.. ,9
y‘.
i.
:

 

5...... $6.. ‘ . ‘

; 5 .. , . .
V . . (xi,
‘5...

 

 

all:

:1» .1
lo I
rm. .

 

 

u: _ kw

m

u LN

w
(

 

 

This is to certify that the
dissertation entitled

LAND-USE PATTERNS AND POPULATION
CHARACTERISTICS OF WHITE-TAILED DEER IN AN
AGRO-FOREST ECOSYSTEM IN SOUTH CENTRAL

MICHIGAN

presented by

TIM L. HILLER

has been accepted towards fulﬁllment
of the requirements for the

PhD. degree in Wildlife EcologL

 

 

khk'K—L 'va)/\\ 99‘ I\\ ;

Major‘ProfessoT’? Signature

\\ DQC7M\3{¥‘LQQ *

 

Date

MSU is an afﬁrmative-action, equal-opportunity employer

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

3

JAN”
:13

I3

 

ﬂ
‘1

"1'
I

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/ClRC/DateDue.indd-p.1

LAND-USE PATTERNS AND POPULATION CHARACTERISTICS
OF WHITE-TAILED DEER IN AN AGRO-FOREST ECOSYSTEM
IN SOUTH CENTRAL MICHIGAN
By

Tim L. Hiller

A DISSERTATION
Submitted to
Michigan State University
In partial fulﬁllment of the requirements
For the degree of
DOCTOR OF PHILOSOPHY
Department of Fisheries and Wildlife

2007

ABSTRACT
LAND-USE PATTERNS AND POPULATION CHARACTERISTICS
OF WHITE-TAILED DEER IN AN AGRO-FOREST ECOSYSTEM
IN SOUTH CENTRAL MICHIGAN
By
Tim L. Hiller
Assessments of demographics and space use are important for habitat and harvest

management of white-tailed deer (Odocoileus virginianus). These population
characteristics often vary across a landscape, and by age (e. g., fawn, yearling, adult) and
sex class. Knowledge of demographics and space use of young fawns is particularly
limited, despite the potential for recruitment having a relatively large inﬂuence on deer
population dynamics. My objectives were to describe age-speciﬁc survival, cause-
speciﬁc mortality, and space use in an agro-forest ecosystem undergoing increasing
urbanization (i.e., increasing housing developments, increasing human population) in
south central Michigan. I captured, radiomarked, and monitored 66 deer during winter
and 34 neonates during spring 2004—2006. Annual survival varied by age class (fawn =
0.51, yearling = 0.94, adult = 0.56), and annually based sources of mortality were
primarily vehicle collisions (fawns) and hunter-harvest (adults). Two- and 6-month post-

capture survival estimates of neonates were 81% and 67%, respectively, and canids

caused most mortalities during both time periods. Yearlings had larger seasonal home
ranges (agricultural growing season: T: = 201.8 ha i 91.1 SE; non-growing season: J-c =
156.9 ha i 28.2 SE) than either fawns (60.2 ha i 14.1; 116.3 ha i 20.6) or adults (77.5

ha i 9.6; 140.4 ha i- 23.4). Home ranges for fawns 0—2 months old averaged 40.9 ha

(range = 27—1668), with conifers and lowland deciduous forests selected in proportions

higher than available on the study area. Adult female deer had relatively small home
ranges compared to deer in other Michigan studies indicating that their habitat
components were readily available. Additionally, this sex-age class is of primary interest
to managers desiring to reduce high deer numbers. To describe cover selection of adult
female white-tailed deer (n = 20), I used a multi-scale approach by varying deﬁnitions of
cover use and availability. The number of cover types assigned as selected decreased
from coarse (landscape) to ﬁne (within home range) scales. Two cover types (conifers,
upland deciduous forests) were consistently ranked as the most important regardless of
scale. I used the concept of usable space (i.e., “ideal” permanent cover situations) to
describe a potentially more accurate biological representation (compared to traditional
home-range estimators) of space use by adult female white-tailed deer. Fixed-kemel
home-range estimates might misrepresent space use by including cover types with no
location estimates (i.e., no evidence of use). Usable space estimates (ha) were
approximately 75% that of kernel home ranges, and were dominated (~87% of area) by
upland deciduous forest, lowland shrub, agriculture, and coniferous cover types. Under
the assumption that deer densities are positively correlated with the amount of usable
space, several cover conversion scenarios (i.e., habitat manipulation) would theoretically
change deer abundance on an area of interest by changing the amount of usable space.
Knowledge of age-speciﬁc deer demographics and addressing deer-habitat management
issues through a multi-scale perspective and a usable-space approach are both relatively
recent but seemingly useful concepts that also have relevance to population ecology of

deer and other wildlife species.

ACKNOWLEDGMENTS

First and foremost, I greatly appreciate the support of my family throughout my
graduate career: Tom, Pat, Troy, Jeanne, and Justin Hiller; Phil and Dorothy Rust; Tom
and Krystal Hennessey; and Danielle Hiller. No doubt they were the primary support that
I relied upon over the past 6 years. Unfortunately, my grandparents, H. A. and Marjorie
Hiller, both passed away while I was in graduate school. This dissertation is dedicated to
their memory.

I thank my advisor, Dr. Henry (Rique) Campa III, for his guidance and patience
throughout my Ph.D. experience, as I learned much through his advice. I also thank Dr.
Barbara Lundrigan, Mr. Brent Rudolph, Dr. Shawn Riley, and Dr. Scott Winterstein for
providing their expertise through service on my graduate committee. Certainly, my
research and writing also beneﬁted greatly from Dr. Fred Guthery, Dr. Jeff Lusk, Dr.
Meredith Gore, Dr. Dwayne Etter, and Kristie Sitar.

Numerous graduate students with the Department of Fisheries and Wildlife
provided me with ﬁiendship, advice, and everything in between, including Alexandra
Felix, Stacy Lischka, Sarah Hamer, Eli Ball, Dan Linden, Dan Walsh, Katrina Mueller,
Nancy Schwalm, Nancy Leonard, Joel Humphries, and many others. I am greatly
indebted to the integrity and perseverance of my technicians under all ﬁeld conditions
and public interactions: Alan Leach, Ed Arrow, Randy Havens, Bill Dodge, Lance
McNew, Damon Haan, Mike Rubley, Ben Gunderson, and Anna Nussbaum.

I also thank the numerous volunteers, graduate students, and MDNR personnel

that assisted with deer capture, and the landowners (especially the Schafers, Tischs, and

iv

Freys) that granted me permission to conduct research on their properties. Several
members of the MDNR made my research efforts much more productive, for which I am
grateful, including Rod Clute, Dr. Shelli Dubay, Shannon Hanna, Fred Davis, Veryl
Tisch, Earl Flegler, Eric Dunton, Kristin Bissell, Marshall Strong, and many others. I
also am indebted to Tom Cooley, MDNR wildlife pathologist, for his necropsy work and
his willingness to answer my questions, and Jordan Burroughs for allowing me to make
comparisons between my project and her past research on deer.

My research was made possible through ﬁnancial support from the Michigan
Agricultural Experiment Station, Michigan State University, the Michigan Department of
Natural Resources through the Federal Aid in Restoration Act under Pittman-Robertson
project W-147—R, Safari Club International, and Whitetails Unlimited. Michigan State
University’s All-University Committee on Animal Use and Care approved all capturing

and handling procedures for my research project (Application No. 01/04-006-00).

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................. ix
LIST OF FIGURES .............................................................................................. xii
ORGANIZATION OF DISSERTATION ............................................................. xv
INTRODUCTION ................................................................................................... I
OBJECTIVES .............................................................................................. 3
LITERATURE CITED ................................................................................ 5
CHAPTER 1: Age-speciﬁc Population Characteristics of White-tailed Deer in Southern
Michigan
INTRODUCTION ................................................................................................... 8
STUDY AREA ...................................................................................................... 10
METHODS ............................................................................................................ l3
Capturing Deer ........................................................................................... 13
Survival and Cause-speciﬁc Mortality ....................................................... 15
Space Use ................................................................................................... 16
RESULTS .............................................................................................................. 18
Capture and Estimation of Locations ......................................................... 18
Survival and Cause-speciﬁc Mortality ....................................................... 18
Space Use ................................................................................................... 19
DISCUSSION ........................................................................................................ 21
Michigan Deer Research ............................................................................ 22
Age-speciﬁc Characteristics ...................................................................... 23
Age-speciﬁc Trends ................................................................................... 26
LITERATURE CITED .......................................................................................... 29
TABLES AND FIGURES ..................................................................................... 34

CHAPTER 2: Survival and Space Use of Fawn White-tailed Deer in Southern Michigan

INTRODUCTION ................................................................................................. 46
STUDY AREA ...................................................................................................... 47
METHODS ............................................................................................................ 50
Survival and Cause-speciﬁc Mortality ....................................................... 51
Space Use ................................................................................................... 52
RESULTS .............................................................................................................. 53
Survival and Cause-speciﬁc Mortality ....................................................... 53
Space Use ................................................................................................... 54

vi

CHAPTER 2 (cont’d)

DISCUSSION ........................................................................................................ 55
Survival and Cause-speciﬁc Mortality ....................................................... 55
Space Use ................................................................................................... 58
LITERATURE CITED .......................................................................................... 60
TABLES ................................................................................................................ 64

CHAPTER 3: A Multi-scale Approach to Describe Cover Selection of White-tailed Deer
Using Use-Availability Data

INTRODUCTION ................................................................................................. 68
STUDY AREA ...................................................................................................... 71
METHODS ............................................................................................................ 73
Capturing Deer ........................................................................................... 73
Estimating Locations ................................................................................. 75
Cover Selection .......................................................................................... 76
RESULTS .............................................................................................................. 82
Capture and Estimation of Locations ......................................................... 82
Cover Selection .......................................................................................... 82
DISCUSSION ........................................................................................................ 85
Single-scale Studies ................................................................................... 87
Issues of Scale ............................................................................................ 89
Conclusions ................................................................................................ 9O
LITERATURE CITED .......................................................................................... 92
TABLES AND FIGURES ..................................................................................... 97

CHAPTER 4: Estimation and Implications of Space Use for White-tailed Deer
Management in Southern Michigan

INTRODUCTION ............................................................................................... 1 15
STUDY AREA .................................................................................................... 1 l6
METHODS .......................................................................................................... 119
Capturing Deer ......................................................................................... 119
Estimating Locations ............................................................................... 120
Assessment of Space Use ......................................................................... 121
Cover Conversion Estimation .................................................................. 125
RESULTS ............................................................................................................ 127
Capture and Estimation of Locations ....................................................... 127
Assessment of Space Use ......................................................................... 128
Cover Conversion Estimation .................................................................. 129
DISCUSSION ...................................................................................................... 130
Cover Conversions ................................................................................... 132
Selection of u,,, ......................................................................................... 134
Cover lnterspersion Rules ........................................................................ 135
Considerations .......................................................................................... 136

vii

CHAPTER 4 (cont’d)

LITERATURE CITED ........................................................................................ 140
TABLES AND FIGURES ................................................................................... 144

CHAPTER 5: Conclusions and Implications

CONCLUSIONS AND IMPLICATIONS ........................................................... 156
Population Characteristics ....................................................................... 157
Resource Selection ................................................................................... 158
Space Use ................................................................................................. 159

LITERATURE CITED ........................................................................................ 162

APPENDICES

APPENDIX A: Assessment of Precision between GPS- and VHF-derived Location

Estimates

INTRODUCTION ............................................................................................... 164

STUDY AREA .................................................................................................... 166

METHODS .......................................................................................................... 166
Stationary VHF Transmitters ................................................................... 167
Animal-bome GPS-VHF Transmitters .................................................... 167
Triangulations .......................................................................................... 168
Cover-type Classiﬁcation ......................................................................... 169

RESULTS ............................................................................................................ 170
Stationary VHF Transmitters ................................................................... 170
Animal-bome GPS-VHF Transmitters .................................................... 171

DISCUSSION ...................................................................................................... 172

MANAGEMENT IMPLICATIONS ................................................................... 174

LITERATURE CITED ........................................................................................ 175

FIGURES ............................................................................................................. 177

APPENDIX B: Capture and Fate Descriptions of White-tailed Deer ................. 183

TABLES .............................................................................................................. 184

viii

LIST OF TABLES

CHAPTER 1: Age-speciﬁc Population Characteristics of White-tailed Deer in Southern

Michigan
Table 1.1. Comparison of 3 winter capture seasons of white-tailed deer
using Clover traps during 2004—2006, south central Michigan.
USA ............................................................................................ 34
Table 1.2. Comparison of 3 spring capture seasons of neonatal white-

tailed deer during 2004—2006, south central Michigan, USA ....35

Table 1.3. Fate assessment of radiomarked female white-tailed deer
(Odocoileus virginianus) based on age class (fawn = <1 yr,
yearling = 21 yr—<2 yr, adult = 22 yr), south central
Michigan, 2004-2006 ................................................................. 36

Table 1.4. Space use (95% ﬁxed-kemel home ranges) and cover
composition of female fawn (<1 yr old)white-tai1ed deer
during the agricultural growing (10 May-7 Oct; n = 28) and
non-growing (8 Oct—9 May; n = 21) seasons, southern
Michigan, 2004—2006 ................................................................. 37

Table 1.5. Space use (95% ﬁxed-kemel home ranges) and cover
composition of female yearling (2 1—<2 yr old) white-tailed
deer during the agricultural growing (10 May—7 Oct; n = 12)
and non-growing (8 Oct—9 May; n = 15), southern Michigan,
2004—2006 ................................................................................... 39

Table 1.6. Space use (95% ﬁxed-kemel home ranges) and cover
composition of female adult (>2 yr old)white—tai1ed deer
during the agricultural growing (10 May-7 Oct; n = 23) and
non-growing (8 Oct—9 May; I: = 25), southern Michigan,
2004—2006 ................................................................................... 41

CHAPTER 2: Survival and Space Use of Fawn White-tailed Deer in Southern Michigan

Table 2.1. Fate assessment of fawn white-tailed deer (n = 34) during
0—60 and 0—180 days post-capture, south central Michigan,
USA, 2004—2006 ......................................................................... 64

ix

CHAPTER 2 (cont’d)

Table 2.2. Space use (95% kernel home ranges) description of
radiomarked fawn white-tailed deer (n = 26) during

approximately 0—2 months of age, south central Michigan,
USA, May—July 2004—2006 ........................................................ 65

CHAPTER 3: A Multi-scale Approach to Describe Cover Selection of White-tailed Deer
Using Use-Availability Data

Table 3.1. Study designs (and deﬁnitions of use and availability in my
study) for collection of use-availability data as described by
Thomas and Taylor (1990, 2006) and Erickson et a1. (2001),
and selection hierarchy as deﬁned by Johnson (1980) ............... 97

Table 3.2. Cover selection of white-tailed deer (n = 20) based on 95%
conﬁdence limits of Euclidean distances (m) between cover
types and randomly generated points (1,000) within the study
area, under design 1 of Thomas and Taylor (1990), south
central Michigan, 2004—2006. Location estimates were
pooled by agricultural growing season (1,626 locations for
growing [10 May—7 Oct], 1,867 locations for non-growing
season [8 Oct—9 May]) ................................................................ 98

Table 3.3. Relative importance of cover types by adult female white-
tailed deer (n = 20) based on selection indices under multiple
study designs (see Thomas and Taylor [1990]), south central
Michigan, 2004—2006. Data were pooled by agricultural
growing season (growing [10 May—7 Oct], non-growing
[8 Oct—9 May]) ......................................................................... 100

Table 3.4. Cover selection of female adult white-tailed deer (n = 20)
under design 2 of Thomas and Taylor (1990), south central
Michigan, 2004—2006. Data were pooled by agricultural
growing season (150 d; 10 May—7 Oct; 23 growing, 25 non-
growing) within an agro-forest ecosystem ................................ 102

Table 3.5. Cover selection of female adult white-tailed deer (n = 20)
pooled by agricultural growing season (150 d; 10 May—7 Oct;
23 growing, 25 non-growing) in an agro-forest ecosystem in
south central Michigan, 2004—2006. Proportional use and
selection indices were unconditional on presence of cover
types within home ranges .......................................................... 106

CHAPTER 3 (cont'd)

Table 3.6.

Cover selection of female adult white-tailed deer (n = 20)

pooled by agricultural growing season (150 d; 10 May—7 Oct;

23 growing, 25 non-growing) in an agro-forest ecosystem in
south central Michigan, 2004—2006. Proportional use and
selection indices were conditional on presence of cover type
within home ranges ................................................................... 109

CHAPTER 4: Estimation and Implications of Space Use for White-tailed Deer

Table 4.1.

Table 4.2.

Management in Southern Michigan

Cover type composition (ha) of 95% ﬁxed kernel home

ranges and usable space of white-tailed deer (n = 20) pooled

by agricultural growing season (23 growing [10 May—7 Oct],

25 non-growing [8 Oct—9 May]) in south central Michigan,
2004-2006 ................................................................................. 144

Estimated cover conversions to maximize usable space (i.e.,

fully usable) and to decrease usable space (e.g., -50%) for
white-tailed deer on an 82,636-ha study area in south central
Michigan, USA, 2004—2006. Estimates were based on
agricultural growing season (growing = 10 May—7 Oct for

23 home ranges; non-growing = 8 Oct—9 May for 25 home
ranges; n = 20 deer) .................................................................. 148

APPENDIX B: Capture and Fate Descriptions of White-tailed Deer

Table 8.1.

Table B.2.

Table B.3.

Capture and fate information of radiomarked female white-
tailed deer captured using Clover traps, south central
Michigan, USA, winters 2004—2006 ........................................ 184

Capture and fate information of ear-tagged male white-tailed
deer captured using Clover traps, south central Michigan,
USA, winters 2004—2006 .......................................................... 186

Capture and fate information of neonate white-tailed deer
captured during May—June 2004—2006 and radiomarked with
break-away radio-collars, south central Michigan, USA .......... 187

xi

LIST OF FIGURES

CHAPTER 1: Age-speciﬁc Population Characteristics of White-tailed Deer in Southern

Figure 1.1.

Figure 1.2.

Michigan

Study area location and distribution of cover types, south
central Michigan, USA, 2004—2006. Study-area boundary

was deﬁned to include any township containing 21
radiomarked white-tailed deer (Odocoileus virginianus)
location estimate. (This dissertation image is presented in

color) ........................................................................................ 43

Pattern of home-range size estimates of white-tailed deer

based on mean size and 95% CLs (— = lower, + = upper) of

3 age classes (fawn = <1 year old, yearling = Z l—<2 year old,
adult = 2 2 years old) pooled through time during the growing
(~May—Oct) and non-growing (~Oct—May) agricultural seasons,
south central Michigan, USA, 2004—2006. Vertical dashed lines
approximate the regular ﬁrearms deer season (15-30 Nov) ..... 44

CHAPTER 3: A Multi-scale Approach to Describe Cover Selection of White-tailed Deer

Figure 3.1.

Figure 3.2.

Using Use-Availability Data

Frequency distribution of landscape-patch size (ha) of all

cover types on the study area, agricultural areas, conifers,

and upland deciduous forests, south central Michigan, USA,
2004—2006 .............................................................................. l 12

Linear relationship (y = 0.52x; r = 0.87; dashed line) of

amount of public land within home range (ha) as a function

of seasonal home-range size (ha) of adult female white-tailed
deer (n = 20), south central Michigan, USA, 2004—2006.
Seasonal home-range estimates were pooled and based on
agricultural growing season (10 May—7 Oct for growing

season, 8 Oct-9 May for non-growing season) ....................... 113

CHAPTER 4: Estimation and Implications of Space Use for White-tailed Deer

Figure 4.1.

Management in Southern Michigan

Frequency distribution of location estimates obtained
through telemetry for white-tailed deer by month, south
central Michigan, USA, 2004—2006 ........................................ 152

xii

CHAPTER 4 (cont’d)

Figure 4.2. Frequency distribution of location estimates obtained

through telemetry for white-tailed deer by time of day,

south central Michigan, USA, 2004—2006 .............................. 153
Figure 4.3. Usable space (ha) as a function of kernel home-range size

(ha) by agricultural growing season (10 May—7 Oct for
growing season, 8 Oct—9 May for non-growing season) for
white-tailed deer in an agro-forest ecosystem, south central

Michigan, USA, 2004—2006 ................................................... 154
APPENDIX A: Assessment of Precision between GPS- and VHF-derived Location
Estimates
Figure A.l. Frequency distribution of error ellipses of stationary VHF

radio-transmitters using triangulation to estimate 64
locations in an agro-forest landscape, south central
Michigan, USA, 2004—2006 ................................................... 177

Figure A.2. Distances between triangulated VHF location estimates and
GPS ﬁxes (n = 64) from a handheld unit of stationary VHF
radio-transmitters as a function of time to complete each
triangulation, south central Michigan, USA, 2004—2006 ....... 178

Figure A.3. Distribution of location estimates of stationary VHF radio-
transmitters using triangulation and point estimates using a
handheld GPS unit. Distribution reﬂects the relationship
between 64 GPS ﬁxes indexed with the coordinate origin
and VHF location estimates .................................................... 179

Figure A.4. Frequency distribution of error ellipses of animal-bome
GPS-VHF radio-transmitters using triangulation to estimate
20 locations of white-tailed deer in an agro-forest landscape,
south central Michigan, USA, 2004—2006 .............................. 180

Figure A.5. Distances between triangulated location estimates and GPS
ﬁxes (n = 20) of animal-bome GPS-VHF radio-transmitters
as a function of time to complete each triangulation, south
central Michigan, USA, 2004—2006. Dashed line represents
linear regression model (y = 41 .4x - 217.8; r = 0.62)
describing relationship ............................................................ 181

xiii

APPENDIX A (cont’d)

Figure A.6.

Distribution of location estimates using animal-home GPS-
VHF radio-transmitters, south central Michigan, 2004—2006.
Distribution reﬂects the relationship between 20 GPS ﬁxes
indexed with the coordinate origin and simultaneous VHF
location estimates of radiomarked white-tailed deer (n = 3)
using triangulation ................................................................... 182

xiv

ORGANIZATION OF DISSERTATION

This dissertation is organized into 5 chapters and 2 appendices that generally
follow formatting guidelines for manuscripts submitted to The Journal of Wildlife
Management; chapters 1—4 were formatted as complete manuscripts, so redundancy may
occur (e. g., study area description) among these chapters. Chapter 1 provides general
results on survival, cause-speciﬁc mortality, and space use of white-tailed deer. Chapter
2 provides similar information, but for white-tailed deer fawns for the time period from
capture to 2 months old. Chapter 3 describes a multi-scale approach to assess resource
selection of adult female white-tailed deer. Chapter 4 builds on Chapter 3 through space-
use assessments and their direct applicability to habitat management in an increasingly
urbanizing area. Finally, Chapter 5 provides an overview of conclusions based on
previous chapters. Appendix A provides an assessment of location-estimate precision
based on telemetry data, and Appendix B provides general identiﬁcation and capture

information on white-tailed deer from this study.

XV

INTRODUCTION

White-tailed deer (Odocoileus virginianus) may be described as a ubiquitous
species, inhabiting numerous vegetation types (e. g., Rolling Plains of Texas, coniferous-
hardwood forests of Michigan’s Upper Peninsula) and distributed throughout most of
North and Central America (Baker 1984). It would seem logical that population
characteristics (e.g., demographics, resource selection, space use) could shift substantially
under differing landscape conditions. These characteristics must be scientiﬁcally
evaluated for effective management of deer populations and their associated habitat.

Movements and demographics (e.g., annual survival estimates) of deer often vary
throughout their distribution, and consequently should be measured at an appropriate
scale (e.g., landscape) or scales for effective management. For example, research in the
United States has shown annual home-range-size estimates ranged from 59 to 740 ha for
nonmigratory deer (Marchinton and Hirth 1984ztable 20; Pusateri 2003). Generally,
home ranges of deer may increase in size from south to north due to climate
(Severinghaus and Cheatum 1956), and home ranges are usually larger in more open
vegetation (i.e., sparse vegetation; Marchinton and Hirth 1984).

Higher population densities in mammals may also result in smaller home ranges
(Sanderson 1966), presumably as a response to maintain spacing among individuals for
increasingly limited resources. In addition, the spatial distribution of an animal is
inﬂuenced to some degree by the distribution of forage and predators within the animal’s
environment (Stephens and Krebs 1986:161—168). The multiple inﬂuences on size of
home ranges of deer suggest that research and management considerations only at the

landscape scale might not be as beneﬁcial as management decisions based on studies

considering >1 spatial scale. For example, habitat manipulation is one tool for wildlife
management and as previously discussed, deer can use a wide variety of habitat
conditions, so understanding the types and quantities of resources selected by deer at
multiple scales should help us understand potential population responses to changes in
habitat conditions.

Research at the state level has supported my assertions about space-use dynamics.
Home ranges in Michigan’s Upper Peninsula have been estimated at 730—1,859 ha
(winter) and 1,255—3,037 ha (summer) for migratory deer (95% adaptive kernel; Van
Deelen 1995), while in the northern Lower Peninsula, home ranges of migratory deer
varied 202-354 ha (winter) and 329—337 ha (summer; 95% harmonic mean; Sitar 1996).
Nonmigratory deer in the southwestern Lower Peninsula of Michigan had an estimated
annual home range size of 50—740 ha (95% ﬁxed kernel; Pusateri 2003). Resource
selection among these study areas must also differ to some degree due to different
ecosystem conditions and weather patterns.

Demographics are also important to estimate for wildlife management. Survival
estimates are a key parameter when building population models to be used for wildlife
management (White and Lubow 2002), as these estimates can describe the growth
potential of a population. Studies in Michigan have included survival estimates of 22%
(males) and 77% (females; Van Deelen 1995), a range of 53—71% (pooled by sex; Sitar
1996), and 40%—58% (pooled by sex; Pusateri 2003) annual survival for adults.
Although some variation may exist among studies due to survival-estimation method
(Mayﬁeld for the former 2, Kaplan-Meier for the latter), generally, survival would seem

to decrease with increasing latitude due in part to stochastic winter-weather events.

Potentially higher estimated neonate survival might be associated with areas of
increasing land-use activities (e.g., agriculture, urban development), although survival
will likely reach an asymptote or decline at some point of land-use activity. Within the
United States, neonate survival (180 days post-capture) has been estimated at 47% (early
urban development) and 96% (post-urban development, Florida Key neonates; Peterson
et a1. 2004), 46% and 59% (forested and agricultural areas, respectively, in Pennsylvania;
Vreeland et a1. 2004), 73% (agricultural, Iowa; Huegel et al. 1985), and 40% (forests and
forest-agriculture borders, New Brunswick; Ballard et a1. 1999). In Oklahoma, 90% of
neonates died within 3 months of capture (prairie and woodland; Bartush and Lewis
1981). Fawn survival from capture as neonates to 220 days post-capture has been
estimated at >76% within Michigan (southwestern Lower Peninsula; Pusateri Burroughs
et a1. 2006). In a suburban Chicago, Illinois forest preserve, fawn mortality was
estimated to be as high as 95% (Piccolo 2002). Different survival estimates are likely
complicated by differing habitat conditions used by deer throughout their distribution.
This suggests the importance of understanding the relationship between population
demographics and habitat when making management decisions.

OBJECTIVES

My goal was to make my ﬁndings applicable and usable for deer management
within south central Michigan and other areas experiencing increasing land-use activities,
such as urbanization. Speciﬁcally, my objectives were to

1. Estimate age-speciﬁc survival, cause-speciﬁc mortality, and space use of female

white-tailed deer,

2. Estimate survival, cause-Speciﬁc mortality, and Space use of young fawns from

time of capture (<2 weeks old) to 2 and 6 months post-capture,

3. Describe resource selection of adult female deer across multiple spatial scales,

4. Quantify and compare space use of adult female deer using two analytical

methods (ﬁxed-kemel and usable space), and

5. Estimate cover conversions, based on the concept of usable space, which will

theoretically affect deer abundance on the study area.

My objectives were met by capturing and radiomarking multiple age classes of
primarily female white-tailed deer within Jackson and Washtenaw counties during 2004—
2006. Deer were monitored using telemetry techniques during this period. Michigan
State University’s All-University Committee on Animal Use and Care approved all

capturing and handling procedures for my study (Application No. 01/04-006-00).

LITERATURE CITED

Baker, R. H. 1984. Origin, classiﬁcation and distribution. Pages 1—18 in L. K. Halls,
editor. White-tailed deer: ecology and management. Stackpole, Harrisburg,
Pennsylvania, USA.

Ballard, W. B., H. A. Whitclaw, S. J. Young, R. A. Jenkins and G. J. Forbes. 1999.
Predation and survival of white-tailed deer fawns in northcentral New Brunswick.
Journal of Wildlife Management 63:574—579.

Bartush, W. S. and J. C. Lewis. 1981. Mortality of white-tailed deer fawns in the
Wichita Mountains. Proceedings of the Oklahoma Academy of Science 61:23—
27.

Huegel, C. N., R. B. Dahlgren and H. L. Gladfelter. 1985. Mortality of white-tailed deer
fawns in south-central Iowa. Journal of Wildlife Management 49:377-380.

Marchinton, R. L. and D. H. Hirth. 1984. Behavior. Pages 129—168 in L. K. Halls,
editor. White—tailed deer: ecology and management. Stackpole, Harrisburg,
Pennsylvania, USA.

Peterson, M. N., R. R. Lopez, P. A. Frank, B. A. Porter, and N. J. Silvy. 2004. Key deer
fawn response to urbanization: is sustainable development possible? Wildlife
Society Bulletin 32:493—499.

Piccolo, B. P. 2002. Behavior and mortality of white-tailed deer neonates in suburban
Chicago, Illinois. Thesis, University of Illinois at Urbana—Champaign, Urbana,
Illinois, USA.

Pusateri, J. S. 2003. White-tailed deer population characteristics and landscape use
patterns in southwestern Lower Michigan. Thesis, Michigan State University,
East Lansing, Michigan, USA.

Pusateri Burroughs, J ., H. Campa, III, S. R. Winterstein, B. A. Rudolph and W. E.
Moritz. 2006. Cause-Speciﬁc mortality and survival of white-tailed deer fawns in
southwestern Lower Michigan. Journal of Wildlife Management 70:743—75 1.

Sanderson, G. C. 1966. The study of mammal movements-a review. Journal of Wildlife
Management 30:215—235.

Severinghaus, C. W., and E. L. Cheatum. 1956. Life and times of the white-tailed deer.
Pages 57—186 in W. P. Taylor, editor. The deer of North America. Stackpole,
Harrisburg, Pennsylvania, USA.

Sitar, K. L. 1996. Seasonal movements, habitat use patterns, and population dynamics of
white-tailed deer (Odocoileus virginianus) in an agricultural region of northern

Lower Michigan. Thesis, Michigan State University, East Lansing, Michigan,
USA.

Stephens, D. W., and J. R. Krebs. 1986. Foraging theory. Princeton University Press,
Princeton, New Jersey, USA.

Van Deelen, T. R. 1995. Seasonal migrations and mortality of white-tailed deer in
Michigan’s Upper Peninsula. Dissertation, Michigan State University, East
Lansing, Michigan, USA.

Vreeland, J. K., D. R. Diefenbach, and B. D. Wallingford. 2004. Survival rates,
mortality causes, and habitats of Pennsylvania white-tailed deer fawns. Wildlife
Society Bulletin 32:542—553.

White, G. C., and B. C. Lubow. 2002. Fitting population models to multiple sources of
observed data. Journal of Wildlife Management 66:300—309.

CHAPTER I

Age-speciﬁc Population Characteristics of White-tailed Deer in Southern Michigan

Age-speciﬁc Population Characteristics of White-tailed Deer in Southern Michigan

INTRODUCTION

Regulated harvest and habitat management for game species are 2 primary
management methods used by state wildlife agencies. Consequently, accurate
descriptions of population characteristics of game species such as estimates of age-
speciﬁc demographics and space use are imperative during the decision-making process.
Many factors inﬂuence survival and cause-Speciﬁc mortality of cervids, including annual
weather patterns (e.g., DelGiudice et al. 2006), predator-prey dynamics (e.g., Labisky and
Boulay 1998), and the effects of hunter-harvest (e.g., Bender et al. 2000). A discussion
of the effects of harvest on game species populations often involves additive versus
compensatory effects of hunting on mortality, especially as the relative abundance of a
species varies. Survival and other age-speciﬁc demographic estimates are important
parameters for managers using simulation modeling to predict species abundance.

Knowledge of space use is relevant for managers to understand the importance of
landscape features to a species. Spatial or temporal changes in cover may affect species
abundance and age and sex structure, and potential future land-use changes (e.g., land
ownership) may impact a state agency’s ability to manage natural resources. Knowledge
of relationships between space use and demographics is also important, as population
characteristics often vary spatially and temporally across the landscape. For example,
lower survival may result from a decrease in suitable cover through space and time. The
complexity of ecological systems makes area-speciﬁc and species-speciﬁc estimates of

these characteristics highly desirable in meeting the challenges of wildlife management.

Further, “home range size likewise varies by sex and age of the individual, habitat, and
season” (Demarais et al. 2000:610), so speciﬁc information about space use may be the
most helpful for guiding management objectives and decisions.

The increasing size of white-tailed deer (Odocoileus virginianus) and human
populations coupled with increasing land-use changes (e.g., urbanization) further add to
the complexity of deer management across much of the United States (Demarais et al.
2000). Because these attributes can vary greatly across a landscape, management
agencies often develop species-speciﬁc management units within a state and apply their
management objectives accordingly. Although management-unit boundaries are often
based on non-ecological components of the landscape, such as roads or county
boundaries, these features are often more transparent to consumptive users of wildlife and
other stakeholder groups. Ecological boundaries could, for example, be based on areas
containing source and sink populations, areas containing low-quality habitat, or
incorporating ecological barriers that might restrict animal movements. Managers often
assume or have evidence that the species of interest has population characteristics that
differ across the landscape, and consequently, the management of this species often
differs by management unit. Management-unit boundaries, even if based on non-
ecological components, are often adapted by state agencies over time to more accurately
reﬂect wildlife population dynamics and land-use patterns.

Management of high deer populations through regulated hunting is probably the
most cost-effective strategy (Demarais et al. 2000). Harvest objectives are normally
implemented at the management-unit level and may be age- and sex-speciﬁc for deer.

For example, harvest objectives for antlerless deer (i.e., males <1 yr old and females) are

often adapted to achieve the agency’s population goals, such as when the deer population
is determined to be higher than desired. Knowledge of population characteristics of deer,
especially female deer, under these circumstances is important as female deer are often
the sex class of primary interest for managers desiring relatively large reductions in
population densities through hunter-harvest (Carpenter 2000).

My objectives were to describe age-speciﬁc survival, cause-speciﬁc mortality,
and space use of female white—tailed deer in south central Michigan, and to compare my
ﬁndings with past research on deer in Michigan. 1 used female white-tailed deer for
analyses, as south central Michigan currently has a relatively high deer density (~27/km2
during fall 2005; Michigan Department of Natural Resources [MDNR] 2005), and the
southern Michigan deer population is generally above the desired population goals of the
MDNR (Clute 2006).

STUDY AREA

My study was conducted in eastern Jackson (Grass Lake, Henrietta, and Waterloo
townships), western Washtenaw (Dexter, Lima, Lyndon, and Sylvan townships), and
southwestern Livingston (Unadilla township) counties in south central Lower Michigan
(Fig. 1.1). The study area (82,636 ha) included publicly owned lands, including the
MDNR Waterloo (8,410 ha; 10.2% of study area) and Pinckney (4,276 ha; 5.2%)
recreation areas, and privately owned lands. South central Michigan has been
characterized by a relatively high deer density, increasing land-use activities (e.g.,
urbanization), and little scientiﬁc study of deer under these conditions. Although the
study area was primarily rural (98% of total land area), the human population increased

16% and housing units increased 22% between 1990 and 2000 (US. Census Bureau

10

2003). Much of the landscape throughout southern Michigan (and in other areas
throughout the Midwest) is expected to experience increasing land-use activities similar
to the study area (Madill and Rustem 2001 ).

The physiographic regions of this area are Hillsdale-Lapeer Hilly Upland, South
Central Rolling Plain, and Southeastern Rolling Plain (Sommers 1977:24) with alﬁsols as
the major soil order (Sommers 1977:36). Surface formations in the study area are the
result of glaciation and include all 4 types present within Michigan (moraine, till plain.
outwash plain, and lacustrine plain; Sommers 1977232). Elevation of the study area
ranged approximately 180—300 In and consisted of relatively limited relief (Sommers
1977:26, 33). The study area received about 81 cm of precipitation annually during
1971-2000 (based on conditions in Chelsea, Michigan; Midwestern Regional Climate
Center, Champaign, Illinois, USA), and had a ISO-day growing season (i.e., the average
annual accumulation of daily mean temperatures >5.6 °C), generally occurring from 10
May to 7 October (Sommers 1977246, 49). During 1971—2000, average annual snowfall
was 99.3 cm and mean monthly temperatures ranged from —5.4 °C (Jan) to 21.8 °C (Jul)
in Jackson County (Midwestern Regional Climate Center, Champaign, Illinois, USA).
Total annual snowfall during my study was highly variable (2004 = 90.9 cm, 2005 =
149.9 cm, 2006 = 40.0 cm; conditions in Chelsea, Michigan; Midwestern Regional
Climate Center, Champaign, Illinois, USA).

Before European settlement, southern Michigan forests consisted of oak (Quercus
spp.) and hickory (Carjya spp.) in well-drained soils and beech (Fagus grandifolia), elm
(Ulmus spp.), maple (Acer spp.), and basswood (T ilia americana) in poorly drained soils

(Sommers 1977:17). Much of the study area is well suited for agriculture (Sommers

ll

1977:38). Most crop agriculture in Jackson and Washtenaw counties (total area =
366,483 ha) is in corn (37,840 ha) and soybean (34,200 ha) production, followed by hay
(17,200 ha), winter wheat (7,970 ha), and oats (930 ha; Michigan Department of
Agriculture 2002).

I generalized land-use, land-cover data (Michigan Center for Geographic
Information 2001) using ArcView GlS v3.2 software (Environmental System Research
Institute, Redlands, California, USA) and Spatial Analyst extension to deﬁne 13 cover
types within the study area (Fig. 1.1): agriculture (non-vegetated farmland, row crops,
forage crops; 52.3% of study area); conifer (pines [Pinus spp.], other upland conifers;
1.5%); herbaceous openland (herbaceous vegetation with <25% woody cover; 2.9%);
lowland deciduous forest (>60% composed of deciduous tree cover; 8.0%); lowland
shrub (with >60% non-water cover; 9.9%); mixed wetland (ﬂoating aquatic vegetation,
emergent wetland, mixed non-forest wetland; 3.1%); northern hardwood (>60% canopy
cover of maple, beech, ash [Fraxinus spp.], cherry [Prunus spp.], birch [Betula spp.];
2.3%); oak association (>60% canopy cover of oak; 1.6%); upland deciduous forest
(>60% canopy cover of upland deciduous trees; 11.6%); upland shrub (>25% woody
cover; <0.1%); urban (low and high intensity, roads, parks, golf courses; 2.8%); water
(surface, ﬂowing; 3.9%); and other (aspen [Populus spp.] association, orchards, bare
ground; 0.1%). Patch size of cover types ranged from <1 ha to >1 1,000 ha (e.g., an
agricultural matrix) and had a mean size of 29.2 ha. Consequently, generalization of
spatial data may have excluded certain ﬁne-scale landscape characteristics (e.g.,

hedgerows, roads) in some instances.

12

METHODS
Capturing Deer

Winter.—Technicians and 1 trapped deer during winter (Dec—Mar) 2004—2006,
using single-door collapsible live traps (Clover 1954). Traps were placed near areas of
deer activity, baited with kernel corn, and checked twice/day to minimize stress and
injury to deer. We restrained captured deer using the collapsed trap and the body weight
of 1 person (Sparrowe and Springer 1970). All trapped deer were blindfolded to reduce
stress and ﬁtted with metal ear tags bearing a unique identiﬁcation number (Style 681;
National Band and Tag, Newport, Kentucky, USA). Individuals were aged as fawn (<1
year old), yearling (21—<2 years old), or adult (22 years old) through general
morphometric differences (e.g., shape and size of head, body size) and dental
characteristics (Severinghaus 1949); ages of necropsied individuals later conﬁrmed
accuracy of ﬁeld observations for deer captured during winter.

Only female deer were ﬁtted with mortality-sensing collar-style radio-
transmitters, either with VHF (Model 500; Telonics, Inc., Mesa, Arizona, USA) or VHF-
GPS (Model G2000; Advanced Telemetry Systems, Isanti, Minnesota, USA) capabilities.
I did not radiomark male deer because I expected low capture success for adult males
when using Clover traps (T. Hiller, unpublished data). Additionally, winter-captured
male deer were not radiomarked because of potential problems associated with their
physiological changes (i.e., neck-swelling) during the breeding season and ﬁtting a radio-
transmitter to deer for these changes. Radio-transmitters had a unique frequency within
the ISO-MHz range, a mass of 270 g (VHF) or 1,100 g (VHF-GPS), and an expected

minimum battery life of 36 months (VHF) or 12 months (VHF-GPS).

l3

Spring—The capture season for neonatal fawns was from mid-May to mid-June
2004—2006. I expected capture success to peak around 1 June and instances of fawns
ﬂushing to increase greatly after mid-June (Pusateri Burroughs et al. 2006), thus reducing
my ability for successful capture; fawns >2 weeks old generally start ﬂushing when
approached by humans (Carroll and Brown 1977). Technicians and I captured neonates
either by hand or in a ﬁsh-landing net (0.5-m-diameter net, 2-m-long extendable handle).
A group of 2—6 people systematically searched potential fawning areas (e.g., areas
transitioning from forest to wetland or grassy ﬁeld) for neonates (Lund 1975, Ballard et
al. 1999). Isolated adult females occasionally provided behavioral signs of a nearby fawn
(e.g., a doe looking toward an area several times/min; Downing and McGinnes 1969).

I weighed, sexed, ear-tagged (Style 681; National Band and Tag, Newport,
Kentucky, USA) and radiomarked each captured neonate. Captured neonates also were
aged using hoof-growth measurements (Haugen and Speake 1958). The mortality-
sensing radio-transmitters (Model M4210; Advanced Telemetry Systems, Isanti,
Minnesota, USA) had an expandable collar to allow for growth and were designed to
drop off after 9—12 months for retrieval (see Diefenbach et al. 2003). Transmitters had a
unique frequency within the 151-MHz range, a mass of 60 g, and an expected minimum
battery life of 12 months. The precise—event transmitter option provided a time-of-
mortality estimate within 30 min for a maximum of ~5 days post-mortality (i.e., 5 days of
no transmitter movement).

Neonates were classiﬁed as fawns at the time of capture, but were reclassiﬁed
when they reached 1 June of the ﬁrst year (yearling) or second year (adult) following

capture. Similarly, winter-captured deer were reclassiﬁed to the next age class each time

14

they reached 1 June following capture. Michigan State University’s All-University
Committee on Animal Use and Care approved all capturing and handling procedures for
my study (Application No. 01/04-006-00).

Survival and Cause-specific Mortality

All deer were monitored 2-5 times/week to estimate survival and assess cause-
speciﬁc mortality, but spring-captured fawns were monitored daily for the ﬁrst 30 days
following capture to potentially increase the accuracy of assessments. I used the
Mayﬁeld method (Mayﬁeld 1961, 1975), modiﬁed by Bunck and Pollock (1993) for
censored individuals, to estimate survival of deer. Daily survival was estimated for 3
time periods for fawns (0—6 months old [Chapter 2], 6—12 months old, annual), and
annually for yearlings and for adults. For analyses, 1 excluded any individual that died
within a 7-day acclimation period following capture. Individuals were censored if I
believed they were alive at the time of transmitter recovery (e.g., the break-away collar
dropped), at the conclusion of ﬁeld data collection, or when they moved into the next age
class.

My assessment of cause-speciﬁc mortality of recovered carcasses was based on
my ﬁeld observations and the necropsies performed by a wildlife pathologist (MDNR
Wildlife Disease Laboratory, East Lansing, Michigan, USA). I classiﬁed mortalities of
individuals into 5 categories: canid (coyote [Cam's latrans], red fox [Vulpes vulpes], gray
fox [Urocyon cinereoargenteus], or domestic dog [C. lupus familiarisD predation, trauma
or malnutrition (e.g., abandonment), vehicle collision, hunter-harvest, and unknown. 1f 1

did not have enough post-mortality evidence to ascribe cause with reasonable certainty

15

(e.g., total consumption of the carcass), then I considered the cause of mortality to be
unknown.
Space Use

Technicians and I estimated locations of deer 2—5 times/week using triangulation
from telemetry signals (White and Garrott 1990:79—1 12) or from visual observations of
known individuals. To increase the accuracy of space-use assessment, we located deer in
a systematic manner during varying time schedules on a diel basis (Beyer and Hauﬂer
1992), with 2 1 nocturnal location/deer/week, except during capture periods. Bearings
were estimated using a 3-element folding Yagi antenna (Advanced Telemetry Systems,
Incorporated, Isanti, Minnesota, USA), portable radio receiver (Model R-1000,
Communications Specialists, Incorporated, Orange, California, USA), and mirror-
sighting compass. A global positioning system (GPS) handheld unit (Model GPS 1V;
GARMIN International, Incorporated, Olathe, Kansas, USA) was used to approximate
the locations from which signals were received.

I used the program LOCATE 111 (Pacer, Truro, Nova Scotia, Canada) to estimate
locations of deer using triangulated data based on the maximum likelihood estimator
(Lenth 1981a,b), as recommended by White and Garrott (1990) and Nams and Boutin
(1991). During analysis, 1 also used LOCATE III to estimate bearing standard deviations
and error ellipses for location estimates. 1 based telemetry-error assessments on the
relationship between mean error-ellipse size and mean landscape-patch size to determine
if I had appropriate sample sizes (see Nams 1989).

Space use is often estimated using 1 of the many home-range estimators (e.g.,

minimum convex polygon. kernel methods, harmonic mean), but each may provide

16

different results using the same data. 1 used the ﬁxed-kernel method, which seems to be
the best estimator currently in use based on the criteria outlined by Kemohan et al.
(2001), and least-squares cross-validation to determine the smoothing parameter (Worton
1995, Seaman et a1. 1999). My description of space use included only individuals with

2 30 locations (Seaman et al. 1999).

Space-use data were pooled by age class and season based on agricultural crop
production (i.e., the growing season on study area [10 May—7 Oct; 150 d], and the non-
growing season [8 Oct—9 May; 215 d]). I assumed that all location estimates that were
classiﬁed in the cover type water were inaccurate and I relocated each to the nearest
alternative cover type. Only location data from VHF signals were used to avoid potential
differences associated with location precision between VHF - and GPS-derived location
estimates (i.e., I used only VHF-derived data from VHF -GPS transmitters).

I used Animal Movement extension in ArcView GIS v3.2 software
(Environmental System Research Institute, Redlands, California, USA) to estimate space
use. For all statistical analyses, 1 used SYSTAT v11 (Systat Software, Inc., San Jose,
California, USA) and ProStat v4.02 (Poly Software International, Inc., Pearl River, New
York, USA). Unless noted otherwise, I used 95% conﬁdence limits (CLs; LCL = lower,
UCL = upper) during my statistical analyses. The use of conﬁdence limits is
advantageous in that an estimate of effect size and a measure of uncertainty are provided

(Johnson 1999).

17

RESULTS
Capture and Estimation of Locations

I captured and radiomarked 42 female deer during winter 2004—2006 (Table 1.1)
and 34 neonates during spring 2004—2006 (Table 1.2). Two neonates and 2 winter-
captured deer died <7 days post-capture and were excluded from analyses. 1 pooled
spring-captured animals by sex (56% male) and assumed behavioral differences were
minimal or nonexistent between male and female fawns (see Ozoga and Verme 1986)
during the agricultural growing season, after which males were excluded from analyses.
I pooled data by age class and season but could not consider year effects due to small
sample sizes.

Survival and Cause-speciﬁc Mortality

Survival was lower for fawns 0—6 months old (0.67 for 6-month survival; 95% CL
= 0.51—0.84; n = 32; Chapter 2) than for fawns 6—12 months old (0.90; 95% CL = 0.78—
1.00; n = 23); annual fawn survival was estimated to be 0.51 (95% CL = 0.37—0.66; n =
48). Annual survival estimates for yearlings and adults were 0.94 (95% CL = 0.85—1.00;
n = 28) and 0.56 (95% CL = 0.38—0.75; n = 28), respectively.

The primary annual source of mortality for fawns was vehicle collisions (64% of
mortalities), and fawns were the only age class depredated by canids (Table 1.3). Of the
12 yearlings radiomarked, only 1 died, with the cause assessed as malnutrition related to
trauma. Eleven of 13 (85%) adults that died were harvested by hunters. Mortalities

related to capture or radio-collar trauma seemed least prevalent with adults (Table 1.3).

18

Space Use

T elemetry-error assessment.—l pooled all location estimates (8,714) from deer of
all ages (n = 66) from my study to estimate overall telemetry error. The mean error-
ellipse size was 10.2 ha. 1 considered telemetry error acceptable, and thus, location

estimates of acceptable precision, given the landscape characteristics of the study area

(e.g., ; = 29.2 ha/patch) and the number of location estimates (Tc = 132/deer). This
conclusion was based on the relationship between telemetry error and patch size to
determine appropriate sample Sizes as described by Nams (1989). Further, given the
patchiness of the study area, I assumed that characteristics of each cover type (e.g.,
differences in foliage densities) did not inﬂuence our telemetry accuracy, resulting in
acceptance of location estimates near cover-type boundaries as accurate even if error
ellipses covered 21 cover type (White and Garrott 1990:200).

Within age class—Seasonal home-range differences existed for fawns, yearlings,
and adults (Tables 1.4, 1.5, and 1.6, respectively). Within the fawn and adult age classes,
the growing season mean home-range size was approximately half that of the non-
growing season mean home-range size (Tables 1.4 and 1.6). Yearlings had similar mean
home-range sizes for the growing (201.8 ha) and non-growing (156.88 ha) seasons (Table
1.5), but variation was large (SE = 91.1 and 28.2, respectively). Ninety-ﬁve percent CLS
of seasonal home ranges overlapped within each age class, but only slightly for adults
(Table 1.6). The mean amount (ha) of upland deciduous forest cover within kernel home
ranges quadrupled for fawns from the growing to the non-growing season (Table 1.4).
Yearlings selected 2 cover types in different proportions (Table 1.5): agricultural areas

were selected at twice the proportion during the growing season (0.26) than during the

19

non-growing season (0.12), while upland deciduous forests were selected in a higher
proportion during the non-growing season (0.33 versus 0.19). Other cover types were
selected in similar proportions within each age class.

Among age classes—Fawn home ranges during the growing season (Table 1.4)
were smaller than both yearling and adult non-growing season estimates (Tables 1.5 and
1.6). The yearling non-growing season estimate was only slightly larger than the adult
growing-season estimate based on 95% CLs, indicating a minor transition in home-range
size through time. Many differences in the amounts (ha) of each cover type in home
ranges existed across age classes. Based on CLs, fawn home ranges during either season
had compositional differences in herbaceous openlands, lowland shrubs, northern
hardwoods, and upland deciduous forests with the composition of 21 seasonal home
range of other age classes. Two other age-speciﬁc differences in home-range cover
composition existed: yearlings during the non-growing season used more northern
hardwoods (8.4 ha) and lowland deciduous forests (25.9 ha) than adults (2.9 and 4.7 ha,
respectively) during the growing season.

Home-range trends—Mean home-range sizes showed an oscillatory pattern
based on age and season, with home-range size greatest for yearlings during the growing
season (Fig. 1.2). Regardless of age, deer movements generally seemed to increase
during fall and peaked soon after hunters expended the most effort for harvest (i.e.,
following regular ﬁrearms season: ~30 Nov) in Lower Michigan. I assumed home-range
size by season was consistent for adults regardless of actual age, and extrapolated the

pattern to be consistent for all deer >2 years old.

20

DISCUSSION

Without research, managers may not have suitable information describing how
population characteristics vary substantially across a landscape. If demographic
information is not available on an area-speciﬁc and age-speciﬁc basis, then estimates
based on the available literature must be used for management. For example, the sex-
age-kill model (SAK; Creed et a1. 1984, Skalski et al. 2005) is often used to estimate deer
abundance within a management unit by using variables associated with demographics
(e.g., annual survival, harvest mortality), various age and sex ratios, harvest data, and
other population characteristics. Of course, the estimates of abundance from SAK are
only as accurate as the input estimates. If population characteristics of deer differ among
management units (or otherwise across landscapes) and through time, then with accurate
data this model should theoretically calculate these differences in abundance and
management objectives (e.g., harvest quotas) may then be adapted for the following year.

Demographics may vary according to many variables (e.g., habitat quality and
seasonal weather patterns [Demarais et al. 2000], population trends [McCullough 2001]),
so management objectives and practices should be adapted accordingly. Although
extrapolation of data is often necessary, as species-speciﬁc research cannot (and probably
should not) be conducted at a ﬁne spatial scale across an entire state due to ﬁnancial
constraints, certainly comparisons among within-state research projects should provide
insight into patterns of wildlife population dynamics, especially in relation to land-use
activities. Additionally, if the spatial scale of study is too ﬁne, the study design may not

reﬂect the natural history of the species under study (e.g., as the deﬁned spatial extent of

21

availability is reduced, resource selection may become more difﬁcult to detect [McClean
et al. 1998]).
Michigan Deer Research

The white-tailed deer is the most popular big-game species in North America
(Smith and Coggin 1984, US. Fish and Wildlife Service 2001). Its status in Michigan is
no different, with about 743,000 hunters harvesting nearly 500,000 deer in 2003 (Frawley
2004). Not surprisingly, Michigan has a strong tradition of deer research that has been
conducted throughout the state, especially during the past decade. Demographics and
space use of free-ranging deer are often most easily estimated using radiomarked
individuals and numerous studies have been conducted in Michigan using telemetry
techniques. Research in the Upper Peninsula has included seasonal migrations and
mortality of deer (Van Deelen et al. 1997, Van Deelen et al. 1998) as well as their habitat
use and browsing effects (Mackey 1996). In the Northern Lower Peninsula, Sitar (1996)
and Sitar et al. (1998) examined the seasonal movements, habitat use, and population
characteristics of deer, while Garner (2001) and Muzo (2003) examined the movements
and behavior of a bovine tuberculosis-infected deer population.

The most recently completed ﬁeld study was a description of the population
characteristics and landscape-use patterns of deer on private lands in southwestern Lower
Michigan (Pusateri 2003, Pusateri Burroughs et a1. 2006). Pusateri (2003) and Pusateri
Burroughs et al. (2006) research on nonmigratory deer in southwestern Lower Michigan
are the most comparable to my results based on deer behavior and landscape similarity;
e.g., deer in the Upper and northern Lower Peninsulas frequently migrate between

summer and winter home ranges for sufﬁcient winter thermal cover (Verme 1973, Van

22

Deelen et al. 1998), while thermal cover may not be necessary for deer in southern Lower
Michigan (Torgersen and Porath 1984). Migratory behavior will affect space-use
assessments and often estimates of demographics; differences in cover types, agricultural
growing season, and landscape characteristics (e.g., patch size and shape) may also affect
population characteristics.
Age-speciﬁc Characteristics

Fawns.—Until now, Pusateri Burroughs et a1. (2006) has been the only study that
described survival, cause-speciﬁc mortality, or space use of young free-ranging white-
tailed deer fawns in Michigan. With data pooled over 2 years, they estimated annual
fawn survival on privately owned lands at 0.75 (95% CL = 0.59—0.91; Kaplan-Meier
method) in southwestern Lower Michigan, much higher than my estimate of 0.51 (95%
CL = 0.37—0.66). Cause-speciﬁc mortality sources for fawns <1 year were also different
between studies, with higher mortality due to hunting (29%) and lower mortality due to

vehicle collisions (29%) compared to my study (18% and 64%, respectively). Potential
reasons for these differences may include higher densities of humans (it = 1.39/ha versus

E = LOO/ha; US. Census Bureau 2006) and deer (~27/km2, MDNR 2005; ~19/ kmz,
Pusateri 2003) on my study area, as well as differences in land—ownership and land-use
patterns (e.g., hunting opportunities might increase with an increase in amount of public
land on an area). Deer habitat quality and the amount of trafﬁc on roadways, both
positively related to deer-vehicle collisions in southern Lower Michigan (Sudharsan
2005), and likely elsewhere, also probably contributed to differences in cause-speciﬁc
mortality between study areas. Other mortality sources contributed little to overall

mortality for deer <1 year old during both studies.

23

Although Pusateri Burroughs et al. (2006) estimated the mean home-range size of
27-week-old fawns (62.7 ha; ~May—Dec), their estimate was similar to my growing
season (~May—Oct) estimate of 60.2 ha. Cover composition within kernel home ranges
varied between studies, as my estimates contained more conifers (22% versus 10%) and
less agriculture (32% versus 46%) and deciduous forests (23% versus 40%). Although
not directly comparable because of differing time intervals, their mean annual home-
range size estimate (75.4 ha) was only 65% of my mean non-growing Season home-range
estimate (1 16.3 ha) for fawns. Differences in cover availability likely contributed to
some cover selection differences between study areas.

Yearlings.—Survival estimates of yearling deer within the Lower Peninsula of
Michigan seem limited. Pooled by sex, annual survival of yearlings for a 2-year study
(Sitar 1996; northern Lower Michigan) ranged from 0.29 (95% CL = 0.10—0.48) to 0.36
(95% CL = 0.18—0.54), which were both much lower than my pooled annual estimate
(95% CL = 0.85—1.00). Sitar (1996) stated that yearling survival may have been
somewhat overestimated due to mild winter conditions during her study, which suggested
that long-term differences between our estimates may be even more pronounced. Long-
term (1971—2000) winter conditions (e.g., mean monthly temperatures, mean annual
snowfall) on her study area (Jan = —7.9 °C, annual snowfall = 142.2 cm; Alpena County;
Midwestern Regional Climate Center, Champaign, Illinois, USA) were generally more
severe than on my study area (Jan = —5.4 °C, annual snowfall = 99.3 cm; Jackson County;
Midwestern Regional Climate Center, Champaign, Illinois, USA). Age- and sex-speciﬁc
survival estimates of migratory deer in Michigan’s Upper Peninsula showed that yearling

females had a high annual survival rate (0.89) in comparison to other age-sex classes

24

(Van Deelen 1995); he also found that non-hunting mortality did not differ between
sexes, but annual hunting mortality was much higher for males. To my knowledge,
cause-speciﬁc mortality and space use have not been assessed speciﬁcally for yearling
white-tailed deer in southern Lower Michigan.

Adults.—Deer >6 months old (i.e., yearlings and adults) in southwestern Lower
Michigan had annual survival estimates (Kaplan-Meier method) that ranged from 0.40
(95% CL =5 0.20—0.60) to 0.77 (95% CL : 0.61—0.93) (data pooled by sex; Pusateri
2003). Using the Mayﬁeld method, Sitar (1996) estimated annual survival of adult deer
in the northern Lower Peninsula of Michigan to be 0.53 (95% CL = 0.37—0.69) and 0.71
(95% CL = 0.50—0.92), depending on year. My survival estimate for adult females (0.56;
95% CL = 0.38—0.75) was generally consistent with these estimates.

Cause-speciﬁc mortality for female deer >6 months old (n = 48; 18 died) included
hunter-harvest (61% of mortalities), vehicle collisions (28%), and trauma-related injuries
(11%) in southwestern Lower Michigan (Pusateri 2003). In northern Lower Michigan,
the 4 most prominent known mortality sources for male and female deer included hunter-
harvest (37% of mortalities), natural causes (e.g., predation, drowning; 24%), illegal
harvest (12%), and vehicle collisions (10%; Sitar 1996). The only signiﬁcant cause of
mortality on my study area for adult female deer was hunter-harvest (85% of mortalities).
Based on radiomarked individuals, adults were also the age class of females most
harvested by hunters, as only 2 fawns and no yearlings were harvested during my 3-year
study. Break-away collars on spring-captured fawns dropped off at an average of 354.7
days (SE = 54.6; Chapter 2), well after the ﬁrst hunting season that these fawns

experienced.

25

I found that home-range Sizes of adult females during the non-growing season
(140.4 ha) were approximately twice the size of home ranges during the growing season
(77.5 ha). Adult female deer often restrict movements near and following parturition
(Marchinton and Hirth 1984), which may explain this difference in space use. Increases
in time spent foraging due to increased energetic demands during the breeding season and
potentially limited food resources during winter may also have caused increased
movements. Interestingly, the mean proportion of each cover type within home ranges
differed little by season for adults, suggesting that although movements increased, cover
selection (i.e., based on selection indices) remained relatively constant.

My non-growing-season adult home-range estimate (140.4 ha) was similar in Size
to the annual estimates of deer >6 months old (157.7 ha; non-dispersers; n = 53; 91%
females) in southwestern Lower Michigan (Pusateri 2003). However, cover composition
within home ranges differed somewhat. Home ranges in southwestern Lower Michigan
consisted of a higher percentage of agricultural areas (39% versus 20%) and deciduous
forests (47% versus 37% [upland and lowland combined]), but less conifer cover (<4%
versus 8%) compared to my results. Lowland-shrub cover composed a large percentage
of my adult home ranges (~23% during both seasons), but this cover type was not deﬁned
by Pusateri (2003). Comparisons of cover use to other studies in Michigan would be less
meaningful, as they were conducted in northern latitudes where deer often exhibit
migratory behavior and cover types become increasingly different.
Age-specific Trends

Although the conﬁdence limits of age-speciﬁc home-range Sizes often overlapped

somewhat, there was useful information in the age-speciﬁc trends. Trends in mean

26

home-range size of individuals over time on my study area were consistent with broad
descriptions of deer movements. “Young fawns have small home ranges, but as they get
older their ranges begin to approximate that of their dams. Yearlings and young adults
may move over larger areas than do older adults, at least in localities where extreme
seasonal range shifts are common” (Marchinton and Hirth 1984:131), insinuating a trend
similar to deer on my study area. Demarais et al. (2000) stated that home-range size
varied by age of individuals of a species and season, which I found to be true; the
comparison of my results to other Michigan deer studies also supported their assertion
that habitat (conditions) and sex of individuals also affects home-range Sizes.

Yearling females had relatively large home-range sizes (Fig. 1.2), as expected
(Marchinton and Hirth 1984). Although it may be counterintuitive that that these
movement patterns would be coupled with the highest survival (0.94) among age classes,
sources of age-speciﬁc mortality may provide an explanation. The primary sources of
mortality for fawns and adults were vehicle collisions and hunter-harvest, respectively.
Deer may be more susceptible to vehicle collisions until some type of avoidance behavior
has been established, which may occur during thejuvenile stage. Further, adults may be
the age class of antlerless deer selected by or more available to hunters, as suggested by
my data (Table 1.3).

I expected differences in survival and cause-speciﬁc mortality among age classes,
but the magnitude of these differences was unknown, as were the relative differences
when making comparisons with other studies of deer ecology in Michigan. The
information that I have provided should help inform deer managers in Michigan and

other areas of the Midwest experiencing increasing urbanization, as the population

27

characteristics of white-tailed deer in an increasingly urbanizing landscape were largely

unknown.

28

LITERATURE CITED

Ballard, W. B., H. A. Whitclaw, S. J. Young, R. A. Jenkins, and G. J. Forbes. 1999.
Predation and survival of white-tailed deer fawns in northcentral New Brunswick.
Journal of Wildlife Management 63:574—579.

Bender, L. C., G. A. Schirato, R. D. Spencer, K. R. McAllister, and B. L. Murphie. 2000.
Survival, cause-speciﬁc mortality, and harvesting of male black-tailed deer in
Washington. Journal of Wildlife Management 68:870—878.

Beyer, Jr., D. E., and J. B. Hauﬂer. 1992. Diurnal versus 24-hour sampling of habitat
use. Journal of Wildlife Management 58:178—180.

Bunck, C. M., and K. H. Pollock. 1993. Estimating survival of radiotagged birds. Pages
51—63 in J. C. Lebreton and P. M. North, editors. Marked individuals in the study
of bird populations. Birkhauser—Verlag, Basel, Switzerland.

Carpenter, L. H. 2000. Harvest management goals. Pages 192—213 in S. Demarais and
P. R. Krausman, editors. Ecology and management of large mammals in North
America. Prentice Hall, Upper Saddle River, New Jersey, USA.

Carroll, B. K., and D. L. Brown. 1977. Factors affecting neonatal fawn survival in
southem-central Texas. Journal of Wildlife Management 41 :63—69.

Clover, M. R. 1954. A portable deer trap and catch-net. California Fish and Game
40:367-373.

Clute, R. 2006. 2006 Michigan deer hunting prospects: statewide forecast. Michigan
Department of Natural Resources Wildlife Division Press Release.

Creed, W. A., F. Haberland, B. E. Kohn, and K. R. McCaffery. 1984. Harvest
management: the Wisconsin experience. Pages 243—260 in L. K. Halls, editor.
White-tailed deer: ecology and management. Stackpole, Harrisburg,
Pennsylvania, USA.

DelGiudice, G. D., J. F ieberg, M. R. Riggs, M. Carstensen Powell, and W. Pan. 2006. A
long-term age-speciﬁc survival analysis of female white-tailed deer. Journal of
Wildlife Management 70: 1 556—1568.

Demarais, S., K. V. Miller, and H. A. Jacobson. 2000. White-tailed deer. Pages 601—
628 in S. Demarais and P. R. Krausman, editors. Ecology and management of

large mammals in North America. Prentice Hall, Upper Saddle River, New
Jersey, USA.

Diefenbach, D. R., C. O. Kochnanny, J. K. Vreeland, and B. D. Wallingford. 2003.

29

Evaluation of an expandable, breakaway radiocollar for white-tailed deer fawns.
Wildlife Society Bulletin 31 :756—761.

Downing, R. L., and B. S. McGinnes. 1969. Capturing and marking white-tailed deer
fawns. Journal of Wildlife Management 33:71 1—714.

F rawley, B. J. 2004. Michigan deer harvest survey report: 2003 seasons. Michigan
Department of Natural Resources Wildlife Report No. 3418.

Garner, M. S. 2001. Movement patterns and behavior at winter feeding and fall baiting
stations in a population of white-tailed deer infected with bovine tuberculosis in
the northeastern Lower Peninsula of Michigan. Dissertation, Michigan State
University, East Lansing, Michigan, USA.

Haugen, A. O., and D. W. Speake. 1958. Determining age of young fawn white-tailed
deer. Journal of Wildlife Management 22:319—321.

Johnson, D. H. 1999. The insigniﬁcance of statistical signiﬁcance testing. Journal of
Wildlife Management 63:763—772.

Kemohan, B. J ., R. A. Gitzen, and J. J. Millspaugh. 2001. Analysis of animal space use
and movements. Pages 125—166 in J. J. Millspaugh and J. M. Marzluff, editors.
Radio tracking and animal populations. Academic Press, San Diego, California,
USA.

Labisky, R. F., and M. C. Boulay. 1998. Behaviors of bobcats preying on white-tailed
deer in the Everglades. American Midland Naturalist 139:275-281.

Lenth, R. V. 19810. Robust measures of location for directional data. Technometrics
23:77—81.

Lenth, R. V. 1981b. On ﬁnding the source of a signal. Technometrics 23:149—154.

Lund, R. C. 1975. The capture and marking of wild, newborn white-tailed deer fawns.
Transactions of the Northeast Section of the Wildlife Society 32:25—33.

Mackey, T. 1996. White-tailed deer movements, habitat use, and browsing effects on
vegetation in the Upper Peninsula of Michigan. Thesis, Michigan State
University, East Lansing, Michigan, USA.

Madill, H., and W. Rustem. 2001. Michigan land resource project. Public Sector
Consultants, Lansing, Michigan, USA.

Marchinton, R. L., and D. H. Hirth. 1984. Behavior. Pages 129-168 in L. K. Halls,
editor. White-tailed deer: ecology and management. Stackpole, Harrisburg,
Pennsylvania, USA.

30

Mayﬁeld, H. 1961. Nesting success calculated from exposure. Wilson Bulletin 73:255—
261.

Mayﬁeld, H. 1975. Suggestions for calculating nest success. Wilson Bulletin 87:456—
466.

McClean, S. A., M. A. Rumble, R. M. King, and W. L. Baker. 1998. Evaluation of
resource selection methods with different deﬁnitions of availability. Journal of
Wildlife Management 62:793—801.

McCullough, D. R. 2001. The George Reserve deer herd. Reprint of ﬁrst edition.
Blackburn Press, Caldwell, New Jersey, USA.

Michigan Center for Geographic Information. 2001. IF MAP land-use, land-cover data.
http://www.mcgi.state.mi.us/mgdl/?rel=ext&action=sext (Accessed 28 Jul 2005).

Michigan Department of Agriculture. 2002. Agricultural statistics: Jackson and
Washtenaw counties. http://www.michigan.gov/mda/ (Accessed 16 Aug 2004).

Michigan Department of Natural Resources. 2005. 2005 deer population estimates for
south central management unit. Michigan Department of Natural Resources,
Wildlife Division Technical Report.

Muzo, D. 2003. Movement of white-tailed deer from a bovine tuberculosis infected
population under baiting and winter feeding restrictions and TB status of black
bears in Michigan. Thesis, Michigan State University, East Lansing, Michigan,
USA.

Nams, V. O. 1989. Effects Of radio-telemetry error on sample size and bias when testing
for habitat selection. Canadian Journal of Zoology 67:1631—1636.

Nams, V. O., and S. Boutin. 1991. What is wrong with error polygons? Journal of
Wildlife Management 55: 172—176.

Ozoga, J. J., and L. J. Verme. 1986. Relation of maternal age to fawn-rearing success in
white-tailed deer. Journal of Wildlife Management 50:480—486.

Pusateri, J. S. 2003. White-tailed deer population characteristics and landscape use
patterns in southwestern Lower Michigan. Thesis, Michigan State University,
East Lansing, Michigan, USA.

Pusateri Burroughs, J., H. Campa, III, S. R. Winterstein, B. A. Rudolph, and W. E.

Moritz. 2006. Cause-speciﬁc mortality and survival of white-tailed deer fawns in
southwestern Lower Michigan. Journal of Wildlife Management 70:743—75 1.

31

Seaman, D. E., J. J. Millspaugh, B. J. Kemohan, G. C. Brundige, K. J. Raedeke, and R.
A. Gitzen. 1999. Effects of sample size on kernel home range estimates. Journal
of Wildlife Management 63:739—747.

Severinghaus, C. W. 1949. Tooth development and wear as criteria of age in white-
tailed deer. Journal of Wildlife Management 13:195—216.

Sitar, K. L. 1996. Seasonal movements, habitat use patterns, and population dynamics of
white-tailed deer (Odocoileus virginianus) in an agricultural region of northern
Lower Michigan. Thesis, Michigan State University, East Lansing, Michigan,
USA.

Sitar, K. L., S. R. Winterstein, and H. Campa III. 1998. Comparison of mortality sources

for migratory and nonmigratory deer in the northern Lower Peninsula of
Michigan. Michigan Academician 30:17—26.

Skalski, J. R., K. E. Ryding, and J. J. Millspaugh. 2005. Wildlife demography: analysis
of sex, age, and count data. Elsevier Academic Press, Burlington, Massachusetts,
USA.

Smith, R. L., and J. L. Coggin. 1984. Basis and role of management. Pages 571—600 in
L. K. Halls, editor. White-tailed deer: ecology and management. Stackpole,
Harrisburg, Pennsylvania, USA.

Sommers, L. M. 1977. Atlas of Michigan. Michigan State University Press, East
Lansing, Michigan, USA.

Sparrowe, R. D., and P. F. Springer. 1970. Seasonal activity patterns of white-tailed
deer in eastern South Dakota. Journal of Wildlife Management 34:420—43 1.

Sudharsan, K. 2005. Environmental factors affecting the frequency and rate of deer-
vehicle crashes (DVCs) in southern Michigan. Thesis, Michigan State University,
East Lansing, Michigan, USA.

Torgersen, 0., and W. R. Porath. 1984. Midwest oak/hickory forest. Pages 411—426 in
L. K. Halls, editor. White-tailed deer: ecology and management. Stackpole,
Harrisburg, Pennsylvania, USA.

US. Census Bureau. 2003. 2000 census of population and housing, population and
housing counts: Michigan. Washington, DC, USA.

US. Census Bureau. 2006. State & county quickfacts: 2006 (Barry, Jackson,
Kalamazoo, Washtenaw counties). Washington, DC, USA.

US. Fish and Wildlife Service. 2001. 2001 national survey of ﬁshing, hunting. and
wildlife-associated recreation. Department of Interior, Washington, DC, USA.

32

Van Deelen, T. R. 1995. Seasonal migrations and mortality of white-tailed deer in
Michigan’s Upper Peninsula. Dissertation, Michigan State University, East
Lansing, Michigan, USA.

Van Deelen, T. R., H. Campa 111, J. B. Hauﬂer, and P. D. Thompson. 1997. Mortality
patterns of white-tailed deer in Michigan’s Upper Peninsula. Journal of Wildlife
Management 61 :903—910.

Van Deelen, T. R., H. Campa 111, M. Hamady, and J. B. Hauﬂer. 1998. Migration and
seasonal range dynamics of deer using adjacent deeryards in northern Michigan.

Journal of Wildlife Management 62:205—213.

Verrne, L. J. 1973. Movements of white-tailed deer in Upper Michigan. Journal of
Wildlife Management 37:545—552.

White, G. C., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data.
Academic Press, San Diego, California, USA.

Worton, B. J. 1995. Using Monte Carlo simulation to evaluate kemel-based home range
estimators. Journal of Wildlife Management 59:794—800.

33

£8-93 mo 598:: :38.

n 3083 wEnEEPe

 

Son @2393 mo Bass: :38.

 

wmod mm 00
vmod m E
Nmod 9 cm
Rod 2 2

NV www—
w 30
mm 5%
NH omm

ESP

ooom .32 mmlcﬂ. a m
moom 32 vmlcﬂ. v

.voom 003 3.1: N

voom SE omlcmm mm ﬂ

 

ammooosm wsamﬁh 8:96 Banged 8:593 50,—.

woo—38068 mhmwdﬁh
moEEom

3mm commom

 

<5 .eewaez 3:8 58... @8358

mats“. £35 535 mean QEEEMRE mammooouQv coon 3:8-833 mo 283m 2393 e253 m we comings—co ._. ._ 03mg.

34

.35 88:8: woe—SE23}

820452539.

 

 

 

 

:2 u .1. 23 em mm Eﬁ
as in: mm 2 e 88 2:: miss: a m
mom Sam 8 2 M: 88 2:: Sta: 2 N
Nam 32 em a M: 38 us: are: : :

eggoo: 8m 30% 2:8 Axe :oﬁmEoEmm b03030 8mm 33%
steam 83:82

 

.<mD Emmi—0:2 35:8 5:8

.coomlwoom wESU QESEMRS SumooonQv Soc 3:86:53 38:8: no 3038 2398 wEEm m we :omtmaﬁoo .NA 03:...

35

Table 1.3. Fate assessment of radiomarked female white-tailed deer (Odocoileus
virginianus) based on age class (fawn = <1 yr, yearling = 2 1 yr—<2 yr, adult = 22 yr),

south central Michigan, 2004—2006.

 

 

 

Age classa
Fate Fawn (n = 32L Yearling (n = 12) Adult (n = 28)
Mortality
Hunter-harvest 2 0 1 1
Vehicle collision 7 0 1
Trauma or malnutrition 0 1 1
Canid predationb 2 0 0
Total 11 1 13
Censored
Collar slipped/removedc 9 6 6
Lost signal 3 1 3
Alive at end of study 6 2 5
Capture myopathy, collar 3 2 I
trauma (mortality)
Total 21 11 15

 

z“At time of fate assessment.

bIncluded coyote (Canis latrans), red fox (Vulpes vulpes), gray fox (Urocyon
cinereoargenteus), and domestic dog (C. lupus familiaris).

cRemoved collars included break-away radio-transmitters designed to drop off after 9—12

months and remotely removed GPS collars after 1 yr of use.

36

Table 1.4. Space use (95% ﬁxed-kernel home ranges) and cover composition of female

fawn (<1 yr old)white-tai1ed deer (Odocoileus virginianus) during the agricultural

growing (10 May—7 Oct; n = 28) and non-growing (8 Oct—9 May; n = 21) seasons,

southern Michigan, 2004—2006.

 

 

 

Season Kernel home range size (ha)
Cover type ; 95% LCL 95% UCL

Growing 60.24 31.23 89.26
Agriculture 19.20 8.16 30.24
Conifer 13.15 5.67 20.64
Herbaceous openland 0.19 0.00 0.44
Lowland deciduous forest 9.33 3.14 15.53
Lowland shrub 4.61 1.93 7.30
Mixed wetland 4.42 0.00 9.47
Northern hardwood 2.29 1.25 3.32
Oak association 0.69 0.00 1.57
Upland deciduous forest 4.50 2.51 6.48
Upland shrub 0.00
Urban 0.18 0.00 0.43
Water 1.49 0.00 3 .25
Other 0.00

Non-growing 116.26 73.33 159.19
Agriculture 21.55 12.70 30.41

 

37

Table 1.4. (cont’d)

 

Season Kernel home range Size (ha)

 

Cover type x 95% LCL 95% UCL

 

Non-growing

Conifer

Herbaceous openland
Lowland deciduous forest
Lowland shrub

Mixed wetland
Northern hardwood

Oak association

Upland deciduous forest
Upland shrub

Urban

Water

Other

15.89

4.99

16.49

18.47

7.75

4.24

2.98

22.55

0.00

0.11

0.83

0.04

6.47

0.79

7.33

3.85

0.00

2.77

1.29

13.44

0.00

0.00

0.00

25.31

9.18

25.65

33.09

19.27

5.72

4.67

31.67

0.34

1.87

0.10

 

38

Table 1.5. Space use (95% ﬁxed-kernel home ranges) and cover composition of female

yearling (2 1—<2 yr old) white-tailed deer (Odocoileus virginianus) during the

agricultural growing (10 May—7 Oct; n = 12) and non-growing (8 Oct—9 May; it = 15),

southern Michigan, 2004—2006.

 

 

 

Season Kernel home range size (ha)
Cover type ,1 95% LCL 95% UCL

Growing 201.82 1.25 402.38
Agriculture 51.74 0.00 113.62
Conifer 19.22 0.00 40.56
Herbaceous openland 2.86 0.51 5.20
Lowland deciduous forest 23.78 0.00 52.86
Lowland shrub 36.27 3.31 69.22
Mixed wetland 10.24 0.00 20.51
Northern hardwood 11.19 0.00 26.04
Oak association 3.70 0.00 10.25
Upland deciduous forest 37.62 1.10 74.14
Urban 0.00
Water 5.10 3.95 13.79
Other 0.10 0.00 0.32

Non-growing 156.88 96.35 217.41
Agriculture 18.34 5.10 31.59
Conifer 13.59 1.46 25.72

 

39

Table 1.5. (cont’d)

 

Season Kernel home range size (ha)

 

Cover type x 95% LCL 95% UCL

 

Non-growing

Herbaceous openland
Lowland deciduous forest
Lowland shrub

Mixed wetland

Northern hardwood

Oak association

Upland deciduous forest
Upland shrub

Urban

Water

Other

4.26

25.89

23.33

3.48

8.42

4.02

52.19

0.00

0.09

3.21

0.08

0.72

8.99

9.83

0.00

5.01

0.92

29.07

0.00

0.00

0.00

7.79

42.80

36.82

7.05

11.82

7.11

75.30

0.27

6.56

0.25

 

40

Table 1.6. Space use (95% ﬁxed-kernel home ranges) and cover composition of female

adult (>2 yr old) white-tailed deer (Odocoileus virginianus) during the agricultural

growing (10 May—7 Oct; n = 23) and non—growing (8 Oct—9 May; it = 25), southern

 

 

 

Michigan, 2004—2006.

Season Kernel home range size (ha)
Cover type 3; 95% LCL 95% UCL

Growing 77.51 57.46 97.56
Agriculture - 11.23 4.57 17.90
Conifer 8.18 2.28 14.08
Herbaceous openland 0.28 0.00 0.63
Lowland deciduous forest 4.69 1.52 7.85
Lowland shrub 17.50 9.47 25.53
Mixed wetland 2.74 0.00 6.90
Northern hardwood 2.92 1.45 4.39
Oak association 1.42 0.32 2.52
Upland deciduous forest 27.12 15.26 38.99
Upland Shrub 0.00
Urban 0.12 0.00 0.29
Water 1.32 0.02 2.61
Other 0.00

Non—growing 140.39 91.58 189.20
Agriculture 27.77 13.76 41.77

 

41

Table 1.6. (cont’d)

 

 

 

Season Kernel home range size (ha)
Cover type _; 95% LCL 95% UCL

Non-growing
Conifer 10.83 4.35 17.31
Herbaceous openland 1.26 0.00 2.61
Lowland deciduous forest 8.47 4.14 12.80
Lowland shrub 33.50 17.87 49.14
Mixed wetland 6.05 0.10 11.99
Northern hardwood 3.86 1.98 5.74
Oak association 2.25 1.23 3.27
Upland deciduous forest 43.89 25.47 62.31
Upland shrub 0.00
Urban 0.14 0.00 0.34
Water 2.39 0.61 4.17
Other 0.00

 

42

t

It
i

E...
Iii
i

3%

iiiit
ii
i

WIN“!!!
?‘ ‘ wt.

A
Ar

 

 

 

 

 

 

 

 

 

 

Figure 1.1. Study area location and distribution of cover types, south central Michigan,
USA, 2004—2006. Study-area boundary was deﬁned to include any township containing

21 radiomarked white-tailed deer (Odocoileus virginianus) location estimate.

43

250 —

 

 

: E : : :
I g I I I
~ ' . : : :
i H : : :
' I I I
AZOO— I I I I I

g E , 2+ 5+ E+
_ r . I I I
B : : I I 0
a I+ ' I I I
0150— I i I : I
w ' i I : :
g i I 0 I I
e: ” i : , :
a) I i g g I

$100 T4. : E- + I- + i- + :_
m ' . I I I
l— :_ I : I I
i : I i i
I g — I — I — I
5° 5 5 E E E
- I r I I I
: 5 : : :

O IIIIIllllllIlIllllIJIlllJlLilIIIIIIIILIJ
o 6 12 I8 24 30 36 42 48 54 60

May Nov May Nov May Nov May Nov May Nov
Age (months)

Figure 1.2. Pattern of home-range size estimates of white-tailed deer based on mean size
and 95% CLs (- = lower, + = upper) of 3 age classes (fawn = 0—12 months old, yearling
= 12—24 months old, adult = >24 months old) pooled through time during the growing
(~May—Oct) and non-growing (~Oct—May) agricultural seasons, south central Michigan,
USA, 2004—2006. Vertical dashed lines approximate the regular ﬁrearms deer season

(15—30 Nov).

44

CHAPTER 2

Survival and Space Use of Fawn White-tailed Deer in Southern Michigan

45

Survival and Space Use of Fawn White-tailed Deer in Southern Michigan

INTRODUCTION

Most members of the deer family (Cervidae) and other ungulates have similar
patterns of behavior near and following parturition. Neonatal ungulate behavior includes
either following the dam or hiding to reduce their conspicuousness to predators; most
cervids exhibit the latter behavior (Lent 1974). Moose (Alces alces), elk (Cervus
canadensis), mule deer (Odocoileus hemionus) and white-tailed deer (0. virginianus)
adult females often isolate themselves from intraspeciﬁcs, the young rely on cover and
cryptic coloration as their primary defenses against predation, and the dam visits
offspring periodically for feeding and grooming. After a certain degree of development,
young cervids progressively increase activity and movements, primarily while
accompanying the dam; adult male deer generally exhibit behaviors and space use
independent of the young (see F ranzmann 1981, Marchinton and Hirth 1984, Smith 1991,
Demarais and Krausman 2000).

Fawn white-tailed deer (hereafter, deer) and their dams express consistent
behaviors through the ﬁrst 2 months postpartum. F awns 52 months old are sedentary
and associate infrequently with their dams (Schwede et al. 1994); Jackson et al. (1972)
found that fawns in this age class had activity patterns dissimilar to adult females, and
that fawns oﬁen were active 5 20% of the time. Twins are maintained at separate
localities during the first 3—6 weeks (Ozoga et al. 1982, Marchinton and Hirth 1984) and
adult females often establish territories surrounding their fawns for approximately the

first month (Ozoga et al. 1982, Marchinton and Hirth 1984). Fawns are functional

46

ruminants around 2 months of age (Short 1964), suggesting decreasing dependence on
the dam. Fawn survival may increase signiﬁcantly after the ﬁrst 8 weeks (Carroll and
Brown 1977), as they may be better able to elude predators (Nelson and Woolf 1987).
Low survival of fawns may decrease recruitment enough to affect deer population
dynamics (Cook et al. 1971; Gaillard et al. 2000), so knowledge of fawn survival should
increase understanding of the population ecology of deer in diverse landscapes.

My objectives were to estimate survival, assess cause-speciﬁc mortality and
quantify space use of fawn white-tailed deer in an agro-forest ecosystem in south central
Michigan. 1 estimated survival at 2- and 6-months postpartum to describe risks during
the sedentary period of fawns and to allow comparison of my ﬁndings to other studies,
respectively. I assessed cause-speciﬁc mortality for the same time periods to describe
what mortality risks existed during development (0—2 months old) and during increased
independence from the dam (e. g., increased movements during hunting season; 0—6
months old), respectively. I also estimated space use and cover selection to describe
cover used by fawns during the ﬁrst 2 months postpartum.

STUDY AREA

My study was conducted in eastern Jackson (Grass Lake, Henrietta, and Waterloo
townships), western Washtenaw (Dexter, Lima, Lyndon, and Sylvan townships), and
southwestern Livingston (Unadilla township) counties in the south central Lower
Peninsula of Michigan (hereafter, Lower Michigan). The study area (82,636 ha) included
publicly owned lands, including the Michigan Department of Natural Resources (MDNR)
Waterloo (8,410 ha; 10.2% of study area) and Pinckney (4,276 ha; 5.2%) recreation

areas, and. privately owned lands. South central Michigan has been characterized by a

47

relatively high deer density (~27/km2 during fall 2005; MDNR 2005), increasing land-use
activities (e.g., urbanization) and little scientiﬁc study of deer under these conditions.
Although the study area was primarily rural (98% of total land area), the human
population increased 16% and housing units increased 22% between 1990 and 2000 (US.
Census Bureau 2003). Much of the landscape throughout southern Michigan (and in
other areas throughout the Midwest) is expected to experience increasing land-use
activities similar to the study area (Madill and Rustem 2001).

The physiographic regions of this area are Hillsdale-Lapeer Hilly Upland, South
Central Rolling Plain, and Southeastern Rolling Plain (Sommers 1977:24) with alﬁsols as
the major soil order (Sommers 1977:36). Surface formations in the study area are the
result of glaciation and include all 4 types present within Michigan (moraine, till plain,
outwash plain, and lacustrine plain; Sommers 1977:32). Elevation of the study area
ranged approximately 180—300 m and consisted of relatively limited relief (Sommers
1977:26, 33). The study area received about 81 cm of precipitation annually during
1971—2000 (based on conditions in Chelsea, Michigan; Midwestern Regional Climate
Center, Champaign, Illinois, USA), and had a ISO-day growing season (i.e., the average
annual accumulation of daily mean temperatures >5.6 °C), generally occurring from 10
May to 7 October (Sommers 1977:46, 49). During 1971—2000, average annual snowfall
was 99.3 cm and mean monthly temperatures ranged from —5.4 °C (Jan.) to 21.8 °C (Jul.)
in Jackson County (Midwestern Regional Climate Center, Champaign, Illinois, USA).

Before European settlement, southern Michigan forests consisted of oak (Quercus
spp.) and hickory (Carya spp.) in well-drained soils and beech (Fagus grandifolia), elm

(Ulmus spp.), maple (Acer spp.), and basswood ( T ilia americana) in poorly drained soils

48

(Sommers 1977:17). Much of the study area is well suited for agriculture (Sommers
1977:38). Most crop agriculture in Jackson and Washtenaw counties (total area =
366,483 ha) is com (37,840 ha) and soybean (34,200 ha) production, followed by hay
(17,200 ha), winter wheat (7,970 ha) and oats (930 ha; Michigan Department of
Agriculture 2002).

I generalized land-use, land-cover data (Michigan Center for Geographic
Information 2001) using ArcView GIS v3.2 software (Environmental System Research
Institute, Redlands, California, USA) and Spatial Analyst extension to deﬁne 13 cover
types within the study area: agriculture (non-vegetated farmland, row crops, forage crops;
52.3% of study area); conifer (pines [Pinus spp.], other upland conifers; 1.5%);
herbaceous openland (herbaceous vegetation with <25% woody cover; 2.9%); lowland
deciduous forest (>60% composed of deciduous tree cover; 8.0%); lowland shrub (with
>60% non-water cover; 9.9%); mixed wetland (floating aquatic vegetation, emergent
wetland, mixed non-forest wetland; 3.1%); northern hardwood (>60% canopy cover of
maple, beech, ash [Fraxinus spp.], cherry [Prunus spp.], birch [Betula spp.]; 2.3%); oak
association (>60% canopy cover of oak; 1.6%); upland deciduous forest (>60% canopy
cover of upland deciduous trees; 11.6%); upland shrub (>25% woody cover; <0.1%);
urban (low and high intensity, roads, parks, golf courses; 2.8%); water (surface, ﬂowing;
3.9%); and other (aspen [Populus spp.] association, orchards, bare ground; 0.1%). Patch
size of cover types ranged from <l ha to >l 1,000 ha (e.g., an agricultural matrix) and had
a mean size of 29.2 ha. Consequently, generalization of spatial data may have excluded

certain ﬁne-scale landscape characteristics (e.g., hedgerows, roads) in some instances.

49'

METHODS

The capture season for neonatal fawns was from mid-May to mid-June 2004—
2006. I expected capture success to peak around 1 June and instances of fawns flushing
to increase greatly after mid-June, thus reducing the ability for successful capture; fawns
>2 weeks old generally start ﬂushing when approached by humans (Carroll and Brown
1977). My technicians and I captured neonates either by hand or in a ﬁsh-landing net
(0.5-m-diameter net, 2-m-long extendable handle). A group of 2—6 people systematically
searched potential fawning habitat (e.g., areas transitioning from forest to wetland or
grassy ﬁeld) for neonates (Lund 1975, Ballard et al. 1999). Isolated adult females
occasionally provided us with behavioral signs of a nearby neonatal fawn (e.g., a doe
looking toward an area several times/min; Downing and McGinnes 1969).

I weighed, sexed, ear-tagged (Style 681; National Band and Tag, Newport,
Kentucky, USA) and radiomarked each captured neonate. Captured neonates also were
aged using hoof grth measurements (Haugen and Speake 195 8). The mortality-
sensing radio-transmitters (Model M4210; Advanced Telemetry Systems, Isanti,
Minnesota, USA) had an expandable collar to allow for growth and were designed to
drop off after 9—l2 months for retrieval (see Diefenbach et al. 2003). Transmitters had a
unique frequency within the lSl-MHz range, a mass of 60 g, and an expected minimum
battery life of 12 months. The precise-event transmitter option provided a time-of-
mortality estimate within 30 min for a maximum of ~5 d post-mortality (i.e., 5 days of no
transmitter movement). I used SYSTAT v11 (Systat Software, Inc., San Jose, California,
USA) and ProStat v4.02 (Poly Software International, Inc., Pearl River, New York, USA)

for my statistical analyses. Michigan State University’s All-University Committee on

50

Animal Use and Care approved all capturing and handling procedures for my study
(Application No. 01/04-006-00).
Survival and Cause-speciﬁc Mortality

My technicians and I monitored fawns daily for the ﬁrst 30 days following
capture and 2—5 times/week thereafter to estimate survival and assess cause-speciﬁc
mortality. I used the Mayﬁeld method (Mayﬁeld I961, 1975), as modiﬁed by Bunck and
Pollock (1993) for censored individuals, to estimate survival of fawns during 2- and 6-
month periods post-capture. Period survival was based on fates and exposure days during
each period to estimate the daily survival rate, as I assumed daily survival would vary by
time period. For analyses, I excluded any fawn that did not exceed a 7-day acclimation
period following capture, but I also estimated survival without an acclimation period.
Individuals were censored if I believed they were alive at the time of transmitter recovery
(e.g., the break-away collar dropped, collection of ﬁeld data concluded). I also
performed a hazard analysis based on the probability of death (i.e., the number of
individuals dying/number of individuals at risk) by weekly time periods for a total of 24
weeks (see Winterstein et al. 2001).

My assessment of cause-speciﬁc mortality of recovered carcasses was based on
my ﬁeld observations and the necropsies performed by a wildlife pathologist (MDNR
Wildlife Disease Laboratory, East Lansing, Michigan, USA). I classiﬁed mortalities of
individuals into 5 categories: canid- (coyote [Cam's latrans], red fox [Vulpes vulpes], gray
fox [Urocyon cinereoargenteus], or domestic dog [C. IupusfamiliarisD kill, malnutrition,

vehicle collision, hunter-harvest and unknown. If I did not have enough post-mortality

51

evidence to ascribe cause with reasonable certainty (e.g., total consumption of the
carcass), then I considered the cause of mortality to be unknown.
Space Use

I estimated locations of fawns 2—5 times/week using triangulation from telemetry
signals (White and Garrott 1990). To increase the accuracy of habitat-use assessment,
my technicians and I located fawns in a systematic manner during varying time schedules
on a diel basis, with 21 nocturnal location/deer/week (Beyer and Hauﬂer 1992), except
during capture periods. I used a 3-element Yagi antenna (Advanced Telemetry Systems,
Incorporated, Isanti, Minnesota, USA), portable radio receiver (Model R-IOOO,
Communications Specialists, Incorporated, Orange, Califomia, USA), handheld global
positioning system (GPS) unit (Model GPS IV; GARMIN International, Incorporated,
Olathe, Kansas, USA) and mirror-sighting compass to estimate bearings.

I used the program LOCATE III (Pacer, Truro, Nova Scotia, Canada) to estimate
locations of fawns using triangulated data based on the maximum likelihood estimator
(Lenth 1981a, b), as recommended by White and Garrott (1990) and Nams and Boutin
(1991). I used Animal Movement extension in ArcView GIS v3.2 software
(Environmental System Research Institute, Redlands, California, USA) to estimate 95%
kernel home ranges and to quantify cover composition within home ranges using my
cover-type classiﬁcation system.

My description of fawn cover included only individuals with Z 30 locations
(Seaman et al. 1999) and that survived Z2-months post-capture. I assumed this time
period approximated the age of fawns with low activity (Jackson et al. I972), and

consequently, avoidance of predators through cover selection. F awns may also have

52

selected cover for thermal properties, although I could not assess these speciﬁc selection
strategies due to my spatial scale of study. Relationships between cover use (the mean
proportion of each cover type within individual kernel home ranges) and availability (the
proportion of each cover type within the study area) characterized cover selection indices
(Ivlev 1961). I followed the equations of Strauss (1979) to estimate 95% conﬁdence
limits of selection indices.
RESULTS

My technicians and I observed 55 neonatal fawns during the capture seasons, and
captured and radiomarked a total of 34 (2004 = 9, 2005 = 10, 2006 = 15). Capture effort

averaged 25.0 observer-hr/radiomarked neonate, but effort decreased by year (2004 =
39.2, 2005 = 30.8, 2006 = 12.3). I pooled age (J-C = 7.2 d i 0.2 SE, range = 4.9—10.3),

weight (E = 4.6 kg :1: 0.2 SE, range = 3.1—6.7) and sex (56% male) data for calculations.
Throughout my analyses, I pooled fawns by sex and assumed behavioral differences were
minimal or nonexistent between male and female fawns during the early stages of life
(see Ozoga and Verme 1986).
Survival and Cause-speciﬁc Mortality

I pooled survival data for all 3 capture seasons (n = 32 fawns) because no single
capture season sample size exceeded 20 individuals, a minimum value recommended by
Hensler and Nichols (1981) based on the uncertainty resulting from small sample sizes.
My survival analyses both included and excluded 2 radiomarked fawns that died during
the acclimation period (<7 days post-capture). With an acclimation period, daily survival
by period using the Mayﬁeld method was 0.9965 (60 d) and 0.9978 (180 d), resulting in

the survival probabilities of 0.81 (60 d; SE = 0.07) and 0.67 (180 d; SE = 0.08). Six of 32

53

fawns died during the ﬁrst 60 days, while none were censored; 10 of 32 died during the
ﬁrst 180 days, while 3 were censored (2 break-away collars dropped at <1 80 d; l fawn
was censored from survival analysis because a wildlife rehabilitator removed the
individual from the ﬁeld for an unknown injury). Without a 7-day acclimation period. 2
additional fawns were included in my survival analysis (n = 34; 60 d = 0.75, 95% CL =
0.61—0.90; 180 d = 0.62, 95% CL = 0.45—0.79), slightly reducing survival estimates.

Mortality from canids was the highest cause of mortality (3 of 34 fawns; mean

age = 33.7 d) for fawns S 60 days (Table 2.1). Fawns 60—180 days (J_c = 125.4)
experienced relatively high mortalities from vehicle collisions (Table 2.1). Breakaway
transmitters had an approximate life span of 1 year during my study, which was expected
based on manufacturer speciﬁcations.

I pooled data and considered probability of death over time based on weekly
estimates. During weeks 2—4, and approximately week 16, the probability of death
peaked (0.50 and 0.35, respectively). The probability of death was minimal and constant
during other time periods.

Space Use

Twenty-six of 34 radiomarked fawns survived 22 months post-capture and were
available for space-use analysis (i.e., had 2 30 location estimates). Locations/fawn for
the ﬁrst 2 months averaged 37.5 (SE = 0.9). Because of small annual sample sizes, I
pooled location data of fawns to assess space use. F awns were radiomarked for an
average of 203.8 days before being removed from the study (i.e., mortality, censored).

Approximately half of the fawns that lived to 2 months were captured in

agriculture and upland deciduous forest cover types (Table 2.2). Home-range estimates

54

averaged 40.9 ha and were highly variable (SE = 6.8; range = 2.7—l 66.8 ha) among fawns
during their ﬁrst 2 months. About 50% of the mean composition of home ranges
included agricultural areas and conifers; no home ranges included the categories upland
shrub and other, although each existed in very small proportions (~0. 1%) on the study
area. Only 2 cover types (conifer, lowland deciduous forest) were used in proportions
greater than available, whereas 4 cover types (agriculture, herbaceous openland, other,
upland shrub) were used less than expected; all other cover types (e.g., mixed wetland,
northern hardwood) were neutrally used (i.e., CLs bracketed 0). Conifers were strongly
selected for, and also comprised 26% of the mean home range area during the ﬁrst 2
months.
DISCUSSION
Survival and Cause-speciﬁc Mortality

Relatively little published information is available describing demographics and
cover types that provide cover speciﬁc to the ﬁrst 2 months of life for white-tailed deer
fawns. Vreeland et al. (2004) estimated 9—week survival of fawns at 72.4% (agricultural
cover types) and 57.2% (forested cover types) in Pennsylvania, both lower than my
estimate (81%). Eight-week-old fawns had low survival in a suburban environment in
Alabama (33%; Saalfeld and Ditchkoff 2007). Pusateri Burroughs et al. (2006) estimated
survival (Kaplan-Meier estimator; Winterstein et al. 2001) for each of 2 years at 290%
(interpolated from ﬁg.l) for 2-month-old fawns in southwestern Lower Michigan; my
180-day survival estimate (67%) was less than their 220-day lowest annual estimate
(76%). Other survival estimates of fawns S2 months old included 90% (4 weeks post-

capture, Minnesota; Brinkman et al. 2004), 89% (1-30 d, Massachusetts; Decker et al.

55

1992), 86% (<30 d) and 84% (<60 d, Iowa; Huegel et al. 1985), and 80% (0—30 d) and
100% (31—60 (1, Mississippi; Bowman et al. 1998). Generally, published fawn survival
estimates by time period were lowest earliest in life, similar to my results.

My survival estimates may be overestimated to some degree (assuming that radio-
transmitters and capture did not affect survival; see Ozoga and Clute 1988), as mortality
may be higher during the ﬁrst several days postpartum (i.e., I discovered the remains of
several dead unmarked neonates S 1 week old on the study area). My radiomarked fawns
had an estimated mean age of 7.2 days (SE = 0.2) at capture, also suggesting an
overestimation of survival. When comparing survival estimates, spatial (e.g., vegetation
structure) and temporal (e.g., weather patterns) differences should be considered.

Pooling data over years, as 1 did, may result in some lost information (e.g., potential
correlation between survival and weather), but may still provide a reasonably accurate
mean survival estimate over time. Aspects related to timing of births, intragenerics (not
present on my study area), and predation may greatly inﬂuence fawn survival (e.g.,
Whittaker and Lindzey 1999) within the geographic range of white-tailed deer, as well as
the proportion of does in the study that are primiparous (Ozoga and Verme 1986).
Survival estimates may also be inﬂuenced by the use of an acclimation period, although
the use of such periods for neonate white-tailed deer has recently come into question
(e.g., Carstensen Powell et al. 2005). Based on 95% CLs, an acclimation period made no
difference in my survival estimates.

Because young fawns are sedentary, predation and malnutrition are often the
causes of mortality, although I witnessed several fawns on or near paved roads. My data

indicated that approximately 10% of fawns S2 months old were killed by predators

56

(speciﬁcally, canids) and 3% died of malnutrition on my study area. Sixty-nine percent

of fawn mortalities in southern Illinois were attributed to canids (Nelson and Woolf

1987); most ofthese fawns were 27—47 days, similar to my result (E = 34 d; Table 2.1).
In 2 suburban areas in Chicago, Illinois, canids caused most (66% and 100%) mortalities
of fawns between capture and 1 July (Piccolo 2002), whereas in suburban Alabama, 42%
of mortalities of 8-week-old fawns were caused by canids (Saalfeld and Ditchkoff 2007).
Huegel et al. (1985) found that almost 77% of fawn mortalities in Iowa were canid-
related, with fawns <30 days seemingly most susceptible. Predation accounted for 67%
of fawn mortality through a 12-week post-capture in southwest Minnesota, with 50% of
predation attributed to coyotes and 50% unknown (Brinkman et al. 2004). At lower (e.g.,
Oklahoma, Bartush and Lewis 1981; Texas, Cook et al. 1971) and higher (e.g., New
Brunswick, Canada, Ballard et al. 1999) latitudes, coyotes may also be the primary
predator of fawns.

Before European settlement, coyotes existed primarily west of the Mississippi
River, but currently their geographic distribution includes nearly all of the continental
United States and Canada (Moore and Parker 1992). The increasing geographical
distribution of coyotes may cause a change in deer demographics, especially in areas
containing low-quality habitat for deer. Cause-speciﬁc mortality sources can be highly
dependent on predator species and densities. No predator density estimates for canids
(e.g., coyote, red fox) were available for my study area, but information from harvest
surveys suggested an increasing trend in the relative abundance of coyotes in Michigan
(D. Etter, MDNR, personal communication), which may inﬂuence the proportion of

young fawns killed by predators.

57

The land-use changes on my study area included increasing urbanization; it would
be logical to suspect that the number of domestic dogs increases and the proportion of
available fawn cover decreases under these circumstances. Consequently, a potential
increase in fawn depredation (or some level of compensatory shift in mortality) by canids
seems likely if canid" densities continue to increase, and some upper bound of urban
development likely exists to reduce habitat suitability and negatively affect fawn survival
(see Piccolo 2002). However, the effects that various levels of canid predation of fawns
would have on deer population characteristics seem largely unknown.

Space Use

Although the relationship between survival and home-range composition may not
be clear (e.g., Nixon and Etter 1995, Vreeland et al. 2004), and white-tailed deer can
persist in a wide range of habitat conditions, I did have evidence of cover selection by 2-
month-old fawns. Pusateri Burroughs et al. (2006) found cover selection (use)
proportional to the composition of their Michigan study area (availability) for 27-week-
old fawns. Cover selected on my study area consisted of conifers and lowland deciduous
forests; however, home-range composition (ha) was highest for agricultural areas and
conifers, respectively (Table 2.2). Although south central Texas had a different
landscape composition than my study area, Carroll and Brown (1977) identiﬁed hiding
cover of 2-month-old fawns at a ﬁne spatial scale consisting of 2 70% grasses, sedges,
vines and forbs. Piccolo (2002) discussed the inverse relationship between hiding cover
availability and risk to predation due to increased fawn movements where hiding cover is

less available.

58

Fawn home-range sizes during my study were highly variable (2.7—166.8 ha).
Although their analysis was for a period of 27 weeks, Pusateri Burroughs et al. (2006)
also found a high variation in size (153—173.3 ha) of home ranges in southwestern Lower
Michigan. In Michigan’s Upper Peninsula, fawns occupied home ranges of 12.6 ha with
little variation in size (range = 10.7—14.2) during the ﬁrst 2 months, although the
population studied was in a 252-ha enclosure containing a high deer population (5 107
deer; Ozoga et al. 1982). I found no other published results describing fawn space use
during their ﬁrst 2 months of life to make additional comparisons.

Although deer are often characterized as habitat generalists, the amount and
quality of cover for fawns should have some relationship with survival. Additionally,
because parturient and post-parturient female white-tailed deer are often territorial, and
an age-related hierarchy may exist in white-tailed deer (Marchinton and Hirth 1984), a
correlation between survival and cover types selected by fawns seems likely. The
potential for fawn recruitment to affect deer population dynamics provides support for
further study of this relationship. Certainly, high-quality habitat, by deﬁnition, would
provide for reduced predation risk as a consequence of increased resource (i.e., food,

cover) availability.

59

LITERATURE CITED

Ballard, W. B., H. A. Whitclaw, S. J. Young, R. A. Jenkins, and G. J. Forbes. I999.
Predation and survival of white-tailed deer fawns in northcentral New Brunswick.
Journal of Wildlife Management 63:574—579.

Bartush, W. S., and J. C. Lewis. 1981. Mortality of white-tailed deer fawns in the
Wichita Mountains. Proceedings of the Oklahoma Academy of Science 61:23—
27.

Beyer, Jr.., D. E., and J. B. Hauﬂer. I992. Diurnal versus 24-hour sampling of habitat
use. Journal of Wildlife Management 58:178—180.

Bowman, J. L., H. A. Jacobson, and B. D. Leopold. 1998. Fawn survival on Davis
Island, Mississippi, after an early spring ﬂood. Proceedings of the Annual
Conference of the Southeast Association of Fish and Wildlife Agencies 52:397—
402.

Brinkman, T. J., J. A. Jenks, C. D. DePemo, B. S. Haroldson, and R. G. Osborn. 2004.
Survival of white-tailed deer in an intensively farmed region of Minnesota.
Wildlife Society Bulletin 32:726—731.

Bunck, C. M., and K. H. Pollock. I993. Estimating survival of radiotagged birds. Pages
51—63 in J. C. Lebreton and P. M. North, editors. Marked individuals in the study
of bird populations. Birkhauser-Verlag, Basel, Switzerland.

Carroll, B. K., and D. L. Brown. 1977. Factors affecting neonatal fawn survival in
southem-central Texas. Journal of Wildlife Management 41:63—69.

Carstensen Powell, M., G. D. DelGiudice, and B. A. Sampson. 2005. Low risk of
marking-induced abandonment in free-ranging white-tailed deer neonates.
Wildlife Society Bulletin 33:643—655.

Cook, R. S., M. White, D. 0. Trainer, and W. C. Glazener. 1971. Mortality of young
white-tailed deer fawns in South Texas. Journal of Wildlife Management 35:47—
56.

Decker, T. A., W. M. Healy, and S. A. Williams. 1992. Survival of white-tailed deer
fawns in western Massachusetts. Northeast Wildlife 49:28—35.

Demarais, S., and P. R. Krausman. 2000. Ecology and management of large mammals
in North America. Prentice-Hall, Upper Saddle River, New Jersey, USA.

Diefenbach, D. R., C. O. Kochnanny, J. K. Vreeland, and B. D. Wallingford. 2003.

Evaluation of an expandable, breakaway radiocollar for white-tailed deer fawns.
Wildlife Society Bulletin 31 :756—761.

60

Downing, R. L., and B. S. McGinnes. I969. Capturing and marking white-tailed deer
fawns. Journal of Wildlife Management 33:711—714.

Franzmann, A. W. 1981. Alces alces. American Society of Mammalogists, Mammalian
Species 154:1—7.

Gaillard, J .-M., M. F esta-Bianchet, N. G. Yoccoz, A. Loison, and C. Toi'go. 2000.
Temporal variation in ﬁtness components and population dynamics of large
herbivores. Annual Review of Ecology and Systematics 31:367-393.

Haugen, A. O., and D. W. Speake. 1958. Determining age of young fawn white-tailed
deer. Journal of Wildlife Management 22:319—321.

Hensler, G. L., and J. D. Nichols. 1981. The Mayﬁeld method of estimating nesting
success-a model, estimators and simulation results. Wilson Bulletin 93:42—53.

Huegel, C. N., R. B. Dahlgren, and H. L. Gladfelter. 1985. Mortality of white-tailed
deer fawns in south-central Iowa. Journal of Wildlife Management 49:377-380.

Ivlev, V. S. 1961. Experimental ecology of the feeding of ﬁshes. Yale University Press,
New Haven, Connecticut, USA.

Jackson, R. M., M. White, and F. F. Knowlton. 1972. Activity patterns of young white-
tailed deer fawns in South Texas. Ecology 53:262—270.

Lent, P. C. 1974. Mother-infant relationships in ungulates. Pages 14—55 in V. Geist and
F. Walther, editors. The behaviour of ungulates and its relationship to
management. lntemational Union for the Conservation of Nature and Natural
Resources, Morges, Switzerland.

Lenth, R. V. I98la. Robust measures of location for directional data. Technometrics
23:77—81.

Lenth, R. V. l98lb. On ﬁnding the source of a signal. Technometrics 23: 149-154.

Lund, R. C. 1975. The capture and marking of wild, newborn white-tailed deer fawns.
Transactions of the Northeast Section of the Wildlife Society 32:25—33.

Madill, H., and W. Rustem. 2001. Michigan land resource project. Public Sector
Consultants, Lansing, Michigan, USA.

Marchinton, R. L., and D. H. Hirth. 1984. Behavior. Pages 129—168 in L. K. Halls (ed.).
White-tailed deer: ecology and management. Stackpole, Harrisburg,
Pennsylvania, USA.

61

Mayﬁeld, H. 1961. Nesting success calculated from exposure. Wilson Bulletin 73:255—
261.

Mayﬁeld, H. 1975. Suggestions for calculating nest success. Wilson Bulletin 87:456—
466.

Michigan Center for Geographic Information. 2001. IFMAP land-use, land-cover data.
http://www.mcgi.state.mi.us/mgdl/?rel=ext&action=sext (Accessed 28 Jul. 2005).

Michigan Department of Agriculture. 2002. Agricultural statistics: Jackson and
Washtenaw counties. http://www.michigan.gov/mda/ (Accessed 16 Aug. 2004).

Michigan Department of Natural Resources. 2005. 2005 deer population estimates for
south central management unit. Michigan Department of Natural Resources,
Wildlife Division Technical Report.

Moore, G. C., and G. R. Parker. 1992. Colonization by the eastern coyote (Cam's
latrans). Pages 23—37 in A. H. Boer, editor. Ecology and management of the
eastern coyote. Wildlife Research Unit, University of New Brunswick,
Fredericton, New Brunswick, Canada.

Nams, V. 0., and S. Boutin. 1991. What is wrong with error polygons? Journal of
Wildlife Management 55:172—176.

Nelson, T. A., and A. Woolf. 1987. Mortality of white-tailed deer fawns in southern
Illinois. Journal of Wildlife Management 51:326—329.

Nixon, C. M., and D. Etter. 1995. Maternal age and fawn rearing success for white-
tailed deer in Illinois. American Midland Naturalist 133:290—297.

Ozoga, J. J ., and R. K. Clute. 1988. Mortality rates of marked and unmarked fawns.
Journal of Wildlife Management 52:549—55 1.

Ozoga, J. J ., and L. J. Verme. 1986. Relation of maternal age to fawn-rearing success in
white-tailed deer. Journal of Wildlife Management 50:480—486.

Ozoga, J, J., L. J. Verme, and C. S. Bienz. 1982. Parturition behavior and territoriality in
white-tailed deer: impact on neonatal mortality. Journal of Wildlife Management
46:1-11.

Piccolo, B. P. 2002. Behavior and mortality of white-tailed deer neonates in suburban
Chicago, Illinois. Thesis, University of Illinois at Urbana-Champaign, Urbana,

Illinois, USA.

Pusateri Burroughs, J., H. Campa, III, S. R. Winterstein, B. A. Rudolph, and W. E.

62

Moritz. 2006. Cause-speciﬁc mortality and survival of white-tailed deer fawns in
southwestern Lower Michigan. Journal of Wildlife Management 70:743—75 1.

Saalfeld, S. T., and S. S. Ditchkoff. 2007. Survival of neonatal white—tailed deer in an
exurban population. Journal of Wildlife Management 71 :940—944.

Schwede, G., H. Hendrichs, and C. Wemmer. 1994. Early mother-young relations in
white-tailed deer. Journal of Mammalogy 75:438—445.

Seaman, D. E., J. J. Millspaugh, B. J. Kemohan, G. C. Brundige, K. J. Raedeke, and R.
A. Gitzen. 1999. Effects of sample size on kernel home range estimates. Journal
of Wildlife Management 63:739—747.

Short, H. L. 1964. Postnatal stomach development of white-tailed deer. Journal of
Wildlife Management 28:445—458.

Smith, W. P. 1991. Odocoileus virginianus. American Society of Mammalogists,
Mammalian Species 388: 1—1 3.

Sommers, L. M. 1977. Atlas of Michigan. Michigan State University Press, East
Lansing, Michigan, USA.

Strauss, R. E. 1979. Reliability estimates for Ivlev’s electivity index, the forage ratio,
and a proposed linear index of food selection. Transactions of the American
Fisheries Society 108:344—352.

US. Census Bureau. 2003. 2000 census of population and housing, population and
housing counts: Michigan. Washington, DC, USA.

Vreeland, J. K., D. R. Diefenbach, and B. D. Wallingford. 2004. Survival rates,
mortality causes, and habitats of Pennsylvania white-tailed deer fawns. Wildlife
Society Bulletin 32:542—553.

White, G. C., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data.
Academic Press, San Diego, California, USA.

Whittaker, D. G., and F. G. Lindzey. 1999. Effect of coyote predation on early fawn
survival in sympatric deer species. Wildlife Society Bulletin 27:256—262.

Winterstein, S. R., K. H. Pollock, and C. M. Bunck. 2001. Analysis of survival data
from radiotelemetry studies. Pages 351—380 in J. J. Millspaugh and J. M.
Marzluff, editors. Radio tracking and animal populations. Academic Press, San
Diego, California, USA.

63

Table 2.1. Fate assessment of fawn white-tailed deer (n = 34) during 0—60 and 0—180

days post-capture, south central Michigan, USA, 2004—2006.

 

 

 

Age (d) at assessmenta Proportion
Fate Eiss S60d SISOd
Mortality
Canid 33.7 i 11.1 0.09 0.12
Malnutrition 15.4 0.03 0.03
Vehicle collision 125.4 :t 26.4 0.00 0.09
Hunter harvest 192.4 i 5.5 0.00 0.03
Unknown 256.5 :t 240.7 0.03 0.03
Total 122.8 d: 33.0 0.15 0.30
Censored
Dropped transmitterb 354.7 i 54.6 0.00 0.06
End of study 233.2 d: 36.4 0.00 0.00
Other° 56.7 i 49.8 0.06 0.09
Total 259.4 :t 34.3 0.06 0.15
Alive after time interval 0.79 0.55

 

aAge was deﬁned as estimated age at capture plus accumulated days until fate was
assessed.

bFawns were radiomarked with break-away transmitters with an expected use of 9—1 2
months.

cIncluded mortality from capture stress and an injured fawn removed from study by a

wildlife rehabilitator.

64

 

 

 

 

2; 85 3o 6.2% 2%:

o _ .o mod! we; ovd Ed had $28 32.28% 253:
8.. oo. _I 94.0 and m _ .o mod cosmmoamm £5
8.. mmdl cud unzm mod cod @0363; 82:82
mod ondl mud VN; _m.o cod 98:25 352
mod 361 o: mm.m Nod cod 9.5? 2.2304
omd nod mm; mod mod Ed $28 283203 23354
Ed! @061 o _ .o h _ .o Rd cod was—:30 £88.38:
9: mud SM Ea: ovd m _ .o #5on
and! _m.ol n _ .m Ah: mod vmd 83.:otw<
40: $3 404 e\cmm mm m A3 AS 093 5.50

axooE cosoﬂom £5 :oEmoQEOo owcﬂ oEo: owcﬁ oEo: ~86 2393
E 8:80.:—

 

acomtaom 2:142 em; .5955 35:8 5:8 .oma a 2:55 To

£0823an wES—o Gm H 5 Loop 3:86:53 :Bﬂ cofanEE mo :ountomow Amowcﬁ 2:0: 3on .Xmav om: 8QO .N.~ 033.

65

$36: oocovccg 303V Loan: 98 3015.630. 09:05 was foo 3 3.3 so woman 803 $065 :ocoﬂomn
.33 >25 2: go 830 9:2 wEntomon 88m? Stash—£5 oEQEwoow a m5? 036 mm; EoEcmﬁmm cabs—38

new 335908 2.5.. mm 3282 203 $5 2:in .Qm u 5 2398-63 320:. mA 3:380 35 83:55:. 52¢ 32:05“

 

 

 

 

.550
cod cod cod
cant:
ovdl oo. T _ _.o o _ .o mod mod
153 $3 404 $3 mm m A3 A3 089 .550
3x02: Ego—om $5 :oEmomEoo owcﬁ oEo: owcﬁ oEo: ~36 2393

E 8:305

 

.5an .3 2.3

66

CHAPTER 3

A Multi-scale Approach to Describe Cover Selection of White-tailed Deer

Using Use-Availability Data

67

A Multi-scale Approach to Describe Cover Selection of White-tailed Deer

Using Use-Availability Data

INTRODUCTION

Selectivity, as described by Johnson (1980), relates to an animal’s use of some
resource (e.g., land-cover types) disproportionately to its availability; this assumes that
resources selected in a proportion higher than their availability are more important than
those resources selected against (i.e., “avoided”, or proportional use < proportional
availability). The assumption is that highly selected resources contribute positively to an
animal’s ﬁtness (Garshelis 2000), with the obverse assumed true for highly avoided
resources. But how availability is deﬁned (e.g., spatial extent), including whether it is
truly accessible to an individual or individuals, can impact how selection and avoidance
are assigned (Buskirk and Millspaugh 2006). Also, as the deﬁned spatial extent of
availability is reduced (i.e., coarse to ﬁne [landscape to home range] scale), selection may
become more difﬁcult to detect (McClean et al. 1998). Another potential problem is that
different statistical techniques selected by researchers may provide different selection
results (Garshelis 2000). “In many cases, interest will be in examining how habitat
selection changes as availability changes” (Alldredge and Griswold 2006:337), which
may provide insight into these potential problems.

By considering the hierarchical selection process used by organisms to meet their
habitat requirements (Johnson 1980, Morrison et al. 1992), researchers can make
inferences about habitat selection at multiple spatial scales; i.e., “the speciﬁcation of

habitat availability is equivalent to speciﬁcation of a spatial scale at which to study the

68

selection process” (Otis 1997:1016). Using a multi-scale approach while considering the
natural history of an organism should improve our understanding of the ecology of that
organism and lead to more informed and effective management decisions. For example,
within a landscape, what coarse-scale features do white-tailed deer (Odocoileus
virginianus) select (i.e., home-range characteristics), and what landscape features
potentially affect population dynamics (e.g., avoidance of roads or urbanized areas, land-
ownership patterns)? Within the home range, what habitat components do deer select
(e.g., within-patch characteristics), and what is the relationship between these
components and deer demographics (e.g., deer with adequate thermal cover have higher
winter survival)? If we assume deer generally avoid urbanized areas within a landscape
(coarse-scale approach), but deer exist locally in high densities within certain urbanized
areas, then what are the differences in resource selection, movements, demographics, and
cause-speciﬁc mortality between urban and rural areas? How should urban deer be
managed differently from other deer populations? The patchy distribution of a species at
various scales, from geographical distribution to metapopulation to family unit and
ﬁnally, to individual, may be best examined by varying deﬁnitions of resource use and
availability at different spatial scales. These deﬁnitions are, of course, human-derived.
Suggestions to reduce errors of assignment while conducting resource selection
studies have been proposed, but often have not been implemented. For example,
“researchers should consider studying selection at more than one scale” (Manly et al.
20025), but published research does not always address this consideration. Otis (1997)
suggested that multi-scale approaches describing resource selection seemed to be

becoming more prevalent; however, Thomas and Taylor (2006), in reviewing methods

69

used in 87 papers in The Journal of Wildlife Management during 2000—2004, found that
57% considered only 1 spatial scale to assess resource selection. With highly speciﬁc
research objectives, a single-scale study may be appropriate, however.

One potential method to address selection at various spatial scales is to vary the
deﬁnitions of resource use and availability (Table 3.1). Challenges exist when deﬁning
resource availability; often, availability is most easily deﬁned based on political
boundaries or management units that may or may not consider ecological processes and
animal movements. Deﬁning study-area boundaries under these conditions may produce
potentially spurious results depending on landscape-patch arrangement (Porter and
Church 1987). Considering a multiple-scale approach would seem appropriate to provide
support for management decisions based on selection.

My objectives were to quantify and compare cover selection of adult female
white-tailed deer in south central Michigan at multiple spatial scales, and attempt to
describe selection at multiple scales based on landscape-scale characteristics (e.g., patch
size). My decision to use only adult females was related to describing relatively high-
quality habitat for deer, as pooling age classes or considering only animal densities may
confound issues related to habitat quality (Van Home 1983), and therefore, resource
selection. I assumed that habitat quality will be higher for adult deer than forjuveniles
because of a social dominance hierarchy (Marchinton and Hirth 1984), and that adult
females may select areas with higher quality forage outside of the breeding season in
comparison to males (Stewart et al. 2003). Further, evidence suggests that adult female
white-tailed deer have higher reproductive success than juvenile females (e.g., Ozoga and

Verme I986, Mech and McRoberts 1990), and females are often the sex class of primary

70

concern for managers desiring relatively large reductions in population densities through
strategies such as hunter-harvest (Carpenter 2000).
STUDY AREA

My study was conducted in eastern Jackson (Grass Lake, Henrietta, and Waterloo
townships), western Washtenaw (Dexter, Lima, Lyndon, and Sylvan townships), and
southwestern Livingston (Unadilla township) counties in south central Lower Michigan.
The study-area boundaries were deﬁned speciﬁcally to include all townships containing
21 radiomarked deer location estimate, a method based on animal distribution suggested
by McClean et a1. (1998) and Erickson et al. (2001) to avoid the somewhat arbitrary
deﬁnition of study-area boundaries. The study area (82,636 ha) included publicly owned
lands (17.8% of study area), including the Michigan Department of Natural Resources
(MDNR) Waterloo (8,410 ha; 10.2%) and Pinckney (4,276 ha; 5.2%) recreation areas,
and privately owned lands (82.2%). South central Michigan has been characterized by an
increase in urbanization and development, and a relatively high deer density (~27/km2
during fall 2005; MDNR 2005). Little scientiﬁc information exists about deer under
these landscape conditions. Although the study area was primarily rural (98% of total
land area), the human population increased 16% and housing units increased 22%
between 1990 and 2000 (US. Census Bureau 2003). Much of the landscape throughout
southern Michigan (and in other areas throughout the Midwest) is expected to experience
increasing land-use activities similar to the study area (Madill and Rustem 2001).

The physiographic regions of south central Michigan are Hillsdale-Lapeer Hilly
Upland, South Central Rolling Plain, and Southeastern Rolling Plain (Sommers 1977:24)

with alﬁsols as the major soil order (Sommers 1977:36). Surface formations in the study

71

area are the result of glaciation and include all 4 types present within Michigan (moraine,
till plain, outwash plain, and lacustrine plain; Sommers 1977232). Elevation of the study
area ranged from approximately 180 to 300 m and consisted of relatively limited relief
(Sommers I977 :26, 33). The study area received about 81 cm of precipitation annually
during 1971—2000 (based on conditions in Chelsea, Michigan; Midwestern Regional
Climate Center, Champaign, Illinois, USA), and had a ISO-day growing season (i.e., the
average annual accumulation of daily mean temperatures >5.6° C), generally occurring
from 10 May to 7 October (Sommers 1977:46, 49). During 1971—2000, average annual
snowfall was 99.3 cm and mean monthly temperatures ranged from —5.4 °C (Jan) to 21.8
°C (Jul) in Jackson County (Midwestern Regional Climate Center, Champaign, Illinois,
USA). Annual snowfall during my study was highly variable (2004 = 90.9 cm, 2005 =
149.9 cm, 2006 = 40.0 cm; Chelsea, Michigan; Midwestern Regional Climate Center,
Champaign, Illinois, USA).

Before European settlement, southern Michigan forests consisted of oak (Quercus
spp.) and hickory (Carya spp.) in well-drained soils and beech (Fagus grandifolia), elm
(Ulmus spp.), maple (Acer spp.), and basswood (T ilia americana) in poorly drained soils
(Sommers 1977:17). Much of the study area is well suited for agriculture (Sommers
1977:38). Most crop agriculture in Jackson and Washtenaw counties (total area =
366,483 ha) is com (37,840 ha) and soybean (34,200 ha) production, followed by hay
(17,200 ha), winter wheat (7,970 ha), and oats (930 ha; Michigan Department of
Agriculture 2002).

I generalized land-use, land-cover data (Michigan Center for Geographic

Information 2001) using ArcView GIS v3.2 software (Environmental System Research

72

Institute, Redlands, California, USA) and Spatial Analyst extension to deﬁne 13 primary
cover types within the study area: agriculture (non-vegetated farmland, row crops, forage
crops; 52.3% of study area); conifer (pines [Pinus spp.], other upland conifers; 1.5%);
herbaceous openland (herbaceous vegetation with <25% woody cover; 2.9%); lowland
deciduous forest (>60% composed of lowland deciduous tree cover; 8.0%); lowland
shrub (with >60% non-water cover; 9.9%); mixed wetland (ﬂoating aquatic vegetation,
emergent wetland, mixed non-forest wetland; 3.1%); northern hardwood (>60% canopy
cover of maple, beech, ash [F raxinus spp.], cherry [Prunus spp.], birch [Betula spp.];
2.3%); oak association (>60% canopy cover of oak; 1.6%); upland deciduous forest
(>60% canopy cover of upland deciduous trees; 11.6%); upland shrub (>25% woody
cover; <0.1%); urban (low and high intensity, roads, parks, golf courses; 2.8%); water
(surface, ﬂowing; 3.9%); and other (aspen [Populus spp.] association, orchards, bare
ground; 0.1%). Publicly owned lands had different proportions of cover types in relation
to the study area (e.g., public lands have about twice the proportion of conifers and
upland deciduous forests, but one-forth the proportion of agriculture). Patch size of cover
types on the study area ranged from <1 ha to >1 1,000 ha (e.g., an agricultural matrix) and
had a mean size of 29.2 ha (median = 4.7 ha). Consequently, generalization of spatial
data may have excluded certain ﬁne-scale landscape characteristics (e.g., hedgerows,
roads) in some instances.
METHODS
Capturing Deer

Technicians and I trapped deer during winter (Dec—Mar) 2004—2006, using

single-door live traps (Clover 1954). Traps were placed near areas of deer activity, baited

73

with kernel corn, and checked twice/day to minimize stress and injury to deer. We
restrained captured deer using the collapsed trap and the body weight of 1 person
(Sparrowe and Springer 1970). All trapped deer were blindfolded to reduce stress and
ﬁtted with metal ear tags bearing a unique identiﬁcation number (Style 681; National
Band and Tag, Newport, Kentucky, USA). Individuals were aged as fawn (<1 year),
yearling (2 1 year—<2 years), or adult (2 2 years) through general morphometric
differences (e.g., shape and size of head, body size) and dental characteristics
(Severinghaus 1949); ages of necropsied individuals later conﬁrmed accuracy of ﬁeld
observations.

Only female deer were ﬁtted with mortality-sensing collar-style radio-
transmitters, either with VHF (Model 500; Telonics, Inc., Mesa, Arizona, USA) or VHF -
GPS (Model G2000; Advanced Telemetry Systems, Isanti, Minnesota, USA) capabilities.
I did not radiomark male deer because I expected low capture success for adult males
based on previous trapping experience and deer behavior. Additionally, winter-captured
male deer were not radiomarked because of potential problems associated with ﬁtting a
radio-transmitter to deer to accommodate their physiological changes (i.e., neck-swelling)
during the breeding season. Radio-transmitters had a unique frequency within the 150-
MHz range, a mass of 270 g (VHF) or 1,100 g (VHF -GPS), and an expected minimum
battery life of 36 months (VHF) or 12 months (VHF-GPS). Michigan State University’s
All-University Committee on Animal Use and Care approved all capturing and handling

procedures for my study (Application No. 01/04-006-00).

74

Estimating Locations

Deer locations were estimated 2—5 times/week using triangulation methods
associated with telemetry (White and Garrott 1990) or from visual observations of known
individuals. To increase the potential of my analyses to accurately describe cover-
selection behavior, deer were located at varying time schedules in a systematic manner on
a diel basis, with 21 nocturnal location/deer/week except during capture periods; for
species (including deer) that potentially move at any time during a 24-hour period,
“management recommendations developed from habitat use data collected from only a
portion of a 24-hour period may be ineffective” (Beyer and Hauﬂer 1992:180). Bearings
were estimated using a 3-e1ement folding Yagi antenna (Advanced Telemetry Systems,
Incorporated, Isanti, Minnesota, USA), portable radio receiver (Model R-1000,
Communications Specialists, Incorporated, Orange, California, USA), and mirror-
sighting compass. A global positioning system (GPS) handheld unit (Model GPS IV;
GARMIN lntemational, Incorporated, Olathe, Kansas, USA) was used to approximate
the locations from which signals were received.

Locations from triangulated data were estimated using the program LOCATE III
(Pacer, Truro, Nova Scotia, Canada). I used the maximum likelihood estimator (Lenth
1981a,b), as recommended by White and Garrott (1990) and Nams and Boutin (1991).
Bearing standard deviation and error ellipses were estimated using LOCATE III for each
location during analysis. I based telemetry error assessments on the relationship between
mean error-ellipse size and mean landscape-patch size to determine if I had appropriate

sample sizes (see Nams I989).

75

Cover Selection

Female deer that were captured as fawns or yearlings were reclassiﬁed as adults
on either the ﬁrst (for yearlings) or the second (for fawns) 1 June following their capture
and subsequently used in my analyses. I assumed that all location estimates that were
classiﬁed in the cover type water were inaccurate and l relocated each to the nearest
alternative cover type. Cover selection was determined seasonally based on agricultural
crop production (i.e., the growing season [150 d; 10 May—7 Oct], and the non-growing
season [215 d; 8 Oct—9 May]) in the study area. I assumed these 2 time periods
approximated seasonal differences in resource (e.g., food) availability and certain
behaviors (e.g., parturition and primary fawn-rearing during the period of high food
availability during the growing season, breeding and fasting during the period of low
food availability during the non-growing season). Because there was low use of some
cover types (e.g., proportional use of 6 of 13 cover types in 95% kernel home ranges
during the growing season was <0.02), 1 did not consider compositional analysis as an
appropriate method to describe resource selection (Thomas and Taylor 2006).

The hierarchical selection process proposed by Johnson (1980) described 4
resource selection scales, from coarse-scale (ﬁrst-order selection: geographical range) to
ﬁne-scale (fourth-order selection: selection of particular food items at a site). This
hierarchy describes a continuum of habitat selection, where ﬁrst-order selection implies
coarse-scale habitat requirements are present based on the geographic distribution of a
species (e.g., a comparison of current and past land-cover conditions to describe the
increase in distribution and abundance of deer [see McCabe and McCabe 1984]); fourth-

order selection may describe ﬁne-scale habitat components (e.g., characteristics of

76

bedding sites selected by fawns; Huegel et al. 1986) that may contribute to increased
ﬁtness of an individual. Studies including intermediate orders of selection are perhaps
the most common in wildlife science as a result of data collection methods and
management objectives.

1 described second- (e.g., home range selection within the study area) and third-
order (e.g., cover selected within a home range) selection (Table 3.1) for deer, as my data
were not appropriate to describe ﬁrst- and fourth-orders of selection. I used SYSTAT
v11 (Systat Software, Inc., San Jose, California, USA) for my data analyses and followed
study designs 1—3 of Thomas and Taylor (1990) to describe cover selection processes of
deer and the magnitude of their differences at different spatial scales. I assumed, for
example, that if a cover type was highly selected at multiple spatial scales, then its
relative importance to deer was higher than a cover type selected at 1 spatial scale; other
factors may also affect cover selection, however, such as the distribution of hunters or
availability of private and public land on the study area. Unless noted otherwise, I used
95% conﬁdence limits (CLs; LCL = lower, UCL = upper) in my statistical analyses. The
use of conﬁdence limits is advantageous in that an estimate of effect size and a measure
of uncertainty are provided (Johnson 1999).

Design 1: Population-level use and availability—Use and availability under
design I of Thomas and Taylor (1990) were quantiﬁed using Euclidean distances to
deﬁne use and availability of cover types on my study area. Although this analysis
method is most often used for linear or point features, it is equally valid for spatial
features such as cover types (i.e., features described as polygons; Conner and Plowman

2001). I pooled all location data (i.e., individuals were not identiﬁed) and determined the

77

Euclidean distance between each location to each of the nearest cover types within my
classiﬁcation system. To describe availability, within the study-area boundaries I used
1,000 randomly generated points from a uniform distribution and measured Euclidean
distances using methods identical to those used for location estimates.

Design 2: Individual use, population-level availability—Use is estimated for
each individual while availability for all individuals is identical under design 2, the
seemingly dominant approach in selection studies (Thomas and Taylor 1990). I deﬁned
availability as the composition of cover types within the study-area boundaries and I used
95% ﬁxed-kernel home ranges to assess selection under this study design. Kernel home
ranges were estimated using the Animal Movement extension (Alaska Biological
Sciences Center, Glacier Bay, Alaska, USA) in ArcView GIS v3.2 software. I used only
deer with 2 30 locations/seasonal home range to describe cover selection (Seaman et al.
1999). I also estimated the amount of publicly owned land within seasonal home ranges.
I used zero-intercept linear regression models to determine if there was a relationship
between home-range size and amount of public land within home ranges.

The ﬁxed-kernel method was chosen over other estimators for several reasons.
First, the kernel methods (fixed and adaptive) met more criteria of importance (e.g.,
nonparametric estimation, sensitivity of outlying locations) than other home-range
estimators (Kemohan et al. 2001:132—140). Secondly, the ﬁxed-kernel method has lower
bias when estimating the outer contours of the utilization distribution when compared to
the adaptive-kemel method (Seaman et al. 1999). This is due to the smoothing parameter
(h) being held constant with the ﬁxed-kernel method as opposed to a variable h with the

adaptive kernel method (i.e., a smaller h in areas of more dense utilization and a larger h

78

in areas with less dense utilization; Worton 1989, Powell 2000). I used the least-squares
cross-validation method for smoothing-parameter selection (Worton I995, Seaman et al.
1999)

Design 3: Individual use and availability—Both use and availability are deﬁned
at the individual level with design 3 (Thomas and Taylor 1990). I deﬁned use as the
proportion of location estimates of an individual within each cover type, and availability
as the proportional area of each cover type within the 95% kernel home range of that
individual. Design 3 seems an appropriate measure of availability at ﬁner spatial scales
because kernel home ranges may overestimate the space used by an animal (i.e., cover
types containing no location estimates are often included in a home-range estimate;
Guthery et al. 2005). Consequently, I developed 2 methods to characterize availability of
cover types within kernel home ranges: conditional and unconditional presence of cover
types. Conditional analyses excluded cover types not present within an individual’s
home range, while unconditional analyses included all cover types within the study area
for the estimation of selection indices.

Selection Indices.—I used Ivlev’s (I961) electivity index as a measure of cover
selection for each individual and calculated the mean for each cover type by season. This
ratio provides an index ranging from -I (implying avoidance; proportion used <
proportion available) to 1 (implying selection; proportion used > proportion available),
with 0 (proportion used = proportion available) suggesting random use. Conﬁdence
limits for Ivlev’s index were truncated at -I (LCL) or +1 (UCL) when appropriate. I
compared selection within and among study designs based on the conﬁdence limits of

selection indices (see below) to investigate cover use by deer at multiple spatial scales.

79

Because point estimates alone may not provide an accurate estimate of resource
selection (Hobbs and Bowden 1982), I based selection assignment (cover types selected,
avoided, or randomly used) on conﬁdence limits. Conﬁdence limits were calculated
using individuals as the sample unit (data were pooled by season so that a deer may have
been included in 21 growing or 21 non-growing season). For design I, conﬁdence
limits based on the means and standard errors of Euclidean distances were used to assign
selection, but to compare selection with designs 2 and 3, I calculated Ivlev’s electivity
index based on ratios of mean Euclidean distances. Because using mean Euclidean
distances as a ratio to assign selection results in mathematical signs opposite (i.e.,
negative values imply selection) to the results normally obtained using other analytical
methods, I reversed the mathematical signs of the selection indices of design I to remain
consistent with selection assignment under designs 2 and 3. For each cover type within
each season under design 1, I assigned selection (used UCL < random LCL), avoidance
(used LCL > random UCL), and random use (overlapping CLs) based on conﬁdence
limits of Euclidean distances.

I followed Strauss (1979) to estimate conﬁdence limits of selection indices under
designs 2 and 3. Under design 3, I assessed selection based on conditional and
unconditional analyses; conditional analysis included only those cover types present
within an individual’s home range, while unconditional analysis included all 13 cover
types (i.e., a non-represented cover type had a proportional availability = 0).

Ranking Cover T ypes.—I used lvlev’s electivity indices to rank selection both
within and among study designs by season. This ranking method would probably be best

described as a relative ranking system as opposed to an absolute rank of cover types (see

80

Johnson 1980). Speciﬁcally, “Absolute statements about preference or avoidance should
be guarded against. Relative statements are possible because their nature invokes the
concept of selection order” (Johnson 1980:69). Essentially, a relative ranking system
does not make assignments of selection, avoidance, or random use for cover types, but
rather ranks cover types based on selection index values. This may be advantageous
when comparing indices derived from use-availability data collected or analyzed using
different methods.

1 ranked cover types within a study design and growing season by selection
indices (i.e., the cover type with the highest selection index within a set received a rank
of 1, and so on). Cover types with identical selection indices within a design and season

were assigned identical ranks. To assess overall relative importance, 1 assigned the

average rank (E) among study designs for each cover type i within a season based on

the equation

Fr: [design 1 + design 2 + (0.5 x design 3U) + (0.5 x design mm,

which incorporated the weighted average of the 2 design 3 methods (U = unconditional,
C = conditional). Ranking of cover types among designs was ordered from the lowest 7?:
(assigned the rank of l) to the highest E (assigned the highest rank value within the set
of cover types). Cover types with identical values of Ri were assigned the same rank.
For example, if cover types I and 2 each had 7?: = 5 (i.e., E = R; = 5), and there were 3
cover types with 73 < 5, cover types 1 and 2 were both assigned the rank of 4; if cover
type 3 had the next highest E , it was assigned a rank of 6 because there were 5 cover

types with E < E.

81

RESULTS
Capture and Estimation of Locations

I captured and radiomarked a total of 42 female deer during 3 winter seasons. A
subset of 20 radiomarked deer that were either aged as adult during capture or that moved
into the adult age class during the study was available for analysis (i.e., individuals with
230 locations; Seaman et al. 1999). This subset contained a total of 3,493 location
estimates and a mean of 71 locations/seasonal home range. I pooled data by season but
not by year due to small sample sizes. The growing seasons of 2004, 2005, and 2006
included 7, l4, and 3 deer, respectively. The non-growing seasons of 2004, 2005, and
2006 included 7, 11, and 7 deer respectively. Five deer were included in similar seasonal
categories for >1 year.

I pooled location estimates (3,493) from all adult deer (n = 20) from my study to
estimate overall telemetry error. I considered telemetry error acceptable, and thus,

location estimates of acceptable precision, given the landscape characteristics of the study
area (e.g., ; = 29.2 ha/patch), the mean telemetry error-ellipse size (10.2 ha), and the

number of location estimates (J-c = 71/deer). This conclusion was based on the
relationship between telemetry error and patch size (diameter ratio = 0.59) to determine
appropriate sample sizes as described by Nams (1989).
Cover Selection

Design I: Population-level use and availability.—Under design 1, selection was
assigned for 8 and 9 cover types for the growing and non-growing seasons, respectively
(Table 3.2). Selection indices for both seasons ranked conifers and upland deciduous

forests as the most highly selected (each I > 0.40), while agriculture was the most highly

82

avoided cover type. Although indices differed somewhat between seasons, the patterns
of selection were similar when assessed under this design (Table 3.3). Two cover types
moved 1 rank value (lowland deciduous forest, other), while 1 cover type (oak
association) moved from a rank of 8 during the growing season to a rank of 6 during the
non-growing season. All other cover types were ranked consistently between seasons,
suggesting no change in relative selection under this study design.

Design 2: Individual use, population-level availability. —The cover types upland
shrub and other were absent (i.e., received no use) from kernel home ranges under design
2; consequently, I considered both as highly avoided cover types during both seasons
(Table 3.4). Urban areas were the most highly avoided, but used, cover type during both
seasons. Design 2 analysis assigned fewer cover types as selected (i.e., based on my
deﬁnition of selection) in comparison to design I. Conifers were highly selected, while
lowland shrub and upland deciduous forest were moderately selected, with similar
selection indices for all 3 cover types during both seasons. Ranking of cover types by
season was similar under design 2 (Table 3.3). Herbaceous openlands and water each
changed in their relative importance between seasons by a value of 2; herbaceous
openlands became more important and water less important during the non-growing
season. Although locations estimated to be in water were moved to the nearest
alternative cover type, water was often included within kernel home-range boundaries.

When I plotted amount of public land within home ranges (ha) as a function of
seasonal home-range size (ha), my models for both seasons showed a positive linear
relationship. For the growing season, the relationship was described by y = 0.54x (95%

CL = 0.41—0.68 for x-coefﬁcient; r = 0.87), while for the non-growing season, the

83

relationship was y = 0.52:: (95% CL = 0.40—0.64; r = 0.88). Because 95% CLs
overlapped for the x-coefﬁcients, I pooled all data, which resulted in a relationship
described byy = 0.52x (95% CL = 0.44—0.61; r = 0.87; Fig. 3.2).

Design 3: Individual use and availability.—Unconditional analysis under design 3
assigned low but positive values for selection to 1 cover type during each season (upland
deciduous forests for growing, conifers for non-growing; Table 3.5). All other cover
types using conditional analysis were randomly used. The relative importance of cover
types changed substantially between seasons (Table 3.3). Herbaceous openland, mixed
wetland, and urban greatly decreased in importance (i.e., rank decreased 24) during the
non-growing season; lowland shrub, oak association, and northern hardwood increased in
importance (i.e., rank increased 23) during the non-growing season.

Similarly, under conditional analysis, proportional use was greater than
proportional availability for upland deciduous forests (growing) and conifers (non-
growing) under design 3; all other cover types were used randomly (Table 3.6). The
difference in relative importance between seasons (growing minus non-growing) was
greatest for urban (-9), oak association (+7), and mixed wetland (-4); moderate for
herbaceous openland (-3), lowland shrub (+3), and northern hardwood (+3); all other
cover types had a difference in rank of S 2.

Ranking Cover T ypes.—Using my equation to estimate the mean rank among the
study designs, most (77%) cover types had a similar rank between seasons (i.e., rank
changed 5 1; Table 3.3). During the growing season, mixed wetland (+3) and urban (+2)
increased in relative importance, while oak association decreased (-3) in relative

importance. Conifers and upland deciduous forests were the 2 most important cover

84

types irrespective of season, while agriculture, other, upland shrub, and urban generally
were of low relative importance. Although use and availability of water were somewhat
inconsistently deﬁned among study designs, it seemed to have little effect on relative
importance of cover types.
DISCUSSION

Garshelis (2000) discussed 2 assumptions relevant to selection studies, both
seemingly related to human perception of resource selection by animals. How humans
perceive resource availability could be different from how the species under study
perceives availability (Litvaitis et al. 1996). Researchers may never fully understand
these differences in perception, but this difference should be considered while designing
studies (e.g., hierarchical selection process by animals [Johnson 1980]) and as a potential
problem when interpreting results. Results that misrepresent the biology of an organism
or the ecological processes throughout a landscape (e.g., vegetation succession) may lead
to both cover assignment and management errors, especially under habitat manipulation
strategies at the species level. If resource selection patterns emerge over multiple spatial
scales, it would be logical to assume strong evidence exists for selection of certain
resources. Conversely, depending on management objectives, we may question the
efﬁcacy of management decisions based on a single spatial scale, if that scale does not ﬁt
selection patterns.

The use of Johnson’s (1980) system of ranking has been implemented in several
ungulate studies. Lopez et al. (2004) used a multi-scale process (ﬁrst-, second-, and
third-order selection) to examine habitat use by Florida Key deer (0. v. clavium) in an

increasingly urban environment; they found that Key deer generally selected for upland

85

vegetation types regardless of spatial scale, and hypothesized that uplands provided
preferred foods, cover, and freshwater. In Oregon, the relative importance of 11 plant
communities were ranked and compared seasonally based on the feeding activities of
mule deer (0. hemionus; individual animals were not identiﬁed) and the relative
importance of each plant community varied substantially by season (Bodurtha et a1.
l989:table 2). These examples provide evidence that resource selection studies should
consider the selection behaviors and natural history of the species of interest. Failure to
do so may provide an incomplete understanding of selection processes, resulting in less
effective management.

Under the hierarchical selection process, selection at ﬁner scales is dependent on
selection at more coarse scales (Johnson 1980). To illustrate, third-order selection (e.g.,
the selection of conifers as thermal cover within a home range) is dependent on second-
order selection (e.g., the selection of conifers across a landscape). My analyses showed
strong patterns of selection regardless of spatial scale, but there were also some
inconsistencies among certain cover types based on changes in their relative importance
across multiple spatial scales. Conifers and upland deciduous forests were ranked as the
2 most important cover types on my study area regardless of the spatial (i.e., study
design) or temporal (i.e., season) scale studied. Several cover types changed their
relative importance across spatial scales. For example, during the growing season, urban
areas shifted from unimportant at the landscape scale, but were increasingly important as
a habitat component within home ranges; urban areas were unimportant during the non-
growing season regardless of spatial scale. Deer near urban areas may have utilized these

areas as fawning cover, although this is based on conjecture.

86

Single-scale Studies

I based my interpretation of selection patterns on past research, but white-tailed
deer habitat is diverse across their geographic distribution, making comparisons
somewhat difﬁcult, especially as spatial scale decreases. Also, cover-type classiﬁcations
are not consistent among studies, further increasing the difﬁculty of comparisons,
especially across the geographic distribution of the white-tailed deer.

In Arkansas, Miranda and Porter (2003ztable 1) used 2 general habitat suitability
classes (food and cover) to model landscape-scale habitat suitability based on cover
types. Although they did not speciﬁcally deﬁne cover, I assumed they primarily
considered security as opposed to (winter) thermal cover given the mild climate of their
study area. Their food and cover suitability values (an index of 0 to I, with 0 being
unsuitable) were both high for shrublands, deciduous forests, low-intensity residential,
and woody wetlands; evergreen forests provided primarily cover, grassland-herbaceous
and agricultural cover types provided primarily food; and water, high-intensity
residential, and various other cover types were generally considered unsuitable.
Although I found lowland shrubs to be relatively important during the non-growing
season (Table 3.3), this cover type was not as important on my study area as suggested by
Miranda and Porter (2003), and I believe that conifers (e.g., eastern redcedar [Juniperus
virginiana]) on my study area contributed cover and some winter food value (see Bender
and Hauﬂer 1994) based on their relative importance (Table 3.3); the importance of
upland deciduous forests (food and cover) on my study area seemed consistent with

Miranda and Porter (2003) at the landscape level.

87

In Midwestern agricultural areas, white-tailed deer use agricultural crops
throughout the year (Gladfelter I984). Regardless of season, 1 found that agricultural
areas increased in relative importance as spatial scale decreased (i.e., scale of selection
became ﬁner). This suggests that agricultural areas were much less important to deer at
the landscape scale in comparison to providing a habitat component (food) within deer
home ranges. This may explain why crop damage by deer seemed to be localized near
ﬁeld edges bordered by cover, at least for larger ﬁelds; smaller ﬁelds may have crop
damage throughout (K. Bissell, MDNR, personal communication). Approximately half
of the study area was composed of agricultural areas, which was probably at a much
higher proportion than to provide optimal conditions (see Chapter 4), which reduced the
relative importance of this cover type.

I estimated selection indices under design 2 using data from Pusateri (2003) for
white-tailed deer 26 months old in southwestern Lower Michigan for comparison to my
results. Cover types were deﬁned differently for my study and data were not pooled by
season. Cover-type compositions between study areas were similar, but mean patch size
differed (20.3 ha for southwestern Michigan, 29.2 ha for my study), suggesting that other
landscape characteristics (e. g., amount of edge) may also have been different between
study areas. Kernel home ranges (ranked by Ivlev’s electivity index) included the cover
types evergreen forest (0.64), emergent herbaceous wetland (0.55), woody wetland
(0.23), deciduous forest (0.20), and agriculture (—0.16); all other cover types were not
used by deer. The only strong similarity between Pusateri (2003) and my study was the
relative importance of conifers (evergreen forests), although deciduous forests also

comprised a large portion (47%) of kernel home ranges (i.e., proportional use) for

88

Pusateri (2003) and my study (~28% during each season). Wetland cover types seemed
to have a higher relative importance to deer in southwestern Michigan than in my study.
Issues of Scale

Cover selection of deer on my study area showed scale effects. From coarse to
ﬁne scale, or as the spatial extent of availability declined, fewer cover types were
assigned as selected (i.e., proportional use > proportional availability), similar to
McClean et al. (1998). Eight cover types were assigned as selected under design 1
(landscape scale), while 3 were assigned as selected under design 2 (meso-scale), and 2
for each variant of design 3 (ﬁne scale). My landscape-scale deﬁnition of availability
encompassed the entire study area. Movements of individual radiomarked deer on my
study area were limited to much smaller areas than the entire study area, suggesting that
resource availability may have been overestimated.

Design 2 and design 3 (conditional and unconditional) seemed most appropriate
to describe cover selection by deer on my study area, as selection was fairly consistent
among designs. Considering our biological knowledge of white-tailed deer, these designs
supported my expectations of cover use by deer. For example, under design 2, conifers,
lowland shrubs, and upland deciduous forests were selected regardless of season, perhaps
as a result of the proximity of cover types. Under each variant of design 3, conifers were
selected during the non-growing season and upland deciduous forests were selected
during the growing season.

Human perceptions of wildlife cover selection, as deﬁned through use and
availability, may also be affected by landscape characteristics. For example, landscape

characteristics (e.g., patch size, shape, and distribution across a landscape) probably

89

affect the distribution of wildlife species, such as white-tailed deer, in a given area (Porter
and Church 1987). The landscape matrix of my study area consisted of few large patches
(i.e., >120 ha) of agricultural areas (i.e., too few to be identiﬁed by frequency in Fig. 3.1),
which probably were in excess quantity relative to the space-use needs of white-tailed
deer (see Chapter 4). Consequently, at the landscape scale, deer likely avoided large
patches of agriculture. Given that publicly owned areas had about 25% of the proportion
of agriculture (and twice the proportions of conifers and upland deciduous forests) than
privately owned lands, habitat quality for deer may be higher on public lands in the study
area. However, home-range size did not seem to decrease with increasing amounts of
public land (Fig. 3.2), suggesting that parcel sizes of or the patches within public land
may have been too small to reduce deer movements.
Conclusions

1 used use-availability data of white-tailed deer to illustrate how and why a multi-
scale approach (i.e., various methods describing use and availability) can be used to
describe cover selection. If, for example, managers considered selection only under
design 1, they might underestimate the importance of lowland deciduous forest (Table
3.3) when managing for white-tailed deer. Similarly, the oak association cover type may
show no difference in relative importance between seasons (under design 2), but the
relative importance as deﬁned through a multi-scale analysis could show a large
difference among spatial scales (Table 3.3) that may not otherwise be considered.

“Inference of selection patterns is limited to the scale designated by the
researcher” (Erickson et al. 2001:217). Vreeland et al. (2004:542) found no relationship

between home-range characteristics and survival of fawns in Pennsylvania, but

90

mentioned that, “future studies should consider landscape-related characteristics on fawn
survival,” suggesting that a more coarse spatial scale (i.e., describing landscape
characteristics such as patch size and juxtaposition) may have been more appropriate;
interestingly, Erickson et al. (2001) suggested that generalist species, which white-tailed
deer seem to be, might select resources at spatial scales ﬁner than the landscape level.
These seemingly contradictory statements seem to support the multi-scale resource
selection approach for determining the appropriate spatial scale or scales of study.

I suggest using multiple spatial scales when assessing resource selection,
assuming appropriate use-availability data were collected. This should improve the
interpretation of resource selection analyses through stronger evidence of selection or
avoidance through a comprehensive description of cover use. When researchers are
unsure of the appropriate scale of study for an application, they may do well using a
multi-scale selection analysis, ranking cover according to relative importance, then
averaging the rank values as I have done. Relative importance of cover types across
multiple scales should provide insight into cover selection, and therefore value of cover,
for habitat management. Errors of assignment (e.g., effects of inappropriately deﬁned
resource availability) may also be minimized and overall selection patterns should

emerge through such an approach.

91

LITERATURE CITED

Alldredge, J. R., and J. Griswold. 2006. Design and analysis of resource selection

studies for categorical resource variables. Journal of Wildlife Management
70:337—346.

Bender, L. C., and J. B. Hauﬂer. 1994. A white-tailed deer HSI for the Upper Great
Lakes region. Pages 121—1 51 (Appendix D) in G. J. Roloff. Using an ecological
classiﬁcation system and wildlife habitat models in forest planning. Dissertation,
Michigan State University, East Lansing, Michigan, USA.

Beyer, Jr., D. E., and J. B. Hauﬂer. I992. Diurnal versus 24-hour sampling of habitat
use. Journal of Wildlife Management 58:178—180.

Bodurtha, T. S., J. M. Peek, and J. L. Lauer. 1989. Mule deer habitat use related to
succession in a bunchgrass community. Journal of Wildlife Management 53:314—
319. ..

Buskirk, S. W., and J. J. Millspaugh. 2006. Metrics for studies of resource selection.
Journal of Wildlife Management 70:358—366.

Carpenter, L. H. 2000. Harvest management goals. Pages 192—213 in S. Demarais and.
P. R. Krausman, editors. Ecology and management of large mammals in North
America. Prentice Hall, Upper Saddle River, New Jersey, USA.

Clover, M. R. 1954. A portable deer trap and catch-net. California Fish and Game
40:367—373.

Conner, L. M., and B. W. Plowman. 2001. Using Euclidean distances to assess
nonrandom habitat use. Pages 275—290 in J. J. Millspaugh and J. M. Marzluff,
editors. Radio tracking and animal populations. Academic Press, San Diego,
California, USA.

Erickson, W. P., T. L. McDonald, K. G. Gerow, S. Howlin, and J. W. Kern. 2001.
Statistical issues in resources selection studies with radio-marked animals. Pages
209—242 in J. J. Millspaugh and J. M. Marzluff, editors. Radio tracking and
animal populations. Academic Press, San Diego, California, USA.

Garshelis, D. L. 2000. Delusions in habitat evaluation: measuring use, selection, and
importance. Pages 111—164 in L. Boitani and T. K. Fuller, editors. Research
techniques in animal ecology: controversies and consequences. Columbia
University Press, New York City, New York, USA.

Gladfelter, H. L. 1984. Midwest agricultural region. Pages 427—440 in L. K. Halls,

editor. White-tailed deer ecology and management. Stackpole, Harrisburg,
Pennsylvania, USA.

92

Guthery, F. S., A. R. Rybak, W. Russ Walsh, S. D. Fuhlendorf, and T. L. Hiller. 2005.
Quantifying usable space for wildlife with use-availability data. Journal of
Wildlife Management 68:655—663.

Hobbs, N. T., and D. C. Bowden. I982. Conﬁdence intervals on food preference
indices. Journal of Wildlife Management 46:505—507.

Huegel, C. N., R. B. Dahlgren, and H. L. Gladfelter. 1986. Bedsite selection by white-
tailed deer fawns in Iowa. Journal of Wildlife Management 50:474—480.

Ivlev, V. S. 1961. Experimental ecology of the feeding of ﬁshes. Yale University Press,
New Haven, Connecticut, USA.

Johnson, D. H. 1980. The comparison of usage and availability measurements for
evaluating resource preference. Ecology 61:65—71.

Johnson, D. H. 1999. The insigniﬁcance of statistical signiﬁcance testing. Journal of
Wildlife Management 63:763-772.

Kemohan, B. J ., R. A. Gitzen, and J. J. Millspaugh. 2001. Analysis of animal space use
and movements. Pages 125—166 in J. J. Millspaugh and J. M. Marzluff, editors.
Radio tracking and animal populations. Academic Press, San Diego, California,
USA.

Lenth, R. V. 1981a. Robust measures of location for directional data. Technometrics
23:77—81.

Lenth, R. V. 1981b. On ﬁnding the source of a signal. Technometrics 23:149—154.

Litvaitis, J. A., K. Titus, and E. M. Anderson. 1996. Measuring vertebrate use of
terrestrial habitats and foods. Pages 254-274 in T. A. Bookout, editor. Research
and management techniques for wildlife and habitats. Fifth ed., rev. The Wildlife
Society, Bethesda, Maryland, USA.

Loft, E. R., J. W. Menke, and J. G. Kie. 1991. Habitat shifts by mule deer: the inﬂuence
of cattle grazing. Journal of Wildlife Management 55:16—26.

Lopez, R. R., N. J. Silvy, R. N. Wilkins, P. A. Frank, M. J. Peterson, and M. N. Peterson.
2004. Habitat-use patterns of Florida Key deer: implications of urban
development. Journal of Wildlife Management 68:900—908.

Madill, H., and W. Rustem. 2001. Michigan land resource project. Public Sector
Consultants, Lansing, Michigan, USA.

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

93

2002. Resource selection by animals. Second edition. Kluwer Academic,
Dordrecht, The Netherlands.

Marchinton, R. L., and D. H. Hirth. 1984. Behavior. Pages 129—168 in L. K. Halls,
editor. White-tailed deer ecology and management. Stackpole, Harrisburg,
Pennsylvania, USA.

McCabe, R. E., and T. R. McCabe. 1984. Of slings and arrows: an historical
perspective. Pages 19-72 in L. K. Halls, editor. White-tailed deer ecology and
management. Stackpole, Harrisburg, Pennsylvania, USA.

McClean, S. A., M. A. Rumble, R. M. King, and W. L. Baker. 1998. Evaluation of
resource selection methods with different deﬁnitions of availability. Journal of
Wildlife Management 62:793—80 1.

Mech, L. D., and R. E. McRoberts. 1990. Survival of white-tailed deer fawns in relation
to maternal age. Journal of Mammalogy 71 :465—467.

Michigan Center for Geographic Information. 2001. IFMAP land-use, land-cover data.
http://www.mcgi.state.mi.us/mgdl/?rel=ext&action=sext (Accessed 28 Jul 2005).

Michigan Department of Agriculture. 2002. Agricultural statistics: Jackson and
Washtenaw Counties. http://www.michigan.gov/mda/ (Accessed 16 Aug 2004).

Michigan Department of Natural Resources. 2005. 2005 deer population estimates for
south central management unit. Michigan Department of Natural Resources,
Wildlife Division Technical Report.

Miranda, B. R., and W. F. Porter. 2003. Statewide habitat assessment for white-tailed
deer in Arkansas using satellite imagery. Wildlife Society Bulletin 31:715—726.

Morrison, M. L., B. G. Marcot, and R. W. Mannan. 1992. Wildlife-habitat relationships:
concepts and applications. University of Wisconsin Press, Madison, Wisconsin,
USA.

Nams, V. O. 1989. Effects of radio-telemetry error on sample size and bias when testing
for habitat selection. Canadian Journal of Zoology 67:1631—1636.

Nams, V. O., and S. Boutin. 1991. What is wrong with error polygons? Journal of
Wildlife Management 55:172—176.

Otis, D. L. 1997. Analysis of habitat selection studies with multiple patches within
cover types. Journal of Wildlife Management 61:1016—1022.

Ozoga, J. J ., and L. J. Verme. 1986. Relation of maternal age to fawn-rearing success in
white-tailed deer. Journal of Wildlife Management 50:480—486.

94

Porter, W. F., and K. E. Church. 1987. Effects of environmental pattern on habitat
preference analysis. Journal of Wildlife Management 51:681—685.

Powell, R. A. 2000. Animal home ranges and territories and home range estimators.
Pages 65—1 10 in L. Boitani and T. K. Fuller, editors. Research techniques in
animal ecology: controversies and consequences. Columbia University Press,
Columbia, New York, USA.

Pusateri, J. S. 2003. White-tailed deer population characteristics and landscape use
patterns in southwestern Lower Michigan. Thesis, Michigan State University,
East Lansing, Michigan, USA.

Seaman, D. E., J. J. Millspaugh, B. J. Kemohan, G. C. Brundige, K. J. Raedeke, and R.
A. Gitzen. 1999. Effects of sample size on kernel home range estimates. Journal
of Wildlife Management 63:739—747.

Severinghaus, C. W. 1949. Tooth development and wear as criteria of age in white-
tailed deer. Journal of Wildlife Management 13:195—216.

Sommers, L. M. 1977. Atlas of Michigan. Michigan State University Press, East
Lansing, Michigan, USA.

Sparrowe, R. D., and P. F. Springer. 1970. Seasonal activity patterns of white-tailed
deer in eastern South Dakota. Journal of Wildlife Management 34:420—43 1.

Stewart, K. M., T. E. Fulbright, D. L. Drawe, and R. T. Bowyer. 2003. Sexual
segregation in white-tailed deer: responses to habitat manipulations. Wildlife
Society Bulletin 31:1210—1217.

Strauss, R. E. 1979. Reliability estimates for lvlev’s electivity index, the forage ratio,
and a proposed linear index of food selection. Transactions of the American

Fisheries Society 108:344—352.

Thomas, D. L., and E. J. Taylor. 1990. Study designs and tests for comparing resource
use and availability. Journal of Wildlife Management 54:322—330.

Thomas, D. L., and E. J. Taylor. 2006. Study designs and tests for comparing resource
use and availability 11. Journal of Wildlife Management 70:324—336.

US. Census Bureau. 2003. 2000 census of population and housing, population and
housing counts: Michigan. Washington, DC, USA.

Van Horne, B. 1983. Density as a misleading indicator of habitat quality. Journal of
Wildlife Management 47:893—901.

95

Vreeland, J. K., D. R. Diefenbach, and B. D. Wallingford. 2004. Survival rates,

mortality causes, and habitats of Pennsylvania white-tailed deer fawns. Wildlife
Society Bulletin 32:542-553.

White, G. C., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data.
Academic Press, San Diego, California, USA.

Worton, B. J. 1989. Kernel methods for estimating the utilization distribution in home-
range studies. Ecology 70: 164-168.

Worton, B. J. 1995. Using Monte Carlo simulation to evaluate kemeI-based home range
estimators. Journal of Wildlife Management 59:794—800.

96

Table 3.1. Study designs (and deﬁnitions of use and availability in my study) for
collection of use-availability data as described by Thomas and Taylor (1990, 2006) and

Erickson et al. (2001), and selection hierarchy as deﬁned by Johnson (1980).

 

 

 

Scale and Deﬁnition Selection
Study design Use Availability Hierarchy
l Population-level Population-level Second-order
(study area) (study area)
2 Individual Population-level Second-order

(kernel home range) (study area)

3 Individual Individual Third-order
(location estimates) (kernel home range)
48 Individual Individual F ourth-order
(location estimate) (paired with and deﬁned for
each use)

 

aStudy included designs 1—3; my use-availability data did not seem appropriate under

study design 4.

97

 

 

 

mm: 3.3 3% .3 4.3% 3 _ m S a 8mm 5:38.... v.8
one 8mm :3 3o 2% 2mm 3% 2: 80322 52.52
8o 8% 28 m8 8% an: .2} e. 5.»... 2253 852
N3 38 3.3 mg a. ﬂ 2 33 8? 33m see... 2233
“8:8

8: 8.3 N. _ S. 8: w. 5m 32 8% NS... 3828.8 2.233
:5 eaee MES 85 3% 88 2E 28 225% .8833:
one ea _ m 3.3. :3 3.3 3%. 3m: 8%.. 5:8
amen N. _ 2 no? on? 23 £2. E: 3.2 2332.?

N no: no: N do: in: no: 16.. 2.5 e28

mEBEwéoZ mEBEO 80:53—

 

:o88m

 

.382 @180 a :088 wEBBwEO: hoe. 80:82 new; .700 51.32 c: wEBEw :8

80:82 80.: :088 wEBEw 33:32:? .3 :28: 883 888:8 :ocmooq .ooomivoom dawEocz 35:3 538 .833 .578... :8

852:. :o e_ :98: 8:5. .88 >95 2: 55:5 Soc. S 35:: 888:8 .A_Eo::8 :o 888:8 5:82 :5 8%: 850 8838: Coma:

u 403 .832 u 404 ”9:: 88: 882.28 exemov 888$: :8::o=m :o :88: SN n 5 88: 8.5-2:? Mo 8828 8>oU .N.m 038.

98

.A00:m:6>m 8::E_ 3:8: :_ :5 60500—8 w::8ww:m 8::_ :m :_ :38: :603 853$: :8E 80:52

A 853$: :8E :08 ..w.0v 8860.8 6309:: .6 3_m0::0 m_ :39? AI .+v 8:5 .6 :0::025 808208 w:_>_:E_ .o v 3 80:5:

.6 ADD :5. :88 .6 :0: wEm: :086060 83 N £66: :08 .6 30 V 3:6: 80:52 .6 30 .: ”90:00.00. w:_>_:E_ .o A E :08

.6 :UD :5 80:53 :0 :0: mi? 85:28 83 N £66: :08 .6 30 A 3:6: E0:::._ .6 30 a Wood M N .:0::m_8>0 30 gm: .2
88V 3:6: :08 :8 axis—68v 3:6: «:0::8 .6 00586 :88 9:8 8:5 33300—0 23: P83: :0 :08: 8065 6:006?

.:0E::o:_ 6: 0.8 28:36:. 526338 _0>0_-:0:a_::0: :8 08 68763230: 85:82 _ :m0._.8Qﬁ

 

 

 

 

3o :35. 3%: mg 3.5:. 323 was}. 32:. :50
mg 38: 38: owe £3 :5 38: 8Q: Es:
2.? 32.: 38.: 2.? 38: :5: $3.: 0. SN: :35
8.? 33.2 32.: 3.? 3mm: 2%.: 3%.: 3.3.2 95: 22.5
:86:
:0 3:; E: 33 8: 32 89. 3mm 382.0% 2.2::
N :0: :0: k :0: :0: :0: :0: 25 .250
w:_30._m-:02 mEBEO 50:53:
:080m

 

528: .mm 2.5

99

Table 3.3. Relative importance of cover types by adult female white-tailed deer (n = 20)
based on selection indices under multiple study designs (see Thomas and Taylor [1990]),
south central Michigan, 2004—2006. Data were pooled by agricultural growing season

(growing [10 May—7 Oct], non-growing [8 Oct—9 May]).

 

 

 

Season StUd)’ Designa
Relative
Cover type 1 2 3UB 3Cc Importanced
Growing
Agriculture 13 9 7 7 1 1
Conifer 1 l 2 3 1
Herbaceous openland 9 10 5 6 8
Lowland deciduous forest 9 6 3 4 6
Lowland shrub 6 3 7 8 4
Mixed wetland 3 7 3 4 3
Northern hardwood 4 4 10 10 5
Oak association 8 5 9 9 7
Upland deciduous forest 2 2 1 2 2
Upland shrub 11 12 11 11 13
Urban 12 1 1 5 1 10
Water 5 8 11 12 8
Other 7 12 11 12 12
Non-growing
Agriculture 13 9 6 6 10

 

100

Table 3.3. (cont’d)

 

 

 

Season Study Designa
Relative

Cover type 1 2 3Ub 3C ° Importanced

Non-growing
Conifer 1 1 1 1 1
Herbaceous openland 9 8 9 9 9
Lowland deciduous forest 10 6 3 4 7
Lowland shrub 6 3 3 5 3
Mixed wetland 3 7 8 8 6
Northern hardwood 4 4 7 7 5
Oak association 6 5 5 2 4
Upland deciduous forest 2 2 2 2 2
Upland shrub 11 12 10 10 12
Urban 12 11 10 10 12
Water 5 10 10 10 8
Other 8 12 10 10 1 1

 

aDesigns are based on various deﬁnitions of use and availability described in text.
bUnconditional analyses included all cover types within the study area for describing
selection.

°Conditional analyses excluded cover types not present within an individual’s home
range.

dRelative importance (R) = [design 1 + design 2 + (0.5 x design 3U) + (0.5 x design

3C]/3; values shown are ranks based on these values.

101

 

 

 

 

o oo._ oo. _I o _ .o Sod mad nmd 532083 me
o oo._ oo. .1 mmd Sod mvod Ed 836:2 E0582
0 mwd oo. _I n _ .ol m 36 mmod wvd cam—83 Boa—2
+ $6 _N.o mmd wmod vomd mwd £25 9.2261—
0 mmd omdl modl vmod wnod $6 888 28.66% nan—Boi—
o \l _ .o oo. _I mmdl Sod aood mmd cam—=30 £58361
+ 2: Ed Ed 536 «m _ .o 36 5:50
I _m.ol mmdl mmdl Sod we _ .o Ed 23:8:w<
wEBEO
883 403 0154 N Rim & A5 owes.— oEo: 09o 530
E 8:805
$82: :osomxom um: Ecoquoi :Ommom

 

.Eﬁmzmoao 328.0ch am

2:23 AwEBEwéo: mm .wEBEw mm ”SO 5132 o_ ”a cm 3 Emma wEBEm Rea—:0:?“ 3 3.08 295 Sam .ooomuﬁoom .cmeB—z

35:3 538 .800 C 8331 use 3825 no «N cwﬁov have: 8N H 5 Loon 3.5-823 2.6m 038$ no Eton—om B>oU Jim 0331

102

 

 

 

 

o o _ .o 3.9. oool o _ o.o mnoo owo $88 32.30% nan—Box—
0 3o oo. _I Nvol oooo N_o.o ovo 23:30 3083.5:
+ oo._ mmo moo .moo No _ .o vwo 5:50
I ovoI _m.oI omoI m moo Q. _ .o woo 23.no_._m<
wEBEwéoZ
I oo. _I ooo.o oo.o .550
o _ _.o oool Emol mooo foo wvo 583
I Sol oo. _I Sol .ooo Nooo ooo :35
I 8. T 83 23 2:5. 2%.:
+ mmo mmo mvo goo ammo ooo $28 30.66% tam—Q:
9:380
new: :03 0154 N ASmm R EV owcﬁ 0E0; 090 1.260
E 3585
pxooE cosoﬂom «om: Ecoanoi acmmom

 

5:53 i 035

103

.33 >95 05

3 Bosses 893 530 so woman 33 bzinga woman“ 0E0: 3on Rue $3 3 3953 womb 330 no woman 33 8: N cwﬁoﬂa

 

 

 

 

I 8. T 83 cod .550
I 8.? 8. T mm? 33 N _ o.o 93 .355
I 3w? 8. T 3.? :55 53 3o 535
I 8. T 83 88 £25 Baa:
+ 3o mg as 33 $3 2: use 32.20% 2%:
o 2: 8. T n _ .o 83 ~85 NS 8:283“ 8.8
o 8.. 8. T w _ .o Sod 33 33 8365. 52:02
o 86 8. T 8.? :3 Rod 8d 2233 852
+ as ad m3 086 8g 33 £5: €233
wEBEwIcoZ
new: AUD 040: N Rom—m m A3 omce 0E0; 090 .525
5 00:93.5
3x09: goon—om 53: 3.85895 5.3%

 

5:53 {m 25

104

.oo:w_mmm 33
A3 3: E022 .o 39:2: 30 o\omo .: «Em 60:3me mm? To 8520; .ov 30 o\emo .: moocwmmmm mm? At song—om .oA 30 $3 to
.o H 8: .mcoanoE t 388:8 on 8: 2:8 use 4 was T «a 3:8ch 203 oxmov 3E:— oocooccouo

.5 xooE 53:83 Coo—v m.>o_>_ co 3me 80:2: Esau—om;

105

Table 3.5. Cover selection of female adult white-tailed deer (n = 20; design 3a of Thomas

and Taylor [1990]) pooled by agricultural growing season (150 d; 10 May—7 Oct; 23

growing, 25 non-growing) in an agro-forest ecosystem in south central Michigan, 2004—

2006. Proportional use and selection indices were unconditional on presence of cover types

within home ranges.

 

 

 

 

Season Proportional use Selection indexb
Cover type p SE(p) 1 LCL° UCL Used
Growing
Agriculture 0.154 0.046 -0.03 -0.12 0.05 o
Conifer 0.168 0.063 0.04 -0.05 0.14 o
Herbaceous openland 0.009 0.008 0.00 —1.00 1.00 o
Lowland deciduous forest 0.081 0.026 0.02 —0.23 0.27 o
Lowland shrub 0.191 0.041 -0.03 -0.10 0.03 0
Mixed wetland 0.023 0.015 0.02 -1.00 1.00 0
Northern hardwood 0.030 0.009 —0.20 -0.76 0.36 0
Oak association 0.016 0.006 -0.09 —l .00 1.00 o
Upland deciduous forest 0.325 0.057 0.06 0.02 0.10 +
Upland shrub 0.000
Urban 0.002 0.002 0.00 -l .00 1.00 0
Water 0.000
Other 0.000

 

106

Table 3.5. (cont’d)

 

 

 

Season Proportional usea Selection indexb
Cover type p SE(p) 1 LCL° UCL User
Non—growing
Agriculture 0.162 0.038 -0.03 -0.1 1 0.05 o
Conifer 0.192 0.058 0.08 0.01 0.17 +
Herbaceous openland 0.005 0.004 -0.41 —1.00 1.00 o
Lowland deciduous forest 0.074 0.025 0.01 -0.27 0.30 o
Lowland shrub 0.206 0.042 0.01 -0.06 0.06 0
Mixed wetland 0.018 0.009 -0.20 -1.00 1.00 0
Northern hardwood 0.026 0.012 -0.1 1 -1.00 0.80 0
Oak association 0.022 0.01 1 0.00 —1.00 1.00 o
Upland deciduous forest 0.294 0.046 0.02 -0.02 0.06 o
Upland shrub 0.000
Urban 0.00]
Water 0.000
Other 0.000

 

8Design 3 use was based on proportion of locations in each cover type averaged over
individuals; availability was based on mean proportion of area of cover types bounded by
individual 95% ﬁxed kernel home range by season. Cover types absent from a home
range had proportional availability = 0.

bSelection indices based on lvlev’s (1961) electivity index (I).

107

°Conﬁdence limits (95%) were truncated at -l and 1, and could not be estimated if
proportional use = 0.
dlf 95% CLs >0, selection (+) was assigned; if 95% CLs <0, avoidance (-) was assigned;

and if 95% CLs included 0, random use (0) was assigned.

108

Table 3.6. Cover selection of female adult white-tailed deer (n = 20; design 3a of Thomas

and Taylor [1990]) pooled by agricultural growing season (150 d; 10 May—7 Oct; 23

growing, 25 non-growing) in an agro-forest ecosystem in south central Michigan, 2004-

2006. Proportional use and selection indices were conditional on presence of cover type

within home ranges.

 

 

 

Season Proportional use Selection indexb
Cover type p SE(p) I LCLc UCL Use‘T
Growing
Agriculture 0.209 0.062 -0.03 —0.10 0.04 o
Conifer 0.258 0.089 0.04 -0.03 0.12 o
Herbaceous openland 0.043 0.036 0.00 -l .00 1.00 o
Lowland deciduous forest 0.109 0.033 0.01 —0.20 0.22 o
Lowland shrub 0.230 0.044 -0.04 —0.09 0.02 0
Mixed wetland 0.048 0.030 0.01 -1.00 1.00 0
Northern hardwood 0.041 0.012 —0.20 —0.65 0.26 0
Oak association 0.028 0.009 -0.08 -1.00 1.00 o
Upland deciduous forest 0.339 0.058 0.06 0.03 0.09 +
Upland shrub
Urban 0.026 0.026 0.18 -1.00 1.00 0
Water 0.000
Other

 

109

Table 3.6. (cont’d)

 

 

 

 

Season Proportional usea Selection indexb
Cover type p SE(p) 1 LCLc UCL used
Non-growing
Agriculture 0.183 0.041 ~0.03 -0.09 0.03 o
Conifer 0.228 0.067 0.08 0.02 0.14 +
Herbaceous openland 0.013 0.009 -0.38 -l .00 1.00 o
Lowland deciduous forest 0.091 0.030 0.01 -0.20 0.21 o
Lowland shrub 0.245 0.046 0.00 -0.04 0.05 0
Mixed wetland 0.029 0.014 —0.22 -l .00 0.64 0
Northern hardwood 0.031 0.014 -0. l. 1 -0.77 0.55 0
Oak association 0.031 0.015 0.02 -l .00 1.00 o
Upland deciduous forest 0.294 0.046 0.02 -0.01 0.05 o
Upland shrub
Urban 0.00
Water 0.00
Other

 

2‘Design 3 use was based on proportion of locations in each cover type averaged over
individuals; availability was based on mean proportion of area of cover types bounded by
individual 95% ﬁxed kernel home range by season. Cover types absent from a home
range were not considered available to respective individual.

bSelection indices based on Ivlev’s (1961) electivity index (1).

110

CConﬁdence limits (95%) were truncated at -l and 1, and could not be estimated if
proportional use = 0.
dIf 95% CLs >0, selection (+) was assigned; if 95% CLs <0, avoidance (-) was assigned;

and if 95% CLs included 0, random use (0) was assigned.

111

.ooomIeooN .<mD £09522 3.5000 508 £3.88 3032000 0:20: 0:0 £00,350

.3000 Eng—sot?“ .0000 beam 05 00 moms B>00 2 :0 So A05 0% n83-0a00m003 0 00:30:56 accuscem ._.m charm

$5 $0 seen.

ON? o9. om om ov ON 0 our 06.. co co ov cm 0

 

 

 

 

 

 

 

 

 

J
cm a
02 .m
L e
- of U
“00.0“. 0
. com 04
02032000 000.03 8:000 A
. 0mm om \.I
L N
1 com 00 n
09 02 om 00 04 om o om? 02 on 00 04 om o W.
__.IIJ__Iu , O___.I_j_ o e
S J
cm 08 mo:
. on 82 d
. ow . m.
05:30:91 - om 003. >035 O
- 82 We.
.1 8 L S
IH (

 

 

on 1 OOON

112

300 —

 

 

A O
<0 _
5
a) o
m 250
C ,r
(U '— z’
a: . /
m 200 — /,/
s _
I O O . ’/’/
c 150 — ° ,” 0
E I/ .
E
.2 100 i—— . ”/.” C
3 T o ’1' .
0 II, C
3; 50 L" .hol"l O
0. _ 049/
y I I I I
0 W1 I I I I I 1 1 1 1 I 1
0 100 200 300 400 500

Home Range Size (ha)

Figure 3.2. Linear relationship ()2 = 0.52x; r = 0.87; dashed line) of amount of public
land within home range (ha) as a function of seasonal home-range size (ha) of adult
female white-tailed deer (n = 20), south central Michigan, USA, 2004—2006. Seasonal
home-range estimates were pooled and based on agricultural growing season (10 May—7

Oct for growing season, 8 Oct—9 May for non-growing season).

113

CHAPTER 4

Estimation and Implications of Space Use for White-tailed Deer

Management in Southern Michigan

114

Estimation and Implications of Space Use for White-tailed Deer

Management in Southern Michigan

INTRODUCTION

Habitat manipulation and harvest strategies are 2 primary management tools used
to influence deer population size and distribution. Understanding space use by white-
tailed deer (Odocoileus virginianus) is integral to successful habitat management
strategies. McCullough (19872547), in fact, stated, “Deer can be beneﬁtted readily by
habitat manipulation over most of their range, and there can be no question that habitat
management is the keystone to continuing high populations.” However, the success of
any habitat management plan depends on the incorporation of accurate and appropriate
scientiﬁc evidence. Managers can apply habitat management practices grounded in
science as opposed to dogma for higher expectations of success.

Habitat management (e.g., land-cover conversions, altering current vegetation
successional stages) is often done to change or redistribute the abundance of a particular
species, although the effects of management on other wildlife species should be
considered. Habitat manipulation can impact wildlife populations at different spatial
scales, depending on the size, shape, frequency, and intensity of the manipulation or
disturbance. For example, it may be possible to redistribute the individual animals on a
managed area without affecting the overall abundance of that species by manipulating
resources (food, water, cover), particularly those resources that are considered limiting.

The conversion or redistribution of winter thermal cover (i.e., coniferous forest stands), a

115

life requisite in northern latitudes (Ozoga 1968, Demarais et al. 2000), could theoretically
inﬂuence deer behavior and demographics and redistribute deer densities.

With deer and human populations both generally increasing during the past
several decades (McShea et al. 1997), wildlife managers face an increasingly challenging
scenario. Urban and urbanizing areas may impose certain obvious control problems
related to harvest management and habitat manipulation; at times, habitat manipulation
may be the only option for deer management, as the harvest of deer in urban areas may be
socially unacceptable or unsafe. Regardless, management options can become quite
limited under these circumstances. These increasing challenges are the impetus behind
integrating research with management, including space-use information speciﬁcally
addressing issues of deer management in urbanizing areas.

My objectives were to quantify space use of white-tailed deer in an increasingly
urbanizing agro-forest ecosystem in south central Michigan, to compare kernel home
range estimates with usable space estimates, and to discuss the implications of cover
conversions designed to increase or decrease the amount of suitable habitat for deer. I
also developed some general rules of cover interspersion for habitat management of
white-tailed deer based on my space-use assessments. In theory, an increase in habitat
quantity or quality will increase the density or abundance of deer, with the obverse
relationship also assumed to be true.

STUDY AREA

My study took place in eastern Jackson (Grass Lake, Henrietta, and Waterloo

townships), western Washtenaw (Dexter, Lima, Lyndon, and Sylvan townships), and

southwestern Livingston (Unadilla township) counties in south central Lower Michigan.

116

The study area (82,636 ha) included publicly owned lands, including the Michigan
Department of Natural Resources (MDNR) Waterloo (8,410 ha; 10.2% of study area) and
Pinckney (4,276 ha; 5.2%) recreation areas, and privately owned lands. South central
Michigan has been characterized by a relatively high deer density (~27/km2 during fall
2005; MDNR 2005), increasing land-use activities (e.g., urbanization), and little
scientiﬁc study of deer under these conditions. Although the study area was primarily
rural (98% of total land area), the human population increased 16% and housing units
increased 22% between 1990 and 2000 (US. Census Bureau 2003). Much of the
landscape throughout southern Michigan (and in other areas throughout the Midwest) is
expected to experience increasing land-use activities similar to the study area (Madill and
Rustem 2001).

The physiographic regions of this area are Hillsdale-Lapeer Hilly Upland, South
Central Rolling Plain, and Southeastern Rolling Plain (Sommers 1977:24) with alﬁsols as
the major soil order (Sommers 1977:36). Surface formations in the study area are the
result of glaciation and include all 4 types present within Michigan (moraine, till plain,
outwash plain, and lacustrine plain; Sommers 1977:32). Elevation of the study area
ranged approximately 180—300 m and consisted of relatively limited relief (Sommers
1977126, 33). The study area received about 81 cm of precipitation annually during
1971-2000 (based on conditions in Chelsea, Michigan; Midwestern Regional Climate
Center, Champaign, lllinois, USA), and had a ISO-day growing season (i.e., the average
annual accumulation of daily mean temperatures >5.6 °C), generally occurring from 10
May to 7 October (Sommers 1977246, 49). During 1971—2000, average annual snowfall

was 99.3 cm and mean monthly temperatures ranged from —5.4 °C (Jan) to 21.8 °C (Jul)

117

in Jackson County (Midwestern Regional Climate Center, Champaign, Illinois, USA).
Annual snowfall during my study was highly variable (2004 = 90.9 cm, 2005 = 149.9 cm,
2006 = 40.0 cm; Chelsea, Michigan; Midwestern Regional Climate Center, Champaign,
Illinois, USA).

Before European settlement, southern Michigan forests consisted of oak (Quercus
spp.) and hickory (Carya spp.) in well-drained soils and beech (Fagus grandifolia), elm
(Ulmus spp.), maple (Acer spp.), and basswood (Tilia americana) in poorly drained soils
(Sommers 1977:17). Much of the study area is well suited for agriculture (Sommers
1977:38). Most crop agriculture in Jackson and Washtenaw counties (total area =
366,483 ha) was com (37,840 ha) and soybean (34,200 ha) production, followed by hay
(17,200 ha), winter wheat (7,970 ha), and oats (930 ha; Michigan Department of
Agriculture 2002).

l generalized land-use, land-cover data (Michigan Center for Geographic
lnforrnation 2001) using ArcView GlS v3.2 software (Environmental System Research
Institute, Redlands, California, USA) and Spatial Analyst extension to define 13 cover
types within the study area: agriculture (non-vegetated farmland, row crops, forage crops;
52.3% of study area); conifer (pines [Pinus spp.], other upland conifers; 1.5%);
herbaceous openland (herbaceous vegetation with <25% woody cover; 2.9%); lowland
deciduous forest (>60% composed of deciduous tree cover; 8.0%); lowland shrub (with
>60% non-water cover; 9.9%); mixed wetland (ﬂoating aquatic vegetation, emergent
wetland, mixed non-forest wetland; 3.1%); northern hardwood (>60% canopy cover of
maple, beech, ash [Fraxinus spp.], cherry [Prunus spp.], birch [Betula spp.]; 2.3%); oak

association (>60% canopy cover of oak; 1.6%); upland deciduous forest (>60% canopy

118

cover of upland deciduous trees; I 1.6%); upland shrub (>25% woody cover; <0.1%);
urban (low and high intensity, roads, parks, golf courses; 2.8%); water (surface, flowing;
3.9%); and other (aspen [Populus spp.] association, orchards, bare ground; 0.1%). Patch
size of cover types ranged from <1 ha to >1 1,000 ha (e.g., an agricultural matrix) and had
a mean size of 29.2 ha. Consequently, generalization of spatial data may have excluded
certain ﬁne-scale landscape characteristics (e.g., hedgerows, roads) in some instances.
METHODS
Capturing Deer

l trapped deer during winter (Dec—Mar) 2004—2006, using single-door collapsible
live traps (Clover 1954). Traps were placed near areas of deer activity, baited with kernel
com, and checked twice/day to minimize stress and injury to deer. My technicians and I
restrained captured deer using the collapsed trap and the body weight of 1 person
(Sparrowe and Springer 1970). All trapped deer were blindfolded to reduce stress and
ﬁtted with metal ear tags bearing a unique identiﬁcation number (Style 681; National
Band and Tag, Newport, Kentucky, USA). Individuals were aged as fawn (<1 year old),
yearling (Z 1—<2 year old,), or adult (2 2 years old) through general morphometric
differences (e.g., shape and size of head, body size) and dental characteristics
(Severinghaus 1949); ages of necropsied individuals later conﬁrmed accuracy of ﬁeld
observations.

Only female deer were ﬁtted with mortality-sensing collar-style radio-
transmitters, either with VHF (Model 500; Telonics, Inc., Mesa, Arizona, USA) or VHF-
GPS (Model 02000; Advanced Telemetry Systems, Isanti, Minnesota, USA) capabilities.

I did not radiomark male deer because 1 expected low capture success for adult males

119

based on previous trapping experience and deer behavior. Additionally, winter-captured
male deer were not radiomarked because of potential problems associated with their
physiological changes (i.e., neck-swelling) during the breeding season and ﬁtting a radio-
transmitter to accommodate these changes. Radio-transmitters had a unique frequency
within the ISO-MHz range, a mass of 270 g (VHF) or 1,100 g (VHF-GPS), and an
expected minimum battery life of 36 months (VHF) or 12 months (VHF-GPS). Michigan
State University’s All-UniversityI Committee on Animal Use and Care approved all
capturing and handling procedures for my study (Application No. 01/04-006-00).
Estimating Locations

1 located deer 2—5 times/week using triangulation methods associated with
telemetry (White and Garrott 1990:79—112) or from visual observations of known
individuals. To increase the potential of my analyses to accurately describe space use,
deer were located using a systematic sampling approach at varying time schedules on a
diel basis, with 21 nocturnal location/deer/week except during capture periods. I used
this sampling design because some wildlife species (including deer) potentially move at
any time during a 24-hour period, and “management recommendations developed from
habitat use data collected from only a portion of a 24-hour period may be ineffective”
(Beyer and Hauﬂer 1992:180). Bearings were estimated using a 3-element folding Yagi
antenna (Advanced Telemetry Systems, Incorporated, Isanti, Minnesota, USA), portable
radio receiver (Model R-IOOO, Communications Specialists, Incorporated, Orange,
California, USA), and mirror-sighting compass. A global positioning system (GPS)
handheld unit (Model GPS IV; GARMIN lntemational, Incorporated. Olathe, Kansas,

USA) was used to approximate the locations from which signals were received.

120

Locations from triangulated data were estimated using the program LOCATE III
(Pacer, Truro, Nova Scotia, Canada). I used the maximum likelihood estimator (Lenth
1981a, b), as recommended by White and Garrott (1990) and Nams and Boutin (I 991).
Bearing standard deviation and error ellipses were estimated using LOCATE III for each
location during analysis. To assess the precision of location estimates, I considered the
relationship between mean error-ellipse size and landscape-patch size (Nams 1989).
Given the patchiness of the study area, I assumed that characteristics of each cover type
(e.g., differences in foliage densities) did not inﬂuence telemetry accuracy, resulting in
my acceptance of location estimates near cover-type boundaries as accurate even if error
ellipses covered 21. cover type (White and Garrott 1990:200). I used only location data
from VHF signals to avoid potential differences associated with location precision
between VHF- and GPS-derived location estimates (i.e., I used only VHF-derived data
from VHF-GPS transmitters; see Appendix A).
Assessment of Space Use

Because habitat quality and species density are not positively correlated in some
instances (e.g., social dominance factors; Van Home 1983), my space-use analyses were
estimated only for adults; i.e., I assumed that habitat quality would be higher for adults
than forjuveniles because deer often behave according to a social dominance hierarchy
(Hirth I977, Marchinton and Hirth 1984). Deer captured as fawns or yearlings were
reclassiﬁed as adults on either the ﬁrst or the second 1 June following their capture,
respectively, and subsequently were used for analyses. I assumed that all location
estimates that were classiﬁed in the cover type water were inaccurate and l relocated each

to the nearest alternative cover type. Space use was determined seasonally based on

121

agricultural crop production (i.e., the growing season on study area [10 May—7 Oct; 150
d], and the non-growing season [8 Oct—9 May; 215 d]). I used SYSTAT v11 (Systat
Software, Inc., San Jose, California, USA) and ProStat v4.02 (Poly Software
lntemational, Inc., Pearl River, New York, USA) for statistical analyses, and 95%
conﬁdence limits (CLs; LCL = lower, UCL = upper) for space-use assessments. The use
of conﬁdence limits is advantageous in that an estimate of effect size and a measure of
uncertainty are provided (Johnson 1999).

Space use is often estimated using 1 of the many home-range estimators currently
available (e.g., minimum convex polygon, kernel methods, harmonic mean), and each
may provide somewhat different results using the same data. One method in particular,
the ﬁxed kernel, seems to be the best estimator currently in use (see Kemohan et al.
2001). Two common methods of smoothing parameter selection for kernel estimation
include the reference bandwidth (hmf) method and least-squares cross-validation (hey)
method (Kemohan et al. 2001). The ha, method is currently recommended and was used
for my analysis, as the he] method may overestimate home-range size (presumably more
than the hm, method) because of its relatively high bias (Worton 1995, Seaman et al.
1999)

Kernel home ranges may, however, misrepresent the space used by an animal
through the inclusion of cover types containing no location estimates (Guthery et al.
2005); Sanderson (1966:231) noted that, “No doubt the home range can be reduced to
some unknown minimum size.” The inclusion of these apparently unused cover types
may result in misguided habitat management strategies, or management effort being

expended on low-priority habitat components. A new concept of resource selection that

122

might express this minimum size is the quantiﬁcation of usable space (Guthery I997,
Guthery et al. 2005).

Usable space has been described as, “the quantity (ha) of ideal (maximizes
ﬁtness), permanent habitat for a species of interest on an area of interest,” where
“permanent” habitat is best considered on an annual basis (Hiller et al. 2007). A
permanent habitat situation might entail frequent management, depending on rates of
vegetation succession in a given area. Usable space, in a qualitative sense, has been
described for deer as, “the portion of a landscape that is or can be utilized by white-tailed
deer, but not all space within a landscape is useful as habitat” (Fulbright and Ortega-S
2006:28). To date, known analyses fonnalizing the concept of usable space have been
limited to use-availability data for northern bobwhites (Colinus virginianus) in Texas and
Arkansas (Guthery et a1. 2005, Hiller et al. 2007), but the concept should be valid for any
wildlife species (e.g., white-tailed deer), given appropriate use-availability data.
Quantifying usable space from use-availability data results in assessment of the
contributions of selected cover types and the contributions of cover types randomly used
and avoided.

For usable space analyses, 1 followed study design 3 (estimating use and
availability for each individual) of Thomas and Taylor (1990) as recommended and
discussed by Hiller et a1. (2007). In essence, use is deﬁned through the proportion of
location estimates in a given cover type within a kernel home range, and availability as
the proportional area of the given cover type within a kernel home range. Under design
3, there is no assumption of identical resource availability among individuals, and

variance estimates of cover-type availability can be obtained; the use of study designs 1

123

or 2 could provide different results from design 3. Following Guthery et al. (2005), I
based my usable space analysis on the ﬁxed-kernel home-range estimator (Worton 1989).

Usable space (U) is deﬁned by
u = 21.14., (1)
[=1

where w = the number of cover types available (i = l, 2, 3, ..., w), u,- = the unknown
proportion of space that is usable in cover type i, and A,- = the area (ha) of cover type i
(Guthery et a1. 2005).

To solve for U, an assumption of u,,, = l (i.e., the proportion of space that is
usable in cover type m in a set of selected types or for data pooled over selected types
equals 1) was made through support from ﬁeld data (e.g., highly selected cover types
based on selection indices; Hiller et al. 2007). When estimating proportions of usable
space within all available cover types, assuming I of the cover types selected by a species
as containing fully usable space makes the appropriate comparisons of usable space
estimates of other cover types possible. Under this assumption, usable space within
cover type i (u,-) is estimated as

u:- = “MD/(AIM), (2)
where A”, = the area of cover type m, p,- = proportional use of cover type i (number of
location estimates in i/total number of location estimates), and pm = proportional use of
cover type m.

The decisions related to deﬁning cover type u,,, are at the experimental unit (i.e.,
for each individual). If a strong selection pattern, based on selection indices, is present
for l or 2 cover types, then these cover types will dominate usable space analyses as um.

However, other cover types with higher selection indices and high proportions of use and

124

availability may be substituted on an individual basis. The decision of the appropriate
cover type used to describe um can also be supported through our knowledge of the
biology of a species (i.e., cover types known to be crucial or most beneﬁcial to a species
of interest on an area of interest). To illustrate, it would be logical to assume that
permanent emergent vegetation would be much more appropriate to describe cover type
um for muskrats (0ndatra zibethicus) than would open water that lacks vegetation (Allen
and Hoffman 1984).

I tested the relationship between the amount of usable space (y) and kernel home
range size (x) using least-squares linear regression. Under this space-use relationship,
when x = 0, y must = 0; i.e., a home range must contain space for the existence of usable
space, so zero-intercept single-variable models were appropriate (see Guthery and
Bingham 2007). I used the Pearson product-moment correlation coefﬁcient (r) to
quantify relationships between usable space and kernel home ranges.

Cover Conversion Estimation

To quantify the direction and magnitude of change (CI) of each cover type (i) to
describe fully usable space within a home range, I followed Hiller et al. (2007):

C.- = A(Cr'ar). (3)
where A = the area (ha) of interest, c,- = the proportional availability of cover type i in
usable space, and a,- = the proportional availability of cover type i on the area of interest.
Logically, to increase the amount of usable space by, for example, 50% (i.e., to
potentially increase the abundance of a species of interest on an area of interest to some
extent less than maximal), I would modify equation 3 to include the desired amount of

change:

125

C,- = 0.5A(c;-a,-), (4)
where the coefﬁcient of 0.5 in my example estimates a 50% increase in usable space (for
cover type 1'). Further, I could modify equation 3 to decrease the amount of usable space
for a species of interest on an area of interest (i.e., to theoretically decrease population
abundance of that species). To illustrate, suppose a manager desired a substantial
(perhaps 50%) decrease in the number of white-tailed deer on a management area due to
concerns with agricultural crop damage. The appropriate equation would be modiﬁed to
include the magnitude (e.g., 50%) and direction (e.g., -) of the desired change (decrease)
in the amount of usable space. Following my example, this would be expressed by

C: = -O.5A(c,-a;), (5)
where the mathematical operator is reversed (i.e., from + to -) within the equation to
reﬂect the manager’s objectives (i.e., to decrease the amount of usable space). Equations
can be modiﬁed similarly to reﬂect the speciﬁc objectives of direction and magnitude of
change of usable space (or speciﬁc cover types). Note that, in my example, 0.5 does not
need to be constant among cover type is; i.e., a multitude of coefﬁcients could achieve a
50% decrease in usable space, not simply the value of 0.5 constant across all cover types
within the area of interest. However, to illustrate my example, using a constant is much
more intuitive to the reader.

The total amount of cover conversion (C) on the area of interest can be estimated

by

lecI
C = —" 2 , (6)

126

the sum ofthe absolute values of C,- (for all i = l, 2, ..., w) divided by 2. This division
operation will account for the conversion of 1 cover type into another cover type (i.e., the
spatial replacement of cover types).
RESULTS
Capture and Estimation of Locations

I captured and radiomarked 42 female deer. A subset of 20 radiomarked deer that
were either aged as adult during capture or that moved into the adult age class during the
study was available for analysis (i.e., individuals with 2 30 locations; Seaman et al.
1999). This subset contained a total of 3,493 location estimates and a mean of 71.
locations/seasonal home range. I pooled data by season but not by year due to small
sample sizes. The growing seasons of 2004, 2005, and 2006 included 7, 14, and 3 deer,
respectively. The non-growing seasons of 2004, 2005, and 2006 included 7, 11, and 7
deer respectively. Five deer were included in similar seasonal categories for >1 year.

I pooled location estimates (3,493) from all adult deer (n = 20) from my study to
estimate overall telemetry error. I considered telemetry error acceptable, and thus,

location estimates to have acceptable precision, given the landscape characteristics of the
study area (e.g., :t = 29.2 ha/patch), the mean telemetry error-ellipse size (10.2 ha), and

the number of location estimates (E = 71/deer/seasonal home range). This conclusion
was based on the relationship between telemetry error and patch size (diameter ratio =
0.59) to determine appropriate sample sizes as described by Nams (1989).

The frequency distribution of all location estimates showed somewhat uneven
sampling effort by month (Fig. 4.1). Effort decreased during months of deer capture (i.e.,

Jan, Feb, Dec, for winter; May for capture of neonates [see Chapter 2]). Sampling effort

127

by time of day was variable and peaked during mid-day (Fig. 4.2); however, when l
pooled location estimates by diel categories (diurnal = 800—1700 hr, nocturnal =
remaining hours), proportional use (i.e., number of location for each cover type/total
number of locations) for all cover types was similar within time periods (i.e., varied
£0.06) except for agricultural areas (diurnal = 0.24, nocturnal = 0.14).

Assessment of Space Use

Kernel Home Range—Mean kernel home-range sizes for adult female deer
during the growing season (J-C = 77.5 ha i 9.6 SE; n = 20) was about half that of the non-

growing season (it = 140.4 ha i 23.4 SE; n = 19), although 95% conﬁdence intervals
overlapped slightly (57.5—97.6 ha versus 91 .6—189.2 ha). The mean number of cover
types within home ranges was similar between seasons (6.5 i 0.4 SE for growing, 7.4 i
0.4 SE for non-growing). Kernel home ranges were dominated by upland deciduous
forests during both seasons, while the amount of agricultural areas in kernel home ranges
increased almost 2.5 x from the growing to the non-growing season (Table 4.1). No
cover types were used exclusively in 1 season.

Usable Space. —During the growing season, the mean amount of usable space
within a kernel home range was 58.2 ha (SE = 6.9; 95% CL = 438—725); during the non-
growing season, the mean amount of usable space was 103.0 ha (SE = 18.2; 95% CL =
65.4—140.6 ha). Both seasonal estimates of usable space were about three-fourths that of
their respective kernel home—range estimates. The mean number of cover types
containing location estimates within kernel home ranges (i.e., describing usable space)

was similar between seasons (5.0 i 0.2 SE for growing, 5.6 i 0.3 SE for non-growing),

128

but both were about 1 less than the mean number of cover types within kernel home
ranges.

Generally, I used upland deciduous forests or conifers as the cover type that
described fully usable space within a kernel home range during both seasons, based on
selection indices (see Chapter 3). However, I occasionally substituted other cover types
to describe fully usable space within kernel home ranges when these cover types had
higher selection indices and high proportions of use and availability. Lowland shrub (7
occasions) or agriculture (6 occasions) dominated (~70%) these substitutions.

Relationship between Usable Space and Kernel Home Range—My zero-intercept
models showed a strong relationship between usable space (y) and kernel home-range
size (x; Fig. 4.3). For the growing season (23 observations), usable space was a strong
linear function (r = 0.99) of kernel home-range size described by y = 0.73x (95% CL for
x-coefﬁcient = 0.68—0.78). For the non-growing season (25 observations; r = 0.99), the
relationship was y = 0.74x (95% CL for x-coefﬁcient = 0.69—0.79). When I pooled all
seasonal data, the relationship between usable space and kernel home-range size was y =
0.74x (r = 0.99; 95% CL for x-coefﬁcient = 0.70—0.78). Under all 3 scenarios, usable
space was between 68% and 79% of the kernel home-range size, consistent with the
mean amount (75%) of usable space within a kernel home range approximated earlier.
Cover Conversion Estimation

Maximizing Usable Space—Under a management scenario to maximize usable
space (i.e., 100%; theoretically maximizing deer abundance) on my study area,
agricultural areas would require the greatest reduction in area during both seasons (Table

4.2). The cover types water, other, upland shrub, and urban contributed little or nothing

129

to usable space for deer on my study area and theoretically could be converted to other
cover types if achieving fully usable space were the only consideration (Table 4.2).
Based on the comparison of proportional availability (study area and usable space; Table
4.2), upland deciduous forests and, to a lesser degree lowland shrubs and conifers, were
cover types that existed in insufﬁcient amounts to maximize deer abundance. Managers
would need to convert about 41,221 ha (49.9%) of the study area to achieve fully usable
space during the growing season, and about 37,812 ha (45.8%) during the non-growing
season to achieve fully usable space, maximizing deer densities.

Decreasing Usable Space—Following my example of a 50% decrease in usable
space, an increase of 31 or 37% of agricultural areas, and a decrease of 100% of conifers
and upland deciduous forests during both seasons (Table 4.2), could be part of 1 possible
habitat management plan. Note that mathematically, the percent decrease in some
instances (e.g., conifers, upland deciduous forests) would be <-100%; when this
occurred, 1 used the amount (ha) of that cover type within the study area and rounded
cover conversions up to -100%, as >100% of any given cover type cannot be removed.
Increasing the amount of unused cover types (e.g., upland shrub, other) by 50% would
also decrease usable space by 50%. Total cover conversions to meet these management
objectives would be 18,060 ha (21.9% of study area) during the growing season, and
17,596 ha (21.3%) during the non-growing season.

DISCUSSION

“Common knowledge” of deer biology is often “enhanced by anecdotes and

accumulated lore, speculative ideas or hypotheses [that] become transformed into a

dogma that is extremely resistant to change” (McShea et al. 199721), which further

130

complicates issues associated with effective deer management. Certainly, a plethora of
space-use information has been published on white-tailed deer in the Midwest, but how
do managers use this information to assist with habitat management decisions?
Descriptions of deer home-range sizes and their vegetation structure and composition are
useful, but literature may neglect a discussion of their direct relevance to the
manipulation of current habitat conditions to change deer abundance on an area of
interest. It would seem that a more comprehensive look at space-use by deer as it relates
to habitat management would be useful for increasing the efﬁcacy of deer management
programs.

Considerations for habitat management include estimating the contributions of
avoided and randomly used cover types (as determined by selection indices) under the
assumption that an animal’s presence in these cover types somehow contributes to their
ﬁtness. Additionally, quantifying cover conversions (or land-use changes, such as
urbanization), which would theoretically change wildlife abundance, would seem logical
to integrate into wildlife habitat management programs. Before quantifying potential
cover conversions, I described and applied a seemingly more accurate method of
quantifying space use (usable space) in comparison to the ﬁxed-kernel estimator. My
analyses showed that the concept of usable space collapsed kernel estimates into a
potentially more accurate representation of space needs of deer on my study area, given
my estimate of ~75% usable space within kernel home ranges. The mean number of
cover types associated with usable space was about I cover type less than that for kernel
home ranges. This thesis was based on the condition that the kernel (and other) methods

often include cover types containing no location estimates, and therefore, potentially little

131

or no use associated with these unused cover types. The assumption that my sampling
methodology accurately represented deer behavior on my study area is a concern relevant
to any space-use assessment.

My estimation of usable space revealed a pattern of cover use between seasons.
Although the area of the 4 cover types (upland deciduous forest, lowland shrub,
agriculture, conifer) that contributed most to usable space differed substantially by season
(Table 4.1), the cumulative proportional contribution of these 4 cover types to usable
space showed similarity (86% for growing, 88% for non-growing). For example, upland
deciduous forest contributed 22.61 ha (39% of usable space) and 37.42 ha (36%) to
usable space during the growing and non-growing seasons, respectively. Lowland shrub
was estimated to contribute double the area to usable space for the non-growing season
than the growing season (Table 4.1), yet contributed 20—23 % of total area regardless of
season; other cover types also showed this trend. These patterns indicated reduced
movements (i.e., reduced home-range size) of adult female white-tailed deer during and
following parturition (Marchinton and Hirth 1984, Chapter 1), although the use of cover
types based on usable space composition remained roughly constant regardless of season.
Cover Conversions

My study area was dominated by agriculture (52%), a relatively undesirable cover
type for deer, at least in high proportions (>25%; see Cover lnterspersion Rules), based
on my usable space analyses and supported by selection indices calculated from
southwestern Lower Michigan (Pusateri 2003). Woody cover types of varying
successional stages provide suitable habitat for white-tailed deer across their geographic

distribution (Baker 1984, Demarais et al. 2000). Logically, forest types dominated my

132

usable space estimates, and white-tailed deer in the Midwest agricultural region
“probably are affected more severely by forest loss than are Whitetails in other regions of
the country” (Gladfelter 1984:427); I attempted to quantify cover conversions under
several hypothetical scenarios (e.g., agriculture converted to upland deciduous forest to
increase deer abundance, upland deciduous forest converted to urban areas to decrease
deer abundance) that may help managers with habitat manipulation strategies.

Urban areas, as expected, essentially contributed nothing to usable space for deer
on my study area. The effects of urbanization on deer habitat may be inferred from my
results. If cover conversions include increasing urbanization, as in many Midwestern
states, the amount of usable space may be affected less if the converted area included
cover types such as agriculture, herbaceous openland, lowland deciduous forest, northern
hardwoods, oak association, or upland shrub; these cover types seemed to exist in excess
based on my usable space analysis. Conversely, if urbanization included the conversion
of cover types that seemed to be in quantities limiting deer abundance (e.g., conifer,
lowland shrub, upland deciduous forest), then the amount of usable space for deer, and
consequently deer abundance, may decline on that area. These potential conditions
assume a positive relationship between deer abundance on a given area and the amount of
usable space on that area, as described earlier. I do not assume that this positive
relationship is linear, as limited evidence for other species has suggested otherwise (see
Guthery et a1. 2001zﬁg. 6). This evidence suggested that with the addition of usable
space to an area containing near-minimum or near-maximum amounts of usable space,
population abundance response may be much less than when adding usable space to an

area with some intermediate amount of usable space.

133

Selection of u...

I based my usable space analyses on the assumption that upland deciduous forests
best described fully usable space, although conifers were also used to some degree, based
on selection indices (see Chapter 3); i.e., I generally used I of these cover types as um,
and assumed that other cover types described usable space at some fraction of either
upland deciduous forests or conifers. Rarely, I considered cover types other than upland
deciduous forests and conifers to describe fully usable space. Although selection indices
(Ivlev 1961) showed conifers as having similar selection ratios (1 = 0.47 for growing
season, I = 0.50 for non-growing season) to upland deciduous forests (I = 0.44 for
growing season, I = 0.41 for non-growing season; Chapter 3), I felt that upland deciduous
forests were a more appropriate choice based on deer ecology (e.g., conifers may provide
thermal or escape cover, but provide little or no food value; Schmitz 1991). Further, the
mean Euclidean distance of all location estimates of deer to conifers was about 3.5 x the
mean distance to upland deciduous forests, and the relative importance (see Johnson
1980) of each changed depending on the spatial scale used (Chapter 3).

The importance of the cover types contributing the most to usable space for deer
within southern Michigan has been suggested by another study. A space-use assessment
in southwestern Lower Michigan by Pusateri (2003) estimated a mean annual kernel
home-range size for primarily female deer (157.7 ha), similar to my non-growing season
estimate (140.4 ha). Cover composition within home ranges described by Pusateri (2003)
and selection indices (Ivlev 1961) included deciduous forest (47%; I = 0.20), evergreen
(<4%; I = 0.64), and agriculture (39%; I = -0.l6). The composition of my home-range

estimates during the non-growing season were somewhat different (deciduous forest

134

types = 37%, conifer = <8%, agriculture = 20%), but selection indices based on use-
availability data provided by Pusateri (2003) generally suggested that an increase in
deciduous forest and conifer, and a decrease in agriculture would be desirable if
management objectives were to increase usable space for deer, which is consistent with
my results. Although cover types were classiﬁed more broadly by Pusateri (2003), the
general composition of cover between study areas was similar. Mean patch size differed
(20.3 ha for southwestern Michigan, 29.2 ha for my study), however, suggesting that
landscape characteristics and land-ownership patterns were different between study areas.
Selection indices between these 2 study areas, however, were similar (Chapter 3).
Cover lnterspersion Rules

I developed 3 rules of cover interspersion generalized from my results. These
rules are broad enough to be implemented regardless of season, and assume management
objectives that include increasing the amount of fully usable space to potentially increase
deer abundance in an area similar to the agro-forest ecosystem of south central Lower
Michigan.
1. The area of interest should include upland deciduous forests, conifers, and lowland
shrub cover, based on their contributions to usable space and on selection indices.
2. If agricultural cover is present, it should be between 5% and 25% of the total land
area. This range is loosely based on 95% CLs of proportions that agricultural areas
contribute to usable space during the growing and non-growing seasons.
3. Upland deciduous forest cover and conifer cover should ideally be interspersed so that
all points in the area are 100—200 m and 400—600 in from each cover type, respectively.

These numbers are based on the mean Euclidean distance of location estimates from each

135

cover type (95% CL = 122.6—148.6 m for upland deciduous forests, 95% CL = 454.4-
554.6 m for conifers; seasons combined). Because deer are a species of relatively high
mobility, my suggestions for distances likely could range much broader without
signiﬁcantly affecting deer abundance.
Considerations

Guthery et al. (2005) discussed 2 possible explanations for the estimation of
usable space in cover types that were used less than expected by northern bobwhites.
Both hypotheses seem to relate to 2 basic assumptions while evaluating habitat, which
may be problematic: “researchers can discern habitat selection or preference from
observations of habitat use and that such selection, perceived or real, relates to ﬁtness and
hence to population growth rate” (Garshelis 2000:111). Certainly, many factors,
including spatial scale of study (e.g., varying deﬁnitions of resource use and availability),
may affect how use-availability data are interpreted in selection studies (e.g., Chapter 3).

The ﬁrst hypothesis of Guthery et al. (2005) described density-dependent cover
use, where these “avoided” cover types served as population sinks (see Pulliam 1988)
under high population levels. Under high population levels, individuals may be forced
into areas containing less usable space, which may affect survival and reproduction of
those individuals. When populations are at a low level, more individuals are assumed to
exist in areas that maximize ﬁtness, presumably because of less intraspeciﬁc competition.
Although my study area had relatively high deer densities (~27/km2), I believe that
limiting my space-use analyses to adult female deer avoided this potential problem for

several reasons.

136

The logic behind my use of adult female deer to assess usable space was based on
several considerations. First, a potential age-related social hierarchy exists among deer
(Hirth 1977), which may result in differences in space use for each age class. Second, on
my study area, females >1 yr old were the most abundant sex-age class (48% does, 26%
bucks, 26% fawns during pre-hunt fall 2005; B. Rudolph, MDNR, personal
communication). Third, female deer are often the sex class of primary interest for
managers desiring relatively large reductions in population densities through strategies
such as hunter-harvest (Carpenter 2000). Finally, male and female white-tailed deer may
respond differently to a speciﬁc habitat manipulation technique on a management area
through different resource selection strategies and space-use requirements (Stewart et al.
2003). For example, sexual segregation of deer outside of the breeding season
(approximated by the growing season) may result in females selecting areas with higher
quality forage (Stewart et al. 2003), indicating that adult female deer may be the best age-
sex class from which to deﬁne usable space for white-tailed deer. Considering the above
evidence, I believed that my selection of adult female deer effectively described usable
space on my study area (and perhaps other areas of interest).

The second explanation proposed by Guthery et al. (2005) was that bobwhites
may respond more to structural as opposed to compositional elements, given their
relatively wide geographic distribution (Johnsgard 1975:82). The white-tailed deer also
inhabits numerous vegetation types (e.g., Rolling Plains of Texas, coniferous-hardwood
forests of Michigan’s Upper Peninsula) and is distributed throughout most of North and
Central America (Baker 1984). Under this hypothesis, if deer responded more to

structural elements, then some unknown level of interchangeability may exist among

137

cover types, and perhaps population responses to changes in the amount of usable space
on a given area may not be as pronounced. To illustrate, Short (1986) considered escape
cover in the southeastern coastal plains to simply be vegetation structure of a certain size
and density regardless of vegetation species composition. Of course, the only way to test
this hypothesis would be a long-term ﬁeld study that monitors deer population dynamics,
habitat manipulations, and land-use changes, which are often extraordinarily difﬁcult
(Garshelis 2000).

Temporal considerations should also be addressed when assessing usable space
for any species. Although habitat conditions are not permanent, for the quantiﬁcation of
usable space I assumed habitat conditions to be relatively constant during the agricultural
growing and non-growing seasons, and among seasons during a relatively short time
period (e.g., 3 yr). Studies over longer time periods or on areas experiencing rapid land-
use conversions should incorporate a study design to minimize these effects (e.g.,
accounting for successional changes in composition and structure for the existing habitat
types).

I attempted to sample for location estimates of deer on an appropriate temporal
scale relative to deer behavior (i.e., time of year, time of day; see assumption 6 of
Guthery et al. 2005:661). Sampling effort was higher during 0800—1700 hr (diurnal), but
general patterns of proportional use between diurnal and nocturnal time periods were
similar. However, agricultural areas received more proportional use (number of location
in agricultural areas/number of total locations) during diurnal hours (0.24 versus 0.14).
There is evidence that deer use more open areas (e.g., agricultural ﬁelds) during the night

(Montgomery 1963), and this difference in use could have potentially affected the

138

estimated contribution of this cover type to usable space on my study area. Given the
small difference in use, I felt this potential bias only minimally affected the magnitude of
change of cover conversions, certainly not the direction.

I addressed some of the assumptions described by Guthery et al. (2005), as
violation of 21 assumption may render my usable space estimates inaccurate to some
unknown degree. Further, Guthery et al. (2005:662) stated that this “metric should be
regarded as a ﬁrst-generation approach,” suggesting that improvements of estimating
usable space may be forthcoming. Certainly, the concept of usable space should be
scrutinized further by wildlife scientists for potential improvements. Here, I applied the
concept of usable space to describe the space use of a species with a much different life
history than northern bobwhites, the only other species known to be considered thus far.
Further, I believe that my results are applicable to describe the management implications
of cover-type conversions, dependent on management objectives (e.g., decreasing or
increasing deer abundance) and land-use changes. My results are based on direct
evidence of cover use by deer (assuming location estimates accurately represented deer
behavior), which seems to be a more accurate description of space use as compared to
home-range estimators such as the kernel methods. With clear evidence of use (e.g.,
cover types containing location estimates of deer), decisions regarding habitat

management should, in theory, be better supported to achieve management objectives.

I39

LITERATURE CITED

Allen, A. W., and R. D. Hoffman. 1984. Habitat suitability index models: muskrat. US.
Fish and Wildlife Service FWS/OBS-82/ 10.46.

Baker, R. H. 1984. Origin, classiﬁcation and distribution. Pages 1—1 8 in L. K. Halls.
editor. White-tailed deer ecology and management. Stackpole, Harrisburg,
Pennsylvania, USA.

Beyer, Jr., D. E., and J. B. Hauﬂer. 1992. Diurnal versus 24-hour sampling of habitat
use. Journal of Wildlife Management 58:178—180.

Carpenter, L. H. 2000. Harvest management goals. Pages 192—213 in S. Demarais and
P. R. Krausman, editors. Ecology and management of large mammals in North
America. Prentice Hall, Upper Saddle River, New Jersey, USA.

Clover, M. R. 1954. A portable deer trap and catch-net. California Fish and Game
40:367—373.

Demarais, S., K. V. Miller, and H. A. Jacobson. 2000. White-tailed deer. Pages 601—
628 in S. Demarais and P. R. Krausman, editors. Ecology and management of
large mammals in North America. Prentice Hall, Upper Saddle River, New
Jersey, USA.

Fulbright, T. E., and J. A. Ortega-S. 2006. White-tailed deer habitat: ecology and
management on rangelands. Texas A&M University Press, College Station,
Texas, USA.

Garshelis, D. L. 2000. Delusions in habitat evaluation: measuring use, selection, and
importance. Pages 111—164 in L. Boitani and T. K. Fuller, editors. Research
techniques in animal ecology. Columbia University Press, New York, USA.

Gladfelter, H. L. 1984. Midwest agricultural region. Pages 427—440 in L. K. Halls,
editor. White-tailed deer ecology and management. Stackpole, Harrisburg,
Pennsylvania, USA.

Guthery, F. S. 1997. A philosophy of habitat management for northern bobwhites.
Journal of Wildlife Management 61 :291—301.

Guthery, F. S., and R. L. Bingham. 2007. A primer on interpreting regression models.
Journal of Wildlife Management 71 :684—692.

Guthery, F. S., M. C. Green, R. E. Masters, S. J. DeMaso, H. M. Wilson, and F. B.

Steubing. 2001. Land cover and bobwhite abundance on Oklahoma farms and
ranches. Journal of Wildlife Management 65:838—849.

140

Guthery, F. S., A. R. Rybak, W. Russ Walsh, S. D. Fuhlendorf, and T. L. Hiller. 2005.
Quantifying usable space for wildlife with use-availability data. Journal of
Wildlife Management 68:655—663.

Hiller, T. L., F. S. Guthery, A. R. Rybak, S. D. Fuhlendorf, S. G. Smith, W. H. Puckett,
Jr., and R. A. Baker. 2007. Management implications of cover selection data:
northern bobwhite example. Journal of Wildlife Management 71 :195—201.

Hirth, D. H. 1977. Social behavior of white-tailed deer in relation to habitat. Wildlife
Monographs No. 53.

Ivlev, V. S. 1961. Experimental ecology of the feeding of ﬁshes. Yale University Press,
New Haven, Connecticut, USA.

Johnsgard, P. A. 1975. North American game birds of upland and shoreline. University
of Nebraska Press, Lincoln, Nebraska, USA.

Johnson, D. H. 1980. The comparison of usage and availability measurements for
evaluating resource preference. Ecology 61 :65—71.

Johnson, D. H. 1999. The insigniﬁcance of statistical signiﬁcance testing. Journal of
Wildlife Management 63:763—772.

Kemohan, B. J., R. A. Gitzen, and J. J. Millspaugh. 2001. Analysis of animal space use
and movements. Pages 125—166 in J. J. Millspaugh and J. M. Marzluff, editors.
Radio tracking and animal populations. Academic Press, San Diego, California,
USA.

Lenth, R. V. 1981a. Robust measures of location for directional data. Technometrics
23:77—81.

Lenth, R. V. 1981b. On ﬁnding the source of a signal. Technometrics 23:149—154.

Madill, H., and W. Rustem. 2001. Michigan land resource project. Public Sector
Consultants, Lansing, Michigan, USA.

Marchinton, R. L., and D. H. Hirth. 1984. Behavior. Pages 129—168 in L. K. Halls,
editor. White-tailed deer ecology and management. Stackpole, Harrisburg,
Pennsylvania, USA.

McCullough, D. R. 1987. The theory and management of Odocoileus populations.
Pages 535—549 in C. M. Wemmer, editor. Biology and management of the
Cervidae. Smithsonian Institution, Washington, DC.

McShea, W. J., H. B. Underwood, and J. H. Rappole. 1997. Introduction: deer

141

management and the concept of overabundance. Pages 1—7 in W. J. McShea, H.
B. Underwood, and J. H. Rappole, editors. The science of overabundance: deer
ecology and population management. Smithsonian Institution, Washington, DC,
USA.

Michigan Center for Geographic Information. 2001. IFMAP land-use, land-cover data.
http://www.mcgi.state.mi.us/mgdl/?rel=ext&action=sext (Accessed 28 Jul 2005).

Michigan Department of Agriculture. 2002. Agricultural statistics: Jackson and
Washtenaw counties. http://www.michigan.gov/mda/ (Accessed 16 Aug 2004).

Michigan Department of Natural Resources. 2005. 2005 deer population estimates for
south central management unit. Michigan Department of Natural Resources,
Wildlife Division Technical Report.

Montgomery, G. G. 1963. Nocturnal movements and activity rhythms of white-tailed
deer. Journal of Wildlife Management 27:422—427.

Nams, V. O. 1989. Effects of radio-telemetry error on sample size and bias when testing
for habitat selection. Canadian Journal of Zoology 67:1631—1636.

Nams, V. O., and S. Boutin. 1991. What is wrong with error polygons? Journal of
Wildlife Management 55:172—176.

Ozoga, J. J. 1968. Variations in microclimate in a conifer swamp deeryard in northern
Michigan. Journal of Wildlife Management 32:574—585.

Pulliam, R. H. 1988. Sources, sinks, and population regulation. American Naturalist
132:652—661.

Pusateri, J. S. 2003. White-tailed deer population characteristics and landscape use
patterns in southwestern Lower Michigan. Thesis, Michigan State University,
East Lansing, Michigan, USA.

Sanderson, G. C. 1966. The study of mammal movements: a review. Journal of
Wildlife Management 30:215—235.

Schmitz, O. J. 1991. Thermal constraints and optimization of winter feeding and habitat
choice in white-tailed deer. Holarctic Ecology 14:104—1 l 1.

Seaman, D. E., J. J. Millspaugh, B. J. Kemohan, G. C. Brundige, K. J. Raedeke, and R.
A. Gitzen. 1999. Effects of sample size on kernel home range estimates. Journal
of Wildlife Management 63:739—747.

Severinghaus, C. W. 1949. Tooth development and wear as criteria of age in white-
tailed deer. Journal of Wildlife Management 13:195—216.

142

Short, H. L. 1986. Habitat suitability index models: white-tailed deer in the Gulf of
Mexico and South Atlantic coastal plains. US. Fish and Wildlife Service,
Biological Report 82 (10.123).

Sommers, L. M. 1977. Atlas of Michigan. Michigan State University Press, East
Lansing, Michigan, USA.

Sparrowe, R. D., and P. F. Springer. 1970. Seasonal activity patterns of white-tailed
deer in eastern South Dakota. Journal of Wildlife Management 34:420—43 1.

Stewart, K. M., T. E. Fulbright, D. L. Drawe, and R. T. Bowyer. 2003. Sexual
segregation in white-tailed deer: responses to habitat manipulations. Wildlife
Society Bulletin 31:1210—1217.

Thomas, D. L., and E. J. Taylor. 1990. Study designs and tests for comparing resource
use and availability. Journal of Wildlife Management 54:322—330.

US. Census Bureau. 2003. 2000 census of population and housing, population and
housing counts: Michigan. Washington, DC, USA.

Van Horne, B. 1983. Density as a misleading indicator of habitat quality. Journal of
Wildlife Management 47:893—90 1.

White, G. C., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data.
Academic Press, San Diego, California, USA.

Worton, B. J. 1989. Kernel methods for estimating the utilization distribution in home-
range studies. Ecology 70:164—168.

Worton, B. J. 1995. Using Monte Carlo simulation to evaluate kernel-based home range
estimators. Journal of Wildlife Management 59:794—800.

143

 

 

 

 

 

who we. woo 42 5. N2 802:2 50552
:.o 85 SN :2 SN v: 2233 @052
m: 36 m3 8.: 5m on: sea e533
3o ad on; m2 M? $4. 528 32668 e533
:3 86 Ed 2.0 :6 m3 2258 ”882%:
mg m; a: m: a: w; £88
9: 3a SN Mm: 3m mm: 2338?
wEBBO
mm H mm H mm m 25 $26
8:80th 58QO 0333 owcﬂ 2:0: 353— cemmom
683.com Swag:

3550 5:8 E 982 also a wEEﬂwéoc mm .ToO 5132 o: wEBEw m8 .888 953% Bea—:0:?“ 3 @208

SN n 5 Home 3:86an we 8QO 038: can mowcﬂ 080: $on voxm $3 mo Amé coummomﬁoo cab B>oU ._.v 2an

144

 

 

 

 

wmd w: o~.m mod 3 .m mw.o_ .8550

m3 3 .5 2e $8 a; E: 23383
mEBEwéoZ

cod cod cod 550

36 mm.“ cod mod mm; 53>?

mod mod Ed nod mod 2 .o :mﬁD

8d cod 23 8:5 253:

E .N 36 36 3.3 mud S KN 520.“ £53280 2:an

vmd wmd mmd wwd mmd we; 5938mm“ *5
wEBEO

mm m mm H mm m 25 5.50
00:20wa 8an 05mm: omen." 0:5: 380M cemeom

 

528$ .2. 053

145

 

 

 

2 .o 3 .o 23 2 .o E .o 535
cod cod cod sea 2%:
3a 2% m3 9.? New 33. use 38208 8%:
one 2: :3 w: one m3 853083 a5
9; 2 .N as 0: 3o 2:” Bases: 82202
4.2 :2 a: 4.3 $3. 8e agave 852
mm: a; New :2 w: on? naﬁ 2233
£6 3m N3 5 z .m 2 .N saw 528 3828a 2233
who 36 m H .o 3o 86 e3 22:80 38035:
mamaewéoz
mm H mm H mm m 25 530
8:80th 3QO 03me 09:2 080: 350M cemmom

 

8.253 .3 2x3

146

.88: 83¢ Ea
382:. mo m swam—u have: Aomcﬁ 6:8: 3on 5.5» 33 .960 .8 «Ba we coanoa n bzﬁngm .086 page some 55:? 33850

5:32 .«o :oEomoa H 83 Saw tax—Eggnog co 3me mm? Anoom .3 8 BE: .38 .3 Ho boﬁzOv cogﬁzmo 8an 0583“

 

 

 

cod cod cod 550

£6 omd cod owd omen... 55$
mEBEwéoZ

mm x mm R mm R 093 b>oU
coconota 8QO 03mm: owcﬂ 0:8: 3on cemmom

 

5:53 .2. 22¢

147

 

 

 

 

:280280 530 voyaEsmm mo 5:530

tom—528$ 3:05.585

2+ 2? SI «.2 T ..Sa E; 80322 .5552
Z T ”.57 3? 28+ 93 .85 2283 3me
2? 3.2.? 22+ 2%.? 83 god see 2533
a? .33.. we T v.2 SI «85 83 use 30.66% 23.33
4.3+ 28. T 3T o. $1? 83 on; 22:25 80883:
98 T 38. T 25+ 9.6.2.? 33 m :3 £68
3? 38.2.. 2.7 goon? em; 22 2330:?

wEBEO
3e 6 :2: ea 3 35 3 3 25 .28

mo\oOml ao\oOml UENm: %zzh 03mm: b—zn— Down—m OENmD No.5 %Uam :Ommom

 

.Coon om u 2 ”mowcﬁ 0:8; mm n8 >22 @IHoO w u wEBEméo: women. 2:0: mm no.“ 80 5132 o_ n 3523 .838 wEBEw

35:30:? :0 women 26>» Sump—3mm— .ooomlvocm .<mD .cmeBE .8280 538 E 3.8 33.0. 2-03.9 an :0 Bow B:B-8E>>

no.“ AﬁcowI ..wov 3an 03mm: 835% 9 was $33: 3:: rod 3QO 033: 3:5me 8 222328 330 33555 .Né 033:

148

 

 

 

56.5250 830 33835 mo @2530

5:333; 35903.5

odo _I WEN. _ I «.8? Newt? wood m 56 3:50

odm+ ndmﬁm _. + n. El #3981 oomd mmmd 23.30tw<
wEBBméoZ

o.om+ m.$+ ode—I ﬁwm T coed mood 350

93+ m. 50.: ode—I ﬁmomﬁl oood amod .533

xii _.woo,_+ ﬁnal N.ww_.ml Sod wmod :35

93+ 2+ odoT 2: 83 83v 9:5 2%:

92:1 93de Qmmm+ n.m©m.mm+ mwmd 2 ﬁo $88 $82208 2833

m.m+ _.mv+ Vol méwl Sod fed 5:38me V30
wEBEO

Eu 6 35 $3 5 35 AS as 25 $30
AXuoml o\oOml 036m: b—DL 036m: %=:n_ Down—m DENmD $0.3 %Usum =Omaom

 

25:08 .Né 033+

149

 

 

 

206.8200 .530 33835 .3 bucazo

9.5232“ 3:29.38;

98+ 33.: SET 52.? 83 ”So :35
98+ 2+ SET 2: god 53v 2:5 2%:
SET 93?: mm _ N+ $28? $3 2 3 sea 32.20% 25:5
32+ 22+ 0.2+ v. _ MT :3 m _ ed =2§8§ x8
«2+ 22+ 3? $8.. Sod mi. 80322. .5532
moi 2T 5. _+ 2;: 5; SE 2.233 352
+8- :81? 2+ _ + 223 _ + 98° 8% sea 2233
3 _+ NEAT 92: $3.? 83. 89¢ :23 25.66% 2233
5.3+ S _ 3+ v. a: 32.? 83 3; 225% msoooaeom
wEBo..w-:oZ
£9 6 3.: Ga 6 35 E E 2:. 526
.Xéml goml 03am: b?“— 0333 b?“— oomam 0333 no; 32m :83m

 

8.253 .3 033

150

.Qotgcoo on 8:58 093

.630 :on is we o\coo _ A to: .532 2: 3 3985 29$ 33 bus; 2.: :o Eomua $5 EsoEm of 33088 35 30652.8 .5»an

 

 

o.om+ mdo+ ode—I ﬁwm T oood mood 550

93+ m. 50. _+ odo _I ﬁmomél coed 986 $33
wEBEwéOZ

33 G 35 §V G 35 E E 25 380
goml okoml 03am: 3::— oEmm: £3“— 8QO 0333 no.“ 355 cemmom

 

563280 $>oU 33655 90 3:530

bzﬁngu .mcoEon—oi

 

8.253 .3 2%;

151

500—

 

 

 

400—

 

 

 

 

300*-

 

200—

 

 

 

 

100

Frequency of Location Estimates

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 2 3 4 5 6 7 8 9 10 ll 12

Month (1 = Jan)

Figure 4.1. Frequency distribution of location estimates obtained through telemetry for

white-tailed deer by month, south central Michigan, USA, 2004—2006.

152

1200 —-

 

 

 

 

l

 

 

 

 

 

 

 

 

 

 

 

 

 

J

 

m
a)
4.; _
E
m
LlJ _
C
.3 800-
co
0 _
C)
_l
(3
(E l--
a) 400 —-
3
U' _
9
LL 200—
0
0

1

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Hour

Figure 4.2. Frequency distribution of location estimates obtained through telemetry for

white-tailed deer by time of day, south central Michigan, USA, 2004—2006.

153

 

 

300 - A
_ A A
250 — A
”a? _
5
(D 200 —
0
(U H A A
6%
g 150 b . A‘
:3 F .
3 100 ~ 4, 1' Season
0A 0
— 0 Growing
.. ‘ t ‘ N .
50 — g. .: °“9'°“""9
__ A
‘A.
0.1111111111111111]!
0 50 100 150 200 250 300 350 400 450

Kernel Home Range Size (ha)

Figure 4.3. Usable space (ha) as a function of kernel home-range size (ha) by agricultural
growing season (10 May—7 Oct for growing season, 8 Oct—9 May for non-growing
season) for white-tailed deer in an agro-forest ecosystem, south central Michigan, USA,

2004—2006.

154

CHAPTER 5

Conclusions and Implications

155

Conclusions and Implications

Increasing deer densities, coupled with the increasing human population, will
likely result in more deer-human interactions in the future. These interactions will be
perceived as positive (e.g., increased hunting opportunities) or negative (e.g., increased
crop damage and deer-vehicle collisions) depending on the interaction. In fact, within
my study area, deer represented both positive (indicative of natural areas) and negative
(deer-vehicle collisions) feelings among residents (Lischka 2006). Because deer
demographics and space use varied substantially across their geographical distribution,
deer population and habitat management decisions should consider science conducted
locally and at multiple spatial scales because deer seem to select resources based on a
selection hierarchy. However, managers should also consider regional trends to help
predict the efﬁcacy of certain strategies.

The need for proper management of deer populations also stems from their status
as a game species, being the most popular big game species in North America (Smith and
Coggin 1984, US. Fish and Wildlife Service 2001). Michigan is no exception, having
about 743,000 hunters that harvested nearly 500,000 deer in 2003 (Frawley 2004).
Currently, deer densities are relatively high in some areas (e.g., southern Michigan), with
deer populations increasing with increases in some land-use activities (e. g., rural
development, agriculture; Smith and Coggin 1984). Increases in some land-use activities
and the human population further exemplify the need for effective management practices,
as human-deer interactions (e. g., hunting, crop damage, deer-vehicle collisions) also

increase as a result. My research attempted to characterize deer ecology in south central

156

Michigan in an increasingly urbanizing ecosystem to support deer-management decisions
based on scientiﬁc evidence. The methods and results that I used should also have
implications to deer-management issues at the regional and larger scales of management
(see below) due to their uniqueness.
Population Characteristics

Knowledge of population characteristics (e.g., space use, demographics) of white-
tailed deer (and other species) is directly applicable to management strategies. The
efﬁcacy of habitat management depends on accurate space-use assessments, often applied
at the management-unit level, where management objectives may include increasing or
decreasing deer abundance. Accurate estimates of demographics, such as age and sex
structure, of a deer population are important for the accurate estimation of deer
abundance (or an index of abundance) through models such as the sex-age-kill model.
Harvest objectives for antlered and antlerless deer are often based on the results of such
models. Without the aforementioned data, managers may unknowingly be making
management decisions that do not support their objectives. For example, subject to some
uncertainty, yearling female deer in this project had a higher-than-expected survival rate,
and the survival rate of young fawns during their development stage and beyond was
unknown in south central Michigan prior to this project. Additionally, the effects of land-
ownership patterns (e.g., public vs. private) may affect deer demographics through hunter
access and habitat quality, an analysis of such effects was not possible with my data

given my small sample sizes.

157

Resource Selection

When using selection indices, large variations (e. g., large conﬁdence intervals)
are not uncommon, especially for relatively low-use and low-availability combinations of
resources (Hobbs and Bowden 1982). Consequently, resource-selection ranking systems
should be considered as a guide when incorporated into management decisions, not an
inflexible rank of importance. When relatively strong patterns of resource selection
emerge across multiple spatial scales, such as the results I presented, managers can have
more conﬁdence in their habitat-management decisions (e.g., which cover types to
actively manage to improve habitat quality or quantity). Management of white-tailed
deer on my study area, based on my results, would seem to be relatively consistent over
multiple spatial scales. However, differences in selection may not be ﬁJlly realized for
multiple reasons (e. g., our lack of knowledge of deer perception), and therefore,
inconsistencies should be considered depending on the spatial scale, time of year, and
other factors.

At a relatively ﬁne spatial scale (e. g., private landowners with home-range-sized
units), managers may beneﬁt from analyses done at the home-range scale to examine the
life requisites of deer, and base management decisions according to their objectives.
Similarly, management at the landscape scale (e. g., deer-management units within
Michigan) should beneﬁt from a coarse-scale analysis of selection to describe appropriate
cover and composition of vegetation for deer. Certainly, results from a relative-
importance ranking system could yield highly different ranks of resource selection across
multiple spatial scales (and seasons), indicating a relatively strong selection hierarchy to

consider when making management decisions. One example would seem to be the use of

158

agricultural lands by deer on my study area. My results suggested that, at the landscape
level, large patches of agriculture could theoretically limit deer population growth, but
that small patches (or the edges of large patches) may be selected for foraging by deer.
Space Use

Generally, my results supported the assertion that space use varies by age class of
deer; the question perhaps was the magnitude of difference more so than whether a
difference existed. I provided a description of space use that included multiple age
classes, but I also provided the next step by describing potential cover conversions
(speciﬁc to adult female deer) that theoretically would affect deer abundance on the study
area. The new concept of usable space would seem to be an appropriate, although
somewhat untested, method to quantify space use and cover conversions useful for
making habitat-management decisions.

A multitude of combinations exist that would theoretically decrease usable space
by a desired amount. For example, removing all coniferous cover types on a given area
would eliminate all usable space for deer, which theoretically could greatly reduce deer
abundance on that area. This extreme scenario may not be as improbable as one might
think, as areas originally covered by longleaf pine (Pinus palustris) forests in the
southeastern US. have decreased by 98% (Noss et al. 1995). My study area was
composed of a very small proportion of conifers (<2%), with most patches located
primarily on public land. These coniferous patched do not seem to be actively managed,
so succession may cause their proportion to only decrease through time. However, this
illustrates the ﬂexibility of cover conversions and should prove advantageous to

managers with objectives to decrease or redistribute the densities of species on a

159

management area, as many cover conversion limitations exist (e.g., social and edaphic
factors). Certainly, managers should also consider the ecological consequences of any
habitat management decision. Managers desiring a relatively high increase in high-
quality habitat, however, will have to address many challenges (e. g., social, ﬁnancial, and
ecological constraints), and likely will not achieve the theoretical maximum amount of
usable space (i.e., maximizing wildlife abundance) on their management area. The
methods and information I have provided are a useful framework to follow to potentially
change the relative abundance of white-tailed deer through habitat management and land-
use change.

Certainly we must make several assumptions and have certain limitations when
studying deer ecology. As I have discussed, questions of perception and social and
ﬁnancial constraints are but a few of these potential limitations. Despite this fact, I
believe that the information that I have provided will be beneﬁcial during the decision-
making process for deer managers to meet their objectives. I also believe that the
implications of the process that I used and the knowledge gained from this study of land-
use patterns of white-tailed deer in an agro-forest ecosystem in south central Michigan
will be beneﬁcial to wildlife managers at the regional level, and perhaps beyond.

Addressing deer-habitat management issues through a multi-scale perspective and
using a usable-space approach to quantify contributions of cover are both recent but
seemingly useful concepts that also have relevance to population ecology of deer and
other wildlife species. Habitat quality and quantity, for example, in theory is positively
related to species abundance and demographics; resources selected by a species are

dependent on resource availability (i.e., relative quality among a set of resources) and

160

often age and sex structure within a population. Knowledge of deer behavior across
scales and on an area-speciﬁc basis should provide a foundation on which to base deer-

management decisions.

161

LITERATURE CITED

Frawley, B. J. 2004. Michigan deer harvest survey report: 2003 seasons. Michigan
Department of Natural Resources Wildlife Report No. 3418.

Hobbs, N. T., and D. C. Bowden. 1982. Conﬁdence intervals on food preference
indices. Journal of Wildlife Management 46:505—507.

Lischka, S. A. 2006. Enabling impact-based management of acceptance capacity for
white-tailed deer in southern Michigan. Thesis, Michigan State University, East
Lansing, Michigan, USA.

Noss, R. F ., E. T. LaRoe III, and J. M. Scott. 1995. Endangered ecosystems of the
United States: a preliminary assessment of loss and degradation. US. Department
of Interior, National Biological Service, Biological Report 28.

Smith, R. L., and J. L. Coggin. 1984. Basis and role of management. Pages 571—600 in
L. K. Halls, editor. White-tailed deer: ecology and management. Stackpole,
Harrisburg, Pennsylvania, USA.

US. Fish and Wildlife Service. 2001. 2001 national survey of ﬁshing, hunting, and
wildlife-associated recreation. Department of Interior, Washington, DC, USA.

162

APPENDIX A

Assessment of Precision between GPS- and VHF-derived Location Estimates

163

Assessment of Precision between GPS- and VHF-derived Location Estimates

INTRODUCTION

Very-high-frequency (VHF) radio-transmitters have been used in wildlife studies
for approximately 50 years. One of the ﬁrst published descriptions on transmitter design
was by LeMunyan et al. (1959) for use with woodchucks (Marmota monax). Although
the authors stated that the maximum range of this transmitter was only about 25 m, there
should be little doubt of their preference for using these transmitters, as relocation of
woodchucks was oﬁen “extremely difﬁcult and frequently required a considerable
amount of digging, only to discover that the burrow was empty” (LeMunyan et al.
1959:107). Radio-transmitters have not only made the wildlife researcher’s job easier,
but they have also helped researchers answer many questions that previously could not be
answered, at least not with much certainty. Since those early days, radio-transmitters
have become a mainstay in wildlife science.

Animal movements and resource selection are important aspects often studied
using radiomarked individuals. Telemetry techniques are used to estimate locations of
VHF-radiomarked individuals, either using homing (i.e., a visual conﬁrmation of an
animal’s location) or triangulation (White and Garrott 1990). Other ﬁeld data may also
be collected, depending on transmitter capabilities, such as body or operative
temperatures (e.g., Lonsdale et al. 1971), and heart rates (e.g., Kanwisher et al. 1978).
Researchers must consider several assumptions when using marked animals, however,

including that the radiomark does not affect animal behavior (White and Garrott 1990).

164

New technology occasionally but consistently makes its way into wildlife science.
Global Positioning System (GPS) telemetry technology for wildlife studies was
introduced around 1992 (Rodgers et al. 1996) and has become increasingly prevalent in
wildlife science during the past 10 years, especially for studying large mammals.
Although GPS-capable collars often cost S 10x that of conventional VHF radio-
transmitters designed for similar applications, the automatically scheduled ﬁxes obtained
through GPS technology may minimize animal relocation effort after transmitter
deployment.

Before 2000, selective availability (SA) degraded the precision and accuracy of
GPS-derived locations. However, post-SA GPS locations are still not without error (cf.
Cain et al. 2005), and ﬁxes should still be treated as location estimates as opposed to
actual locations of radiomarked individuals. Assessing the accuracy of these systems
under actual ﬁeld conditions has occurred (e.g., Cargnelutti et al. 2007, Sager-Fradkin et
al. 2007), but a comparison of GPS-derived location estimates with simultaneous VHF -
derived estimates (i.e., a measure of precision) in various wildlife habitat conditions
seems lacking.

My objective was to assess location precision between GPS- and VHF -derived
location estimates in an agro-forest landscape in southern Lower Michigan. I assessed
precision using 2 scenarios. For the ﬁrst scenario, 1 used a relatively controlled ﬁeld
experiment by comparing GPS locations estimated using a handheld unit to VHF
locations estimated using triangulation. For the second scenario, 1 used free-ranging
white-tailed deer (Odocoileus virginanus) radiomarked with GPS-VHF radio-transmitters

to compare automatically scheduled GPS ﬁxes to simultaneous VHF locations estimated

165

using triangulation under actual ﬁeld conditions. I also assessed cover-type assignment
consistency between GPS- and VHF-derived location estimates using a GIS.
STUDY AREA

My study took place during 2004—2006 in eastern Jackson (Grass Lake, Henrietta,
and Waterloo townships) and western Washtenaw (Dexter, Lima, Lyndon, and Sylvan
townships) counties in an agro-forest landscape in south central Lower Michigan. The
physiographic regions of this area are Hillsdale-Lapeer Hilly Upland, South Central
Rolling Plain, and Southeastern Rolling Plain (Sommers 1977:24) with alﬁsols as the
major soil order (Sommers 1977236). Surface formations in the study area are the result
of glaciation and include all 4 types present within Michigan (moraine, till plain, outwash
plain, and lacustrine plain; Sommers 1977:32). Elevation of the study area ranged
approximately 180—300 m and consisted of relatively limited relief (Sommers 1977:26,
33).
METHODS

I used the distance between GPS-estimated locations and VHF -estimated
locations to assess location precision. All statistical analyses were performed using
SYSTAT vll (Systat Software, Inc., San Jose, California, USA) and ProStat v4.02 (Poly
Software lntemational, Inc., Pearl River, New York, USA) software. I made
comparisons using 95% conﬁdence limits (CLs; LCL = lower, UCL = upper) and t-tests,
and used linear regression to assess the relationship between time to triangulate VHF-
location estimates and time to complete triangulations. To test for a systematic bias in
direction between GPS and VHF location estimates, 1 arbitrarily used GPS locations as

the origin on a coordinate system and plotted VHF location estimates in relation to this

166

origin. Circular statistics were used to determine mean angle if VHF location estimates
were not uniformly distributed using the Rayleigh test for randomness (Batschelet 1981)
at u = 0.05.
Stationary VHF Transmitters

During 8 time periods throughout the study, I placed 6 transmitters (either 3 each
of 2 models or 6 of 1 model) of VHF radio-transmitters in various cover types. Both
radio-transmitters were suitable for white-tailed deer, 1 for neonates (Model M4210;
Advanced Telemetry Systems, Isanti, Minnesota, USA), and 1 generally for deer 2 6
months old (Model 500; Telonics, Inc., Mesa, Arizona, USA) or similar-sized wildlife
species. I attempted to place radio-transmitters in a relatively systematic manner in
various cover types throughout the study area (e.g., agricultural areas, upland deciduous
forests, lowland shrubs), during several seasons, and at various distances from roadways
in an attempt to simulate actual ﬁeld conditions. Transmitter locations were estimated
using a handheld GPS unit (Model GPS 1V; GARMIN International, Incorporated,
Olathe, Kansas, USA).
Animal-borne GPS-VHF Transmitters

I captured deer during winter (Dec-Mar) 2004—2006, using single-door
collapsible live traps (Clover 1954). Traps were placed near areas of deer activity, baited
with kernel corn, and checked twice daily to minimize stress and injury to deer. My
technicians and 1 restrained captured deer using the collapsed trap and the body weight of
1 person (Sparrowe and Springer 1970). All trapped deer were blindfolded to reduce
stress and ﬁtted with metal ear tags bearing a unique identiﬁcation number (Style 681;

National Band and Tag, Newport, Kentucky, USA). Only adult female deer were ﬁtted

167

with mortality-sensing collar-style GPS-VHF radio-transmitters (Model G2000;
Advanced Telemetry Systems, Isanti, Minnesota, USA) due to transmitter weight (1,100
g). Radio-transmitters had a unique frequency within the ISO-MHz range and an
expected minimum battery life of 12 months. I scheduled GPS-ﬁx intervals for 1 ﬁx
every 9 hours (Jan-Nov) or 3 hours (Nov—Dec), and remotely released transmitters from
deer approximately 11 months post-deployment. Michigan State University’s All-
University Committee on Animal Use and Care approved all capturing and handling
procedures for my study (Application No. 01/04-006-00).
Triangulations

Observers located stationary VHF radio-transmitters and GPS-VHF-marked deer
using triangulation methods. Each of 6 observers was given transmitter frequencies and a
vague general description of where each transmitter signal could be received.
Transmitters used were based on availability at the time of each trial. Azimuths were
estimated using a 3-element folding Yagi antenna (Advanced Telemetry Systems,
Incorporated, Isanti, Minnesota, USA), portable radio receiver (Model R-IOOO,
Communications Specialists, Incorporated, Orange, California, USA), and mirror-
sighting compass. A handheld GPS unit was used to approximate the locations from
which signals were received. Observers also recorded date and time to complete each
triangulation. After triangulations of stationary VHF transmitters were completed,
observers recovered transmitters from the ﬁeld. Locations from triangulated data were
estimated using the program LOCATE III (Pacer, Truro, Nova Scotia, Canada). I used

the maximum likelihood estimator (Lenth 19810, b), as recommended by White and

168

Garrott (1990) and Nams and Boutin (1991). Ninety-ﬁve percent error ellipses were
estimated using LOCATE III for each location during analysis.
Cover-type Classification

I generalized land-use, land-cover data (Michigan Center for Geographic
Information 2001) using ArcView GIS v3.2 software (Environmental System Research
Institute, Redlands, California, USA) and Spatial Analyst extension to deﬁne 13 cover
types within the study area: agriculture (non-vegetated farmland, row crops, forage crops;
52.3% of study area); conifer (pines [Pinus spp.], other upland conifers; 1.5%);
herbaceous openland (herbaceous vegetation with <25% woody cover; 2.9%); lowland
deciduous forest (>60% composed of deciduous tree cover; 8.0%); lowland shrub (with
>60% non-water cover; 9.9%); mixed wetland (ﬂoating aquatic vegetation, emergent
wetland, mixed non-forest wetland; 3.1%); northern hardwood (>60% canopy cover of
maple, beech, ash [F raxinus spp.], cherry [Prunus spp.], birch [Betula spp.]; 2.3%); oak
association (>60% canopy cover of oak; 1.6%); upland deciduous forest (>60% canopy
cover of upland deciduous trees; 11.6%); upland shrub (>25% woody cover; <0.1%);
urban (low and high intensity, roads, parks, golf courses; 2.8%); water (surface, ﬂowing;
3.9%); and other (aspen [Populus spp.] association, orchards, bare ground; 0.1%). Patch
size of cover types ranged from <1 ha to >1 1,000 ha (e.g., an agricultural matrix) and had
a mean size of 29.2 ha.

I used a general vegetation canopy description of either open (agricultural areas
and grasslands) or closed (forest and shrub cover types) at the site of each stationary

transmitter to test for effects of vegetation on location precision. I also used the GIS

I69

cover-type classiﬁcation system to test for differences in cover-type assignment between
VHF and GPS location estimates for both stationary and animal-bome transmitters.
RESULTS

Stationary VHF Transmitters

For stationary transmitters, I found no differences among observers (n = 6
observers for 64 VHF location estimates) based on 95% CLs of distances between VHF
location estimates (triangulation) and GPS estimates of radio-transmitters (handheld GPS
unit). Observers completed each triangulation in 14.3 min (SE = 0.8) on average, and
were positioned an average distance of 465.3 m (SE = 18.7) from transmitters when
azimuths (n = 192) were estimated. I also found no transmitter model (95% CL = 113.1—
217.2 for neonate, n = 21; 95% CL = l36.2—178.0 for adult, n = 43) or season effects
(95% CL = 1309—1843 for spring, n = 24; 95% CL = 1298—1989 for summer, n = 34;
95% CL = 504—2339 for winter, n = 6). Error ellipses based on triangulations averaged
9.5 ha (SE = 2.0) and the distribution of error ellipse size was skewed right (Fig. A.l).
Locations estimated by the handheld GPS unit were contained within 38% of VHF error
ellipses.

The mean overall distance between VHF location estimates and GPS estimates
was 159.7 m (SE = 10.6, 95% CL = 138.5—181.0). When I plotted distance as a function
of the time needed to complete triangulation (Fig. A.2), no relationship existed. Using
the Rayleigh test, I found the angles between GPS and VHF location estimates to be
uniformly distributed (P z 0.8; Fig. A.3), indicating no systematic directional error

between GPS- and VHF-derived estimates.

170

Using my canopy classiﬁcation system, closed (95% CL = 1227—1793, n = 40)
and open (95% CL = 1410—2076, n = 24) canopy coverage did not seem to affect
precision between VHF and GPS location estimates. Cover-type assignments between
GPS- and VHF -derived locations were consistent 86% (55 of 64 instances) of the time.
One GPS-derived location was incorrectly assigned (water) using my GIS cover
classiﬁcation system, which was likely due to error caused by generalization of spatial
data.

Animal-borne GPS-VHF Transmitters

I captured 42 female deer, of which 4 adults were marked with GPS-VHF radio-
transmitters. The GPS-VHF radio-transmitters (n = 4) recorded a total of 2,219 GPS
ﬁxes (77% were 3-dimensional) to estimate locations of deer. One GPS-VHF radio-
transmitter failed to obtain any GPS ﬁxes aﬁer <1 month of deployment due to a faulty
antenna and failed to collect any simultaneous data. From the 3 GPS-VHF transmitters,
observers obtained only 20 VHF triangulations of location estimates that 1 considered
simultaneous (i.e., i 15 min) with GPS ﬁx times, so I pooled locations regardless of
individual or year. Ninety percent of these 20 GPS ﬁxes were 3-dimensional. Scheduled
ﬁxes from GPS-VHF radio-transmitters were routinely delayed, which greatly reduced
the ability to predict and obtain simultaneous VHF triangulations. A mean of 12.6 min
(SE = 0.8) was used by observers to complete VHF triangulations. Mean distance
between GPS ﬁxes from animal-bome GPS-VHF transmitters and simultaneous VHF
location estimates was 304.1 m (SE = 54.5; 95% CL = 190.0—418.3). Error ellipses

averaged 19.6 ha (SE = 7.4) and the frequency distribution of ellipse size was skewed

I71

right (Fig. A.4). Estimates of GPS-derived locations were contained within 30% of VHF
error ellipses.

I found a weak linear relationship when the distance between triangulated location
estimates and GPS ﬁxes (n = 20) of animal-home GPS-VHF radio-transmitters was
plotted as a function of time needed to complete each triangulation (y = 41 .4x - 217.8; r
= 0.62; Fig. A.5). Angles between GPS and VHF location estimates were not distributed
uniformly, however (P = 0.04; Fig. A.6); the mean angle between GPS and VHF
estimates was 9.7°.

Of 20 GPS ﬁxes, 13 (65%) were assigned to cover types consistently with VHF
location estimates. Mean GPS-to-VHF distances between consistently and inconsistently
assigned cover types were 284.6 m and 340.3 m, respectively, although 95% CLs
overlapped (l32.1—437.2 m for consistent assignment, 116.6—564.1 m for inconsistent
assignment); based on my sample size, distance may not have been related to consistency
of cover-type assignment.

DISCUSSION

My objective was to assess location precision between GPS- and VHF -
transmitters, and I consider both to provide estimated, not actual, locations. Locations
based on GPS ﬁxes certainly are not free from error, and this error can be highly variable
(Cargnelutti et al. 2007). Many variables can inﬂuence location precision with the GPS
units, including topography, animal behavior, vegetation, and brand of unit (Cain et al.
2005, Cargnelutti et al. 2007, Hebblewhite et al. 2007), similarly to VHF radio-
transmitters. Perhaps most importantly, observation-rate bias (i.e., systematically failed

GPS ﬁxes) from missing data could bias resource selection assignment more than

172

location errors (Johnson et al. 1998, Sager-Fradkin et al. 2007). Relatively ﬁne-scale
movements of animals may be particularly difﬁcult to ascertain from GPS-derived
location error (Ganskopp and Johnson 2007). This may have contributed to the lower
precision of animal-bome location estimates in comparison to stationary location
estimates, as well as the greater inconsistency of cover-type assignment of animal-borne
transmitters in comparison to stationary transmitters.

It is unclear whether animal movements, imprecise GPS ﬁxes, or vegetation
caused increased location error between VHF and GPS location estimates, but I had no
evidence that animal-home GPS units were free of location error or more accurate than

VHF radio-transmitters; in fact, some level of systematic error may have been present

among animal-home estimates (Fig. A6). The time (J_c = 12.6 min) observers needed to

complete triangulations on animal-bome transmitters was similar to that needed for

stationary transmitters (3c = 14.3 min), but as time to complete triangulations increased,
distances between GPS ﬁxes and VHF location estimates increased (Fig. A.5). This
provides some evidence that animal movements may have contributed to a lack of
precision between GPS- and VHF -derived location estimates, and if time to triangulate
was minimized, this type of error should be minimized. The primary advantage of GPS
units, depending on study objectives and unit model, would seem to be less ﬁeld effort
(e.g., VHF triangulations unnecessary due to automatic GPS ﬁxes) after deployment, and
a potentially larger location data set. Other GPS advantages could include data
describing temperature and altitude, and the remote release of GPS units to allow

recovery when desired.

173

Although I found no difference in precision when considering canopy cover (open
or closed) for stationary VHF transmitters, a larger sample size may provide more
information about canopy effects. The use of a handheld GPS unit by an observer to
estimate a location may provide different location error or bias than an animal-borne
GPS-transmitter with a ﬁx schedule. Canopy cover, especially in forested vegetation
types, has been shown to strongly bias GPS location estimates, which may be dependent
on sampling rates (DeCesare et al. 2005). Very-high-frequency systems are also subject
to the effects of vegetation, as well as the distance between the receiver and the
transmitter during triangulations (Chu et al. 1989). Researchers’ decisions on which
system is most appropriate for their study may be based on research objectives, ﬁnancial
constraints, species under study, cover types within the study area, and ﬁeld effort, as
each system has advantages and disadvantages.

MANAGEMENT IMPLICATIONS

Resource selection studies may be most inﬂuenced by location errors, as the
animal-home units and VHF triangulations were imprecise 35% of the time for cover-
type assignment. The accuracy of assignment, however, is also important and several
studies have assessed this issue. Erroneous assignment of resource use, especially due to
systematic error, could impact management decisions regarding habitat management and
other practices. Tests of location precision, and when possible, location accuracy should
be performed for studies using radiomarked animals, regardless of whether VHF or GPS
units are used. When assessing telemetry precision and accuracy, researchers should
consider study and sampling designs (e.g., when and how often to sample) as well as the

natural history (e.g., when and how often animals move) of the species under study.

174

LITERATURE CITED

Batschelet, E. 1981. Circular statistics in biology. Academic Press, New York, New
York, USA.

Cain, 111, J. W., P. R. Krausman, B. D. Jansen, and J. R. Morgart. 2005. Inﬂuence of
topography and GPS ﬁx interval on GPS collar performance. Wildlife Society
Bulletin 33:926—934.

Cargnelutti, B., A. Coulon, A. J. M. Hewison, M. Goulard, J.-M. Angibault, and N.
Morellet. 2007. Testing Global Positioning System performance for wildlife

monitoring using mobile collars and known reference points. Journal of Wildlife
Management 71 :1380—1387.

Chu, D. S., B. A. Hoover, M. R. Fuller, and P. H. Geissler. 1989. Telemetry location
error in a forested habitat. Pages 188—194 in C. J. Amlaner, Jr., editor.
Proceedings of the Tenth lntemational Symposium on Biotelemetry, University of
Arkansas Press, Fayetteville, Arkansas, USA.

Clover, M. R. 1954. A portable deer trap and catch-net. California Fish and Game
40:367—373.

DeCesare, N. J., J. R. Squires, and J. A. Kolbe. 2005. Effect of forest canopy on GPS-
based movement data. Wildlife Society Bulletin 33:935—941.

Ganskopp, D. C., and D. D. Johnson. 2007. GPS error in studies addressing animal
movements and activities. Rangeland Ecology and Management 60:350-358.

Hebblewhite, M., M. Percy, and E. H. Merrill. 2007. Are all Global Positioning System
collars created equal? Correcting habitat-induced bias using three brands in the
central Canadian Rockies. Journal of Wildlife Management 71:2026—2033.

Johnson, B. K., A. A. Ager, S. L. Findholt, M. J. Wisdom, D. B. Marx, J. W. Kern, and
L. D. Bryant. 1998. Mitigating spatial differences in observation rate of
automated telemetry systems. Journal of Wildlife Management 62:958-967.

Kanwisher, J. W., T. C. Williams, J. M. Teal, and K. O. Lawson, Jr. 1978.
Radiotelemetry of heart rates from free-ranging gulls. Auk 95:288—293.

LeMunyan, C. D., W. White, E. Nyberg, and J. J. Christian. 1959. Design of a miniature
radio transmitter for use in animal studies. Journal of Wildlife Management
23:107—110.

Lenth, R. V. 1981a. Robust measures of location for directional data. Technometrics
23:77—81.

175

Lenth, R. V. 1981b. On ﬁnding the source ofa signal. Technometrics 23:149—154.

Lonsdale, E. M., B. Bradach, and E. T. Thorne. 1971. A telemetry system to determine
body temperature in pronghorn antelope. Journal of Wildlife Management
35:747—751.

Michigan Center for Geographic Information. 2001. IF MAP land-use, land-cover data.
http://www.mcgi.state.mi.us/mgdl/?re|=ext&action=sext (Accessed 28 Jul 2005).

Nams, V. O., and S. Boutin. 1991. What is wrong with error polygons? Journal of
Wildlife Management 55:172—176.

Rodgers, A. R., R. S. Rempel, and K. F. Abraham. 1996. A GPS-based telemetry
system. Wildlife Society Bulletin 24:559—566.

Sager-Fradkin, K. A., K. J. Jenkins, R. A. Hoffman, P. J. Happe, J. J. Beecham, and R. G.
Wright. 2007. Fix success and accuracy of Global Positioning System collars in

old-growth temperate coniferous forests. Journal of Wildlife Management
71:1298—1308.

Sommers, L. M. 1977. Atlas of Michigan. Michigan State University Press, East
Lansing, Michigan, USA.

Sparrowe, R. D., and P. F. Springer. 1970. Seasonal activity patterns of white-tailed
deer in eastern South Dakota. Journal of Wildlife Management 341420—431.

White, G. C., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data.
Academic Press, San Diego, California, USA.

176

50—

40

 

 

 

>.
g 30
0)
:3
c"
2
IL.
20
10
p
olllllllllll [MIIIIMIIIJ
0 10 20 30 40 50 60 70 80 90 100
Area(ha)

Figure A. 1. Frequency distribution of error ellipses of stationary VHF radio-transmitters
using triangulation to estimate 64 locations in an agro-forest landscape, south central

Michigan, USA, 2004—2006.

177

500

400

300

Distance (m)

200

100

 

 

..O . .0. . . . O
I o. o o .
‘0 0 o
' : -;; =. ' '
1'. 1 1 1 1 l 1 1 . 1 I l
5 10 15 20 25 30 35
Time(min)

Figure A.2. Distances between triangulated VHF location estimates and GPS ﬁxes (n =

64) from a handheld unit of stationary VHF radio-transmitters as a function of time to

complete each triangulation, south central Michigan, USA, 2004—2006.

178

300

 

 

 

 

200 — '
_ \ l
100 — l
E :
; 7 \' \\ ;/"/
0 — .L
P /
-100 ~—
-200 *— J
'— D
_300bLIllLIIIJJlllJlllllllLllllllllLl
-500 -400 -300 -200 -l 00 0 100 200 300
X011)

Figure A.3. Distribution of location estimates of stationary VHF radio-transmitters using
triangulation and point estimates using a handheld GPS unit. Distribution reﬂects the
relationship between 64 GPS ﬁxes indexed with the coordinate origin and VHF location

estimates.

179

14—

12

10

Frequency

       

 

[Irllllllll .1.L.|,l.iID/rlrl

0
0 10 20 30 40 50 60 70 80 90 100110120130140150
Area(ha)

Figure A.4. Frequency distribution of error ellipses of animal-borne GPS-VHF radio-
transmitters using triangulation to estimate 20 locations of white-tailed deer in an agro-

forest landscape, south central Michigan, USA, 2004—2006.

180

 

 

_ A ' ’
800 — ,
l— , "
a ‘ A
_ A
E 600 —
V
8 T A . '
§ C
m x
'5 400 P
_ ’A’
_ a," ‘ ‘
200 — j ‘ A ‘
.. ”I A
A A
C A A
0 1 1.!” 1 l 1 l 1 l 1 l 1 J
0 5 10 15 20 25 30
Time (min)

Figure A.5. Distances between triangulated location estimates and GPS ﬁxes (n = 20) of
animal-bome GPS-VHF radio-transmitters as a function of time to complete each
triangulation, south central Michigan, USA, 2004—2006. Dashed line represents linear

regression model (y = 41 .4x - 217.8; r = 0.62) describing relationship.

181

800

600

400

Y (m)

200

TllllITlllllllllllIlll]
\

 

is
[8
O
O
O
L_
8
'_
N
O—
O
Or—
1.
N
C—
O
A
Cr—
C
1..
r—
$__
0

Figure A.6. Distribution of location estimates using animal-home GPS-VHF radio-
transmitters, south central Michigan, 2004—2006. Distribution reﬂects the relationship
between 20 GPS ﬁxes indexed with the coordinate origin and simultaneous VHF location

estimates of radiomarked white-tailed deer (n = 3) using triangulation.

182

APPENDIX B

Capture and Fate Descriptions of White-tailed Deer

183

Table B. 1. Capture and fate information of radiomarked female white-tailed deer

captured using Clover traps, south central Michigan, USA, winters 2004—2006.

 

 

Age Class
Tag Date of Capture at Capturea F ateb
MSU 10 1/31/2005 A capture myopathy
MSU 75 1/29/2004 Y collar slipped
MSU 86 2/24/2004 A hunter-harvest
MSU 87 2/6/2004 Y hunter-harvest
MSU 88 2/16/2004 A unknown
MSU 89 3/12/2004 Y vehicle collision
MSU 91 2/3/2004 F collar slipped
MSU 94 2/3/2004 Y hunter-harvest
MSU 95 3/15/2004 A hunter-harvest
MSU 96 2/4/2004 F collar slipped
MSU 97 1/6/2005 Y transmitter failure
MSU 98 3/10/2004 Y removed GPS collar
MSU 101 1/8/2005 Y trauma (old)
MSU 104 l/11/2005 A hunter-harvest
MSU 105 1/12/2005 F collar trauma
MSU 106 1/21/2005 F trauma, malnutrition
MSU 107 1/11/2005 F transmitter failure
MSU 109 1/25/2005 Y alive
MSU 110 1/25/2005 F unknown
MSU 113 1/8/2005 Y hunter-harvest
MSU 115 1/20/2005 F unknown
MSU 116 1/27/2005 A collar slipped
MSU 117 1/27/2005 F alive at end of study
MSU 120 1/27/2005 Y hunter-harvest
MSU 122 3/3/2005 A removed GPS collar
MSU 123 3/23/2004 A hunter-harvest
MSU 124 1/6/2005 Y capture myopathy
MSU 134 12/21/2004 F collar slipped
MSU 136 1/31/2005 A alive
MSU 138 2/24/2005 Y hunter-harvest
MSU 139 3/2/2005 F alive
MSU 148 2/9/2006 F alive
MSU 159 1/20/2006 A alive
MSU 160 3/16/2006 F alive
MSU 162 2/11/2006 A hunter-harvest

 

184

Table 8.1. (cont’d)

 

 

Age Class

Tag Date of Capture at Capturea F ateb
MSU 163 2/9/2006 F vehicle collision
MSU 166 2/28/2006 A hunter-harvest
MSU 167 1/10/2006 F unknown
MSU 169 2/11/2006 F dropped break-away

collar
MSU 171 2/24/2006 A removed GPS collar
MSU 176 12/20/2004 A hunter-harvest
MSU 178 1/8/2005 F collar slipped
unknown 2/17/2004 F capture myopathy

 

2'F = fawn (<1 yr old), Y = yearling (2 1—<2 yr old), A = adult (22 yr old).

bUnknown included deer that may have dispersed long distances or unconﬁrmed

transmitter failures; alive = alive at conclusion of ﬁeld data collection.

185

 

Table 3.2. Capture and fate information of ear-tagged male white-tailed deer captured

using Clover traps, south central Michigan, USA, winters 2004—2006.

 

 

Age Class

Tag Date of Capture at C apturea Fate
MSU 80 2/11/2004 A unknown
MSU 85 1/28/2004 F unknown
MSU 90 2/25/2004 Y unknown
MSU 92 1/6/2005 F unknown
MSU 93 2/19/2004 F unknown
MSU 99 2/2/2004 A unknown
MSU 102 1/6/2005 F unknown
MSU 103 1/11/2005 F unknown
MSU 108 1/21/2005 F unknown
MSU 111 12/20/2004 F hunter-harvest
MSU 112 1/26/2005 F unknown
MSU 114 1/13/2005 F hunter-harvest
MSU 118 1/26/2005 F hunter-harvest
MSU 125 3/24/2004 F unknown
MSU 135 12/21/2004 F unknown
MSU 140 3/8/2005 F unknown
MSU 142 2/5/2005 Y vehicle collision
MSU 144 2/2/2005 F vehicle collision
MSU 161 1/10/2006 F hunter-harvest
MSU 165 1/9/2006 F unknown
MSU 168 2/24/2006 F unknown
MSU 170 1/25/2006 F unknown
MSU 175 3/17/2006 F hunter-harvest
MSU 177 1/11/2005 Y hunter-harvest

 

aF = fawn (<1 yr old), Y = yearling (21—<2 yr old), A = adult (2 2 yr old).

186

Table 8.3. Capture and fate information of neonate white-tailed deer captured during

May—June 2004—2006 and radiomarked with break-away radio-collars, south central

 

 

Michigan, USA.

Estimated Age at
Tag Date of Capture Sex Capture (d) F atea
MSU 126 5/21/2004 M 8.0 canid-kill
MSU 127 5/25/2004 M 8.4 malnutrition
MSU 128 5/26/2004 M 8.4 collar dropped
MSU 129 5/27/2004 F 7.2 hunter-harvest
MSU 130 5/30/2004 F 6.9 vehicle collision
MSU 131 6/1/2004 M 6.8 unknown
MSU 132 6/4/2004 F 5.8 collar dropped
MSU 133 6/11/2004 F 10.3 collar dropped
MSU 137 5/25/2005 M 8.1 canid-kill
MSU 143 5/28/2005 F 7.4 vehicle collision
MSU 145 5/16/2005 M 7.7 canid-kill
MSU 146 5/25/2005 F 7.2 canid-kill
MSU 147 5/26/2005 F 6.6 collar dropped
MSU 149 5/31/2005 M 9.2 unknown
MSU 150 5/17/2004 M 5.5 hunter-harvest
MSU 151 5/28/2005 M 8.4 collar dropped
MSU 153 5/26/2005 M 7.0 alive
MSU 155 5/31/2005 M 8.5 canid-kill
MSU 156 5/29/2005 F 7.3 collar dropped
MSU 157 5/17/2006 M 6.3 hunter-harvest
MSU 158 5/17/2006 M 6.6 hunter-harvest
MSU 182 5/27/2006 F 8.2 collar dropped
MSU 183 5/17/2006 M 6.0 alive
MSU 184 6/2/2006 M 7.0 alive
MSU 185 5/9/2006 M 6.6 alive
MSU 186 5/17/2006 M 8.7 alive
MSU 187 5/18/2006 F 5.9 capture mortality
MSU 188 5/25/2006 F 8.2 alive
MSU 189 5/25/2006 F 6.1 alive
MSU 190 5/25/2006 F 5.7 alive
MSU 191 6/1/2006 M 7.3 alive
MSU 192 6/1/2006 M 5.5 alive
MSU 193 6/5/2006 F 4.0 vehicle collision
MSU 194 6/5/2006 F 7.5 unknown injuryb

 

187

aUnknown cause of mortality due to consumption of carcass; alive = alive at conclusion
of ﬁeld data collection.

bFound injured by a wildlife rehabilitator and censored from study.

188

 

   

11111111111111111111111