draft...draft duquette et al.|3 46 selection can be influenced by multiple spatially-dependent...

Draft

Black bear functional resource selection relative to

intraspecific competition and human risk

Journal: Canadian Journal of Zoology

Manuscript ID cjz-2016-0031.R2

Manuscript Type: Article

Date Submitted by the Author: 30-Nov-2016

Complete List of Authors: Duquette, Jared; Mississippi State University, Carnivore Ecology Laboratory, Department of Fisheries, Wildlife, and Aquaculture Belant, Jerrold; Mississippi State University, Carnivore Ecology Laboratory, Department of Fisheries, Wildlife, and Aquaculture Wilton, Clay; Mississippi State University, Carnivore Ecology Laboratory, Department of Fisheries, Wildlife, and Aquaculture

Fowler, Nick; Mississippi State University, Carnivore Ecology Laboratory, Department of Fisheries, Wildlife, and Aquaculture Waller, Brittany; Mississippi State University, Carnivore Ecology Laboratory, Department of Fisheries, Wildlife, and Aquaculture Beyer, Jr., Dean; Michigan Department of Natural Resources, Wildlife Division Svoboda, Nathan; Mississippi State University, Carnivore Ecology Laboratory, Department of Fisheries, Wildlife, and Aquaculture Simek, Stephanie; Mississippi State University, Carnivore Ecology Laboratory, Department of Fisheries, Wildlife, and Aquaculture Beringer, Jeff; Missouri Department of Conservation

Keyword: black bear, food, hunting, roads, Ursus americanus, RIPARIAN < Habitat,

mixed models

https://mc06.manuscriptcentral.com/cjz-pubs

Canadian Journal of Zoology

Draft

Duquette et al.|1

Black bear functional resource selection relative to intraspecific competition and human 1

risk 2

3

4

Jared F. Duquette*, Jerrold L. Belant, Clay M. Wilton, Nicholas Fowler, Brittany W. Waller, 5

Dean E. Beyer Jr, Nathan J. Svoboda, Stephanie L. Simek, Jeff Beringer 6

7

8

J.F. Duquette, J.L. Belant ([email protected]), C.M. Wilton ([email protected]), N. 9

Fowler ([email protected]), B.W. Waller ([email protected]), N.J. Svoboda 10

([email protected]), and S.L. Simek ([email protected]). 11

Carnivore Ecology Laboratory, Forest and Wildlife Research Center, Mississippi State 12

University, Mississippi State, Mississippi. 13

14

D.E. Beyer Jr. ([email protected]) 15

Michigan Department of Natural Resources, Wildlife Division, Marquette, Michigan. 16

17

J. Beringer ([email protected]) 18

Missouri Department of Conservation, Columbia, Missouri, United States of America. 19

20

*Corresponding author: Current affiliation: Institute for Wildlife Studies, 2327 Kettner Blvd, San 21

Diego, CA 92101; Email: [email protected]; Phone: 619-291-5892; Fax: 619-291-5869 22

23

Page 1 of 36



Draft

Duquette et al.|2

24

Black bear functional resource selection relative to intraspecific competition and human 25

risk 26

27

J.F. Duquette, J.L. Belant, C.M. Wilton, N. Fowler, B.W. Waller, N.J. Svoboda, and S.L. Simek, 28

D.E. Beyer Jr., and J. Beringer. 29

30

Abstract 31

The spatial scales at which animals make behavioral trade-offs is assumed to relate to the scales 32

at which factors most limiting resources and increasing mortality risk occur. We used global 33

positioning system collar locations of 29 reproductive-age female black bears (Ursus americanus 34

Pallas, 1780) in three states to assess resource selection relative to bear population-specific 35

density, an index of vegetation productivity, riparian corridors, or two road classes of and within 36

home ranges during spring-summer 2009–2013. Female resource selection was best explained 37

by functional responses to vegetation productivity across nearly all populations and spatial 38

scales, which appeared to be influenced by variation in bear density (i.e., intraspecific 39

competition). Behavioral trade-offs were greatest at the landscape scale, but except for 40

vegetation productivity, were consistent for populations across spatial scales. Females across 41

populations selected locations nearer to tertiary roads, but females in Michigan and Mississippi 42

selected main roads and avoided riparian corridors, while females in Missouri did the opposite, 43

suggesting population-level trade-offs between resource (e.g., food) acquisition and mortality 44

risks (e.g., vehicle collisions). Our study emphasizes that female bear population-level resource 45

Page 2 of 36



Draft

Duquette et al.|3

selection can be influenced by multiple spatially-dependent factors, and that scale-dependent 46

functional behavior should be identified for management of bears across their range. 47

48

Key words 49

black bear, food, hunting, mixed models, riparian, roads, Ursus americanus 50

51

Introduction 52

Animals improve their reproductive success by selecting the most beneficial resources available 53

across a landscape (Rettie and Messier 2000; Manly et al. 2002). But resources may be limited 54

by factors including habitat fragmentation (e.g., roads; Shepard and Whittington 2006, land-use 55

change; Hiller et al. 2015a, b), intra- and interspecific competition (Ciarniello et al. 2007; Watts 56

and Holekamp 2009), or direct and indirect mortality risks (Creel et al. 2008; Stayaert et al. 57

2013; Kiffner et al. 2014) which can vary across landscapes (Johnson and Seip 2008). Imperfect 58

knowledge or cues to resource availability or quality can also limit resource accessibility (Basille 59

et al. 2013). Therefore, animals improve their reproductive success by using behavioral trade-60

offs between resource selection and mortality risk avoidance (Martin et al. 2010; Latif et al. 61

2011). 62

The spatial scales at which animals adapt their behavior are generally assumed to relate to 63

the scales where access to resources is maximized and mortality risk is minimized (Rettie and 64

Messier 2000). A factor (e.g., predation risk) that limits resources and increases mortality risk 65

modifies animal behavior at successively finer scales until another factor (e.g., food) becomes 66

more limiting and further modifies behavior. Intraspecific population density can be a major 67

factor influencing the behavioral trade-offs of animals because intraspecific competition 68

Page 3 of 36



Draft

Duquette et al.|4

increases with population density (e.g., Ciarniello et al. 2007). Greater intraspecific competition 69

can decrease the availability of ideal resources and limit animal fitness (Zedrosser et al. 2006; 70

van Beest et al. 2014). While identifying factors that influence the behavioral trade-offs of 71

animals across multiple spatial scales is common (e.g., Basille et al. 2013), generalizing those 72

conclusions across the geographic range of a species can introduce bias from variability in 73

biological and environmental factors found within each population. Assessing behavioral trade-74

offs among population density, resource selection, and mortality risks within and across 75

populations could provide insights into factors affecting the range-wide reproductive success of a 76

species (McLoughlin et al. 2000; Bojarska and Selva 2012). 77

Animals maximize their fitness by adapting to environmental variability in a functional 78

nonlinear manner (Matthiopoulos et al. 2011), often at multiple spatial scales (Moreau et al. 79

2012). For example, animals living in landscapes with patchy and fluctuating food resources and 80

mortality risks may functionally select habitat patches based on a trade-off between perceived 81

resources and mortality risks (Godvik et al. 2009; Moreau et al. 2012). This concept can be 82

extended to variability in other factors, such as resource competition and anthropogenic 83

disturbance, which directly or indirectly influence animal fitness. Understanding the functional 84

behaviors of animals relative to their population density (i.e., competition) can indicate the 85

habitat quality of the landscape (McLoughlin et al. 2010) and assist in guiding population 86

management. 87

Behavioral trade-offs are important for many large carnivores during spring, when young 88

are dependent on maternal nutrition (Elowe and Dodge 1989) and vulnerable to intra- or 89

interspecific competition and predation (Palomares and Caro 1999; Garrison et al. 2007). Large 90

carnivores in human-dominated landscapes often adjust behaviors to reduce mortality risks or 91

Page 4 of 36



Draft

Duquette et al.|5

disturbance associated with roads, including vehicles (Brody and Pelton 1989; Dixon et al. 2006) 92

and hunters (Milner et al. 2007; Stillfried et al. 2015). Movement corridors are important to 93

large carnivores because they facilitate navigation through fragmented landscapes while 94

minimizing mortality risks (Shepard and Whittington 2006). Although large carnivores typically 95

avoid roads (e.g., Simek et al. 2015), roads with low traffic volume (Dixon et al. 2006) may 96

serve as movement corridors or areas with plentiful food (Mace et al. 1996; Roever et al. 2008; 97

Hiller et al. 2015a, b). Additionally, riparian areas can allow large carnivores to move among 98

patches of food and cover (Naiman and Rogers 1997) while often providing food within the 99

corridor itself (Lyons et al. 2003). Investigating frequency of large carnivore use of different 100

movement corridors help reveal the relationship between spatial variation in resources and 101

mortality risks. 102

Black bears (Ursus americanus Pallas, 1780) are omnivorous carnivores found 103

throughout most forested regions of North America (Scheick and McCown 2014) and are 104

thought to use the smallest amount of space necessary to acquire resources for reproduction 105

(Mitchell and Powell 2007). To access resources, bears must contend with variation in 106

intraspecific competition related to bear density, with individuals in denser populations typically 107

having greater despotic behavior (Beckmann and Berger 2003). Bears may reduce intraspecific 108

competition or human-related risks (Ordiz et al. 2011) by using habitat corridors (e.g., riparian 109

areas; Lyons et al. 2003) to move among patches of resources (Dixon et al. 2006) and adjusting 110

their fine-scale resource selection behaviors to those risks. Bears may also occasionally use 111

roads with low anthropogenic risk (Brody and Pelton 1989; Mace et al. 1996; Reynolds-Hogland 112

et al. 2007) as movement corridors, but roads typically hinder bears by introducing mortality 113

risks (e.g., vehicle collisions) and fragmenting resources across the landscape (Reynolds-114

Page 5 of 36



Draft

Duquette et al.|6

Hogland et al. 2007; Waller et al. 2014). Anthropogenic mortality risk may especially influence 115

resource selection of adult females with dependent young while meeting increased nutritional 116

demands and avoiding vehicles (Wilton et al. 2014a), hunters (Reynolds-Hogland et al. 2007), or 117

intra-specific predation (LeCount 1987; Garrison et al. 2007). An assessment of resource 118

selection and mortality risk trade-offs of reproductive-aged female black bears may therefore 119

reveal factors that limit population-level fitness (Mitchell and Powell 2007). While numerous 120

aspects of black bear resource selection relative to mortality risks have been reported for 121

individual populations (e.g., Belant et al. 2010; Waller et al. 2014), little is known about bear 122

trade-offs between resources and mortality risks across populations with varying population 123

density and environmental characteristics. 124

Our objective was to assess reproductive-aged female black bear resource selection 125

relative to variation in population density, vegetation productivity, riparian corridors, and two 126

classes of roads of and within home ranges during spring-summer for populations along a 127

latitudinal gradient of the United States. We focused on reproductive-aged females because this 128

demographic is likely sensitive to behavioral trade-offs between resources and mortality risks 129

(Martin et al. 2010) while maintaining nutritional condition for reproductive success (Elowe and 130

Dodge 1989; Dahle and Swenson 2003; Ayers et al. 2013). We established four hypotheses 131

related to differences in population density, latitude, and spatial-scale. First, we hypothesized 132

that vegetation productivity would best explain female resource selection across populations and 133

spatial-scales because reproductive-age females require increased food intake to support their 134

nutritional condition for breeding, gestation, and care of dependent young (Dahle and Swenson 135

2003; Ayers et al. 2013). Second, we hypothesized that females would reduce their selection for 136

profitable resources as population density increased because females would avoid intraspecific 137

Page 6 of 36



Draft

Duquette et al.|7

competition (van Beest et al. 2014). Third, we hypothesized female resource selection behaviors 138

to be more influential to the establishment of home ranges across the landscape than selection 139

within home ranges because variation in resources and mortality risks would be more predictable 140

at this scale (Rettie and Messier 2000; Ordiz et al. 2011). Fourth, we hypothesized females to 141

exhibit greater resource-risk trade-offs from Mississippi to Michigan due to increasing 142

intraspecific competition (Dixon et al. 2006), shorter vegetation growing season (Ferguson and 143

McLoughlin 2000; Bojarska and Selva 2012), and greater mortality risks from vehicles and 144

hunters (Reynolds-Hogland et al. 2007; Stillfried et al. 2015). 145

Materials and methods 146

Study areas 147

We conducted the study in Michigan, Missouri, and Mississippi, USA. The study area in the 148

south-central Upper Peninsula of Michigan (528.9 km2; 45°34’ N, 87°20’ W; Fig. 1) has a mean 149

elevation of 185 m above sea level and topography is flat. Predominant vegetation included 150

upland and lowland hardwoods, lowland conifer swamps, upland conifers, aspen (Populus spp.) 151

stands, row-crop and livestock agriculture, herbaceous wetlands, and intermittent patches of 152

berry-producing shrubs (e.g., raspberries [Rubus spp.], and blueberries [Vaccinium spp.]). 153

Human development was low density (0.09 km/km2) residential and recreational camps and main 154

and tertiary road density was 0.03 km/km2 and 1.69 km/km

2, respectively (United States Bureau 155

of the Census 2011). Riparian corridor (i.e., rivers and streams) density was 1.17 km/km2

156

(United States Geological Survey 2014). During May-August daily mean temperature ranged 157

from 3.0−32.2°C and rainfall during the study period was 12.2−14.5 cm based on a weather 158

station we deployed in the middle of the study area. Black bear density was 14–19/100 km2 (J. 159

L. Belant, unpublished data). Black bears in Michigan were harvested annually during 160

Page 7 of 36



Draft

Duquette et al.|8

September and October, but only females without dependent young were legal for harvest 161

(Belant et al. 2011). 162

The study area in the Ozark Highlands ecological region of south-central Missouri 163

(8033.8 km2; 35°57’–40°35’N, 89°8’–95°46’W; Fig. 1) consisted of mostly hilly and steep 164

topography with elevation from 70–540 m with highest elevations in the Ozark Highlands 165

(Wilton et al. 2014b). Dominant landcover types in the Ozark Highlands included forest, crop 166

and pasture, grassland, and human developed areas (Fry et al. 2011). Forests were primarily 167

upland oak-hickory (Quercus spp., Carya spp.) and oak-pine (Pinus spp.; Raeker et al. 2010). 168

Landownership included private homesteads and farms and public lands. Main and tertiary road 169

density was 0.18 km/km2 and 1.28 km/km

2, respectively (United States Bureau of the Census 170

2011). Riparian corridor density was 0.08 km/km2

(United States Geological Survey 2014). 171

Mean daily temperature ranged 7.9−33.3°C and mean rainfall was 12.7 cm during the study 172

period (NOAA 2014). Black bear density was 1.7/100 km2 (Wilton et al. 2014b) and are 173

recolonizing the southern and central regions of the state (Puckett et al. 2014; Wilton et al. 174

2014a). 175

The study area in Mississippi (7004.8 km2; 33°20’ N, 90°93’ W; Fig. 1) has elevations 176

from 0–114 m above sea level and topography is generally flat. Land cover included agricultural 177

(39 %), water/wetland (35 %), forested (24 %), and urban areas (2 %; Simek et al. 2015). Main 178

and tertiary road density was 0.14 km/km2 and 0.31 km/km

2, respectively (United States Bureau 179

of the Census 2011). Riparian corridor density was 0.33 km/km2 (United States Geological 180

Survey 2014). Climate is subtropical with average monthly temperature ranging from 7.5–27.0 181

°C (Simek et al. 2015). Bear density was likely <1/100 km2 (R. Rummell, Mississippi 182

Department of Wildlife, Fisheries and Parks, personal communication), but has been increasing 183

Page 8 of 36



Draft

Duquette et al.|9

due to recolonization from expanding populations in adjacent states, increased restoration of 184

native habitat, and no legal harvest of bears (Simek et al. 2012). 185

Capture, handling, and monitoring 186

We live-captured female bears in Michigan from late May-early July 2009–2011, in Missouri 187

from late May-early October 2010–2013, and in Mississippi year-round in 2009–2011 using 188

Aldrich foot snares and culvert traps (Johnson and Pelton 1980; Reagan et al. 2002) baited with 189

commercial lures and attractants. We checked traps at least daily. We immobilized females in 190

the 3 study areas intramuscularly with Telazol® (1:1 mixture of tiletamine hydrochloride [HCL] 191

and zolazepam; Fort Dodge Animal Health, Fort Dodge, Iowa, USA; Kreeger and Arnemo 2007) 192

using a syringe pole or CO2-powered dart pistol or rifle (Pneu-Dart, Inc., Williamsport, 193

Pennsylvania. USA). After induction we recorded morphometrics, determined sex, ear tagged 194

individuals, and collected blood, hair, and tissue samples. We fit females with various Global 195

Positioning System (GPS) telemetry collars in Michigan (Lotek 7000MU, Lotek Wireless, 196

Newmarket, Ontario, Canada), Missouri (Northstar NSG-LD2, RASSL Globalstar, King George, 197

Virginia, USA; Advanced Telemetry Systems [ATS] M2610B, ATS, Isanti, Minnesota, USA; 198

and Lotek Wireless 7000MU, Lotek Wireless, Newmarket, Ontario, Canada), and Mississippi 199

(Telonics, Inc., Mesa, Arizona, USA; Advanced Telemetry Systems Inc., Isanti, Minnesota; 200

Northstar, King George, Virginia, USA). We released all females at their sites of capture. All 201

capture and handling complied with the American Society of Mammalogists guidelines (Sikes et 202

al. 2011) and the Institutional Animal Care and Use Committee at Mississippi State University 203

(protocols 08-052, 09-004, and 10-037). We programmed GPS radiocollars in Michigan, 204

Missouri, and Mississippi to record a location every 15, 10, or 60 minutes from the time of 205

deployment, respectively. 206

Page 9 of 36



Draft

Duquette et al.|10

Spatial and habitat selection 207

We estimated the likelihood of female resource selection of and within home ranges using 208

second- and third-order selection analyses (Manly et al. 2002) with designs 2 and 4, (Thomas 209

and Taylor 2006), respectively. We first selected female GPS locations that were separated by 210

≥60 minutes to reduce spatial autocorrelation, which was the minimum relocation time across 211

populations. We standardized step length estimates from locations >60 minutes apart to a 60 212

minute interval to compare parameter estimates across populations and selection of home ranges 213

across the landscape. We then defined resource use as female locations from one day post den-214

exit or capture to cessation of monitoring (e.g., dropped collar), or 31 August, which we 215

imported into ArcGIS 10.1 (Environmental Systems Research Institute 2011) as a shapefile. We 216

used 31 August as our cut-off date due to senescence of primary vegetation productivity and 217

limited GPS location data after this time. To estimate female selection home ranges across the 218

landscape, we first created a 100% minimum convex polygon around all locations in each 219

population using ArcGIS. We then defined study area availability by using the Generate 220

Conditional Random Points tool in the Geospatial Modeling Environment (Version 0.7.1.0; 221

[Beyer 2012]) to plot the same number of random points as GPS locations for each population 222

within each of the polygons of used locations. 223

To estimate female selection of resources within home ranges, we estimated the mean 224

step length of successive locations for each female in each population using Calculate Movement 225

Path Metrics tool in the Geospatial Modeling Environment and then calculated mean step length 226

across populations. We then defined resource availability using the Generate Conditional 227

Random Points tool in the Geospatial Modeling Environment to establish a paired random point 228

within the determined standardized step distance (1054 m) of each female location across 229

Page 10 of 36



Draft

Duquette et al.|11

populations. We then copied the individual female and year information from used locations to 230

respective random locations for each population for model development. 231

We used the 16-day composite Normalized Difference Vegetation Index (NDVI) data 232

from the MODIS server (250 m resolution; United States Geological Survey 2013) as a metric of 233

vegetation productivity and quality (i.e., food) for all populations, which is assumed to be of 234

great nutritional value during spring and summer (Milakovic et al. 2012; Fowler 2014). We 235

obtained NDVI raster layers with the closest date to the beginning of each month for May–236

August 2009–2013 and used the Raster Calculator in ArcGIS 10.1 to estimate a mean NDVI 237

score for each year. We imported the composite NDVI layers into ArcGIS, extracted the NDVI 238

value of each female location corresponding to the year when the location was recorded, and 239

exported this data to a spreadsheet. 240

We assessed rivers and non-intermittent streams as a metric of riparian movement 241

corridors for all populations using the United States Geological Survey National Hydrography 242

Data (United States Geological Survey 2014). We imported river and stream data for each study 243

area into ArcGIS and then estimated the distance of each female or random location to the 244

nearest river or non-intermittent stream. Similarly, we used distance to primary and secondary 245

(main roads) and tertiary roads as two metrics of anthropogenic mortality risk (Reynolds-246

Hogland et al. 2007; Martin et al. 2010; Waller et al. 2014), including hunting (Ciarniello et al. 247

2007; Czetwertynski et al. 2007; Ordiz et al. 2011). We obtained spatial road data in Michigan 248

from the United States Bureau of the Census (2011), in Missouri from the Missouri Spatial Data 249

Information Service (Curators of the University of Missouri-Columbia 2011), and Mississippi 250

from the Mississippi Automated Resource Information System (2014) and estimated distance of 251

Page 11 of 36



Draft

Duquette et al.|12

each female location or random point to the nearest main and tertiary road in each study area by 252

conducting a spatial join in ArcGIS. 253

We standardized all resource variables to z-scores and centered scores to provide equal 254

weight in multiple regression analyses (Zar 1999). We used variance inflation factor (VIF) 255

analysis to assess multicollinearity (collinearity considered ≥7) among resource variables (Quinn 256

and Keough 2002); no metrics were correlated (VIF = 1.05–1.13). We used package lme4 (Bates 257

et al. 2011) in program R (R core team 2013) to develop six generalized linear mixed-effects 258

models for each spatial scale in each state using a maximum likelihood estimator and binomial 259

distribution using female radiolocations (1) and random points (0) as the response variable. We 260

assessed a null model incorporating only an intercept and a naïve model incorporating main 261

roads, tertiary roads, riparian corridors, and NDVI as fixed effects. We structured the remaining 262

four models to assess a potential functional response to each of the variables using the following 263

notation: 264

g(x) = β0 + β1xij + …βnxnij = γnjxnj + γ0j 265

where β0 is the mean intercept, βn are the fixed regression coefficients with covariates xn, ij are 266

individual females clustered within each year, γnj is the random coefficient of covariate xn for 267

each female (j), and γ0j is the random intercept (Gillies et al. 2006). Each of the four functional 268

models incorporated the fixed parameters used in the naïve model, but incorporated either main 269

roads, tertiary roads, riparian corridors, or NDVI as a random covariate. Individual females were 270

clustered within year as a random effect to account for variation in female behavior among years 271

(Gillies et al. 2006). We considered parameter significance as α = 0.05. We verified model fit 272

by examining standardized versus fitted residual plots and ranked models using Akaike’s 273

Information Criterion corrected for small sample size (AICc; Burnham and Anderson 2002). We 274

Page 12 of 36



Draft

Duquette et al.|13

considered models competing if the difference in their ∆AICc were ≤2 and presented models 275

best explaining bear resource selection. 276

Results 277

We obtained a median of 1677 (range = 658−5051) locations from 7 females in Michigan, 3153 278

(range = 68−6127) locations from 11 females in Missouri, and 3377 (range = 103−10363) 279

locations from 11 females in Mississippi. Females in Mississippi used locations further from 280

tertiary roads than did females in Michigan or Missouri (Table 1). Females in Missouri used 281

locations further from main roads, but closer to riparian corridors, than did females in Michigan 282

or Mississippi. Females in Michigan used locations with NDVI values 14–15% less than 283

females in Missouri or Mississippi which had similar values. 284

Selection of home ranges 285

Variation in resource selection of females in Michigan was best explained (Table 2) by the 286

functional NDVI (i.e., vegetation productivity) model suggesting females selected tertiary and 287

main roads and locations with greater vegetation productivity, but avoided riparian corridors 288

(Table 3; Fig. 2a). Females in Michigan showed functional selection of tertiary roads between 289

about 300–500 m and NDVI of about 7000–8000. Variation in resource selection of females in 290

Missouri was best explained by the functional riparian corridors model suggesting females 291

selected riparian corridors, tertiary roads, and locations with greater vegetation productivity, but 292

avoided main roads (Fig. 2b). Females in Missouri showed functional selection of tertiary roads 293

between about 300–2000 m and riparian corridors between about 500–6500 m. Variation in 294

resource selection of females in Mississippi was best explained the functional NDVI model 295

suggesting females selected tertiary and main roads and locations with greater vegetation 296

productivity, but avoided riparian corridors (Fig. 2c). Females in Mississippi showed functional 297

Page 13 of 36



Draft

Duquette et al.|14

selection of tertiary roads between about 1500–4500 m and riparian corridors between about 298

5000–10000m and NDVI of about 7000–10000. Greater coefficient values indicated the 299

resource selection behaviors of females in Missouri were more pronounced than females in 300

Michigan or Mississippi. Females in Mississippi and Michigan had analogous resource-301

avoidance behaviors, but the resource selection behaviors of females in Michigan were more 302

pronounced. 303

Selection within home ranges 304

Females across populations had similar resource selection behaviors to those they used to 305

establish home ranges across the landscape. But coefficient values suggested female resource 306

selection behaviors were less pronounced within home ranges than those used to establish home 307

ranges across the landscape, with the exception of main roads and riparian corridors in 308

Mississippi. Although the functional NDVI model best explained variation of resource selection 309

across populations (Table 2), females used locations that avoided or were indifferent to areas 310

with greater vegetation productivity (Table 3; Figs. 2d, e, f). Females in Michigan showed 311

functional selection of riparian corridors between about 7000–11000 m and females in Missouri 312

showed functional selection of NDVI of about 4000–7000. 313

Discussion 314

Reproductive-aged female bear resource selection was best explained by a functional response to 315

vegetation productivity across nearly all populations and spatial scales, supporting our first 316

hypothesis. Similar to previous bear studies (Milakovic et al. 2012; Fowler 2014), we presume 317

vegetation productivity correlated with food resources reproductive females needed to maintain 318

nutritional condition for maximizing their reproductive success (Dahle and Swenson 2003; Ayers 319

et al. 2013). Marked support for a functional response to vegetation productivity emphasizes 320

Page 14 of 36



Draft

Duquette et al.|15

females were adjusting their behavior based on variation in vegetation productivity across the 321

landscape, and to a lesser extent within their home ranges. Compared to females in Missouri and 322

Mississippi, females in Michigan had the broadest functional selection of vegetation 323

productivity, which were also the least NDVI values across populations, despite similar or 324

greater availability across all landscapes. We suspect this trend in functional behavior was due 325

greater intraspecific competition for areas with productive vegetation (i.e., food and cover), 326

which increased despotic behavior (Beckmann and Berger 2003) from Mississippi to Michigan. 327

For example, the low density population in Mississippi would have minimal intraspecific 328

competition for those females and allowed them to select choice patches of food and cover 329

(Beyer et al. 2013) as we observed at the landscape scale. While females in Missouri also 330

exhibited a threshold of vegetation productivity selection, the prominence of riparian corridor 331

use by Missouri females was probably due to the steep terrain of the study area and 332

anthropogenic risk (e.g., vehicles; Brody and Pelton 1989) on hilltops, which filtered animals 333

nearer to streams or rivers (Garneau et al. 2006; Waller et al. 2013). Areas near riparian 334

corridors may have also provided suitable patches of food (Lyons et al. 2003) or cover in that 335

landscape. 336

The shift in selection to avoidance of vegetation productivity from the landscape to home 337

range scale further suggests intraspecific competition influenced female resource selection (van 338

Beest et al. 2014). This shift supported our second hypothesis that females would reduce their 339

selection for profitable resources as population density (i.e., intraspecific competition) increased. 340

The negative relationship between profitable resource access and intraspecific competition was 341

verified by greater selection of vegetation productivity by lower bear density populations in 342

Missouri and Mississippi compared to Michigan. With greater bear density, females in Michigan 343

Page 15 of 36



Draft

Duquette et al.|16

would have maximized their life-time reproductive success by occupying areas with reduced 344

mortality risk to themselves or cubs (LeCount 1987; Garrison et al. 2007), even if those habitats 345

had poorer food resources (Ben-David et al. 2004; McDonald Jr. and Fuller 2005). But females 346

in Michigan likely offset these constraints by establishing suitable home ranges across the 347

landscape using prior knowledge of food distribution (McDonald Jr. and Fuller 2005). This 348

strategy can allow females to minimize their space use while providing flexibility to move 349

among patches of food and avoid risks (Mitchell and Powell 2007), as indicated by the functional 350

selection of vegetation productivity by each population. 351

With the exception of vegetation productivity, females in each population established 352

home ranges across the landscape based on similar resource selection patterns as those within 353

home ranges. Although consistency in resource selection behaviors suggests these resources 354

were perceived similarly at each spatial scale, the strength of these behaviors were greater at the 355

landscape scale than within home range, supporting our third hypothesis. Similar to vegetation 356

productivity, broad-scale knowledge of variation in resources (McDonald Jr. and Fuller 2005) 357

and mortality risks was likely more beneficial for females than attempting to monitor and 358

respond to smaller-scale changes in their home ranges (Ordiz et al. 2011, 2012). For example, 359

females in Mississippi increased their selection of main roads and avoidance of riparian areas 360

from the landscape to home range scale, plausibly due to a 35% increase and 40% decrease in the 361

mean distances of those resources at the home range scale, respectively. The scale-dependent 362

behaviors of female bears in our study corroborate those of other large mammals (Rettie and 363

Messier 2000; Latif et al. 2011; Beyer et al. 2013) and similarly suggest females functionally 364

adjusted their behavior to the spatial scale at which each factor most limited their reproductive 365

success (Rettie and Messier 2000). 366

Page 16 of 36



Draft

Duquette et al.|17

The breadth of resource selection trade-offs were greatest for females in Missouri, 367

followed by Michigan and Mississippi, not supporting our fourth hypothesis that females would 368

exhibit greater behavioral trade-offs from Mississippi to Michigan. Variation in resource 369

selection across populations appeared to reflect resource availability in each landscape, rather 370

than bear population density (Bojarska and Selva 2012). While population-level variation in 371

selection of vegetation productivity appeared to be influenced by bear density, females likely 372

maximized their reproductive success by secondarily adapting to other landscape-specific 373

resources or mortality risks. Brown bear (Ursus arctos L., 1758) life-history strategies were 374

similarly shown to be primarily influenced by variation in environmental pressures across a large 375

portion of their range (Ferguson and McLoughlin 2000). Differences in resources and mortality 376

risks among populations may explain why females in Missouri had greater behavioral trade-offs 377

than in Michigan, despite bear density being 8.2–11.2 times greater in Michigan than in 378

Missouri. 379

Within populations, female trade-offs in resources and mortality risks were likely related 380

to landscape-specific variation in those factors. Tertiary road avoidance increased as bear 381

population density increased from Mississippi to Michigan. This trend in avoidance was closely 382

associated with landscape availability, as tertiary road density was 5.5 or 4.0 times greater in 383

Michigan and Missouri, respectively, than in Mississippi. By adjusting avoidance behavior to 384

landscape-specific tertiary road density, females in each population could have maximized 385

accessed resources (e.g., food) between tertiary roads with human mortality risk (Ciarniello et al. 386

2007; Reynolds-Hogland and Mitchell 2007) and interior habitats with greater risk of conflicts 387

with male bears (Garrison et al. 2007). Our results were corroborated by Stillfried et al. (2015) 388

who conducted an independent analysis of female black bear resource selection in the Michigan 389

Page 17 of 36



Draft

Duquette et al.|18

study area. Similarly, avoidance of primary roads by females in Missouri was likely due to main 390

roads being 1.3 to 6.0 times denser than in Mississippi and Michigan, respectively. The 391

functional selection of riparian corridors between about 1000–6000 m by females in Missouri 392

could represent a similar scenario as in Michigan, whereby females used areas (e.g, hillsides) 393

between main roads with human risks (e.g., vehicles) and riparian corridors where conspecifics 394

that tend to concentrate (Lyons et al. 2003). Given the choice, use of areas near riparian 395

corridors were likely safer than areas near main roads for females in Missouri. In contrast, 396

landscape terrain in Michigan and Mississippi was similarly flat and had low-density human 397

development which could have allowed females to use habitats closer to main roads with less 398

human disturbance. These landscape characteristics could have allowed females to access food 399

and cover between main roads with human risk (Jonkel and Cowen 1971; Reynolds-Hogland and 400

Mitchell 2007; Roever et al. 2008) and riparian areas with potentially greater conspecific risk 401

(Garneau et al. 2006; Waller et al. 2013) or thick vegetation that impeded movement. 402

Conclusions 403

We demonstrated variation in spatially-dependent resource selection across black bear 404

populations that allowed females to procure resources while reducing apparent mortality risks. 405

Females in our study adjusted their behavior to establish home ranges with a threshold of 406

vegetation productivity, under the confines of interspecific competition and mortality risk within 407

their population. While this behavior decreased within home ranges, females maintained 408

behavioral trade-offs among resources and mortality risks across two spatial scales. This 409

behavior emphasizes the importance of considering the functional behavior of populations at 410

multiple spatial scales (Rettie and Messier 2000) to provide the best knowledge to guide 411

population management. Similar functional trade-offs have been reported for other large 412

Page 18 of 36



Draft

Duquette et al.|19

mammals (Godvik et al. 2009; Matthiopoulos et al. 2011; Moreau et al. 2012), emphasizing that 413

behavioral flexibility allows animals to meet their life history requirements under spatially-414

variable intraspecific competition and environmental conditions. As black bear populations in 415

Missouri and Mississippi increase, greater competition for food and space will likely greater 416

constrain bear trade-offs between resources and mortality risks, as we observed between bear 417

density and vegetation productivity at the landscape scale. Increased competition from 418

conspecifics may increase female use of areas with greater mortality risks (e.g., food nearer to 419

tertiary roads), which could limit population growth (Simek et al. 2015). We suggest that 420

maintenance and establishment of movement corridors linking high quality food and cover 421

within and across bear populations is important for maintaining or restoring demographic and 422

genetic connectivity of bears (Dixon et al. 2006; Wilton et al. 2014a; Simek et al. 2012, 2015). 423

In addition to yielding important information for guiding black bear management, our study 424

underscores how multiple factors can influence broad variation in spatially-dependent functional 425

resource selection of a species across large portions of its range. Advances in computational 426

abilities should allow greater exploration of models incorporating multiple functional 427

relationships and improved information for management. 428

429

Acknowledgements 430

This project was supported by the Federal Aid in Wildlife Restoration Act under Pittman-431

Robertson. We thank the Michigan Department of Natural Resources; Mississippi Department 432

of Wildlife, Fisheries, and Parks; Missouri Department of Conservation; Safari Club 433

International Foundation; and Safari Club International–Michigan Involvement Committee for 434

financial support. We thank the Mississippi State University Department of Wildlife, Fisheries, 435

Page 19 of 36



Draft

Duquette et al.|20

and Aquaculture; Carnivore Ecology Laboratory; and Forest and Wildlife Research Center for 436

logistical support. Much gratitude to participating landowners for land access and C. Albright, S. 437

Edwards, D. O’Brien, P. Lederle, B. Roell, B. Young and numerous technicians for field and 438

technical support. We thank the Institute for Wildlife Studies for providing financial assistance 439

for primary author time to prepare this paper. We thank A. Bridges for valuable comments on an 440

earlier draft of this paper. 441

442

References 443

Ayers, C.R., Belant, J.L., and Millspaugh, J.J. 2013. Directness of resource use metrics affects 444

predictions of bear body fat gain. Polar Biol. 36: 169-176. 445

Basille, M., Van Moorter, B., Herfindal, I., Martin, J., Linnell, J.D.C., Odden, J., Andersen, R., 446

and Gaillard, J.M. 2013. Selecting habitat to survive: the impact of road density on 447

survival in a large carnivore. PLoS ONE, 8: e65493. 448

Bates, D., Maechler, M., and Bolker, B. 2011. lme4: Linear mixed-effects models using S4 449

classes. R package version 0.999375-42. Available from http://CRAN.R-450

project.org/package= lme4. [accessed 02 June 2015]. 451

Beckmann, J.P., and Berger, J. 2003. Using black bears to test ideal-free distribution models 452

experimentally. J. Mammal. 84: 594–606. 453

Belant, J.L., Griffith, B., Zhang, Y., Follmann, E.H., and Adams, L.G. 2010. Population-level 454

resource selection by sympatric brown bears and American black bears in Alaska. Polar 455

Biol. 33: 31–40. 456

Belant, J.L., Etter, D.R., Mayhew, S.L., Visser, L.G., and Friedrich, P.D. 2011. Improving large 457

scale mark-recapture estimates for American black bear populations. Ursus, 22: 9–23. 458

Page 20 of 36



Draft

Duquette et al.|21

Ben-David, M., Titus, K., and Beier, L.R. 2004. Consumption of salmon by Alaskan brown 459

bears: a trade-off between nutritional requirements and the risk of infanticide? Oecologia, 460

138: 465–474. 461

Beyer, H.L. 2012. Geospatial Modelling Environment (Version 0.7.1.0). Available from 462

http://www.spatialecology.com/gme. [accessed 02 June 2016]. 463

Beyer, H.L., Ung, R., Murray, D.L., and Fortin, M.J. 2013. Functional responses, seasonal 464

variation and thresholds in behavioural responses of moose to road density. J. Appl. Ecol. 465

50: 286–294. 466

Bojarska, K., and Selva, N. 2012. Spatial patterns in brown bear Ursus arctos diet: The role of 467

geographical and environmental factors. Mammal Rev. 42: 120–143. 468

Brody, A.J., and Pelton, M.R. 1989. Effects of roads on black bear movements in western North 469

Carolina. Wildl. Soc. Bull. 17: 5–10. 470

Burnham, K.P, and Anderson, D.R. 2002. Model selection and multimodel inference: a practical 471

information-theoretic approach. Springer, New York. 472

Ciarniello, L.M., Boyce, M.S., Heard, D.C., and Seip, D.R. 2007. Components of grizzly bear 473

habitat selection: density, habitats, roads, and mortality risk. J. Wildl. Manage. 71: 474

1446-1457. 475

Creel, S., Winnie, Jr., J.A., Christianson, D., and Liley, S. 2008. Time and space in general 476

models of antipredator response: tests with wolves and elk. Anim. Behav. 76: 1139–1146. 477

Curators of the University of Missouri-Columbia. 2011. Missouri Spatial Data Information 478

Service. Available from http://msdis.missouri.edu/. [accessed 02 June 2015]. 479

Czetwertynski, S.M., Boyce, M.S., and Schmiegelow, F.K. 2007. Effects of hunting on 480

demographic parameters of American black bears. Ursus, 18: 1–18. 481

Page 21 of 36



Draft

Duquette et al.|22

Dahle, B., and Swenson, J.E. 2003. Seasonal range size in relation to reproductive strategies in 482

brown bears Ursus arctos. J. Anim. Ecol. 72: 660-667. 483

Dixon, J.D., Oli, M.K., Wooten, M.C., Eason, T.H., McCown, J.W., and Paetkau, D. 2006. 484

Effectiveness of a regional corridor in connecting two Florida black bear populations. 485

Conserv. Biol. 20: 155-162. 486

Elowe, K.D., and Dodge, W.E. 1989. Factors affecting black bear reproductive success and cub 487

survival. J. Wildl. Manage. 53: 962-968. 488

Environmental Systems Research Institute, Inc. 2011. ArcGIS 10.0. Redlands, California, USA. 489

Ferguson, S.H., and McLoughlin, P.D. 2000. Effect of energy availability, seasonality, and 490

geographic range on brown bear life history. Ecography, 23: 193-200. 491

Fowler, N.L. 2014. Expression of facultative den behavior in Ursidae. Thesis, Mississippi State 492

University, Mississippi State, USA. 493

Fry, J.A., Coan, M., Homer, C., Meyer, D., and Wickham, J. 2011. Completion of the 2006 494

National Land Cover Database for the conterminous United States. Photogramm. Eng. 495

Remote Sens. 77: 858–864. 496

Garneau, D.E., Boudreau, T., Keech, M., and Post, E. 2006. Habitat use by black bears in 497

relation to conspecifics and competitors. Mamm. Biol. 73: 48–57. 498

Garrison, E.P., McCown, J.W., and Oli, M.K. 2007. Reproductive ecology and cub survival of 499

Florida black bears. J. Wildl. Manage. 71: 720–727. 500

Gillies, C.S., Hebblewhite, M., Nielsen, S.E., Krawchuk, M.A., Aldridge, C.L., Frair, J.L., Saher, 501

D.J., Stevens, C.E., and Jerde, C.L. 2006. Application of random effects to the study of 502

resource selection by animals. J. Anim. Ecol. 75: 887–898. 503

Page 22 of 36



Draft

Duquette et al.|23

Godvik, I.M.R., Loe, L.E., Vik, J.O., Veiberg, V., Langvatn, R., and Mysterud, A. 2009. 504

Temporal scales, trade-offs, and functional responses in red deer habitat selection. 505

Ecology, 90: 699–710. 506

Hiller, T.L., and Belant, J.L. 2015a. Sexual-size dimorphism mediates effects of resource 507

dispersion on American black bear space use. J. Zool. (Lond) 295: 200–207. 508

Hiller, T.L., Belant, J.L., Beringer, J., and Tyre, A.J. 2015b. Resource selection by recolonizing 509

American black bears in a fragmented forest landscape. Ursus, 26: 116–128. 510

Johnson, C.J., and Seip, D.R. 2008. Relationship between resource selection, distribution, and 511

abundance: a test with implications to theory and conservation. Popul. Ecol. 50: 145–157. 512

Johnson, K.G., and Pelton, M.R. 1980. Prebaiting and snaring techniques for black bears. Wildl. 513

Soc. Bull. 8: 46–54. 514

Jonkel, C.J., and Cowan, I.M. 1971. The black bear in the spruce-fir forest. Wildl. Monogr. 27: 515

3–57. 516

Kiffner, C., Kioko, J., Kissui, B., Painter, C., Serota, M., White, C., and Yager, P. 2014. 517

Interspecific variation in large mammal responses to human observers along a 518

conservation gradient with variable hunting pressure. Anim. Conserv. 17: 603-612. 519

Kreeger, T.J., and Arnemo, J.M. 2007. Handbook of wildlife chemical immobilization, 3rd edn. 520

Wildlife Pharmaceuticals. Sunquest, Shanghai, China. 521

Latif, Q.S., Heath, S.K., and Rotenberry, J.T. 2011. An ‘ecological trap’ for yellow warbler nest 522

microhabitat selection. Oikos, 120: 1139–1150. 523

LeCount, A.L. 1987. Causes of black bear cub mortality. Int. Conf. Bear Res. Manage. 7: 75–82. 524

Lyons, A.L. 2003. Black bear resource selection in the northeast Cascades, Washington. Biol. 525

Conserv. 113: 55–62. 526

Page 23 of 36



Draft

Duquette et al.|24

Mace, R.D., Waller, J.S., Manley, T.L., Lyon, L.J., and Zuuring, H. 1996. Relationships among 527

grizzly bears, roads and habitat in the Swan Mountains Montana. J. Appl. Ecol. 33: 528

1395–1404. 529

Manly, B.F.J., McDonald, L.L., Thomas, D.L., McDonald, T.L., and Erickson, W.P. 2002. 530

Resource selection by animals. 2nd

edition. Kluwer Academic Publishers, Boston, MA. 531

Martin, J., Basille, M., Van Moorter, B., Kindberg, J., Allainé, D., and Swenson, J.E. 2010. 532

Coping with human disturbance: spatial and temporal tactics of the brown bear (Ursus 533

arctos). Can. J. Zool. 88: 875-883. 534

Matthiopoulos, J., Hebblewhite, M., Aarts, G., and Fieberg, J. 2011. Generalized functional 535

responses for species distributions. Ecology, 92: 583–589. 536

McDonald, Jr., J.E., and Fuller, T.K. 2005. Effects of spring acorn availability on black bear diet, 537

milk consumption, and cub survival. J. Mammal. 86: 1022–1028. 538

McLoughlin, P.D., Ferguson, S.H., and Messier, F. 2000. Intraspecific variation in home range 539

overlap with habitat quality: a comparison among brown bear populations. Evol. Ecol. 540

14: 39-60. 541

McLoughlin, P.D., Morris, D.W., Fortin, D., Vander Wal, E., and Contasti, A.L. 2010. 542

Considering ecological dynamics in resource selection functions. J. Anim. Ecol. 79: 4–543

12. 544

Milakovic, B., Parker, K.L., Gustine, D.D., Lay, R.J., Walker, A.B.D., and Gillingham, M.P. 545

2012. Seasonal habitat use and selection by grizzly bears in northern British Columbia. J. 546

Wildl. Manage. 76: 170–180. 547

Milner, J.M., Nilsen, E.B., and Andreassen, H.P. 2007. Demographic side effects of selective 548

hunting in ungulates and carnivores. Conserv. Biol. 21: 36-47. 549

Page 24 of 36



Draft

Duquette et al.|25

Mississippi Automated Resource Information System. 2014. Available from 550

http://www.maris.state.ms.us/HTM/DownloadData/Statewide.html. [accessed 02 June 551

2015]. 552

Mitchell, M.S., and Powell, R.A. 2007. Optimal use of resources structures home ranges and 553

spatial distribution of black bears. Anim. Behav. 74: 219-230. 554

Moreau, G., Fortin, D., Couturier, S., and Duchesne, T. 2012. Multi-level functional responses 555

for wildlife conservation: the case of threatened caribou in managed boreal forests. J. 556

Appl. Ecol. 49: 611–620. 557

Naiman, R.J., and Rogers, K.H. 1997. Large animals and system-level characteristics in river 558

corridors. BioScience 47: 521–529. 559

National Oceanic and Atmospheric Administration (NOAA). 2014. National Climatic Data 560

Center. Available from http://cdo.ncdc.noaa.gov/qclcd/. [accessed 02 June 2015]. 561

Ordiz, A., Støen, O.G., Delibes, M., and Swenson, J.E. 2011. Predators or prey? Spatio-temporal 562

discrimination of human-derived risks by brown bears. Oecologia, 166: 59-67. 563

Ordiz, A., Støen, O.G., Sæbø, S., Kindberg, J., Delibes, M., and Swenson, J.E. 2012. Do bear 564

know they are being hunted? Biol. Conserv. 152: 21-28. 565

Palomares, F., and Caro, T.M. 1999. Interspecific killing among mammalian carnivores. Am. 566

Nat. 153: 492-508. 567

Puckett, E.E., Kristensen, T.V., Wilton, C.M., Lyda, S.B., Noyce, K.V., Holahan, P.M., Leslie, 568

Jr., D.M., Beringer, J., Belant, J.L., White, Jr., D, and Eggert, L.S. 2014. Influence of 569

drift and admixture on population structure of American black bears (Ursus americanus) 570

in the Central Interior Highlands, USA 50 years after translocation. Mol. Ecol. 23: 2414–571

2427. 572

Page 25 of 36



Draft

Duquette et al.|26

Quinn, G.P., and Keough, M.J. 2002. Experimental design and data analysis for biologists. 573

Cambridge University Press, New York, NY. 574

R Core Team. 2013. R: A language and environment for statistical computing website. R 575

Foundation for Statistical Computing. Available from http://www.R-project.org/. 576

Raeker, G., Fleming, J., Morris, M., Moser, K., and Treiman, T. 2010. Missouri’s forest resource 577

assessment and strategy, seeking a sustainable future for Missouri’s forest resources. 578

Missouri Department of Conservation and US Forest Service, Northern Research Station. 579

Reagan, S.R., Ertel, J.M., Stinson, P., Yakupzack, P., and Anderson, D. 2002. A passively 580

triggered foot snare design for American black bears to reduce disturbance by non-target 581

animals. Ursus, 13: 317–320. 582

Rettie, W.J., and Messier, F. 2000. Hierarchical habitat selection by woodland caribou: its 583

relationship to limiting factors. Ecography 23: 466-478. 584

Reynolds-Hogland, M.J., and Mitchell, M.S. 2007. Effects of roads on habitat quality for bear in 585

the southern Appalachian: a long-term study. J. Mammal. 88: 1050-1061. 586

Roever, C.L., Boyce, M.S., and Stenhouse, G.B. 2008. Grizzly bears and forestry I: road 587

vegetation and placement as an attractant to grizzly bears. Forest Ecol. Manag. 256: 588

1253–1261. 589

Scheick, B.K., and McCown, W. 2014. Geographic distribution of American black bears in 590

North America. Ursus, 25: 24-33. 591

Shepherd, B., and Whittington, J. 2006. Response of wolves to corridor restoration and human 592

use management. Ecol. Soc. 11: 1. 593

Page 26 of 36



Draft

Duquette et al.|27

Sikes, R.S., Gannon, W.L., and Animal Care and Use Committee of the American Society of 594

Mammalogists. 2011. Guidelines of the American Society of Mammalogists for the use 595

of wild mammals in research. J. Mammal. 92: 235–253. 596

Simek, S.L., Belant, J.L., Young, B.W., Shropshire, C., and Leopold, B.D. 2012. History and 597

status of the American black bear in Mississippi. Ursus, 23: 159–167. 598

Simek, S.L., Belant, J.L., Fan, Z., Young, B., Leopold, B.D., Fleming, J., and Waller, B. 2015. 599

Source populations and roads affect American black bear recolonization. Eur. J. Wildl. 600

Res. 61: 583–590. 601

Stillfried, M., Belant, J.L., Svoboda, N.J., Beyer, Jr., D.E., and Kramer-Schadt, S. 2015. When 602

top predators become prey: large carnivores alter movement behaviour in response to 603

hunting pressure. Behav. Processes, 120: 30–39. 604

Thomas, D.L., and Taylor, E.J. 2006. Study designs and tests for comparing resource use and 605

availability II. J. Wildl. Manage. 70: 324–336. 606

United States Bureau of the Census. 2011. Topologically Integrated Geographic Encoding and 607

Referencing (TIGER) system files website. Available from 608

http://www.census.gov/geo/www/tiger/. [accessed 02 June 2015]. 609

United States Geological Survey. 2013. 250 m MODIS normalized difference vegetation index. 610

Available from http://earthexplorer.usgs.gov/. [accessed 02 June 2015]. 611

United States Geological Survey. 2014. National Hydrography Data. Available from 612

http://nhd.usgs.gov/. [accessed 02 June 2015]. 613

Waller, B.W., Belant, J.L., Leopold, B.D., Evans, D.L., Young, B.W., and Simek, S.L. 2013. 614

Scale-dependent den-site selection by American black bears in Mississippi. Mamm. 615

Study 38: 211–216. 616

Page 27 of 36



Draft

Duquette et al.|28

Waller, B.W., Belant, J.L., Leopold, B.D., Young, B.W., Simek, S.L., and Evans, D.L. 2014. 617

Influence of landscape attributes on American black bear den-site selection in 618

Mississippi. Mamm. Study 39: 115–119. 619

Watts, H.E., and Holekamp, K.E. 2009. Ecological determinants of survival and reproduction in 620

the spotted hyena. J. Mammal. 90: 461-471. 621

Wilton, C.W., Belant, J.L., and Beringer, J. 2014a. Distribution of American black occurrences 622

and human-bear incidents in Missouri. Ursus, 25: 53–60. 623

Wilton, C.M., Puckett, E.E., Beringer, J., Gardner, B., Eggert, L.S., and Belant, J.L. 2014b. Trap 624

array configuration influences estimates and precision of black bear density and 625

abundance. PLoS ONE, 9: e111257. 626

Zar, J.H. 1999. Biostatistical Analysis. 4th edn. Prentice Hall, New Jersey. 627

Zedrosser, A., Dahle, B., and Swenson, J.E. 2006. Population density and food conditions 628

determine adult female body size in brown bears. J. Mammal. 87: 501–518. 629

van Beest, F.M., Uzal, A., Vander Wal, E., Laforge, M.P., Contasti, A.L., Colville, D., and 630

McLoughlin, P.D. 2014. Increasing density leads to generalization in both coarse-grained 631

habitat selection and fine-grained resource selection in a large mammal. J. Anim. Ecol. 632

83: 147–156. 633

634

Page 28 of 36



Draft

Table 1. Mean and standard deviation (SD) of variables related to locations of reproductive female black bear (Ursus americanus) in

Michigan, Missouri, and Mississippi, USA, during spring-summer 2009–2013.

Michigan Missouri Mississippi

Variable Definition Representation Estimate Mean SD Mean SD Mean SD

Bear density

(bears/100 km2)

Michigan: J.L. Belant, unpub. data. Resource competition Population density 14–19 - 1.7 - < 1 -

Missouri: Wilton et al. 2014b.

Mississippi: R. Rummell, MDWFP,

pers. comm.

Distance to nearest

tertiary road (m)

Local county or city roads, off-road

vehicle trails, and private roads

(United States Bureau of the Census

2011)

Mortality risk or

movement corridor Density (km/km

2) 1.69 - 1.28 - 0.31 -

Used locations 496 386 511 316 3702 3251

LS availability 653 518 823 767 5476 5637

HR availability 625 414 611 325 3799 3256

Distance to nearest

main road (m)

Primary and secondary (main) roads

are generally divided, limited-access

highways or main arteries and have

one or more lanes of traffic in each

direction, and typically have

intersections with many other roads

(United States Bureau of the Census

2011)

Mortality risk Density (km/km2) 0.03 - 0.18 - 0.14 -

Used locations 2142 1323 7689 2686 1716 1293

LS availability 5808 2948 1914 2156 4792 3585

HR availability 5914 3058 4172 1799 6460 2685

Distance to nearest

riparian corridor (m)

Streams/rivers, braided streams,

canals, ditches, and pond and lake

edges (United States Geological

Survey 2014)

Movement corridors or

areas of food resources

Density (km/km2) 1.17 - 0.08 - 0.33 -

Used locations 5833 3097 4064 1784 6366 2683

LS availability 3260 2125 9429 5961 4542 3709

HR availability 2271 1349 7792 2690 1814 1301

NDVI 16-day composite Normalized

Difference Vegetation Index

Food resources Used locations 6449 985 7478 503 7576 1049

LS availability 6160 1240 6143 1925 5169 2248

HR availability 6522 1007 7583 530 7595 1228

Page 29 of 36



Draft

Table 2. Model rankings of generalized linear mixed-effect models assessing resource selection of reproductive female black bear (n =

29; Ursus americanus) across the study landscape and within home ranges in Michigan, Missouri, and Mississippi, USA, during

spring-summer 2009–2013. Models included individual bear and year as random effects on the intercept. Models presented Akaike’s

Information Criterion corrected for small sample size (AICc), difference in AIC relative to minimum AIC (∆AICc), AIC weight of

model relative to other models (AICc Wt), number of parameters (K), log-likelihood (LL). The model NDVI is the Normalized

Difference Vegetation Index.

Landscape Home range

State Model K AICc ∆AICc AICc Wt LL Model K AICc ∆AICc AICc Wt LL

Michigan NDVI 8 14233.0 0.0 1.0 -7108.5 NDVI 8 12522.6 0.0 1.0 -6253.3

(n = 7) Riparian corridor 8 20368.1 6135.1 0.0 -10176.0 Riparian corridor 8 19403.0 6880.4 0.0 -9693.5

Main roads 8 20838.6 6605.6 0.0 -10411.3 Main roads 8 19395.8 6873.2 0.0 -9689.9

Tertiary roads 8 20882.9 6649.9 0.0 -10433.5 Tertiary roads 8 19391.6 6869.0 0.0 -9687.8

Naïve 7 20949.1 6716.1 0.0 -10467.6 Naïve 7 19489.0 6966.5 0.0 -9737.5

Null 3 42886.8 28653.8 0.0 -21440.4 Null 3 42864.4 30341.8 0.0 -21429.2

Missouri Riparian corridor 8 18270.1 0.0 1.0 -9127.1 NDVI 8 12522.6 0.0 1.0 -6253.3

(n = 11) NDVI 8 18365.8 95.7 0.0 -9174.9 Tertiary roads 8 19391.6 6869.0 0.0 -9687.8


Tertiary roads 8 25841.9 7571.8 0.0 -12912.9 Riparian corridor 8 19403.0 6880.4 0.0 -9693.5

Naïve 7 26346.3 8076.2 0.0 -13166.2 Naïve 7 19489.0 6966.5 0.0 -9737.5

Null 3 95957.0 77686.9 0.0 -47975.5 Null 3 42864.4 30341.8 0.0 -21429.2

Mississippi NDVI 8 43575.4 0.0 1.0 -21779.7 NDVI 8 30313.8 0.0 1.0 -15148.9

(n = 11) Riparian corridor 8 44157.3 581.9 0.0 -22070.7 Riparian corridor 8 38815.4 8501.6 0.0 -19399.7


Page 30 of 36



Draft

Tertiary roads 8 51588.6 8013.2 0.0 -25786.3 Tertiary roads 8 40040.5 9726.8 0.0 -20012.3

Naïve 7 60549.4 16974.0 0.0 -30267.7 Naïve 7 40520.6 10206.9 0.0 -20253.3

Null 3 128997.9 85422.5 0.0 -64496.0 Null 3 95957.0 65643.2 0.0 -47975.5

Page 31 of 36



Draft

Table 3. Generalized linear mixed-effect models best explaining resource selection of

reproductive female black bear (n = 30; Ursus americanus) across the study landscape or within

home ranges in Michigan, Missouri, and Mississippi, USA, during spring-summer 2009–2013.

Models presented with standardized parameter estimates (β), standard errors (SE) or deviations

(SD), probability values (P), and the variance (Var) of the functional response parameter

(Normalized Difference Vegetation Index [NDVI] or riparian), individual bear, and year

included as random effects on the intercept.

Michigan

Landscape Home range

β0 SE P β0 SE P

-4.223 0.344 < 0.001 -3.501 0.358 < 0.001

βtertiary roads SE P βtertiary roads SE P

-5.286 0.249 < 0.001 -2.568 0.213 < 0.001

βmain roads SE P βmain roads SE P

-3.717 0.072 < 0.001 -3.275 0.078 < 0.001

βriparian SE P βriparian SE P

3.232 0.073 < 0.001 2.919 0.074 < 0.001

βNDVI SE P βNDVI SE P

1.766 0.102 < 0.001 -0.065 0.071 < 0.001

VarNDVI SD VarNDVI SD

7.901 2.811 6.226 2.495

Varbear SD Varbear SD

0.539 0.734 0.294 0.543

Varyear SD Varyear SD

0.049 0.221 0.170 0.412

Missouri

β0 SE P β0 SE P

-0.511 -0.170 0.865 -2.979 0.283 < 0.001


-4.316 1.143 < 0.001 -2.872 0.162 < 0.001


8.959 0.602 < 0.001 2.021 0.029 < 0.001


Page 32 of 36



Draft

-9.321 0.493 < 0.001 -2.226 0.030 < 0.001


8.378 0.569 < 0.001 -0.104 0.074 0.160

Varriparian SD VarNDVI SD

6.204 2.560 7.281 2.698


0.869 0.932 0.054 0.233


0.441 0.664 0.176 0.419

Mississippi

β0 SE P β0 SE P

-2.241 0.330 < 0.001 -0.981 0.299 0.001


-0.215 0.015 < 0.001 0.038 0.016 0.019


-2.525 0.031 < 0.001 -4.477 0.053 < 0.001


1.583 0.024 < 0.001 4.180 0.051 < 0.001


2.668 0.048 < 0.001 -0.142 0.029 < 0.001

VarNDVI SD VarNDVI SD

5.367 2.317 3.154 1.776


1.104 1.051 0.387 0.622


0.029 0.169 0.307 0.554

Page 33 of 36



Draft

Figure captions

Figure 1. Reproductive-age female black bear (Ursus americanus) study areas (black polygons)

and locations within study areas in Michigan, Missouri, and Mississippi, USA, 2009–2013.

Figure 2. Coefficient relationships of the best generalized linear mixed-effect model explaining

the resource selection of reproductive-age female black bears (Ursus americanus) across the

study landscape or within home ranges in Michigan (a, d), Missouri (b, e), and Mississippi (c, f),

USA, during spring–summer 2009–2013. Solid bold line is main roads, solid thin line is tertiary

roads, long-dashed line is riparian corridors, and dotted line is the Normalized Difference

Vegetation Index. Gray fill around lines represent ±1 standard error of the mean of each

resource.

Page 34 of 36



Draft

52x68mm (600 x 600 DPI)

Page 35 of 36



Draft

793x806mm (96 x 96 DPI)

Page 36 of 36



draft...draft duquette et al.|3 46 selection can be influenced by multiple spatially-dependent...

Documents