in Minnesota (Mech, 1989) and 

4 

 
 
 
 
 
 
 
 
 
 
 
 
 
Wisconsin (Mladenoff et al., 2009) wolves occupied areas with higher road densities than earlier 

in recolonization (Thiel, 1985). Landscape use of the western Great Lakes region wolf 

population appears dynamic during recolonization which complicates our understanding of their 

potential distribution. 

We used snow tracking data collected during 1989–2020 to assess temporal variation in 

wolf distribution in response to landscape characteristics of the western Great Lakes region, 

USA. We predicted wolves would increasingly occupy areas with greater human disturbance 

during recolonization and range expansion. We also predicted a decline in the rate of 

recolonization, and that wolves currently occupy most accessible suitable areas suggesting 

wolves are approaching a local equilibrium.  

2.3 METHODS 

Study area 

The western Great Lakes region (Figure 2.1) comprises the states of Minnesota (220,185 

km2), Wisconsin (145,594 km2), and Michigan (150,648 km2), USA. This region is dominated by 

forest (44%) in the north, and agricultural land (37%) in the south, with 86% of the area within 

10 km of natural water features (NLCD, 2016; Gantchoff et al., 2021). Elevations are 174–701 

meters above sea level (USGS, 1996). Human populations are concentrated in the southern part 

of each state, with densities > 100 km², and declining sharply across a northern gradient (US 

Census Bureau, 2010; NLCD, 2016). 

When wolves were protected in 1974 through the ESA, the regional population 

comprised about 750 individuals in northeastern Minnesota and a small, isolated population on 

Isle Royale, Michigan (Erb and DonCarlos, 2009). Legal protection facilitated population growth 

in Minnesota, resulting in the recolonization of northern Wisconsin by 1975 (Wydeven et al., 

2009) and the Upper Peninsula of Michigan by 1989 (Beyer et al., 2009). Recent estimates of 

population size are about 4,200 individuals, with 2,600 in Minnesota, 900 in Wisconsin, and 700 

in the Upper Peninsula of Michigan (USFWS, 2020a).  

Data collection 

We used long-term winter track survey data on gray wolves collected by the Departments 

of Natural Resources (DNR) of Minnesota, Wisconsin, and Michigan during 1989–2020. In 

Minnesota, trained natural resource officers were instructed to record locations of all wolf 

5 

 
 
 
 
 
 
 
 
 
 
 
 
  
 
 
 
 
 
 
 
 
 
 
 
sightings and signs (e.g., tracks, scat) observed during work hours from November until 

snowmelt, usually around mid-May (Gantchoff et al., 2022). Participants could record locations 

on forms or maps, but more recently primarily used a web-based GIS application. The final 

dataset was combined with other presence data recorded during other surveys coordinated by the 

DNR (e.g., furbearer survey, carnivore scent station survey). In Wisconsin, staff and trained 

volunteers conducted surveys throughout known wolf range, primarily during December–April 

(Stauffer et al, 2020). The survey area included 164 survey blocks of about 500 km²; delineated 

using waterways, roads, and state boundaries; ensuring each block could be surveyed within a 

day. Employees of the DNR, tribal biologists, and trained volunteers attempted to survey most 

snow-covered roads within a block 1–3 days after snowfall. In the Upper Peninsula of Michigan, 

surveys are conducted along roads and trails using truck or snowmobile (Michigan Department 

of Natural Resources, 2008). Efforts to estimate wolf abundance began in 1989 and initially 

focused on following up on reports of wolf sign observed by MDNR staff and the public and 

searching areas where the last wolf packs occurred in the 1940s–1950s. By 1995, staff surveyed 

suitable habitat across the Upper Peninsula and during 2000–2006, at least 25% of available 

roads and trails were surveyed at least once annually. Since 2007, due to increasing wolf 

abundance, the Upper Peninsula was divided into 21 survey units of which 12–13 were randomly 

selected and surveyed annually during 2007–2011 and 2013–2014, then every other year starting 

2016, ensuring ≥ 60% coverage of the Upper Peninsula. Portions of the Lower Peninsula of 

Michigan were periodically monitored for wolf occurrence based on reported sightings, but there 

are no systematic surveys as wolves have not established there (Beyer et al., 2009). Sampling 

area and methodology varies between states and over the three decades of data collection, though 

we are confident sampling consistently covered the approximate distribution of wolves in each 

state. Monitoring in Minnesota was well-established in 1989 as wolves were not extirpated, 

while the approximate distribution of wolves in Wisconsin and Michigan was well-known in 

early years of data collection, based on public sightings and extensive agency work resulting 

from high public interest in wolves. 

Data selection and processing 

Because areas states surveyed varied across years, we grouped data into six periods 

(1989–1994, 1995–2000, 2001–2005, 2006–2010, 2011–2015, and 2016–2020) that represented 

the shortest intervals of complete spatial coverage. To account for spatial autocorrelation, we 

6 

 
 
 
 
 
 
 
 
 
 
 
 
filtered each dataset to a maximum of one presence point per 3-km cell (Guisan et al., 2017 van 

den Bosch et al., 2022) and randomly generated two pseudo-absence points for each presence 

point, ensuring presence and pseudo-absence points grouped by period did not co-occur within a 

cell. We excluded Isle Royale National Park, Michigan, as we had no survey data available for 

this area. We chose random sampling despite potential sampling bias (Phillips et al., 2009) 

because the regional distribution of wolves was well-known from monitoring during the period 

of data collection. 

Modeling 

To model temporal dynamics in the relationship between wolf habitat suitability and 

human landscape disturbance, we used the National Land Cover Database (30-m resolution) 

(Yang et al., 2016; Dewitz, 2019; NLCD, 2001; 2006; 2008; 2011; 2016; 2020) to derive 

proportion of agricultural land cover (classes planted/cultivated and pasture/hay) (Figure 2.1), 

and layers of human population density (per km2) from 5 yearly Gridded Population of the 

World V4.11 (30 arc-second resolution) datasets (CIESIN, 2000; 2005; 2010; 2015; 2020). We 

extracted these variables from the datasets most temporally aligned with each period of data 

collection. As the earliest and most recent datasets were highly correlated for proportion of 

agricultural land cover (r = 0.97) and human population density (r = 0.98), we assumed 

similarity of these metrics in earlier years and used values from the 2000 dataset for periods for 

which data were unavailable. We rescaled covariates to 3-km resolution to reduce spatial 

mismatch between species data and our covariates (Guisan and Thuiller, 2005). This was the 

finest possible resolution as data contained presence points collected with accuracy to the nearest 

mile (1.61 km). We removed cells classified as water from analysis then scaled continuous 

variables (-1 to 1) to facilitate effect comparison. We did not use variables related to developed 

land cover (e.g. road density, proportion of developed cover, etc.) as surveys were conducted 

along roads which causes positive bias between wolf presence and such variables (van den Bosch 

et al., 2022). We used variance inflation factors (VIF) and pairwise correlations to test for 

multicollinearity between variables. We created two models for pairwise correlation > 0.70, each 

with one of the correlated variables, and removed the variable which resulted in greater AIC 

from the final model (Burnham and Anderson, 2003; Guisan et al., 2017).   

We used a binomial generalized linear model in program R 3.6.2 (R Core Team, 2020), 

with proportion of agricultural land cover and human population density as continuous variables 

7 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
which were not correlated (r =-0.09– -0.10 across six periods), and interactions with the six 

periods of data. We used the area under the curve (AUC) of the receiver-operating characteristic 

(ROC), along with associated sensitivity and specificity scores, to evaluate model performance. 

We considered scores of AUC ≥ 0.90 as excellent, 0.90 > x ≥ 0.80 as good and 0.80 > x ≥ 0.70 as 

fair (Araújo et al., 2005). We used the DHARMa package in R (Hartig et al., 2017) to assess 

model fit diagnostics by testing for data dispersion, distribution, and outliers.  

To assess temporal dynamics in wolf distribution we created binary habitat predictions 

for each period. We transformed habitat suitability maps for each period to binary format by 

defining a threshold suitability that maximized sensitivity and specificity with the ROCR 

package in R (Sing et al., 2005; Guisan et al., 2017). We then estimated total habitat area for 

each period. To quantify uncertainty associated with habitat area estimates, we used a bootstrap 

approach which incorporated random variability in the selection of pseudo-absence points and 

uncertainty around the model parameter estimates. We ran 1,000 bootstrap simulations, whereby 

each simulation fit a model using a randomly generated set of pseudo-absence points. From each 

fitted model, we randomly sampled a set of parameter estimates based on the point estimates and 

associated variance-covariance matrices. We used these estimates to predict habitat suitability. 

This bootstrap approach yielded a distribution of total habitat area for each period, from which 

we calculated a 95% confidence interval. 

We used response curves of the continuous variables to indicate effects on habitat 

suitability for each period (Elith et al., 2005), and binary maps from the previous step to 

calculate changes in estimated habitat area between the bootstrapped predictions for all periods 

(Thuiller et al., 2009). An overall decline in increases of estimated wolf habitat would suggest 

the wolf population is approaching spatial equilibrium. We conducted a post-hoc evaluation of 

mean road density in gained or lost wolf habitat between periods to compare our results to 

previous studies (Thiel, 1985; Mech, 1989, Mladenoff et al., 2009).  

2.4 RESULTS 

The number of presence points for the six periods were 542 (1989–1994), 1,253 (1995–

2000), 1,163 (2001–2005), 1,584 (2006–2010), 1,608 (2011–2015), and 1,588 (2016–2020). Our 

species distribution model had good model performance (AUC = 0.82). There were no 

indications of poor model fit relating to data dispersion, distribution, or outliers (Appendix: 

figure A).  

8 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Our model indicated a negative relationship between wolf habitat suitability and human 

population density, as well as proportional agricultural land cover (Table 2.1, Figure 2.2). This 

negative relationship between habitat suitability and human population density was stronger in 

the first (1989–1994) period than in the fourth (2006–2010), fifth (2011–2015), and sixth (2016–

2020) periods. The negative relationship between habitat suitability and proportional agricultural 

land cover was stronger in the first (1989–1994) period than in the fifth (2011–2015) and sixth 

(2016–2020) periods. Overall, the strength of the negative relationships between habitat 

suitability and our environmental variables decreased over time, particularly for human 

population density. 

Our binary habitat classification estimated about 148 500 km2 of wolf habitat in the first 

(1989–1994) period, increasing 35% to about 201 000 km2 by the last (2016–2020) period 

(Figure 2.3). Estimated wolf habitat increased between periods except between the second 

(1995–2000) and third (2001–2005) periods and increases in habitat overall declined across 

periods. Habitat gains occurred primarily in peripheral occupied wolf range, whereas habitat 

losses (500 km2; 0.003%) were dispersed throughout wolf range (Figure 2.4). Finally, mean road 

densities were greater (1.44 km/km2) in predicted habitat gained between the fifth (2011–2015) 

and last (2016–2020) period than in habitat present in the first period (1989–1994; 0.87 km/km2).  

2.5 DISCUSSION 

We identified a negative relationship between gray wolf habitat suitability and two 

indices of human disturbance, human population density and proportions of agricultural land 

cover, supporting our prediction. Our prediction of wolves expanding their range to include areas 

with higher human disturbance was also supported. We estimated a 35% increase in occupied 

habitat between the first (1989–1994) and last (2016–2020) periods. However, wolf range 

expansion into areas with more human disturbance appeared to slow during more recent periods, 

most notably in response to human population density, indicating wolves may be approaching 

spatial equilibrium within the study area. If wolves have neared spatial equilibrium, we can infer 

the approximate upper thresholds of human disturbance that wolves can occur in the western 

Great Lakes region. For example, using wolf presence locations from habitat gained between the 

last period (2016–2020) and the previous period (2011–2015), mean proportions of agriculture 

and human population densities were 20.73 and 11.40/km2, respectively.  

9 

 
 
 
 
 
 Gray wolves in the Great Lakes region have expanded their range from relatively 

undisturbed areas of northeastern Minnesota to include areas of the Great Lakes region with 

higher human landscape disturbance, where maximum road densities of occupied areas increased 

from 0.53 km/km2 during 1926–1960 (Thiel, 1985) to 0.73 km/km2 during 1969–1986 (Mech, 

1989) and 0.93 km/km2 in 2007 (Mladenoff et al., 2009). This increased use of areas with higher 

human disturbance has continued, given the greater mean road densities in habitat gained during 

2016–2020 (1.44 km/km2).  

We note several limitations of our study. We omitted cells classified as water from 

analysis, as the suitability of water can vary spatially and temporally. The suitability of 

waterbodies depends on their size, as wolves can cross waterbodies up to 2 km by swimming 

(Darimont and Paquet, 2002), while the extent and duration of freeze-over of larger waterbodies 

determines whether they are permeable to wolves (e.g., Orning et al., 2020). Our model predicted 

a decrease in wolf habitat in the third (2001–2005) period compared to the second (1995–2000), 

apparently relating to a temporarily increased negative relationship between habitat suitability 

and proportions of agricultural land cover. This was presumably caused by an unknown bias 

related to sampling effort or changes in sampling protocol, as the strength of the negative 

relationship between habitat suitability and human population density consistently decreased 

across periods.  

The western Great Lakes region wolf population expanded rapidly in numbers from its 

legal protection in 1974 to about 2010, after which population growth decreased markedly 

(USFWS, 2020a). Though our estimates of wolf habitat area increased with each period after 

2010, lower increases between recent periods compared to increases between earlier periods 

suggest that along with lower wolf population growth, wolf range expansion within the Great 

Lakes region has declined. Most areas currently estimated to be unsuitable for wolves have 

markedly higher levels of human disturbance than currently occupied areas except the northern 

Lower Peninsula of Michigan, which has not been recolonized as the Straits of Mackinaw limit 

dispersal, due to periods of limited or no ice formation or when ships have disrupted ice 

formation (Stricker et al., 2019; van den Bosch et al., 2022). The next closest area suitable for 

recolonization in the USA is in western North Dakota and South Dakota, but connectivity to this 

area is limited due to high human disturbance surrounding current wolf range in Minnesota (van 

den Bosch et al., 2022). While the upper thresholds of human disturbance wolves can co-exist 

10 

 
 
 
with are ultimately unknown, most areas suitable for recolonization within the western Great 

Lakes region appear occupied, suggesting the wolf population is approaching spatial equilibrium. 

This stage of equilibrium occurred about 30 years following initial wolf recolonization in the 

region, demonstrating the importance of long-term planning in wolf and other large carnivore 

management during recolonization of historical range.  

2.6 CONCLUSIONS 

Gray wolves have increasingly occupied areas with higher levels of human disturbance 

during recolonization, but the extent of areas reoccupied has declined since about 2010. Due to 

temporal variation in the relationship between environmental variables and the occurrence of a 

recolonizing species, dynamic landscape use of recolonizing species needs to be considered 

during development and interpretation of species distribution models. Further wolf 

recolonization of the western Great Lakes region appears largely constrained by human 

disturbances, resulting in a probable stabilization of current range. This information can aid 

policymakers and managers to determine the course of wolf conservation and management 

within and beyond the western Great Lakes region. 

11 

 
 
 
 
 
  
 
 
TABLES AND FIGURES 

Table 2.1 Model results from a binomial generalized linear model comparing presences and 

pseudo-absences within the western Great Lakes distinct population segment of gray wolves 

(Canis lupus), USA, 1989–2020. Variables included proportion of agricultural land cover (Prop. 

agriculture), human population density (Human pop. dens., per km2), and their interaction with 

six periods of data (reference level: first period = 1989–1994). Continuous variables were scaled 

(-1 to 1), parameter estimates are reported with standard error (SE) and p-values (α < 0.05). 

Parameter 

Intercept 

Estimate 

SE 

P-value 

-2.254 

0.168 

< 0.001 

Prop. agriculture 

-2.043 

0.155 

< 0.001 

Prop. agriculture : 1995–2000 

0.188 

0.184 

Prop. agriculture : 2001–2005 

-0.133 

0.196 

Prop. agriculture : 2006–2010 

0.264 

0.173 

Prop. agriculture : 2011–2015 

0.515 

0.167 

0.306 

0.498 

0.126 

0.002 

Prop. agriculture : 2016–2020 

0.611 

0.167 

< 0.001 

Human pop. dens. 

-2.241 

0.435 

< 0.001 

Human pop. dens. : 1995–2000 

0.304 

0.499 

Human pop. dens. : 2001–2005 

0.913 

0.472 

Human pop. dens. : 2006–2010 

1.464 

0.446 

0.542 

0.053 

0.001 

Human pop. dens. : 2011–2015 

1.503 

0.441 

< 0.001 

Human pop. dens. : 2016–2020 

1.608 

0.442 

< 0.001 

12 

 
 
 
 
 
 
Figure 2.1 Land cover in Minnesota, Wisconsin, and Michigan, USA including adjacent parts of 

the Great Lakes (from National Land Cover Database 2019). 

13 

 
 
 
 
 
 
Figure 2.2 Relationship between predicted habitat suitability for gray wolves (Canis lupus) and 

human population density (per km2, left panel) and proportion of agriculture (%, right panel) in 

Minnesota, Wisconsin, and Michigan, USA, 1989–2020, including 95% confidence intervals 

(shaded). 

14 

 
 
 
 
 
 
Figure 2.3 Predicted change in estimated habitat area for gray wolves (Canis lupus) in 

Minnesota, Wisconsin, and Michigan, USA, 1989–2020. The figure indicates predicted change 

in estimated habitat area (km2). Confidence intervals were obtained using a bootstrap approach. 

15 

 
 
 
 
Figure 2.4 Predicted change in estimated habitat for gray wolves (Canis lupus) in Minnesota, 

Wisconsin, and Michigan, USA, 1989−1994 to 2016−2020. 

16 

 
 
 
 
 
 
CHAPTER 3: HABITAT SELECTION OF RESIDENT AND NON-RESIDENT 

GRAY WOLVES: IMPLICATIONS FOR HABITAT CONNECTIVITY  

3.1 ABSTRACT 

Habitat selection studies facilitate assessing and predicting species distributions and 

habitat connectivity, but habitat selection can vary temporally and among individuals, which is 

often ignored. We used GPS telemetry data from 96 gray wolves (Canis lupus) in the western 

Great Lakes region of the USA to assess differences in habitat selection while wolves exhibited 

resident (territorial) or non-resident (dispersing or floating) movements and discuss implications 

for habitat connectivity. We used a step-selection function (SSF) to assess habitat selection by 

wolves exhibiting resident or non-resident movements, and modeled circuit connectivity 

throughout the western Great Lakes region. Wolves selected for natural land cover and against 

areas with high road densities, with no differences in selection among wolves when resident, 

dispersing, or floating. Similar habitat selection between resident and non-resident wolves may 

be due to similarity in environmental conditions, when non-resident movements occur largely 

within established wolf range rather than near the periphery or beyond the species range. 

Alternatively, non-resident wolves may travel through occupied territories because higher food 

availability or lower human disturbance outweighs risks posed by conspecifics. Finally, an 

absence of differences in habitat selection between resident and non-resident wolf movements 

may be due to other unknown reasons. We recommend considering context-dependency when 

evaluating differences in movements and habitat use between resident and non-resident 

individuals. Our results also provide independent validation of a previous species distribution 

model and connectivity analysis suggesting most potential wolf habitat in the western Great 

Lakes region is occupied, with limited connectivity to unoccupied habitat.  

3.2 INTRODUCTION 

Understanding how animals select habitat is necessary to explain and predict species 

distributions, facilitating population management and species conservation (Rodríguez et al., 

2007, Guisan et al., 2013). Characterizing species-habitat relationships can inform where 

populations can establish (Fahrig et al., 2003) and identify linkages between habitat patches 

suitable for dispersal (Correa Ayram et al., 2016). Whereas habitat selection and associated 

connectivity studies are valuable (Guisan et al., 2013, Correa Ayram et al., 2016), processes 

17 

 
 
 
 
 
underlying habitat selection are often poorly understood (McLoughlin et al., 2010). Drivers of 

habitat selection can differ among life stages or individuals and understanding these differences 

can improve our understanding of habitat selection and connectivity (McLoughlin et al., 2010; 

Benz et al., 2016).  

Mismatches between landscape connectivity analyses and focal species ecology can be 

mitigated by accounting for behavioral aspects that can influence movement (Chetkiewicz et al., 

2006; Benz et al., 2016). Processes underlying animal movements are relevant to connectivity 

analyses as they influence the behavior and movement patterns of dispersing animals (Laundré et 

al., 2010). Human landscape disturbances, including high human population densities and 

associated activities, can substantially alter large carnivore movements (Oriol-Cotterill et al., 

2015; Smith et al., 2016), but avoidance of human disturbance can be lower for non-resident 

individuals than for residents (Rio-Maior et al., 2019). This can affect accuracy of connectivity 

models, as studies generally consider habitat selection across individuals as being equivalent 

(Vasudev et al., 2015, Abrahms et al., 2016).  

Reduced avoidance of human disturbance by non-resident animals has been documented 

for dispersing red wolves (Canis rufus; Hinton et al., 2016) and lions (Panthera leo; Elliot et al., 

2014), which avoided areas near roads and with higher human population densities less strongly 

than residents. Non-resident gray wolves (C. lupus) similarly displayed lower selection against 

human disturbance compared to resident wolves (Dondina et al., 2022; Morales-González et al., 

2022). Alternatively, dispersers may not avoid higher human disturbance at all; resident brown 

bears (Ursus arctos) avoided public roads and resident Iberian lynx (Lynx pardinus) avoided 

low-traffic roads, while dispersers did not (Gastón et al., 2016; Thorsen et al., 2022). 

Gray wolves historically occupied the Northern Hemisphere north of 11–20° N, though 

by 1970 wolves were extirpated from most of their historical range in the contiguous USA 

(Boitani et al., 2003). Following federal protection in 1974, wolves recolonized additional areas 

of Minnesota, and former range in Wisconsin (Wydeven et al., 2009) and the Upper Peninsula of 

Michigan (Beyer et al., 2009). The western Great Lakes population appears to have stabilized at 

around 4,200 wolves (USFWS, 2020a). Unoccupied habitat within former wolf range has been 

identified in the eastern USA, with apparent limited connectivity to current wolf range in the 

Great Lakes region (van den Bosch et al., 2022). However, estimates of habitat availability and 

connectivity should further consider factors underlying habitat selection including potential 

18 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
differences between resident and non-resident movements.   

Gray wolves are territorial, though most wolves disperse from their natal territory and 

become residents of different territories (Mech and Boitani, 2003). Other wolves do not establish 

new territories or join existing territories and remain nomadically “floating” across the 

landscape, constrained by conspecific territories (Boitani, 2003; Mancinelli et al., 2018). 

Preceding dispersal, wolves can make extraterritorial excursions (i.e., predispersal movements) 

of varying distance and duration (Messier 1985; Mech, 2020). Lower avoidance of human 

disturbance by wolves when dispersing or floating could result from avoiding existing wolf 

territories in less human-disturbed areas or decreased site familiarity that reduces their ability to 

avoid human disturbances, compared to when they are resident of a territory (Laundré et al., 

2010; Imbert et al., 2016). Alternatively, disturbances such as roads may facilitate efficient travel 

for non-residents (Hill et al., 2021), while areas with high livestock densities may provide food 

when lower site familiarity or prey abundance limits acquisition of wild prey (Imbert et al., 

2016). 

We investigated habitat use by gray wolves in the western Great Lakes region 

exhibiting resident (territorial) or non-resident (dispersing or floating) movements relative to 

human disturbance. We predicted wolves would select for areas with greater natural land cover 

and against areas of greater human disturbance as indexed by road densities and proportions of 

agricultural land cover, with stronger selection during resident than non-resident movements. We 

also quantified habitat selection and connectivity throughout the western Great Lakes region and 

evaluated these results against an existing connectivity map for wolves in the eastern USA. We 

expected a strong correlation between a previous connectivity map developed using winter track 

surveys (van den Bosch et al., 2022) and one resulting from this habitat selection analysis based 

on telemetry data.  

3.3 METHODS 

Study area 

The study area (Figure 3.1) included the area representing the western Great Lakes 

distinct population segment of gray wolves (hereafter, western Great Lakes region; USFWS, 

2008), including Minnesota (220,185 km²), Wisconsin (145,593 km²), and Michigan (151,279 

km²), and parts of North Dakota (108,193 km²), South Dakota (93,571 km²), Iowa (99,971 km²), 

and Illinois (27,190 km²). 

19 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
The study area also included southern Ontario, Canada (515,966 km²), delineated by the 

Area of the Undertaking (the area in Ontario under forest management; Ouellette et al., 2020), 

and southern Manitoba (84,920 km²). The climate is predominantly humid continental, with 

warm summers and cold winters (Beck et al., 2018). Average summer (June–September) minima 

are 7–17° C and maxima are 17–30° C while average winter (December–March) minima are -

25– -6° C and maxima are -10–4° C (Scott and Huff, 1996). Elevations are 30–757 m above sea 

level (USGS, 1996). The study area, excluding the Great Lakes, contains 46% natural land cover 

(primarily various forest types and wetlands) and 18% water, while agricultural and urban areas 

comprise 32% and 4%, respectively (Homer, 2017).  

Data collection and processing 

We used gray wolf GPS telemetry data collected during 2017–2021 by state, federal, and 

tribal agencies of Minnesota, Wisconsin, and Michigan, following their respective animal 

capture and handling protocols. We excluded the first five days of post-capture data from each 

wolf to reduce potential capture effects (Powell et al., 2003). We created a dataset of 96 wolves 

(51 males, 39 females, 6 unknown) collared in Michigan (44), Wisconsin (31), and Minnesota 

(21), with 13- or 16-hour relocation intervals. We compared two preliminary models using 

datasets with 13- or 16-h relocation intervals, found no notable differences, and pooled these 

datasets for analysis.    

We separated resident (territorial) from non-resident (dispersing or floating) annual wolf 

movements by calculating relative net squared displacement (rNSD), which represents the 

squared Euclidian distance between consecutive locations, where a change from low to high 

rNSD values implies dispersal from a territory (Bunnefeld et al., 2011). For each wolf during 

each biological year (starting 15 April), we used the MigrateR package (Spitz et al., 2017) in 

program R (R Core Team, 2021) to fit data to three a priori non-linear models representing 

resident, dispersing, and floating (named ‘nomadic’ in the MigrateR package) movements 

(Bunnefeld et al., 2011). We selected the best supported movement type based on the lowest AIC 

score (Burnham and Anderson, 2004).  

We further separated data by movement type when individual wolves displayed multiple 

movement categories within a biological year, based on visual inspection of the rNSD-plots and 

raw GPS data (Spitz et al., 2017). Visual inspection of movement data to confirm rNSD 

classifications is recommended to override rNSD classifications when suspected to be incorrect 

20 

 
 
 
  
 
 
 
(Spitz et al. 2017). We then calculated 90% bivariate normal kernel utilization distributions to 

approximate annual range size using the ‘kernelUD’ function in AdehabitatHR package 

(Calenge, 2011). We used an independent source to set the maximum annual range size for 

wolves to be considered resident and the associated annual range to be considered a territory 

from state reports during 2017–2021 (561 km2; Erb and Humpal, 2020). We classified 

movements within annual ranges ≤ 561 km2 as resident and reclassified movements within larger 

annual ranges initially classified as resident movements as floating movements. We classified 

extraterritorial movements between a territory and a non-overlapping territory as dispersal 

movements, from the first movement beyond the initial territory to the last movement before 

entering the subsequent territory. We included potential predispersal movements (Messier, 1985; 

Mech, 2020) as dispersal movements by including extraterritorial movements leaving from and 

returning to the same territory, with a duration ≥ 10 days, based on visual inspection.  

Modeling landscape use 

We used step-selection functions (SSF), linking consecutive animal locations and 

contrasting each observed step with three random available steps (Thurfjell et al., 2014). To 

obtain random steps, we pooled individual movements by movement type (Thurfjell et al., 2014) 

then randomly sampled the length and angle of random steps from the distribution of observed 

steps for each type. We used road density, proportion of natural land cover, and proportion of 

agricultural cover as continuous variables, whereby proportional land covers were calculated as 

the percentage of respective land cover types within a cell. We used road data from TIGER/line 

shapefiles (50-m resolution; US Census Bureau, 2020) and the Canadian National Road Network 

(5-m resolution; Statistics Canada, 2020), the most comprehensive road databases for these 

countries including categories ranging from highways to service roads, and roads only accessible 

by four-wheel drive vehicles. We used the North American land change monitoring system 

(NALCMS; 30-m resolution) (Homer, 2017) to calculate proportional land cover. We 

reclassified land covers as natural (managed and unmanaged ‘forest’ classes, ‘shrubland’, 

‘grassland’, ‘barren land’, and ‘wetland’), agricultural (class ‘cropland’), developed, and water 

(Figure 3.1). We resampled rasters to 300-m resolution to reduce spatial mismatch between 

species and environmental data (Guisan and Thuiller, 2005), and rescaled continuous variables (-

1 to 1) to facilitate effect comparison.  

We fit the SSF using a conditional Poisson regression model, which yields equivalent 

21 

 
 
 
 
 
 
 
 
 
 
 
 
  
estimates to the conditional logistic regression model typically used for SSFs (Muff et al., 2020). 

We included random slopes for the continuous variables (Muff et al., 2020) to account for 

individual variation among wolves. We fit the model using the glmmTMB package (Brooks et 

al., 2017) in program R. We used variance inflation factors (VIF) and pairwise correlations to 

test for multicollinearity of variables with thresholds of 10 and 0.70, respectively (Dormann et 

al., 2013). We selected from two candidate models based on the lowest AIC, or the competing 

model (ΔAIC < 2) with fewer terms (Burnham and Anderson, 2004): one model contained road 

density and proportion of natural land cover, and another model contained these variables 

interacting with movement type. We created used-habitat calibration plots (UHC) to visualize 

how well model predictions characterize used locations by plotting the distribution of an 

explanatory variable at used locations and overlaying this with the distribution of explanatory 

variable values predicted by the model (Fieberg et al., 2018). As a measure of ecological 

importance of statistical estimates, we spatially predicted relative strength of selection (RSS; 

Avgar et al., 2017) throughout the study area by calculating RSS as the probability of selecting a 

given point over a point with average variable values in our study area, scaling probabilities from 

0 to 1. 

Modeling connectivity  

To assess landscape-level connectivity without the assumption of animals having 

landscape knowledge, we used Circuitscape software (McRae and Shah, 2009). We inverted the 

RSS surface raster to obtain an estimate of movement resistance (Buchholtz et al., 2020) and 

then replaced each cell from the movement resistance surface with nodes connected by resistors, 

translating connectivity to ‘current flow’. We limited connectivity analysis to non-resident 

movements if the results of our SSF indicated significant differences (α < 0.05) in habitat 

selection for resident and non-resident wolf movements. We incorporated part of Indiana (8 795 

km²) into the study area to avoid a spatial interruption that would bias the circuit connectivity 

model. We assigned 154 points at about 40-km intervals along the perimeter of the study area 

and calculated connectivity between all pairs of points, providing an omnidirectional 

connectivity map for animals moving randomly through the landscape (McRae and Shah, 2009). 

Connectivity between perimeter points is prone to edge effects as connectivity increases near 

these points. We therefore placed perimeter points at the midpoint of a 15-km buffer bordering 

the study area edge, filled cells within this buffer with the average movement resistance value of 

22 

 
 
 
 
 
 
 
 
 
 
 
 
 
the study area, and removed this buffer after analysis (Buchholtz et al. 2020). To assess how the 

connectivity map compared to a previous species distribution model (SDM) and circuit 

connectivity analysis for gray wolves in the eastern USA and southern Canada (van den Bosch et 

al., 2022), we resampled our RSS and circuit connectivity maps to 1-km resolution and 

calculated Pearson’s correlation coefficients between the RSS and SDM rasters and their 

respective circuit connectivity rasters. 

3.4 RESULTS 

We retained 24,540 steps, with a median of 878 steps per wolf (range = 156–3,811 steps). 

Of these steps, 16,668 were classified as resident, 1,656 as dispersing, and 6,216 as floating 

movements, with median step-lengths of 609, 1,011, and 655 meters for resident, dispersing, and 

floating movements, respectively. Resident, dispersing, and floating movements were found for 

72, 20, and 24 wolves, respectively, and 17 of 96 wolves displayed multiple movement types. 

We used a maximum annual range size of 561 km2 for resident wolves to support movement type 

classification using the NSD-method and visual inspection, though average estimated annual 

range sizes were smaller for wolves classified as resident (mean = 195 km2, StDev = 116) than 

for dispersers (9,294 km2, StDev= 15,065) or floaters (8,608 km2, StDev= 11,119). Short-

distance dispersal events occurred (n = 16, median = 54 km), with the longest dispersal 615 km. 

Proportions of natural (VIF = 3.36) and agricultural (VIF = 3.03) cover were correlated (r = -

0.80), and we retained proportion of natural cover as it is inverse to the combined proportions of 

agricultural and urban cover, thus a stronger proxy of human disturbance.  

The model retaining movement type as a factor interacting with proportion of natural 

land cover and road density indicated no habitat selection differences among resident, dispersing, 

and floating wolves (Table 3.1), due to high overlap in used habitat characteristics (Figure 3.2). 

Our final model included road density and proportion of natural cover without 

interactions with movement type, and had a lower AIC value than the model including these 

interactions (ΔAIC = 5.2). This final model suggested wolves avoided areas with greater road 

density and selected for areas with greater proportions of natural cover (Figure 3.3). Calibration 

of our top-ranked model was successful for road density and reasonable for proportions of 

natural cover, based on visual inspection of overlap between predicted and used values in the 

UHC plots (Figure 3.4). The resampled RSS raster (Figure 3.5) was positively correlated 

(r = 0.78) with a previous habitat suitability raster (van den Bosch et al., 2022). 

23 

 
 
 
Areas previously estimated suitable that were not selected by wolves in this study 

included parts of the Lower Peninsula of Michigan, and isolated and fragmented areas east of the 

Missouri river in North and South Dakota. Non-resident movements in our dataset were limited 

to established wolf range. Because we detected no differences in habitat selection between 

resident and non-resident movements, we retained all data for connectivity analysis. Our circuit 

connectivity map indicated highest connectivity for wolves in the northern and eastern parts of 

Ontario and was positively correlated (r = 0.75) with the circuit connectivity raster of the SDM 

(Figure 3.6).  

3.5 DISCUSSION 

Our estimate of gray wolf habitat selection in the western Great Lakes region of the USA 

and southern Ontario and Manitoba, Canada, supported our prediction that wolves avoid areas 

with high road densities and select for areas with high proportions of natural cover. Contrary to 

our prediction, we found no differences among habitat selection of wolves exhibiting resident, 

dispersing, or floating movements. We also confirmed unoccupied habitat in the western Great 

Lakes region is limited, and connectivity between occupied and unoccupied habitat is 

constrained by the Great Lakes, areas of extensive agriculture, and urban areas. 

Habitat selection of resident and non-resident wolves in our study was similar, though 

previous studies found it can differ by disturbance type. Gray wolves in Portugal displayed 

increased tolerance toward roads and settlements during dispersal, but not towards areas with 

higher livestock densities or windfarms (Rio-Maior et al., 2019). The absence of increased 

tolerance for high road densities or proportions of urban or agricultural land cover during non-

residency may be due to non-resident movements in our study occurring within established wolf 

range, causing high similarity or overlap in habitat characteristics to which residents and non-

residents are exposed. Only four dispersing wolves exceeded estimated mean dispersal distances 

(range = 29–148 km) for the western Great Lakes region (Treves et al., 2009). The prevalence of 

short-distance dispersals (median = 54 km), and dispersal and floating movements being largely 

limited to established wolf range, may be caused by increased opportunities for non-residents to 

join existing packs due to high pack densities (Brainerd et al., 2008; Jimenez et al., 2017). 

Alternatively, with little unoccupied habitat available, non-residents may traverse conspecific 

territories when higher prey availability or lower human disturbance in occupied habitat 

outweighs risks of encountering conspecifics (Morales-González et al., 2022).  

24 

 
 
 
 
 
 
 
The relative selection strength (RSS) map was positively correlated with a habitat 

suitability map for the same area based on wolf winter track surveys (van den Bosch et al., 

2022), while derived connectivity maps were also positively correlated. This similarity provides 

validation of current wolf range and habitat connectivity throughout the western Great Lakes 

region from independent data. Our results also suggest that accounting for differences between 

resident and non-resident movements using data transformations (Keeley et al., 2016), or by 

limiting connectivity analyses to non-resident movement data (Elliot et al., 2014; Dondina et al., 

2022), may be unnecessary when non-resident movements are largely limited to established 

range or otherwise similar to resident individuals. 

We note several limitations to our study. Similar to how movement resistance rasters are 

calculated from habitat suitability maps (van den Bosch et al., 2022), we calculated a resistance 

raster by inverting the RSS map resulting from our SSF (Osipova et al., 2019). This is not ideal 

because RSS values are conditional probabilities (Avgar et al., 2017), though no alternative 

approach is available. Similarly, an entirely mechanistic approach to classifying movement types 

is unavailable, and the migrateR package was developed primarily for migratory animals, so 

classification depends in part on visual interpretation of data (Spitz et al., 2017). Also, wolves 

can swim up to 2 km (Darimont and Paquet, 2002) but can cross larger waterbodies during 

freeze-over (Orning et al., 2020). Our approach resulted in waterbodies having above average 

resistance to movement due to low natural, terrestrial cover, but as the Great Lakes are roadless 

they have a lower resistance than areas with low natural cover and high road density. Year-round 

estimates of connectivity are imperfect due to seasonal changes in movement resistance of water. 

Using GPS locations collected at shorter intervals could be used to assess finer-scale wolf 

movements, and may reveal differences in habitat selection among movement types (Rio-Maior 

et al., 2019). Analysis including non-resident movements beyond established range also is 

needed to confirm whether differences in habitat use between resident and non-resident wolves 

depend on differences in the range of conditions they occur in. Finally, wolves generally avoid 

higher road densities but can select for minor, lower traffic roads for efficient travel 

(Zimmermann et al., 2014). The road databases used here generally group unpaved rural roads, 

that wolves are known to use, with roads in suburban areas that wolves would likely avoid, thus 

testing the response to road densities classified by road type was not possible. 

We suggest potential for further recolonization of the western Great Lakes region is low as 

25 

 
 
 
 
unoccupied habitat and habitat connectivity are limited. The Straits of Mackinac can connect 

current range in the Upper Peninsula of Michigan with the Lower Peninsula during freeze-over, 

though recent crossings of the straits have been too infrequent for population establishment 

(Stricker et al., 2019). Recolonization of potential habitat in North and South Dakota is limited 

by low dispersal frequencies and high anthropogenic mortality (Licht and Fritts, 1994), and 

connectivity with current range may be higher through Manitoba than through Minnesota. 

3.6 CONCLUSIONS 

We offer further support that gray wolves in the western Great Lakes region select for 

areas with high proportions of natural cover, and against human disturbance as indexed by road 

densities. We found no differences in habitat selection among wolves that were resident, 

dispersing, or floating. , suggesting the need to limit connectivity analyses to non-resident 

movements, or to apply transformations to data of primarily resident wolf movements, depends 

on the magnitude of differences in habitat characteristics experienced by resident and non-

resident individuals. As most wolf habitat in the western Great Lakes region appears occupied 

and because limited habitat connectivity between currently occupied range and limited 

unoccupied range in the USA part of the western Great Lakes region, further recolonization 

appears most likely through Canada to connect with wolf habitat in North Dakota, and across the 

Straits of Mackinaw to connect with habitat in the Lower Peninsula of Michigan. 

Interjurisdictional cooperation will be important to improve landscape connectivity for gray 

wolves between Canada and the USA. If recolonization of areas beyond current wolf range in the 

Great Lakes region is desired, promoting human-wolf co-existence in areas most likely to be 

recolonized is pertinent, though further natural recolonization within and beyond the Great Lakes 

region appears limited by the dominance of urban and agricultural areas surrounding current 

range. 

26 

 
 
 
 
 
 
TABLES AND FIGURES 

Table 3.1 Model selection results comparing used and available steps within the western Great 

Lakes distinct population segment of gray wolves (Canis lupus), USA and southern Ontario and 

Manitoba, Canada, 2017–2021. Models were ranked using AIC; variables included road density 

(km/km2) and proportion of natural cover, and their interactions with movement type (reference 

level: Resident). Continuous variables were scaled (-1 to 1) and included random slopes for 

continuous variables to account for individual variation among wolves. Parameter estimates are 

reported with standard error (SE) and p-values (α < 0.05). 

Top model                                       AIC = 454,149.5 

Parameter 

Estimate 

SE 

P-value 

Prop. natural cover 

0.335 

0.033 

< 0.001 

Road density 

-0.175 

0.025 

< 0.001 

Second model                                 ΔAIC = +5.2                                        

Parameter 

Estimate 

SE 

P-value 

Prop. natural cover 

0.335 

0.033 

< 0.001 

Road density 

- 0.175 

0.025 

< 0.001 

Road density x Floating 

0.031 

0.052 

Prop. natural cover x Floating 

- 0.069 

0.064 

Road density x Dispersing 

0.037 

0.061 

Prop. natural cover x Dispersing 

0.017 

0.050 

0.557 

0.275 

0.548 

0.734 

27 

 
 
 
  
 
 
Figure 3.1 Land cover within the western Great Lakes distinct population segment of gray 

wolves (Canis lupus), USA and southern Ontario and Manitoba, Canada. 

28 

 
 
 
 
 
Figure 3.2 Characteristics of used and available steps for road density (km/km2, top panel) and 

proportion of natural cover (0–100, bottom panel) for 96 gray wolves (Canis lupus) in the 

western Great Lakes region, USA, and southern Ontario and Manitoba, Canada, 2017–2021. 

Circles represent average selection of wolves by movement type, and error bars are 95% 

confidence intervals of these averages.  

29 

 
 
 
 
 
 
 
 
 
Figure 3.3 Predictions from a step-selection function for gray wolves (Canis lupus) in the 

western Great Lakes region, USA, and southern Ontario and Manitoba, Canada, 2017–2021. 

Predicted values are probabilities of selection relative to the average variable value of used and 

available steps (dashed lines, average proportion of natural land cover = 90.42%, average road 

density = 0.83 km/km2).  

30 

 
 
 
 
 
 
Figure 3.4 Used-habitat calibration plots for proportion of natural cover (top panel) and road 

density (km/km2, bottom panel) used in a step-selection function for gray wolves (Canis lupus) 

in the western Great Lakes region, USA, and southern Ontario and Manitoba, Canada, 2017–

2021. 

31 

 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.5 Predicted relative selection strength (RSS) for the western Great Lakes distinct 

population segment of gray wolves (Canis lupus), USA and southern Ontario and Manitoba, 

Canada based on a step-selection function, along with an approximate southern border of wolf 

range. 

32 

 
 
 
 
 
 
Figure 3.6 Circuit connectivity for the western Great Lakes distinct population segment of gray 

wolves (Canis lupus), USA and southern Ontario and Manitoba, Canada, 2017–2021. Figure 

based on a step-selection function (main figure) and circuit connectivity map derived from the 

same study area based on snow track data (van den Bosch et al., 2022; inset).  

33 

 
 
 
 
 
 
CHAPTER 4: IDENTIFYING POTENTIAL GRAY WOLF HABITAT AND 

CONNECTIVITY IN THE EASTERN USA 

4.1 ABSTRACT 

Following federal protection in 1974, gray wolves (Canis lupus) partially recolonized 

former range in the western Great Lakes region, USA, yet remain absent from most of the 

eastern USA. Understanding potential for further recolonization requires quantifying remaining 

wolf habitat and habitat connectivity. We used recent snow tracking data from the western Great 

Lakes region to create an ensemble distribution model to estimate areas of habitat large enough 

to support gray wolf populations in the eastern USA. We then modeled cost-weighted distances 

between these areas and circuit connectivity to identify potential linkages. Our final distribution 

model had good performance (Receiver Operator Characteristic = 0.87) and suggests wolves 

selected against areas with greater human population densities and proportions of agricultural 

land. Gray wolves currently occupy about 4% of their historical range in the eastern USA, which 

represents 12% of the area estimated to remain suitable. We estimated 35% of range is currently 

suitable for wolves, and 18% of these suitable areas are protected, mostly under state and federal 

jurisdictions. We identified five unoccupied areas where wolves could establish viable 

populations ranging from 18,110 to 725,488 km2. Connectivity between these areas and current 

wolf range is limited primarily by the Great Lakes and extensive agriculture in the Midwest 

USA. Most core habitat areas and priority linkages cross state or country borders, highlighting 

the importance of interjurisdictional cooperation. Our estimates of remaining suitable range and 

the potential for recolonization provide a baseline for the development of policies on gray wolf 

conservation in the eastern USA. 

4.2 INTRODUCTION 

Worldwide, geographic ranges of large carnivores have contracted markedly since the 

1700s due to anthropogenic causes (Ripple et al., 2014; Wolf and Ripple, 2017), primarily 

persecution (Musiani and Paquet, 2004) and habitat and connectivity loss (Crooks et al., 2011), 

which are linked to increases in human populations and land use change (Woodroffe, 2000). 

Range contractions of over 20% have occurred for 80% of large carnivore species, particularly in 

regions with high livestock densities (Wolf and Ripple, 2017). Human encroachment on large 

carnivore habitat also increases potential for conflict with humans (Boudreau et al., 2021), which 

34 

 
 
 
 
can result in direct carnivore mortality and reduce public support for their conservation (Treves 

and Karanth, 2003; Treves and Bruskotter, 2014). 

A better understanding of ecological top-down effects has increased recognition of the 

importance of large carnivores (Ripple et al., 2014). Together with improvements in public 

perceptions (Chapron et al., 2014; Gompper et al., 2015), policymaking has shifted and 

coexistence has become a more prevalent management objective (Linnell et al., 2001). Improved 

coexistence with large carnivores has contributed to partial reoccupation of former ranges in the 

United States (Gompper et al., 2015) and Europe (Chapron et al., 2014), though range 

contractions continue for many species, particularly in Africa and Asia (Wolf and Ripple, 2017). 

Recolonization of former range by large carnivores depends in part on habitat connectivity 

(Hemmingmoore et al., 2020). However, connectivity can be reduced by habitat fragmentation 

(Crooks et al., 2017), to which these species are especially vulnerable due to their low densities 

and large territories (Crooks et al., 2011). Maintaining viable populations of large carnivores 

requires large areas, often thousands of square kilometers (USFWS, 1992; Wielgus, 2002), yet 

only 5.2% of their habitat worldwide is estimated to have protected status (Crooks et al., 2011), 

requiring dispersal through, or persistence in, non-protected landscapes with high human 

disturbance (Boron et al., 2016). This suggests that land sparing, by protecting areas relatively 

free from human disturbance, but also land sharing, by tolerating large carnivores in unprotected 

areas, are critical for recolonization of former range by large carnivores (Chapron et al, 2014; 

Gompper et al, 2015).  

In Europe and North America, species such as gray wolf (Canis lupus), Eurasian lynx 

(Lynx lynx), and American black bear (Ursus americanus) have recolonized former range 

following increased legal protection of large carnivores and their habitat (Linnell et al., 2001; 

Smith et al., 2016). Since 1974 gray wolves have received federal protection in the contiguous 

USA through the Endangered Species Act (ESA) (Ruid et al., 2009), with periods wherein 

protection was removed for some populations (Bergstrom et al., 2009; Olson et al., 2015). 

Currently, the largest wolf population in the contiguous USA is the western Great Lakes distinct 

population segment, which occurs in portions of Minnesota, Wisconsin, and Michigan, and 

contained around 4,200 gray wolves in 2020 (USFWS, 2020a). Wolf habitat in other eastern 

USA states has been identified in New York, Vermont, New Hampshire, and Maine (Mladenoff 

et al., 1998), and 15 states in the Midwest region (Smith et al. 2016), but there are no 

35 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
assessments of remaining habitat throughout former range across the eastern USA (Nowak, 

1995; USFWS, 2009). Gray wolf populations are estimated to require 25,600 km² or 12,800 km² 

for viable independent or immigration-dependent populations, respectively (USFWS, 1992). 

However, there has been no assessment of where natural recolonization of former range is 

plausible, considering habitat availability and landscape connectivity.  

Most gray wolf mortality in the USA is human-caused (Hill et al., 2022), and 

recolonization of former range in the eastern USA may depend largely on human tolerance, 

highlighting the need for conflict prevention and mitigation (Treves et al., 2004). Estimating 

habitat for gray wolf establishment and identifying linkages that might facilitate movement could 

assist development of management strategies and policy. Assessing the role of protected and 

non-protected areas in wolf recolonization is particularly important to management and policy 

planning, as human-wildlife conflicts are more likely in non-protected areas with greater human 

disturbance (Treves and Karanth, 2003, but see Reinhardt et al., 2019). Additionally, evaluating 

natural recolonization potential can help identify areas where re-establishment within former 

range is unlikely without human assistance, or where former range has become unsuitable. 

We developed a distribution model to estimate current wolf range in the western Great 

Lakes region and used it to predict habitat availability throughout former wolf range in the 

eastern USA, expecting wolf habitat to occur mostly in areas with low human disturbance and 

greater proportions of natural land cover. We also estimated habitat connectivity, including 

southern Canada, to identify potential linkages that could connect unoccupied areas with 

currently occupied areas. Finally, we evaluated the ownership status of protected suitable areas, 

potential linkages, and identified areas likely suitable to maintain viable wolf populations. 

4.3 METHODS 

Study area 

Our study included the former range of wolves in the eastern USA (Nowak, 1995), and 

adjacent southern Canada of which most is currently occupied by wolves (Wolf and Ripple, 

2017). We used the Area of the Undertaking of Ontario (Hunt et al., 2005), the zone where 

forests are managed by the province, as the northern limit of the study area (Figure 4.1). The 

overall area excluding water is 4,187,681 km2, of which about 60% is natural land cover (e.g. 

forests, grasslands, shrubland), 33% is agricultural land, and 7% developed (Homer, 2017). 

36 

 
 
 
 
 
 
 
Elevations are 1–2,436 meters above sea level (USGS, 1996), with higher elevations in the 

western part of the study area and the Appalachian range in the east. 

Data collection and processing 

We used winter wolf track survey data collected by the Departments of Natural 

Resources (DNR) of Minnesota (2018), Wisconsin (2018–2020), and Michigan (2018, 2020). In 

Minnesota, trained natural resource professionals were instructed to record locations of all wolf 

sightings and signs (e.g. tracks, scat) observed from November until snowmelt, usually mid-

April (Erb and Sampson, 2013). Participants could record locations on forms or maps, but most 

used a web-based GIS application. The final dataset was combined with presence data recorded 

during other surveys coordinated by the DNR (e.g. furbearer survey, carnivore scent station 

survey). In Wisconsin, DNR staff, tribal biologists, and trained volunteers conducted surveys 

throughout known wolf range, primarily during December–April (Stauffer et al, 2020). The 

survey area included 164 survey blocks of approximately 500 km² each, delineated using 

waterways, roads, and state boundaries, ensuring each block could be surveyed within a day. 

Surveyors attempted to survey most snow-covered roads within a block 1–3 days after snowfall 

and blocks were surveyed on average 2.8 times. In Michigan, the Upper Peninsula is divided into 

21 survey blocks from which a stratified random sample of 12–13 survey blocks are selected for 

surveys every other year, representing ≥ 60% of the total area (Michigan Department of Natural 

Resources, 2008). Michigan Department of Natural Resources staff were assisted by US 

Department of Agriculture Wildlife Services personnel, and surveys occurred during December–

April. Searches for wolf tracks and other sign occurred along trails and roads by truck or 

snowmobile. Only the Upper Peninsula was surveyed as wolves are not established in the Lower 

Peninsula of Michigan (Michigan Department of Natural Resources, 2008).  

We filtered snow tracking data to one presence point per 1 km² across all years to reduce 

spatial autocorrelation (Gantchoff et al., 2021), resulting in 3 689 presence points. Within the 

entirety of each state, we randomly generated twice as many pseudo-absence points as there were 

presence points (Barbet-Massin et al., 2012; Figure 4.2). Assigning pseudo-absences throughout 

former range would imply a mismatch between the environment currently unoccupied and the 

ecological niche of wolves, yet wolf absence in most former range is thought to be primarily a 

consequence of historic anthropogenic activities (Wolf and Ripple, 2017) rather than a 

consequence of differences in physical landscape features (Nowak, 1995). We then removed 

37 

 
 
 
 
 
 
 
 
 
 
 
 
 
duplicate points, pseudo-absence points located in water, and pseudo-absence points on Isle 

Royale National Park, Michigan (USA), where wolves are present but we did not have survey 

data. 

Ensemble modeling 

We used six variables to model wolf presence in the western Great Lakes region. We 

used the North American Land Change Monitoring System (30-m resolution; Homer, 2017) to 

derive proportions of, and distances to, agricultural land (class cropland), and natural land (all 

other classes except water and ice and snow) (Figure 4.1). We excluded areas of the land cover 

class water from the model, as water bodies are typically not considered wolf habitat. We 

combined all natural land covers into a single class, because vegetation types in current wolf 

range, where presence points for the model originate, differ from parts of former wolf range. 

This mismatch could result in former range being estimated unsuitable based on vegetation 

differences, despite wolves being adaptable to most ecoregions (Nowak et al, 1995). For human 

population density we used the Gridded Population of the World (v4.11) database (30 arc-

seconds resolution; CIESIN, 2020). We used the USGS GTOPO30 digital elevation model for 

elevation (30 arc-seconds resolution; USGS, 1996). Before calculating proportional land cover 

variables, we rescaled variables to 1-km resolution to reduce spatial mismatch between species 

data and environmental data (Guisan and Thuiller, 2005). Because wolf surveys were mostly 

conducted along roads, leading to positive bias between wolf presence and variables related to 

developed land cover and road density, we did not include distance to, and proportion of, 

developed land cover, or road density (Gantchoff et al., in revision). As road density has been 

found negatively related to wolf habitat selection in distribution models (Jędrzejewski et al., 

2008; Mladenoff et al., 2009), we used alternative variables relating to human populations and 

agricultural activities as proxies for human landscape disturbance (Smith et al., 2016). We used 

variance inflation factors (VIF) and pairwise correlations to test for multicollinearity of 

variables. For variables with pairwise correlation > 0.70, we created test models with only one of 

the correlated variables, and removed the variable which resulted in lower fit from the final 

model. This was done until all pairwise correlations were < 0.70 and VIF scores were < 10 

(Guisan et al., 2017).   

To reduce potential for overprediction of wolf habitat using single models (Marmion et 

al, 2009), we developed an ensemble model with 10 submodels (Thuiller et al, 2009): artificial 

38 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
neural network (ANN), random forest (RF), mixture discriminant analysis (MDA), maximum 

entropy (MaxEnt), generalized linear model (GLM), generalized additive model (GAM), 

generalized boosted model (GBM), multivariate adaptive regression splines (MARS), surface 

range envelope (SRE), and classification tree analysis (CTA). We created the ensemble model 

and extrapolation beyond current range using the biomod2 package (Thuiller et al, 2009) in 

program R 3.6.2 (R Core Team, 2020). We used 70% of data to calibrate the model and the 

remaining 30% to evaluate model performance, repeating this procedure three times (Guisan et 

al., 2017). We used the area under the curve (AUC) of the receiver-operating characteristic 

(ROC) and the True Skill Statistic (TSS), and their associated sensitivity and specificity scores, 

as evaluation metrics (Allouche et al., 2006). We considered AUC scores > 0.90 as “excellent”, 

0.90 > x > 0.80 as “good” and 0.80 > x > 0.70 as “fair” (Araújo et al, 2005) and limited the 

ensemble model to submodels with a TSS score ≥ 0.5 (Gantchoff et al., 2021). We assigned 

weights to submodels that were retained, proportional to their respective evaluation scores, and 

averaged them using an ensemble model for the western Great Lakes region, and a projection to 

former wolf range in the eastern US. We assessed variable importance by calculating correlations 

between fitted values and three randomly permutated values of each variable (Thuiller et al, 

2009). As resulting values are automatically inverted, low correlation numbers imply low 

contribution of each variable to the model. We used response curves of the variables with 

greatest importance to assess their effects on likelihood of wolf presence (Elith et al, 2005). 

Estimating connectivity 

 We first classified core areas of habitat by transforming the landscape suitability map to 

a binary format, using an optimized probability threshold resulting in maximized TSS scores in 

Biomod2 (Thuiller et al., 2009). We defined core areas as habitat patches ≥ 100 km², 

representing the smallest plausible annual home range (Stauffer et al., 2020). Patches separated ≤ 

2 km were merged (Gantchoff et al., 2020), considered reasonable as wolves can travel up to 72 

km in 24 hours (Mech and Boitani, 2003). Because wolves might use lower-quality habitat than 

residents (Keeley et al., 2016), we created a resistance surface raster using a c8 exponential 

transformation (Zeller et al. 2018) rather than assuming an inverse relationship between habitat 

suitability and landscape resistance. Though wolves cross frozen lakes and rivers during winter 

(Orning et al., 2020), these conditions occur only several months each year in northern portions 

of our study area. We therefore assigned maximum resistance scores to 1-km cells classified as 

39 

 
 
 
 
water to better represent overall natural barriers to dispersal, as the maximum distance recorded 

for swimming wolves is about 2 km (Darimont and Paquet, 2002).   

To map linkages, we calculated least-cost paths (LCP) between core areas using the 

LinkageMapper toolbox in ArcGIS (McRae and Kavanagh, 2011), and created linkages by 

buffering 1 km on either side of LCPs. We calculated cost-weighted distances (CWD) of 

linkages by summing resistance values we derived from the resistance surface raster, using cells 

that intersected with linkages (Zeller et al, 2018, Gantchoff et al. 2021). Linkages between core 

areas are categorized as having low, medium or high connectivity using equal intervals of their 

total CWD, whereby a lower CWD implies higher landscape connectivity. We used the Linkage 

Priority tool to assign linkage importance, equally weighted using linkage permeability, and the 

proximity and value of core areas connected by this linkage (Gallo and Greene, 2018). We 

calculated core area values based on area size, centrality, landscape resistance, and 

area/perimeter ratio. To further assess landscape-level connectivity without the assumption of 

animal landscape knowledge, we used Circuitscape software (McRae and Shah, 2020). We 

assigned 310 points at 40-km intervals around the perimeter of the study area and calculated 

connectivity between all pairs of points, providing an omnidirectional connectivity map for 

animals moving randomly through the landscape (Gantchoff et al., 2021) 

Core area protection status and size   

We assessed the ownership of core areas and linkages with protection status using the 

Protected Areas Database of the United States (USGS, 2018) and the Canadian Protected and 

Conserved Areas Database (Environment Canada, 2020), which include private conservation 

lands and public lands at all jurisdictional levels (Table 4.1; Table 4.2). Only terrestrial parts of 

protected areas were used in analyses. We examined core area size to estimate their potential to 

maintain a viable population alone or with immigration, using areas of > 25,600 and 12,800–25, 

600 km², respectively (USFWS, 1992), and defined these as population-sized core areas (PCAs).  

4.4 RESULTS 

From the initial 3,238 presence points (1,601 in Minnesota, 1,002 in Wisconsin, and 635 

in Michigan) we created a filtered dataset of 2,832 wolf presence points (1,095, 1,141, and 596, 

respectively) and 6,928 pseudo-absence points (Figure 4.2). The proportion of natural land cover 

(VIF = 13.06) and agricultural land cover (VIF = 12.07) were correlated (r = -0.92) and we 

retained proportion of agriculture due to better model fit. No variables used in analyses had VIF 

40 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
> 1.82 or pairwise correlation > 0.57.  

All submodels except the SRE had TSS scores ≥ 0.5 and were included in the ensemble 

model. The ensemble model had ROC and TSS scores of 0.87 and 0.59, respectively, and the 

suitability map had a sensitivity and specificity of 90.7 and 68.3, respectively. Averaged variable 

importance suggested proportion of agricultural cover as most influential (0.32), followed by 

human population density (0.22), and distance to agricultural land cover (0.15). Elevation (0.05) 

and distance to natural land cover (0.03) were of least importance. Our model estimated a strong 

decline in wolf presence likelihood at human population densities 50–75/km2, and an inverse 

relationship with proportions of agricultural land (Figure 4.3). 

Overall, we estimated 1,174,839 km², or 34.7% of former range in the eastern USA 

(3,385,780 km²) as potential wolf habitat. Wolves currently occupy 4.1%, or 139,462 km², of 

their former range, which is 11.9% of the area we estimated suitable. We identified 161 core 

areas comprising 1,760,835 km2 overall, with the largest 5 comprising 94.8% of the total (Figure 

4.4). We identified 6 population-sized core areas (PCAs) including the currently occupied Great 

Lakes PCA (135,060 km2). The 5 unoccupied PCAs (18,110 to 725,488 km²) included 3 which 

could maintain independent populations and 2 estimated as immigration-dependent (Figure 4.5). 

Among core areas, 308,028 km2 (17.5%) had protected area status, most commonly national 

forests (21.3%) (Table 4.1). Protected areas within PCAs were managed primarily by state 

(34.5%) and federal governments (33.6%). 

We identified 242 least-cost paths (LCPs) 3.8–573.6-km long (Figure 4.5), and a total 

linkage area of 26,871 km2 in the eastern USA, of which 3,887 km2 (14.5%) was within 

protected areas. The most common land designation for linkage protection was conservation 

easement (21.9%), and protected areas within linkages were primarily state-managed (35.6%) 

(Table 4.1). The circuit connectivity map suggested highest connectivity within the Great Plains 

PCA, the Great Lakes PCA, and the Appalachian PCA, and in Canada north of Lake Superior 

(Figure 4.6). Direct linkages between the Great Plains and Great Lakes PCAs have medium 

connectivity, while areas south of the Great Lakes PCA have low connectivity (Figure 4.5). 

There is high connectivity between the Great Lakes PCA and currently occupied habitat in 

Canada, extending east to the St. Lawrence River.  

Assuming wolves use the lowest-CWD linkage to cross from the Great Lakes PCA to a 

core area in central Wisconsin where wolves are present (Thiel et al. 2009; Wisconsin DNR, 

41 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2021), we consider this a baseline suitable linkage (Figure 4.5). The only direct linkages between 

PCAs with a CWD equal to or less than this linkage is the Straits of Mackinac which connects 

the Great Lakes PCA with the Lower Peninsula Michigan PCA, and a linkage between the New 

England and Adirondacks PCAs. Four smaller core areas between Montreal and Quebec City, 

Canada, connect large areas of wolf habitat between Manitoba and Ontario with a larger core 

area east of the St. Lawrence River which in turn connects to the New England PCA. This 

potential route contains four short linkages with individual CWDs less than the baseline linkage. 

4.5 DISCUSSION 

We estimated gray wolf habitat across former range in the eastern USA, supporting our 

prediction that wolves occupy areas with low human disturbance and high natural land cover 

(Mladenoff et al., 1998; Smith et al., 2016). We estimated 65.3% of former range in the eastern 

USA is currently unsuitable for wolves. Following USFWS criteria (USFWS, 1992), we 

identified 6 core areas in the eastern USA that could maintain viable wolf populations after 

natural recolonization or reintroduction, of which only the Great Lakes PCA (population-sized 

core area) is currently occupied. Overall, 17.5% of core areas have protected status, primarily 

under state and federal jurisdiction. Connectivity between current wolf range and other PCAs 

appears limited due to extensive agriculture, high human populations, and presence of the Great 

Lakes. Consequently, recolonization of the three easternmost PCAs appears most plausible via 

dispersal through or from current wolf range in southern Canada. 

Gray wolves are habitat generalists, thus climatic, geological, or biological landscape 

features including snow cover, elevation, or vegetation types may not strongly limit their 

distribution (Fechter and Storch, 2014). Habitat generalism formerly allowed wolves to occupy 

most of North America (Nowak, 1995) and suggests the diversity of natural landscapes in the 

eastern USA would not limit recolonization. Our model suggests landscape suitability for wolves 

is primarily limited by high human population density and extensive agriculture, with wolves 

occupying areas with greater proportions of natural cover. Because large parts of the eastern 

USA have been altered by humans (Ellis et al., 2010), we estimate only 34.7% of former wolf 

range (Nowak, 1995) is currently suitable for wolves. 

The Great Plains, Appalachian, and New England PCAs could likely support independent 

populations of wolves whereas the Adirondacks and Lower Peninsula Michigan PCAs appear 

dependent on connectivity with other populations. We acknowledge the criteria we used for areas 

42 

 
 
 
 
 
 
 
 
 
 
 
 
 
required to sustain populations (see USFWS, 1992) may be conservative. We estimated 3 840 

km2 of habitat for a previously disjunct wolf population in central Wisconsin established around 

1993 (Theil et al., 2009) that was at least initially dependent on connectivity with the Great 

Lakes PCA. Wolf populations considered viable can occupy areas 1,500–3,000 km2 within 

protected areas in North America (Fritts and Carbyn, 1995). However, protected areas cover only 

21.6% of unoccupied PCAs, compared to 55.6% in the Great Lakes PCA (Table 4.2). Wolf 

territory size may be inversely related to habitat quality (Kittle et al., 2015), and habitat quality 

in protected areas may benefit from lesser human disturbance and greater natural cover (Bassi et 

al., 2015), so the minimum area requirement for maintaining a population may be greater with 

lower protection status and habitat quality. While human tolerance and conflict mitigation are 

important where humans and large carnivores co-exist, it may be particularly important for wolf 

recolonization and persistence in unprotected areas, which represented most unoccupied wolf 

habitat in our study (Smith et al. 2016; Gantchoff et al., 2020).  

Wolves in the Great Lakes PCA can disperse > 800 km in straight line distance 

(Michigan DNR, unpublished data), considerably greater than the longest linkage (574 km) we 

identified, yet typical dispersals in the Great Lakes region range from 20 to 100 km (Treves et 

al., 2009). Based on least-cost paths and circuit theory, we suggest the low suitability of areas 

connecting current wolf range with other PCAs likely limits recolonization more than linkage 

length does. Considering the CWD of the linkage in central Wisconsin, we suggest human 

disturbance and associated land use limit connectivity of many linkages within the USA, 

particularly in agricultural areas west and south of the Great Lakes PCA. Consequently, most 

linkages between current range and other PCAs incorporate areas in southern Canada, where 

there is reduced human disturbance (Homer et al., 2017). An exception may be the Straits of 

Mackinac, which could connect the Great Lakes and Lower Peninsula Michigan PCAs. Wolves 

can cross these straits when ice-covered, though crossings thus far appear too infrequent for 

population establishment (Stricker et al., 2019). Despite higher connectivity in southern Canada, 

human development along the St. Lawrence River may limit connectivity with the New England 

and Adirondacks PCAs, while anthropogenic mortality in Canada could reduce the frequency of 

dispersal into the eastern USA (Wydeven et al., 1998). Most linkages between current and 

potential range span multiple states or countries, so maintaining or improving connectivity for 

wolves in the eastern USA will depend on interjurisdictional cooperation, particularly in the 

43 

 
 
 
 
absence of federal protection (USFWS, 2020b). Wolves were listed under the Endangered 

Species Act to conserve the species in “a significant portion of its range”, yet controversy has 

arisen over the interpretation of this phrase, and whether this goal has been met (USFWS, 2020b; 

Defenders of Wildlife et al. v. USFWS et al., 2022). Our study provides a framework for 

interpreting current gray wolf range in the eastern USA in relation to potential for further 

recolonization and habitat loss throughout former range. Policy and management toward further 

recolonization of gray wolves should consider potential overlap with former range of the 

federally endangered red wolf (C. rufus) (Hinton et al., 2013). 

Although further recolonization of former range by gray wolves in the eastern USA 

appears restricted due to limited connectivity, our estimates may be conservative because wolf 

populations exhibit behavioral plasticity and can select for areas previously considered 

unsuitable (Mladenoff et al., 2009). Species distribution and connectivity models assume species 

are in equilibrium with their environment, yet recolonizing species are not (Guisan and Thuiller, 

2005). Our models therefore represent only the current state of wolf distribution and potential for 

recolonization, and future changes are difficult to predict due to variation in dispersal rates, 

distances, and directions (Mech and Boitani, 2003). Recolonization of current range in the Great 

Lakes PCA required over 30 years (Treves et al., 2009), and European countries with high 

habitat fragmentation and human population densities are documenting wolf re-establishment 

after 100–200 years of absence (Reinhardt et al., 2019; Van Der Veken et al., 2021). Further, our 

linkages are theoretical least-cost paths and wolves undoubtedly move through areas we have not 

identified. Finally, our analysis does not account for occasional long-distance dispersals over 

frozen lakes (e.g., Orning et al., 2020). Given the conservativeness of our analysis and 

unpredictability of wolf dispersal, natural recolonization beyond the Great Lakes PCA appears 

possible long term.  

Complete prey density data for the eastern USA were unavailable, however, our model 

performed well in identifying suitable areas and provided output similar to previous gray wolf 

distribution models (Mladenoff and Sickley, 1998; Smith et al., 2016). Land cover variables, 

such as natural land cover used in our study, may serve as a coarse surrogate for prey density 

(Hanberry, 2021), on which large carnivores depend (Wolf and Ripple, 2016). Wolves are 

ungulate specialists and white-tailed deer (Odocoileus virginianus) is the primary prey of wolves 

in the Great Lakes PCA (DelGiudice et al., 2009). Conservative estimates suggest PCAs 

44 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
generally support deer densities > 5.8/km² (Hanberry, 2021), except the Adirondacks PCA, with 

typical densities ≤ 2 deer/km2 (Hinton et al., in revision), parts of the Great Plains PCA 

(Hanberry, 2021), and possibly northern Maine (MDIFW, 2007). However, northern Maine has 

2.7–4.0 moose (Alces alces)/km², also an important prey species for wolves (DelGiudice et al., 

2009; Kantar and Cumberland, 2013). Finally, several wolf-occupied areas in the Great Lakes 

PCA have deer densities < 5.8/km² (Hanberry and Hanberry, 2020; Gable et al., 2017); low 

ungulate densities can result in increased use of alternative prey such as beaver (Castor 

canadensis) and smaller mammals (Newsome et al., 2016).   

The suitability of areas for large carnivore recolonization depends not only on their 

environmental conditions, but also human willingness to co-exist with them (Treves and 

Karanth, 2003; Gompper et al, 2015), as recolonization can be limited by human persecution 

(Mech et al., 2019, Recio et al., 2020). Human tolerance depends on real and perceived risks and 

benefits associated with large carnivore presence (Bruskotter and Wilson, 2014). Science-based 

public education on wolves could aid in promoting co-existence with humans (Slagle et al., 

2013; Treves and Bruskotter, 2014), while increased regulations and conflict mitigation 

programs may help when more direct action is required (Musiani and Paquet, 2004; Boudreau et 

al., 2021), particularly in rural areas where humans and wolves share landscapes and interact 

more frequently (Smith et al., 2014). Our results allow for prioritization of areas most likely to 

be reoccupied and consequently, where promoting wolf co-existence with humans is most 

relevant.  

4.6 CONCLUSIONS 

Despite continuing land use change, suitable areas for large carnivore recolonization 

remain available (Smith et al., 2016). Many of those species, such as gray wolves, cougars 

(Puma concolor), and American black bears, have recolonized former range, but only partially 

(Ripple et al., 2014). Nevertheless, we estimate 65% of former wolf range in the eastern USA is 

currently unsuitable, and limited connectivity between occupied and unoccupied habitat may 

limit further wolf recolonization. Moreover, linkages for large carnivores often cross 

jurisdictional borders, highlighting the need for interjurisdictional cooperation if recolonization 

or reintroduction is desired. Our results suggest that while several parts of the eastern USA may 

not be recolonized naturally, there is sufficient remaining wolf habitat for reintroduction. Beyond 

the availability of habitat and linkages for large carnivores, their recolonization and persistence 

45 

 
 
 
 
 
 
 
will require human willingness to share landscapes with them. Knowing where and how large 

carnivore recolonization may occur can aid in prioritizing areas for conservation and promoting 

successful co-existence with humans (Olson et al., 2021).  

46 

 
 
 
 
 
TABLES AND FIGURES 

Table 4.1 Management jurisdictions and land designations of protected areas within core areas 

(i.e. suitable areas > 100 km²) and linkages (i.e. 1 km buffer around least-cost paths connecting 

core area pairs) for gray wolves (Canis lupus) throughout former range in the eastern USA 

(Nowak, 1995). 

Protected status of core areas 

Protected status of linkage 

Management 

jurisdiction 

Area 

(km2) 

% 

Management 

jurisdiction 

Area 

(km2) 

% 

State 

Federal 

106,238 

34.5  State 

1,342 

35.6 

104,589 

34.0  Unknown 

694 

18.4 

Land designation 

33,415 

10.9  Federal 

Unknown 

26,873 

8.7  Private 

537 

14.2 

451 

12.0 

Private 

17,367 

5.6  Land designation 

391 

10.4 

Local government 

10,824 

3.5  Local government 

188 

5.0 

Non-governmental 

organization 

5,644 

1.8 

Non-governmental 

organization 

128 

3.4 

Other 

3,078 

1.0  Other 

42 

1.1 

Land designation 

Area 

(km2) 

% 

Land designation 

Area 

(km2) 

% 

National forest 

65,574 

21.3  Conservation easement 

825 

21.9 

47 

 
 
 
 
 
 
Table 4.1 (cont’d) 

State resource 

management area 

55,329 

18.0  State conservation area 

521 

13.8 

State conservation 

area 

44,808 

14.6 

State resource 

management area 

502 

13.3 

Conservation 

easement 

25,228 

8.2  National forest 

257 

6.8 

State wilderness 

11,459 

3.7  State park 

222 

5.9 

National grassland 

9,550 

3.1 

Local conservation 

area 

9,407 

3.0 

Recreation 

management area 

National Wildlife 

Refuge 

198 

5.3 

180 

4.8 

Wilderness area 

9,221 

3.0  Military land 

147 

3.9 

Unknown easement 

9,007 

2.9  Marine protected area 

136 

3.6 

National park 

7,301 

2.4  Private conservation 

109 

2.9 

State park 

7,133 

2.3  Local park 

91 

2.4 

Other 

51,256 

16.6  Other 

583 

15.5 

48 

 
 
 
 
 
 
Great Plains PCA 

Federal 

State 

Non-governmental 

organization 

Other 

Total area 

Great Lakes PCA 

State 

Federal 

Table 4.2  

Management jurisdictions and protected area land designations within gray wolf (Canis lupus) 

population-sized core areas (> 12 800 km², PCAs) in the eastern USA (Nowak, 1995). 

PROTECTED STATUS OF POPULATION-SIZED CORE AREAS (PCAs) 

MANAGEMENT JURISDICTION 

LAND DESIGNATION 

Area 

(km2) 

%  Great Plains PCA 

State resource management 

Area 

(km2) 

% 

49,384  64.3 

area 

17,984  23.4 

24,079  31.3  Conservation easement 

17,335  22.6 

Local government 

547  0.7  National forest 

2,684  3.5  National grassland 

132  0.2  National park 

76,826 

Other 

Area 

(km2) 

%  Great Lakes PCA 

9,278  12.1 

5,959  7.8 

4,630  6.0 

21,640  28.1 

Area 

(km2) 

% 

30,513  40.6  National forest 

24,196  32.2 

State resource management 

25,965  34.6 

area 

16,541  22.0 

Local government 

7,976  10.6  State conservation area 

13,057  17.4 

Land designation 

7,549  10.1  Local conservation area 

7,745  10.3 

Other 

Total area 

Appalachian PCA 

3,073  4.1  Wilderness area 

75,076 

Other 

Area 

(km2) 

%  Appalachian PCA 

4,208  5.6 

9,329  12.4 

Area 

(km2) 

% 

49 

 
 
 
 
 
Table 4.2 (cont’d) 

Federal 

State 

Joint 

Non-governmental 

organization 

Other 

Total area 

Non-governmental 

organization 

Federal 

State 

38,715  52.3  National forest 

25,209  34.1 

31,175  42.1  State conservation area 

15,321  20.7 

State resource management 

2,351  3.2 

area 

13,538  18.3 

1,259  1.7  Inventoried roadless area 

3,651  4.9 

506  0.7  Wilderness area 

74,006 

  Other 

Area 

10,972  37.3  National forest 

9,143  31.1  Unknown easement 

8,682  29.5  Forest stewardship easement 

3,973  13.5 

State resource management 

2,575  3.5 

13,712  18.5 

Area 

(km2)  % 

4,986  16.9 

4,435  15.1 

New England PCA 

(km2)  %  New England PCA 

Local government 

566  1.9 

area 

Other 

Total area 

64  0.2  Conservation easement 

29,427 

  Other 

Area 

Adirondacks PCA 

(km2)  %  Adirondacks PCA 

State 

13,968  93.5  State wilderness 

Non-governmental 

organization 

Federal 

443  3.0  Unknown easement 

374  2.5  State conservation area 

Local government 

138  0.9  Military land 

3,561  12.1 

3,156  10.7 

9,316  31.7 

Area 

(km2)  % 

9,492  63.5 

2,768  18.5 

1,147  7.7 

369  2.5 

50 

 
 
 
 
Table 4.2 (cont’d) 

State resource management 

Other 

Total area 

16  0.1 

area 

14,939 

  Other 

350  2.3 

813  5.5 

Lower Peninsula Michigan 

Area 

Lower Peninsula Michigan 

Area 

PCA 

Federal 

State 

Private 

Non-governmental 

organization 

Other 

Total area 

(km2)  % 

PCA 

7,781  50.0  State conservation area 

7,493  48.1  National forest 

246  1.6  Military land 

(km2)  % 

7,370  47.3 

6,920  44.5 

578  3.7 

40  0.2  Wild and scenic river 

156  1.0 

Private recreation or 

15  0.1 

education 

15,574 

  Other 

125  0.8 

425  2.7 

51 

 
 
 
 
 
 
Figure 4.1 Land cover throughout former range of gray wolves (Canis lupus) in the eastern USA 

(Nowak, 1995) and southern Canada approximating the northern limit of forest management in 

Ontario.  

52 

 
 
 
 
 
 
Figure 4.2 Binary landscape suitability for gray wolves (Canis lupus) throughout former range in 

the eastern USA (Nowak, 1995) and southern Canada, and presence and pseudo-absence points 

in Minnesota, Wisconsin, and Michigan, USA, used for modeling. 

53 

 
 
 
 
 
 
 
Figure 4.3 Relationship between the three most important variables of an ensemble model to 

predict habitat suitability for gray wolves (Canis lupus) in Minnesota, Wisconsin, and Michigan, 

USA. Submodels are listed in order of relative contribution to the ensemble model: generalized 

boosted model (GBM), maximum entropy (MaxEnt), artificial neural network (ANN), 

multivariate adaptive regression splines (MARS), generalized additive model (GAM), 

generalized linear model (GLM), random forest (RF), flexible discriminant analysis (FDA), 

classification tree analysis (CTA), and surface range envelope (SRE). 

54 

 
 
 
 
 
Figure 4.4 Landscape resistance for gray wolves (Canis lupus) throughout former range in the 

eastern USA (Nowak, 1995) and southern Canada, and the top third linkages in terms of priority 

based on cost-weighted distance and qualities of core areas they connect. Landscape resistance is 

overlaid with core areas of habitat in the USA, including six population-sized core areas (PCAs) 

that could host viable wolf populations (> 12,800 km²) (USFWS, 1992).  

55 

 
 
 
 
 
 
 
Figure 4.5 Core habitat areas (i.e. suitable patch area > 100 km²) for gray wolves (Canis lupus) 

throughout former range in the eastern USA (Nowak, 1995) and southern Canada. Core areas in 

dark green, intersection with protected areas in light yellow. Linkages between core areas are 

categorized using equal intervals of their cost-weighted distance (CWD).  

56 

 
 
 
 
 
 
 
 
 
Figure 4.6 Circuit connectivity for gray wolves (Canis lupus) throughout former range in the 

eastern USA (Nowak, 1995) and southern Canada, representing habitat connectivity for a gray 

wolf moving randomly through the landscape. 

57 

 
 
 
 
 
 
 
LITERATURE CITED 

Abrahms, B., Sawyer, S.C., Jordan, N.R., McNutt, J.W., Wilson, A.M., Brashares, J.S., 2017. 

Does wildlife resource selection accurately inform corridor conservation? J. Appl. Ecol. 54, 
412–422. https://doi.org/10.1111/1365-2664.12714 

Allouche, O., Tsoar, A., Kadmon, R., 2006. Assessing the accuracy of species distribution 

models: prevalence, kappa and the true skill statistic (TSS). J. Appl. Ecol. 43, 1223–1232. 
https://doi.org/10.1111/j.1365-2664.2006.01214.x 

Araújo, M.B., Pearson, R.G., Thuiller, W., Erhard, M., 2005. Validation of species-climate 

impact models under climate change. Glob. Chang. Biol. 11, 1504–1513. 
https://doi.org/10.1111/j.1365-2486.2005.01000.x 

Avgar, T., Lele, S.R., Keim, J.L., Boyce, M.S., 2017. Relative selection strength: Quantifying 

effect size in habitat‐and step‐selection inference. Ecol. Evol. 7, 5322–5330. 
https://doi.org/10.1002/ece3.3122 

Barbet-Massin, M., Jiguet, F., Albert, C.H., Thuiller, W., 2012. Selecting pseudo-absences for 
species distribution models: how, where and how many? Methods Ecol. Evol. 3, 327–338. 
https://doi.org/10.1111/j.2041-210X.2011.00172.x 

Bassi, E., Willis, S.G., Passilongo, D., Mattioli, L., Apollonio, M., 2015. Predicting the spatial 
distribution of wolf (Canis lupus) breeding areas in a mountainous region of Central Italy. 
PloS one 10, e0124698. https://doi.org/10.1371/journal.pone.0124698 

Beck, H.E., Zimmermann, N.E., McVicar, T.R., Vergopolan, N., Berg, A., Wood, E.F., 2018. 

Present and future Köppen-Geiger climate classification maps at 1-km resolution. Sci. Data 
5, 180214. https://doi.org/10.1038/sdata.2018.214 

Benz, R.A., Boyce, M.S., Thurfjell, H., Paton, D.G., Musiani, M., Dormann, C.F., Ciuti, S., 
2016. Dispersal ecology informs design of large-scale wildlife corridors. PLoS One 11, 
e0162989. https://doi.org/10.1371/journal.pone.0162989 

Bergstrom, B. J., Vignieri, S., Sheffield, S. R., Sechrest, W., Carlson, A. A., 2009. The Northern 

Rocky mountain gray wolf is not yet recovered. BioSci. 59, 991–999. 
https://doi.org/10.1525/bio.2009.59.11.11 

Beyer, D.E., Peterson, R.O., Vucetich, J.A., Hammill, J.H, 2009. Wolf population changes in 
Michigan. In: In: Wydeven, A.P., Van Deelen, T.R., Heske, E. (Eds.), Recovery of gray 
wolves in the Great Lakes region of the United States. Springer, New York, NY, pp. 65–85. 
http://doi.org/10.1007/978-0-387-85952-1_5 

Biek, R., Akamine, N., Schwartz, M.K., Ruth, T.K., Murphy, K.M., Poss, M., 2006. Genetic 
consequences of sex-biased dispersal in a solitary carnivore: Yellowstone cougars. Biol. 
Lett. 2, 312–315. https://doi.org/10.1098/rsbl.2005.0437 

Boitani, L., 2003. Wolf conservation and recovery. In: Mech, M. and Boitani, L. (Eds.), Wolves: 
behavior, ecology, and conservation. University of Chicago Press, Chicago, IL, pp. 317–
341. https://doi.org/10.14430/arctic540 

Boron, V., Tzanopoulos, J., Gallo, J., Barragan, J., Jaimes-Rodriguez, L., Schaller, G., Payán, E., 
2016. Jaguar densities across human-dominated landscapes in Colombia: the contribution of 

58 

 
 
 
unprotected areas to long term conservation. PLoS One 11, e0153973. 
https://doi.org/10.1371/journal.pone.0153973 

Boudreau, M., Gantchoff, M., Ramirez-Reyes, C., Conlee, L., Belant, J., Iglay, R. 2022. Using 

habitat suitability and landscape connectivity in the spatial prioritization of public outreach 
and management during carnivore recolonization. J. App. Ecol., 59, 757–767. 
https://doi.org/10.1111/1365-2664.14090 

Brainerd, S.M., Andrén, H., Bangs, E.E., Bradley, E.H., Fontaine, J.A., Hall, W., …, Wydeven, 

A.P., 2008. The effects of breeder loss on wolves. J. Wildl. Manag. 72, 89–98. 
https://doi.org/10.2193/2006-305 

Brooks, M.E., Kristensen, K., Van Benthem, K.J., Magnusson, A., Berg, C.W., Nielsen, A., ..., 
Bolker, B. M., 2017. glmmTMB balances speed and flexibility among packages for zero-
inflated generalized linear mixed modeling. R. J. 9, 378–400. https://doi.org/10.3929/ethz-
b-000240890 

Bruskotter, J.T., Wilson, R.S., 2014. Determining where the wild things will be: using 

psychological theory to find tolerance for large carnivores. Conserv. Lett. 7, 158–165. 
https://doi.org/10.1111/conl.12072 

Buchholtz, E.K., Stronza, A., Songhurst, A., McCulloch, G., Fitzgerald, L.A., 2020. Using 

landscape connectivity to predict human-wildlife conflict. Biol. Conserv. 248, 108677. 
https://doi.org/10.1016/j.biocon.2020.108677 

Bunnefeld, N., Börger, L., van Moorter, B., Rolandsen, C.M., Dettki, H., Solberg, E.J., Ericsson, 
G., 2011. A model-driven approach to quantify migration patterns: individual, regional and 
yearly differences. J. Anim. Ecol. 80, 466–476. https://doi.org/10.1111/j.1365-
2656.2010.01776.x 

Burnham, K.P., Anderson, D.R., 2004. Model selection and multimodel inference: a practical 

information-theoretic approach. Springer, New York, NY. https://doi.org/10.1007/b97636 

Calenge, C., 2011. Home range estimation in R: the adehabitatHR package. Office national de la 

classe et de la faune sauvage: Saint Benoist, Auffargis, France. 

Carroll, C., Phillips, M.K., Schumaker, N.H., Smith, D.W., 2003. Impacts of landscape change 

on wolf restoration success: planning a reintroduction program based on static and dynamic 
spatial models. Conserv. Biol. 17, 536–548. https://doi.org/10.1046/j.1523-
1739.2003.01552.x 

Ceballos, G., Ehrlich, P.R., Barnosky, A.D., García, A., Pringle, R.M., Palmer, T.M., 2015. 

Accelerated modern human–induced species losses: Entering the sixth mass extinction. Sci. 
Adv. 1, e1400253. https://doi.org/10.1126/sciadv.1400253. 

Chapron, G., Kaczensky, P., Linnell, J.D.C., von Arx, M., Huber, D., Andrén, H., López-Bao, 

J.V., Adamec, M., Álvares, F., Anders, O., … , D., Boitani, L., 2014. Recovery of large 
carnivores in Europe’s modern human-dominated landscapes. Science 346, 1517–1519. 
https://doi.org/10.1126/science.1257553 

Chetkiewicz, C.B., St. Clair, C.C., Boyce, M.S., 2006. Corridors for conservation: integrating 

pattern and process. Annu. Rev. Ecol. Evol. Syst. 37, 317–342. 
https://doi.org/10.1146/annurev.ecolsys.37.091305.110050 

59 

 
 
 
CIESIN (Center for International Earth Science Information Network), 2020. Gridded population 
of the world, version 4 (GPWv4.11): population density adjusted to match 2015 revision of 
UN WPP country totals, revision 11. https://doi.org/10.7927/H4F47M65. 

Clavero, M., García-berthou, E., 2005. Invasive species are a leading cause of animal 

extinctions. Trends Ecol. Evol. 20, 110–110. https://doi.org/10.1016/j.tree.2005.01.003 

Core Team, 2020. R: A language and environment for statistical computing. R foundation for 
statistical computing, Vienna, Austria. Available at: https://www. R-project.org/. 

Correa Ayram, C.A., Mendoza, M.E., Etter, A., Salicrup, D.R.P., 2016. Habitat connectivity in 
biodiversity conservation: A review of recent studies and applications. Prog. Phys. Geogr. 
40, 7–37. https://doi.org/10.1177/0309133315598713 

Crooks, K.R., Burdett, C.L., Theobald, D.M., Rondinini, C., Boitani, L., 2011. Global patterns of 
fragmentation and connectivity of mammalian carnivore habitat. Philos. Trans. R. Soc. B 
Biol. Sci. 366, 2642–2651. https://doi.org/10.1098/rstb.2011.0120 

Darimont, C.T., Paquet, P.C., 2002. Gray wolves, Canis lupus, of British Columbia's central and 
north coast: distribution and conservation assessment. Can. Field-Nat. 116, 416–422. 

Defenders of Wildlife et al v. U.S. Fish and Wildlife Service et al., 2022. No. 21-CV-00344-

JSW, 2022 WL 499838 (N.D. Cal. Feb. 10, 2022). United States District Court, Northern 
District of California. Available 
at: https://biologicaldiversity.org/species/mammals/pdfs/Wolf-Order-2022-02-10.pdf 

DelGiudice, G.D., McCaffery, K.R., Beyer, D.E., Nelson, M.E., 2009. Prey of wolves in the 

Great Lakes region. In: Wydeven, A. P., Van Deelen, T. R., Heske, E. (Eds.), Recovery of 
Gray wolves in the Great Lakes region of the United States. Springer, New York, pp. 155–
173. https://doi.org/10.1007/978-0-387-85952-1_10 

Dewitz, J., 2019, National Land Cover Database (NLCD) 2016 Products: U.S. geological survey 

data release. https://doi.org/10.5066/P96HHBIE. 

Dondina, O., Meriggi, A., Bani, L., Orioli, V., 2022. Decoupling residents and dispersers from 
detection data improve habitat selection modelling: the case study of the wolf in a natural 
corridor. Ethol. Ecol. Evol. 34(6), 617–635. 
https://doi.org/10.1080/03949370.2021.1988724 

Dormann, C.F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G., ..., Lautenbach, S., 2013. 
Collinearity: a review of methods to deal with it and a simulation study evaluating their 
performance. Ecography 36, 27–46. https://doi.org/10.1111/j.1600-0587.2012.07348.x 

Elith, J., Ferrier, S., Huettmann, F., Leathwick, J., 2005. The evaluation strip: a new and robust 

method for plotting predicted responses from species distribution models. Ecol. Modell. 
186, 280–289. https://doi.org/10.1016/j.ecolmodel.2004.12.007 

Elith, J., Leathwick, J. R., 2009. Species distribution models: ecological explanation and 
prediction across space and time. Annu. Rev. Ecol. Evol. Syst. 40, 677–697. 
https://doi.org/10.1146/annurev.ecolsys.110308.120159 

Elliot, N.B., Cushman, S.A., Macdonald, D.W., Loveridge, A.J., 2014. The devil is in the 
dispersers: predictions of landscape connectivity change with demography. J. Appl. 
Ecol. 51, 1169–1178. https://doi.org/10.1111/1365-2664.12282 

60 

 
 
 
Ellis, E.C., Klein Goldewijk, K., Siebert, S., Lightman, D., Ramankutty, N., 2010. 

Anthropogenic transformation of the biomes, 1700 to 2000. Glob. Ecol. Biogeogr. 19, 589–
606. https://doi.org/10.1111/j.1466-8238.2010.00540.x 

Environment Canada, 2020. Canadian protected and conserved areas database. Ottawa, Canada: 

Environment and climate change Canada. Available at: 
https://www.canada.ca/en/environment-climate-change/services/national-wildlife- 
areas/protected-conserved-areas-database.html 

Erb, J., DonCarlos, M.W., 2009. An overview of the legal history and population status of 

wolves in Minnesota. In: Wydeven, A.P., Van Deelen, T.R., Heske, E. (Eds.), Recovery of 
gray wolves in the Great Lakes region of the United States. Springer, New York, NY, pp. 
49–64. http://doi.org/10.1007/978-0-387-85952-1_4 

Erb, J.D., Humpal, C., 2020. Minnesota wolf population update 2019. Minnesota Department of 

Natural Resources, St. Paul.  

Erb, J., Sampson, B., 2013. Distribution and abundance of wolves in Minnesota, 2012-13. 

Minnesota Department of Natural Resources. Available at: 
http://files.dnr.state.mn.us/fish_wildlife/wildlife/wolves/2013/wolfsurvey_2013.pdf 

Fahrig, L., 2003. Effects of habitat fragmentation on biodiversity. Annu. Rev. Ecol. Evol. Syst. 

34, 487–515. https://doi.org/10.1146/annurev.ecolsys.34.011802.132419 

Fechter, D., Storch, I., 2014. How many wolves (Canis lupus) fit into Germany? The role of 
assumptions in predictive rule-based habitat models for habitat generalists. PloS one 9, 
e101798. https://doi.org/10.1371/journal.pone.0101798 

Fieberg, J.R., Forester, J.D., Street, G.M., Johnson, D.H., ArchMiller, A.A., Matthiopoulos, J., 
2018. Used-habitat calibration plots: a new procedure for validating species distribution, 
resource selection, and step-selection models. Ecography 41, 737–752. 
https://doi.org/10.1111/ecog.03123 

Fritts, S.H., Carbyn, L.N., 1995. Population viability, nature reserves, and the outlook for gray 

wolf conservation in North America. Restor. Ecol. 3, 26–38. https://doi.org/10.1111/j.1526-
100X.1995.tb00072.x 

Gable, T., Windels,S., Olson, B., 2017. Estimates of white-tailed deer density in Voyageurs 

National Park: 1989-2016. NPS/VOYA/NRR—2017/1427 

Gallo, J.A., Greene, R., 2018. Connectivity analysis software for estimating linkage priority. 

Available at: https://doi.org/10.6084/m9.figshare.5673715.v6 

Gantchoff, M. G., Beyer Jr, D. E., Erb, J. D., MacFarland, D. M., Norton, D. C., Roell, B. J., ..., 
Belant, J. L., 2022. Distribution model transferability for a wide-ranging species, the Gray 
Wolf. Sci. Rep. 12, 13556. 

Gantchoff, M.G., Erb, J.D., MacFarland, D.M., Norton, D.C., Price Tack, J.L., Roell, B.J., 

Belant, J.L., 2021. Potential distribution and connectivity for recolonizing cougars in the 
Great Lakes region, USA. Biol. Conserv. 257, 109144. 
https://doi.org/10.1016/j.biocon.2021.109144 

Gantchoff, M., Laura, C., Belant, J.L., 2020. Planning for carnivore recolonization by mapping 

sex-specific landscape connectivity. Glob. Ecol. Conserv. 21, e00869. 

61 

 
 
 
 
https://doi.org/10.1016/j.gecco.2019.e00869 

Gastón, A., Blázquez‐Cabrera, S., Garrote, G., Mateo‐Sánchez, M.C., Beier, P., Simón, M.A., 

Saura, S., 2016. Response to agriculture by a woodland species depends on cover type and 
behavioural state: insights from resident and dispersing Iberian lynx. J. Appl. Ecol. 53, 814–
824. https://doi.org/10.1111/1365-2664.12629 

Gaynor, K.M., Hojnowski, C.E., Carter, N.H., Brashares, J.S., 2018. The influence of human 

disturbance on wildlife nocturnality. Science 360, 1232–1235. 
https://doi.org/10.1126/science.aar7121 

Gompper, M.E., Belant, J.L., Kays, R., 2015. Carnivore coexistence: America’s recovery. 

Science 347, 382–383. https://doi.org/10.1126/science.347.6220.382-b 

Guisan, A., Theurillat, J.-P., 2000. Equilibrium modeling of alpine plant distribution: how far 
can we go? Phytocoenologia 30, 353–384. https://doi.org/10.1127/phyto/30/2000/353 

Guisan, A., Thuiller, W., 2005. Predicting species distribution: offering more than simple habitat 

models. Ecol. Lett. 8, 993–1009. https://doi.org/10.1111/j.1461-0248.2005.00792.x 

Guisan, A., Thuiller, W., Zimmermann, N.E., 2017. Environmental predictors: issues of 
processing and selection, in: habitat suitability and distribution models. Cambridge 
University Press, Cambridge, pp. 61–109. https://doi.org/10.1017/9781139028271.011 

Guisan, A., Tingley, R., Baumgartner, J.B., Naujokaitis‐Lewis, I., Sutcliffe, P.R., Tulloch, A.I., 
Regan, T.J.,..., Buckley, Y.M., 2013. Predicting species distributions for conservation 
decisions. Ecol. Lett. 16, 1424–1435. https://doi.org/10.1111/ele.12189 

Guisan, A., Zimmermann, N.E., 2000. Predictive habitat distribution models in ecology. Ecol. 

Modell. 135, 147–186. https://doi.org/10.1016/S0304-3800(00)00354-9 

Hanberry, B.B., 2021. Addressing regional relationships between white‐tailed deer densities and 

land classes. Ecol. Evol. 11, 13570–13578. https://doi.org/10.1002/ece3.8084 

Hanberry, B.B., Hanberry, P., 2020. Rapid digitization to reclaim thematic maps of white-tailed 

deer density from 1982 and 2003 in the conterminous US. PeerJ, 8, e8262. 
https://doi.org/10.7717/peerj.8262 

Hemmingmoore, H., Aronsson, M., Åkesson, M., Persson, J., Andrén, H., 2020. Evaluating 

habitat suitability and connectivity for a recolonizing large carnivore. Biol. Conserv. 242, 
108352. https://doi.org/10.1016/j.biocon.2019.108352 

Hill, J.E., Boone, H.M., Gantchoff, M.G., Kautz, T.M., Kellner, K.F., Orning, E.K, Parchizadeh, 
J., Petroelje, T.R., Wehr, N.H., Finnegan, S.P., Fowler, …, Belant, J.L., 2022. Quantifying 
anthropogenic wolf mortality in relation to hunting regulations and landscape attributes 
across North America. Ecol. Evol. 12, e8875. https://doi.org/10.1002/ece3.8875 

Hill, J.E., DeVault, T.L., Belant, J.L., 2021. A review of ecological factors promoting road use 

by mammals. Mamm. Rev. 51, 214–227. https://doi.org/10.1111/mam.12222 

Hinton, J.W., Chamberlain, M.J., Rabon Jr, D.R., 2013. Red wolf (Canis rufus) recovery: a 

review with suggestions for future research. Animals 3, 722–744. 
https://doi.org/10.3390/ani3030722 

Hinton, J.W., Proctor, C., Kelly, M.J., van Manen, F.T., Vaughan, M.R., Chamberlain, M.J., 

62 

 
 
 
 
 
 
 
2016. Space use and habitat selection by resident and transient red wolves (Canis 
rufus). PLoS One 11, e0167603. https://doi.org/10.1371/journal.pone.0167603 

Homer, C., Colditz, R.R., Latifovic, R., Llamas, R.M., Pouliot, D., Danielson, P., …, Vulpescu, 
M., 2017. Developing a new North American land cover product at 30m resolution: 
methods, results and future plans. AGU Fall Meeting Abstracts Vol. 2017, pp. GC52C-01. 

Hunt, L.M., Boxall, P., Englin, J., Haider, W., 2005. Remote tourism and forest management: a 

spatial hedonic analysis. Ecol. Econ. 53, 101–113. 
https://doi.org/10.1016/j.ecolecon.2004.06.025 

Imbert, C., Caniglia, R., Fabbri, E., Milanesi, P., Randi, E., Serafini, M., ... Meriggi, A., 2016. 
Why do wolves eat livestock? Factors influencing wolf diet in northern Italy. Biol. 
Conserv. 195, 156–168. https://doi.org/10.1016/j.biocon.2016.01.003 

Jędrzejewski, W., Jędrzejewska, B., Zawadzka, B., Borowik, T., Nowak, S., Mysłajek, R.W., 

2008. Habitat suitability model for Polish wolves based on long‐term national census. 
Anim. Conserv. 11, 377–390. 

Jimenez, M.D., Bangs, E.E., Boyd, D.K., Smith, D.W., Becker, S.A., Ausband, D.E., ..., Laudon, 
K. (2017). Wolf dispersal in the Rocky Mountains, Western United States: 1993–2008. J. 
Wildl. Manag. 81, 581–592. https://doi.org/10.1002/jwmg.21238 

Jiménez-Valverde, A., Peterson, A.T., Soberón, J., Overton, J.M., Aragón, P., Lobo, J.M., 2011. 

Use of niche models in invasive species risk assessments. Biol. Invasions 13, 2785–2797. 
https://doi.org/10.1007/s10530-011-9963-4 

Kantar, L.E., Cumberland, R.E., 2013. Using a double-count aerial survey to estimate moose 

abundance in Maine. Alces 49, 29–37. 

Keeley, A.T.H., Beier, P., Gagnon, J.W., 2016. Estimating landscape resistance from habitat 
suitability: effects of data source and nonlinearities. Landsc. Ecol. 31, 2151–2162. 
https://doi.org/10.1007/s10980-016-0387-5 

Kittle, A.M., Anderson, M., Avgar, T., Baker, J.A., Brown, G.S., Hagens, J., Iwachewski, E., 

Mofatt, S., Mosser, A., Patterson, B.R., Douglas, E.B., ..., Fryxell, J.M., 2015. Wolves adapt 
territory size, not pack size to local habitat quality. J. Anim. Ecol. 84, 1177–1186. 
https://doi.org/10.1111/1365-2656.12366 

Laliberte, A.S., Ripple, W.J., 2004. Range contractions of North American carnivores and 

ungulates. Bioscience. https://doi.org/10.1641/0006-
3568(2004)054[0123:RCONAC]2.0.CO;2 

Lamb, C.T., Mowat, G., Reid, A., Smit, L., Proctor, M., McLellan, B.N., ..., Boutin, S., 2018. 

Effects of habitat quality and access management on the density of a recovering grizzly bear 
population. J. Appl. Ecol. 55(3), 1406–1417. https://doi.org/10.1111/1365-2664.13056 

Laundre, J.W., Hernandez, L., Ripple, W.J., 2010. The landscape of fear: ecological implications 
of being afraid. Open Ecol. J. 3, 1–7. https://doi.org/10.2174/1874213001003030001 

Licht, D.S., Fritts, S.H., 1994. Gray wolf (Canis lupus) occurrences in the Dakotas. Am. Midl. 

Nat. 132, 74–81. https://doi.org/10.2307/2426202 

Linnell, J.D.C., Swenson, J.E., Anderson, R., 2001. Predators and people: conservation of large 

63 

 
 
 
carnivores is possible at high human densities if management policy is favourable. Anim. 
Conserv. 4, 345–349. https://doi.org/10.1017/S1367943001001408 

Lockwood, J.L., Hoopes, M.F., Marchetti, M.P., 2013. Invasion ecology. Wiley and Sons, New 

Jersey, NJ. 

Maiorano, A.L., Boitani, L., Chiaverini, L., Ciucci, P., 2017. Uncertainties in the identification 
of potential dispersal corridors: the importance of behaviour, sex, and algorithm. Basic 
Appl. Ecol. 21, 66–75 https://doi.org/10.1016/j.baae.2017.02.005 

Mancinelli, S., Falco, M., Boitani, L., Ciucci, P. (2019). Social, behavioural and temporal 

components of wolf (Canis lupus) responses to anthropogenic landscape features in the 
central Apennines, Italy. J. Zool. 309, 114–124. https://doi.org/10.1111/jzo.12708 

Manitoba Natural Resources and Northern Development (MNRND), 2022. Manitoba 2022 

hunting guide. Available at: https://www.gov.mb.ca/nrnd/fish-
wildlife/pubs/fish_wildlife/huntingguide.pdf  

Marmion, M., Parviainen, M., Luoto, M., Heikkinen, R.K., Thuiller, W., 2009. Evaluation of 

consensus methods in predictive species distribution modelling. Divers. Distrib. 15, 59–69. 
https://doi.org/10.1111/j.1472-4642.2008.00491.x 

Martínez‐Meyer, E., González‐Bernal, A., Velasco, J.A., Swetnam, T.L., González‐Saucedo, 

Z.Y., Servín, J., ..., Heffelfinger, J.R., 2021. Rangewide habitat suitability analysis for the 
Mexican wolf (Canis lupus baileyi) to identify recovery areas in its historical 
distribution. Divers. Distrib. 27(4), 642–654. https://doi.org/10.1111/ddi.13222 

McLoughlin, P.D., Morris, D.W., Fortin, D., Vander Wal, E., Contasti, A.L., 2010. Considering 

ecological dynamics in resource selection functions. J. Anim. Ecol. 79, 4–12. 
https://doi.org/10.1111/j.1365-2656.2009.01613.x 

McRae, B.H., Kavanagh, D.M., 2011. Linkage mapper connectivity analysis software. The 

Nature Conservancy, Seattle, WA. Available at: https://linkagemapper.org/ 

McRae, B. H., Shah, V. B., 2009. Circuitscape user's guide. The University of California, Santa 

Barbara, CA. Available at: http://www.circuitscape.org 

MDIFW (Maine Department of Inland Fisheries and Wildlife), 2007. Deer population 

management system and database. Augusta, ME: MDIFW. Available at: 
https://www.maine.gov/ifw/docs/deer-managementsystem2007.pdf 

Mech, L.D., 1989. Wolf population survival in an area of high road density. Am. Midl. Nat. 121, 

387–389. https://doi.org/10.2307/2426043 

Mech, L.D., 2017. Where can wolves live and how can we live with them? Biol. Conserv. 210, 

310–317. https://doi.org/10.1016/j.biocon.2017.04.029 

Mech, L.D., 2020. Unexplained patterns of grey wolf Canis lupus natal dispersal. Mamm. 

Rev. 50, 314–323. https://doi.org/10.1111/mam.12198 

Mech, L.D., Boitani, L., 2003. Wolf social ecology. In: Mech, M. and Boitani, L. (Eds.), Wolves: 
behavior, ecology, and conservation. University of Chicago Press, Chicago, IL, pp. 317–
341. https://doi.org/10.14430/arctic540 

Mech, L.D., Isbell, F., Krueger, J., Hart, J., 2019. Gray wolf (Canis lupus) recolonization failure: 

64 

 
 
 
a Minnesota case study. Can. Field-Nat. 133, 60–65. 
https://doi.org/10.22621/cfn.v133i1.2078  

Messier, F. (1985). Solitary living and extraterritorial movements of wolves in relation to social 

status and prey abundance. Canadian Journal of Zoology, 63(2), 239-245. 
https://doi.org/10.1139/z85-037 

Michigan Department of Natural Resources (2008). Estimating wolf abundance in Michigan. 

Available at: 
http://www.mich.gov/documents/dnr/Estimating_Wolf_Abundance_in_Michigan_060208_
239125_7.pdf 

Mladenoff, D.J., Clayton, M.K., Pratt, S.D., Sickley, T.A., Wydeven, A.P., 2009. Change in 

occupied wolf habitat in the northern Great Lakes region. In: In: Wydeven, A.P., Van 
Deelen, T.R., Heske, E. (Eds.), Recovery of gray wolves in the Great Lakes region of the 
United States. Springer, New York, pp. 119–138. https://doi.org/10.1007/978-0-387-85952-
1_8 

Mladenoff, D.J., Sickley, T.A., Haight, R.G., Wydeven, A.P., 1995. A regional landscape 

analysis and prediction of favorable gray wolf habitat in the northern Great Lakes region. 
Conserv. Biol. 9, 279–294. https://doi.org/10.1046/j.1523-1739.1995.9020279.x 

Mladenoff, D., Sickley, T.A., 1998. Assessing potential gray wolf restoration in the northeastern 

United States : A spatial prediction of favorable habitat and potential population levels. J. 
Wildl. Manage. 62, 1–10. https://doi.org /10.2307/3802259  

MNRF (Ministry of Natural Resources and Forestry), 2020. Wolf and coyote hunting 

regulations. Available at: https://www.ontario.ca/document/ontario-hunting-regulations-
summary/wolf-and-coyote 

Morales‐González, A., Fernández‐Gil, A., Quevedo, M., Revilla, E., 2022. Patterns and 
determinants of dispersal in grey wolves (Canis lupus). Biol. Rev. 97, 466–480. 
https://doi.org/10.1111/brv.12807 

Muff, S., Signer, J., Fieberg, J. (2020). Accounting for individual‐specific variation in habitat‐

selection studies: Efficient estimation of mixed‐effects models using Bayesian or frequentist 
computation. J. Anim. Ecol. 89, 80–92. https://doi.org/10.1111/1365-2656.13087 

Musiani, M., Paquet, P.C., 2004. the practices of wolf persecution, protection, and restoration in 

Canada and the United States. Bioscience 54, 50–60. https://doi.org/10.1641/0006-
3568(2004)054[0050:TPOWPP]2.0.CO;2 

Newsome, T.M., Boitani, L., Chapron, G., Ciucci, P., Dickman, C.R., Dellinger, J.A., …, Ripple, 

W.J., 2016. Food habits of the world's grey wolves. Mamm. Rev. 46, 255–269. 
https://doi.org/10.1111/mam.12067 

Nowak, R.M. 1995. Another look at wolf taxonomy. In: Carbyn, L.N., S.H. Fritts, and D.R. Seip 
(Eds.). Ecology and conservation of wolves in a changing world. Canadian Circumpolar 
Institute, University of Alberta, Canada, pp. 375–397. 

O’Neil, S.T., Beyer, D.E., Bump, J.K., 2019. Territorial landscapes: incorporating density-

dependence into wolf habitat selection studies. R. Soc. Open Sci. 6, 190282. 
https://doi.org/10.1098/rsos.190282 

65 

 
 
 
Olson, E.R., Goethlich, J., Goudos-Weisbecker, B., 2021. Attitudes towards a transient carnivore 
prior to recolonization. Wildl. Soc. Bull. 45, 191–201. https://doi.org/10.1002/wsb.1166 

Olson, E.R., Stenglein, J.L., Shelley, V., Rissman, A.R., Browne‐Nuñez, C., Voyles, Z., ..., Van 
Deelen, T., 2015. Pendulum swings in wolf management led to conflict, illegal kills, and a 
legislated wolf hunt. Conserv. Lett. 8, 351–360. https://doi.org/10.1111/conl.12141 

Oriol-Cotterill, A., Valeix, M., Frank, L.G., Riginos, C., Macdonald, D.W., 2015. Landscapes of 
coexistence for terrestrial carnivores: the ecological consequences of being downgraded 
from ultimate to penultimate predator by humans. Oikos 124, 1263–1273. 
https://doi.org/10.1111/oik.02224 

Orning, E.K., Romanski, M.C., Moore, S., Chenaux-Ibrahim, Y., Hart, J., Belant, J.L., 2020. 

Emigration and first-year movements of initial wolf translocations to Isle Royale. Northeast. 
Nat. 27, 701–708. https://doi.org/10.1656/045.027.0410 

Osipova, L., Okello, M.M., Njumbi, S.J., Ngene, S., Western, D., Hayward, M.W., Balkenhol, 
N., 2019. Using step‐selection functions to model landscape connectivity for African 
elephants: Accounting for variability across individuals and seasons. Anim. Conserv. 22, 
35–48. https://doi.org/10.1111/acv.12432 

Ouellette, M., Remmel, T.K., Perera, A.H., 2020. A spatial database of historical wildfire and 

timber harvesting in the boreal Area of the Undertaking of Ontario: the methodological 
framework. Ontario MNRF, TR 37. 

Phillips, S.J., Dudík, M., Elith, J., Graham, C.H., Lehmann, A., Leathwick, J., Ferrier, S., 2009. 
Sample selection bias and presence-only distribution models: implications for background 
and pseudo-absence data. Ecol. Appl. 19, 181–197. https://doi.org/10.1890/07-2153.1 

Piper, W.H., 2011. Making habitat selection more “familiar”: a review. Behav. Ecol. Sociobiol. 

65, 1329–1351. https://doi.org/10.1007/s00265-011-1195-1 

Powell, R.A., Proulx, G., 2003. Trapping and marking terrestrial mammals for research: 

integrating ethics, performance criteria, techniques, and common sense. ILAR J. 44, 259–
276. https://doi.org/10.1093/ilar.44.4.259 

R Core Team, 2021. R: A language and environment for statistical computing. R foundation for 

statistical computing, Vienna, Austria. Available at: https://www. R-project.org/. 

Recio, M.R., Singer, A., Wabakken, P., Sand, H., 2020. Agent-based models predict patterns and 

identify constraints of large carnivore recolonizations, a case study of wolves in 
Scandinavia. Biol. Conserv. 251, 108752. https://doi.org/10.1016/j.biocon.2020.108752 

Reinhardt, I., Kluth, G., Nowak, C., Szentiks, C.A., Krone, O., Ansorge, H., Mueller, T., 2019. 

Military training areas facilitate the recolonization of wolves in Germany. Conserv. Lett. 12, 
e12635. https://doi.org/10.1111/conl.12635 

Rio-maior, H., Nakamura, M., Álvares, F., Beja, P., 2019. Designing the landscape of 

coexistence : integrating risk avoidance , habitat selection and functional connectivity to 
inform large carnivore conservation. Biol. Conserv. 235, 178–188. 
https://doi.org/10.1016/j.biocon.2019.04.021 

Ripari, L., Premier, J., Belotti, E., Bluhm, H., Breitenmoser-Würsten, C., Bufka, L.,…, Heurich, 
M, 2022. Human disturbance is the most limiting factor driving habitat selection of a large 

66 

 
 
 
carnivore throughout Continental Europe. Biol. Conserv. 266, 109446. 
https://doi.org/10.1016/j.biocon.2021.109446 

Ripple, W.J., Beschta, R.L., 2012. Trophic cascades in Yellowstone: The first 15 years after wolf 

reintroduction. Biol. Conserv. 145, 205–213. https://doi.org/10.1016/j.biocon.2011.11.005 

Ripple, W.J., Estes, J.A., Beschta, R.L., Wilmers, C.C., Ritchie, E.G., Hebblewhite, M., …, 

Wirsing, A.J., 2014. Status and ecological effects of the world’s largest carnivores. Science 
343, 151–162. https://doi.org/10.1126/science.1241484 

Rodríguez, J.P., Brotons, L., Bustamante, J., Seoane, J., 2007. The application of predictive 

modelling of species distribution to biodiversity conservation. Divers. Distrib. 13, 243–251. 
https://doi.org/10.1111/j.1472-4642.2007.00356.x 

Ruid, D.B., Paul, W.J., Roell, B.J., Wydeven, A.P., Willging, R.C., Jurewicz, R.L., Lonsway, 
D.H., 2009. Wolf–human conflicts and management in Minnesota, Wisconsin, and 
Michigan. In: Wydeven, A.P., Van Deelen, T.R., Heske, E. (Eds.), Recovery of gray wolves 
in the Great Lakes region of the United States. Springer, New York, pp. 279–295. 
https://doi.org/10.1007/978-0-387-85952-1_18 

Scott, R.W., Huff, F.A., 1996. Impacts of the Great Lakes on regional climate conditions. J. 
Great Lakes Res. 22, 845–863. https://doi.org/10.1016/S0380-1330(96)71006-7 

Sing, T., Sander, O., Beerenwinkel, N., Lengauer, T., 2005. ROCR: visualizing classifier 

performance in R. Bioinformatics, 21, 3940–3941. 

Slagle, K., Zajac, R., Bruskotter, J., Wilson, R., Prange, S., 2013. Building tolerance for bears: a 

communications experiment. J. Wildl. Manag. 77, 863–869. 
https://doi.org/10.1002/jwmg.515 

Smith, J.A., Duane, T.P., Wilmers, C.C., 2019. Landscape and urban planning moving through 
the matrix: promoting permeability for large carnivores in a human-dominated landscape. 
Landsc. Urban Plan. 183, 50–58. https://doi.org/10.1016/j.landurbplan.2018.11.003 

Smith, J.B., Nielsen, C.K., Hellgren, E.C., 2014. Illinois resident attitudes toward recolonizing 

large carnivores. J. Wildl. Manag. 78, 930–943. https://doi.org/10.1002/jwmg.718 

Smith, J.B., Nielsen, C.K., Hellgren, E.C., 2016. Suitable habitat for recolonizing large 

carnivores in the midwestern USA. Oryx 50, 555–564. 
https://doi.org/10.1017/S0030605314001227 

Spitz, D.B., Hebblewhite, M., Stephenson, T.R., 2017. ‘MigrateR’: extending model‐driven 

methods for classifying and quantifying animal movement behavior. Ecography, 40, 788–
799. https://doi.org/10.1111/ecog.02587 

Statistics Canada, 2020. Road Network File, Catalogue no. 92-500-X. Available at: 

https://www150.statcan.gc.ca/n1/en/catalogue/92-500-X 

Stauffer, G., Roberts, N.M., MacFarland, D., van Deelen, T.R., 2020. Evaluation of alternative 
methods for estimating wolf abundance in Wisconsin, USA. Wisconsin Department of 
Natural Resources, Wisconsin. 

Stricker, H.K., Gehring, T.M., Donner, D., Petroelje, T., 2019. Multi-scale habitat selection 

model assessing potential gray wolf den habitat and dispersal corridors in Michigan, USA. 

67 

 
 
 
Ecol. Modell. 397, 84–94. https://doi.org/10.1016/j.ecolmodel.2018.12.021 

Suraci, J.P., Clinchy, M., Zanette, L.Y., Wilmers, C.C., 2019. Fear of humans as apex predators 
has landscape‐scale impacts from mountain lions to mice. Ecol. Lett., 22, 1578–1586. 
https://doi.org/10.1111/ele.13344 

Svenning, J.-C., Skov, F., 2004. Limited filling of the potential range in European tree species. 

Ecol. Lett. 7, 565–573. https://doi.org/10.1111/j.1461-0248.2004.00614.x 

Thiel, R.P., 1985. Relationship between road densities and wolf habitat suitability in Wisconsin. 

Am. Midl. Nat. 113, 404. https://doi.org/10.2307/2425590 

Thiel, R.P., Hall, W., Heilhecker, E., Wydeven, A.P., 2009. An isolated wolf population in 

central Wisconsin. In: Wydeven, A.P., Van Deelen, T.R., Heske, E. (Eds.), Recovery of 
gray wolves in the Great Lakes region of the United States. Springer, New York, pp. 107–
117. http://dx.doi.org/10.1007/978-0-387-85952-1 

Thomas, C.D., Cameron, A., Green, R.E., Bakkenes, M., Beaumont, L.J., Collingham, Y.C, ..., 

Williams, S.E., 2004. Extinction risk from climate change. Nature 427, 145–148. 
https://doi.org/10.1038/nature02121 

Thorsen, N.H., Hansen, J.E., Støen, O.G., Kindberg, J., Zedrosser, A., Frank, S.C., 2022. 

Movement and habitat selection of a large carnivore in response to human infrastructure 
differs by life stage. Mov. Ecol. 10, 1–14. https://doi.org/10.1186/s40462-022-00349-y 

Thuiller, W., Lafourcade, B., Engler, R., Araújo, M.B., 2009. BIOMOD - a platform for 

ensemble forecasting of species distributions. Ecography 32, 369–373. 
https://doi.org/10.1111/j.1600-0587.2008.05742.x 

Thurfjell, H., Ciuti, S., Boyce, M.S., 2014. Applications of step-selection functions in ecology 

and conservation. Mov. Ecol. 2, 4. https://doi.org/10.1186/2051-3933-2-4 

Treves, A., Bruskotter, J., 2014. Tolerance for predatory wildlife. Science 344, 476–477. 

https://doi.org/10.1126/science.1252690 

Treves, A., Karanth, K.U., 2003. Human-carnivore conflict and perspectives on carnivore 

management worldwide. Conserv. Biol. 17, 1491–1499. https://doi.org/10.1111/j.1523-
1739.2003.00059.x 

Treves, A., Martin, K.A., Wiedenhoeft, J.E., Wydeven, A.P., 2009. Dispersal of gray wolves in 
the Great Lakes region. In: Wydeven, A.P., Van Deelen, T.R., Heske, E. (Eds.), Recovery 
of gray wolves in the Great Lakes region of the United States. Springer, New York, NY, pp. 
191–204. http://doi.org/10.1007/978-0-387-85952-1_12 

Treves, A., Naughton-Treves, L. , Harper, E.K., Mladenoff, D.J., Rose, R.A., Sickley, T.A., 

Wydeven, A.P., 2004. Predicting human‐carnivore conflict: a spatial model derived from 25 
years of data on wolf predation on livestock. Conserv. Biol. 18, 114–125. 
https://doi.org/10.1111/j.1523-1739.2004.00189.x 

United States Census Bureau, 2010. Decennial census of population and housing. Available at: 

https://www.census.gov/programs-surveys/decennial-census/decade/decennial-
publications.2010.html  

United States Census Bureau, 2020. TIGER/Line shapefiles 2020: roads. Available at: 

68 

 
 
 
https://www.census.gov/geographies/mapping-files/time-series/geo/tiger-line-file.html. 

United States Geological Survey, 1996. USGS EROS archive: digital elevation global 30 arc-

second elevation (GTOPO30). https://doi.org/10.5066/F7DF6PQS  

United States Geological Survey, 2016. National hydrography dataset. Available at: 

https://www.usgs.gov/core-science-systems/ngp/national-hydrography/access-national-
hydrography-products  

United States Geological Survey Gap Analysis Project, 2018. Protected Areas Database of the 

United States (PAD-US): U.S. Geological Survey Data Release. https://doi.org/ 
10.5066/P955KPLE. 

USFWS (U.S. Fish and Wildlife Service), 1992. Recovery plan for the eastern timber wolf. Twin 

Cities, Minnesota. 

USFWS (U.S. Fish and Wildlife Service), 2008. Post-delisting monitoring plan for the western 
Great Lakes distinct population segment of the gray wolf. U.S. Fish and Wildlife Service, 
Twin Cities Field Office and Midwest Region, Bloomington, MN and Ft. Snelling, MN. 

USFWS (U.S. Fish and Wildlife Service), 2009. threatened and endangered species system: 

delisting report. US Government Printing Office, Washington, DC 

USFWS (U.S. Fish and Wildlife Service), 2020a. Gray wolf biological report: information on the 
species in the lower 48 United States. US Government Printing Office, Washington, DC. 

USFWS (U.S. Fish and Wildlife Service), 2020b. Endangered and threatened wildlife and plants; 
removing the gray wolf (Canis lupus) from the list of endangered and threatened wildlife. 
Federal Register, 85, 69778–69895. 

Václavík, T., Meentemeyer, R.K., 2012. Equilibrium or not? modelling potential distribution of 

invasive species in different stages of invasion. Divers. Distrib. 18, 73–83. 
https://doi.org/10.1111/j.1472-4642.2011.00854.x 

van den Bosch, M., Beyer Jr, D.E., Erb, J.D., Gantchoff, M.G., Kellner, K.F., MacFarland, D. 
M., Norton, D.C., ..., Belant, J.L., 2022. Identifying potential gray wolf habitat and 
connectivity in the eastern USA. Biol. Conserv. 273, 109708. 
https://doi.org/10.1016/j.biocon.2022.109708 

Van Der Veken, T., Van Den Berge, K., Gouwy, J., Berlengee, F., Schamp, K., 2021. Diet of the 

first settled wolves (Canis lupus) in Flanders, Belgium. Lutra 64, 45–56. 

Vasudev, D., Fletcher, R.J., Goswami, V.R., Krishnadas, M., 2015. From dispersal constraints to 
landscape connectivity: lessons from species distribution modeling. Ecography 38, 967–
978. https://doi.org/10.1111/ecog.01306 

Wielgus, R.B., 2002. Minimum viable population and reserve sizes for naturally regulated 

grizzly bears in British Columbia. Biol. Conserv. 106, 381–388. 
https://doi.org/10.1016/S0006-3207(01)00265-8 

Wisconsin Department of Natural Resources, 2021. Wisconsin gray wolf monitoring report 15 

april 2020 through 14 april 2021. Available at: 
https://widnr.widen.net/s/vlfpznrl2w/wisconsin_gray_wolf_2020_2021_final 

Wolf, C., Ripple, W.J., 2016. Prey depletion as a threat to the world's large carnivores. R. Soc. 

69 

 
 
 
Open Sci. 3, 160252. https://doi.org/10.1098/rsos.160252 

Wolf, C., Ripple, W.J., 2017. Range contractions of the world’s large carnivores. R. Soc. Open 

Sci. 4, 170052. https://doi.org/10.1098/rsos.170052 

Woodroffe, R., 2000. Predators and people: using human densities to interpret declines of large 
carnivores. Anim. Conserv. 3, 165–173. https://doi.org/10.1017/S136794300000086X 

Wydeven, A.P., Fuller, T.K., Weber, W., MacDonald, K., 1998. The potential for wolf recovery 
in the northeastern United States via dispersal from southeastern Canada. Wildl. Soc. Bull. 
26, 776–784. 

Wydeven, A.P., Wiedenhoeft, J.E., Schultz, R.N., Thiel, R.P., Jurewicz, R.L., Kohn, B.E., Van 

Deelen, T.R., 2009. History, population growth, and management of wolves in 
Wisconsin. In: Wydeven, A.P., Van Deelen, T.R., Heske, E. (Eds.), Recovery of gray 
wolves in the Great Lakes region of the United States. Springer, New York, NY, pp. 87–
105. http://doi.org/10.1007/978-0-387-85952-1_6 

Yang, L., Jin, S., Danielson, P., Homer, C., Gass, L., Bender, S.M., ..., Xian, G., 2018. A new 
generation of the United States national land cover database: requirements, research 
priorities, design, and implementation strategies. ISPRS J. Photogramm. Remote Sens. 146, 
108–123. https://doi.org/10.1016/j.isprsjprs.2018.09.006 

Zeller, K.A., Jennings, M.K., Vickers, T.W., Ernest, H.B., Cushman, S.A., Boyce, W.M., 2018. 
Are all data types and connectivity models created equal? Validating common connectivity 
approaches with dispersal data. Divers. Distrib. 24, 868–879. 
https://doi.org/10.1111/ddi.12742 

Zimmermann, B., Nelson, L., Wabakken, P., Sand, H., Liberg, O., 2014. Behavioral responses of 

wolves to roads: scale-dependent ambivalence. Behav. Ecol. 25(6), 1353–1364. 
https://doi.org/10.1093/beheco/aru134 

70 

 
 
 
 
 
 
 
 
APPENDIX 

Figure A DHARMa residual tests for a binomial general linearized model, used in Chapter 3 to 

predict the relationship between anthropogenic disturbances and gray wolf (Canis lupus) habitat 

suitability in Minnesota, Wisconsin, and Michigan, USA, 1989–2020. From left to right: 

Kolmogorov-Smirnov test, DHARMa nonparametric dispersion test, and outlier test.  

71