wig:
» a... run. rant?» .

7.:
a, Em?
. .

I

t ....
:1.
I.

c 9.3..
3 - 3.9

(ﬂ

ﬁn,
d a...
A?

“WI-‘1‘

;
.
A L”.
‘ x... a “ﬂaunt! .

,... .mmwuﬁ

.
p
n
.
..
v
n
.0

, . (11:317.
29. ... ..a;
..H £3... Winn?
.. 4...?! .... 3.
£334”?
argfaz’

‘-I~n. a. ...

..tc‘. .

7.
t r.

ﬂaunt“.

..
.... 19,.
.1! 2.21.5.3 ...:
. . .t t . 5....
BE. .39..
.{

 

: ; ......» (.3; x! ., vylmuxmasynkxm 33.x}.
. * V

 

 

Hit-$13
/.

£2» 3
{(5, l/ 5/ $3

This is to certify that the
dissertation entitled

GENEALOGICAL RELATIONSHIPS INFLUENCE THE
PROBABILITY OF INFECTION WITH BOVINE
TUBERCULOSIS AND MICROGEOGRAPHIC GENETIC
STRUCTURE IN FREE-RANGING WHITE-TAILED DEER

ate ’

'1

presented by

UDHAHY
lchigan St

Uni' *‘

 

W

JULIE ANNE BLANCHONG '

has been accepted towards fulfillment
of the requirements for the

DOCTORAL degree in FISHERIES & WILDLIFE;

ECOLOGY, EVOLUTIONARY
BIOLOGY & BEHAVIOR

Major Professor’s Signature

(43%);

Date

 

 

 

MSU is an Afﬁrmative Action/Equal Opportunity Institution

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

 

DATE DUE DATE DUE DATE DUE

[21111115

(’1')

 

U

‘MAY 07 2005

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DatoDue.p65-p.15

 

GENEALOGICAL RELATIONSHIPS INFLUENCE
THE PROBABILITY OF INFECTION WITH BOVINE TUBERCULOSIS
AND MICROGEOGRAPHIC GENETIC STRUCTURE
IN FREE-RANGING WHITE-TAILED DEER

By

Julie Anne Blanchong

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;
Ecology, Evolutionary Biology, and Behavior Program

2003

ABSTRACT

GENEALOGICAL RELATIONSHIPS INFLUENCE
THE PROBABILITY OF INFECTION WITH BOVINE TUBERCULOSIS
AND MICROGEOGRAPHIC GENETIC STRUCTURE
IN FREE-RANGING WHITE-TAILED DEER
By

Julie Anne Blanchong

Zoonoses are of increasing importance to wildlife conservation and human health.
It has become increasingly recognized that wildlife ecology plays a key role in disease
transmission in wildlife populations. In domestic populations, contacts among individuals
are controlled by humans and disease transmission is often density dependent. Unlike
domestic animals, wildlife populations often have complex social systems in which
contacts among individuals are not solely density dependent, and can play an important
role in the transmission and maintenance of disease.

White-tailed deer (Odocoileus virginianus) in the northeast lower peninsula of
Michigan (MI) are infected with bovine tuberculosis (Mycobacterium bovis) (TB). The
practice of artiﬁcial feeding that brings large numbers of deer into contact likely
facilitated the transmission of TB. Artiﬁcial feeding was banned in order to reduce
contacts among individuals at artiﬁcial feeding sites and thereby reduce the risk of TB
infection. White-tailed deer ecology may also play an important role in the probability of
infection with TB. Deer have a complex social system in which females live in related
groups (matrilines). The rate of contact among individuals within matrilines (relatives) is
high relative to contact rates among individuals from different matrilines. Estimates of

genealogical relationships were used to infer the role that white-tailed deer’s social

structure played in the risk of TB infection. TB-infected deer were signiﬁcantly more
closely related than were non-infected deer. Relatedness between pairs of TB-positive
deer was higher than relatedness between non-infected deer independent of the
occurrence of artiﬁcial feeding. The probability of TB infection was not random, but
rather a function of genetic relatedness to TB-infected individuals. It can be inferred from
these data that matrilines of deer likely serve as reservoirs of TB within the deer
population.

White-tailed deer’s matrilineal social structure would be expected to result in
heterogeneity in allele frequencies among matrilines across space. Artiﬁcial feeding of
the deer population in MI, however, results in the congregation of large numbers of
individuals at artiﬁcial feeding sites. Molecular markers were used to characterize the
impact of artiﬁcial feeding on deer spatial genetic structure in the northeast lower
peninsula of MI. Spatial autocorrelation analyses indicated that when artiﬁcial feeding
was occurring there was no Signiﬁcant relationship between the degree to which quadrats
of deer were genetically differentiated and their geographic distance from one another.
The aggregation of multiple matrilines at feeding sites likely homogenized spatial genetic
structure. Following the ban on artiﬁcial feeding, there was signiﬁcant heterogeneity in
allele frequencies among groups of deer as a function of geographic distance. The
signiﬁcant microgeographic genetic structure that exists within the deer population
following the ban on artiﬁcial feeding indicates that transmission of TB across

genetically differentiated groups is likely to be limited.

ACKNOWLEDGMENTS

I would like to thank a number of individuals who made this research possible. I
thank my advisor, Dr. Kim T. Scribner, and the members of my committee, Dr. Bryan K.
Epperson, Dr. Jack Liu, Dr. Barbara L. Lundrigan, and Dr. Scott R. Winterstein for their
guidance. I thank Kristine Bennett, Brandi Hughey, and the numerous Michigan State
University undergraduates for their assistance in sample collection. I thank Dr. Scott
Fitzgerald and the Animal Health Diagnostic Laboratory for providing space for sample
collection. I thank Dr. Steve Schmitt, Jean Fierke, and the Wildlife Division of the
Michigan Department of Natural Resources for providing funding, materials to collect
samples, maps, and access to databases. I thank Scot Libants, Jennifer Warrillow,
Kristine Bennett, Laura Garzel, Veronica Lopez, and several undergraduate assistants for
their contributions to laboratory data collection. I thank Dr. Sasha Kravchenko for
assistance in analyzing portions of data in Chapter 2. I thank Dr. Arnold Saxton from the
University of Tennessee for providing SAS programs used to analyze portions of data in
Chapter 2. Finally, I would like to thank all of the members of Dr. Scribner's laboratory

for their contributions to data collection, data analysis, and overall moral support.

iv

TABLE OF CONTENTS

LIST OF TABLES .................................................................................. vii

LIST OF FIGURES ....................................... x

CHAPTER 1

INTRODUCTION .................................................................................... 1
Historical background of white-tailed deer and bovine tuberculosis in Michigan..l
Models of disease transmission ............................................................ 7
Utility of molecular markers for studying disease in wildlife populations ......... 12
Sample collection and general methods ................................................. 14

CHAPTER 2

GENEALOGICAL RELATIONSHIPS INFLUENCE THE PROBABILITY OF
INFECTION WITH BOVINE TUBERCULOSIS H\l FREE-RANGING WHITE-

TAILED DEER ...................................................................................... 15
Methods ...................................................................................... 21
Relatedness Among TB-positive and Non-infected Deer Across All of
DMU452 ............................................................................ 26
Relatedness Among TB-positive and Non-infected Deer at Distances
Consistent With Deer Movement in Northeast Michigan .................... 28
Results ....................................................................................... 28
Relatedness Among TB-positive and Non-infected Deer Across All of
DMU452 ........................................................................... 29
Relatedness Among TB-positive and Non-infected Deer at Distances
Consistent With Deer Movement 1n Northeast Michigan .................... 35
Discussion ................................................................................... 37
CHAPTER 3
WHITE-TAILED DEER SPATIAL GENETIC STRUCTURE PRIOR TO AND
COIN CIDENT WITH CHANGES IN ARTIFICIAL FEEDING REGULATIONS ...... 46
Spatial Structure in White-Tailed Deer .................................................. 46
White-Tailed Deer in Michigan .......................................................... 48
Methods ...................................................................................... 52
Results ....................................................................................... 56
Discussion ................................................................................... 65
CHAPTER 4

A SIMPLE MODEL HIGHLIGHTING THE IMPORTANCE OF SOCIAL
STRUCTURE IN THE MAINTENANCE OF BOVINE TUBERCULOSIS IN FREE-

RANGING WHITE-TAILED DEER ............................................................ 69
Methods ...................................................................................... 73
Results ....................................................................................... 78
Discussion ................................................................................... 8 1

CONCLUSIONS .................................................................................... 86

Genealogical Relationships and the Probability of TB Infection .................... 86
Microgeographic Genetic Structure in Deer ............................................ 87
Future Research Directions ................................................................ 88
Contributions to Management ...................... _ ...................................... 89
General Contributions ...................................................................... 90
APPENDIX I
MICROSATELLITE LOCI SCREENED IN WHITE-TAILED DEER FROM
DMU452 .............................................................................................. 92
APPENDIX 11
POLYMERASE CHAIN REACTION PROTOCOLS FOR OPTIMIZED WHITE-
TAILED DEER PRIMERS ......................................................................... 93
REFERENCES ...................................................................................... 95

vi

LIST OF TABLES

TABLE 1. Number of harvested deer screened for TB and number of TB-positive deer in

Michigan from 19954000...................................; ....................................... 6

TABLE 2. Number of harvested deer screened for TB and number of TB-positive deer in

Michigan from 1995-2000 ......................................................................... 17

TABLE 3. Sampling area and number of deer used in the estimation of genealogical
relationships among TB-positive deer (TB+) and among non-infected deer (TB-) from

1998-2000 ............................................................................................ 22

TABLE 4. Mean heterozygosity and allelic diversity(:SE) among TB-positivc deer (+)

and among non-infected deer (-) for 1998-2000 ................................................ 30

TABLE 5. Mean pairwise estimates of relatedness among TB-positive deer (+) and
among non—infected deer (-) during 1998-2000 as measured by Queller and Goodnight’s
(1989) QGr,‘y (iSE), Bowcock et al.’s (1994) Similarity, and Lynch and Ritland’s (1999)
LRrxy. Associated p-values are based on permutation tests of differences in mean
relatedness between TB-positive deer and non-infected deer for each measure for each

year .................................................................................................... 3 1

vii

TABLE 6. Results of Mantel tests of correlations between genetic and geographic
distance including results of permutation tests of signiﬁcance and associated p-values for

TB-positive deer (+) and for non—infected deer (-) during 1998-2000 ....................... 36

TABLE 7. Mean Qery and associated p-values between TB-positive deer (+) and

between TB-negative deer (-) at different distances of geographic separation (km) ...... 40

TABLE 8. Mean similarity and associated p-values between TB-positive deer (+) and

between non-infected deer (-) at different distances of geographic separation (km) ...... 41

TABLE 9. Number of deer genotyped and number of quadrats in each network for the

North and South regions during 1998, 1999, and 2000 ........................................ 53

TABLE 10. Heterozygosity and allelic richness (iSE) for deer genOtyped in the North

and South regions during 1998, 1999, and 2000 ................................................ 57

TABLE 11. Measures of allelic variance within and among quadrats for the network of

26-ka quadrats for the North and South regions during 1998, 1999, and 2000 .......... 59

TABLE 12. Results of Mantel tests of correlations between genetic differentiation
between quadrats and their geographic distance including results of permutation tests of
signiﬁcance and associated p-values for the network of 233-ka quadrats for the North

and South regions during 1998, 1999, and 2000 ............................................. 6O

viii

TABLE 13. Results of Mantel tests of correlations between genetic differentiation
between quadrats and their geographic distance including results of permutation tests of
signiﬁcance and associated p-values for the network of 26-ka quadrats for the North

and South regions during 1998, 1999, and 2000 ................................................ 63

TABLE 14. Initial values of parameters included in the model of TB in deer in northeast

Michigan ............................................................................................. 76

TABLE 15. Model predictions of population density, number of unrelated interactions,
and number of TB-positive deer in northeast Michigan over 20 years. Years in which

Speciﬁc prevalence goals were set by management are 2003 (0.5%) and 2010 (0%). . ....79

TABLE 16. Number of TB-positive deer in northeast Michigan as predicted by the model

and empirical data from 1998—2001 ............................................................... 80

TABLE 17. Number of TB-positive deer in northeast Michigan in 2010 when initial

values of parameters were varied ............................................................... 82

LIST OF FIGURES

FIGURE 1. Map of the lower peninsula of Michigan highlighting the 5 county region

where the deer population is infected with bovine tuberculosis (DMU452) .................. 4

FIGURE 2. Map of the lower peninsula of Michigan highlighting the 5 county region

where the deer population is infected with bovine tuberculosis (DMU452) ............... 16

FIGURE 3. Distribution of Queller and Goodnight’s estimates of relatedness (Qery)
among TB-positive deer and among non-infected deer in the year prior to the ban on
artiﬁcial feeding (a), in the year in which artiﬁcial feeding was banned (b), and in the

year following the ban on artiﬁcial feeding (c) ................................................. 32

FIGURE 4. Distribution of Bowcock et al.’s Similarity values among TB-positive deer
and among non-infected deer in the year prior to the ban on artiﬁcial feeding (a), in the
year in which artiﬁcial feeding was banned (b), and in the year following the ban on

artiﬁcial feeding (c) .................................................................................. 33

FIGURE 5. Distribution of Lynch and Ritland’s estimates of relatedness (LRrxy) among
TB-positive deer and among non-infected deer in the year prior to the ban on artiﬁcial
feeding (a), in the year in which artiﬁcial feeding was banned (b), and in the year

following the ban on artiﬁcial feeding (c) ....................................................... 34

FIGURE 6. Queller and Goodnight estimates of relatedness (Qery) among TB-positive
deer and among non-infected deer at different distances of geographic separation in the
year prior to the ban on artiﬁcial feeding (a), in the year in which artiﬁcial feeding was
banned (b), and in the year following the ban on artiﬁcial feeding (c). Asterisks indicate
signiﬁcant differences in mean pairwise relatedness between TB-positive and non-

infected deer (p<0.05) .............................................................................. 38

FIGURE 7. Bowcock et a1. Similarity values among TB-positive deer and among non-
infected deer at different distances of geographic separation in the year prior to the ban
on artiﬁcial feeding (a), in the year in which artiﬁcial feeding was banned (b), and in the
year following the ban on artiﬁcial feeding (c). Asterisks indicate signiﬁcant differences

in mean pairwise relatedness between TB-positive and non-infected deer (p<0.05) ...... 39

FIGURE 8. Map of northeast MI (DMU452) and networks for the ”analysis of spatial
genetic structure among “subpopulations” of deer. In the North region is a depiction of
the network of 233-ka quadrats. Each point indicates the center of the 233-ka quadrat
and the box is an example of a single 233-ka quadrat. In the South region is an example

of the network of 26-1an quadrats ............................................................... 50

FIGURE 9. Correlation between quadrats of deer separated by different geographic
distances in the North and South regions for the network of 233-1an quadrats in the year
prior to the ban on artiﬁcial feeding (1998) (a), the year in which feeding was banned

(1999) (b), and the year following the ban on feeding (2000) .............................. 61

FIGURE 10. Correlation between quadrats of deer separated by different geographic
distances in the North and South regions for the network of 26-ka quadrats in the year
prior to the ban on artiﬁcial feeding (1998) (a), the year in which feeding was banned

(1999) (b), and the year following the ban on feeding (2000) ................................. 64

FIGURE 11. A conceptual model of TB transmission among deer in the northeast lower

peninsula of Michigan .............................................................................. 74

xii

CHAPTER 1

INTRODUCTION

Historical background of white-tailed deer and bovine tirberculosis in Michigan

The white-tailed deer (Odocoileus virginianus) is the oldest deer species in North
America, ﬁrst appearing in the late Pliocene before the glaciations of the Pleistocene
(Geist et a1. 2000). Prior to European settlement, populations of white-tailed deer were
widespread throughout what is currently the United States, and may have approached 26
million animals (Geist et a1. 2000). In Michigan (MI), before human settlement, white-
tailed deer were most abundant in the southern part of the state where the habitat
primarily consisted of open oak and hickory forests ideal for deer (Langenau 1994,
Michigan Department of Natural Resources (MDNR) 1993). In the northern lower
peninsula of MI, forests were primarily pine and hardwood that did not provide optimal
habitat for deer (MDNR 1993). When settlers ﬁrst arrived in MI in the 18508, their
clearing of land through logging and burning initially increased deer habitat quality
(Langenau 1994, NHDNR 1993). However, as settlement progressed and the human
population grew larger, over-clearing, and buming of forests increased the openness of
the land and decreased its quality for deer (Langenau 1994). In addition, increased
numbers of settlers resulted in an over-harvest of the deer population (Langenau 1994). In
the 1870s, the deer population in MI was estimated to be one million and had
dramatically declined in southern MI as the result of decreased habitat quality and

excessive commercial harvest (Langenau 1994, MDNR 1993). Most deer in the late

18705 were found in the northern lower peninsula where white pine logging created
openings in the forest resembling habitat found in presettlement southern MI (MDNR
1993). At the end of the 19‘h century, the deer population in MI had declined to fewer
than 500 thousand due to heavy market hunting and uncOntrolled logging (MDNR 1993).
Extensive ﬁres on land after logging slowed the recovery of forests, and deer were not
able to survive in open grasslands that did not provide sufﬁcient food or winter cover
(MDNR 1993).

During the early 20th century, the Lacy Act ending commercial deer harvest
together with Pittman-Robertson funds for habitat restoration allowed the deer herd in M1
to increase to about 1.5 million animals by 1949 (MDNR 1993). Most of these deer were
found in the northern lower peninsula where habitat was best and the human population
was smaller than in the southern part of the state (MDNR 1993). However, deer habitat
quality declined again in the 19505 as second growth forest matured leaving few
openings and poor food resources (MDNR 1993). As a result of habitat decline and an
increase in antlerless hunting, the deer population declined in MI such that by 1972, only
approximately 85 thousand deer remained in southern MI with fewer than 500 thousand
in the entire state (MDNR 1993). In recent years, an increase in the timber industry along
with mild winters and supplemental feeding has allowed the deer population to increase
(MDNR 1993). By the mid-19903 deer numbers neared 2 million and densities in
northeastern MI averaged >35/ km2 (Schmitt et al. 1997, Hickling 2002).

Carrying capacity for white-tailed deer in forested habitats in the northeast lower
peninsula of MI was historically low. During the early 20th century, numerous hunting

clubs were established in northeast MI. Due to high hunting demand, shortage of natural

browse was alleviated through artiﬁcial feeding to enhance deer population numbers.
Artiﬁcially high deer densities resulting from artiﬁcial feeding practices has the potential
to cause damage to the natural vegetation and to have negative effects on the deer
population itself by increasing the transmission and prevalence of disease (Chouinard and
Filion 2001, Miller et a1. 2003).

Currently, the white-tailed deer population in MI, primarily in the northeast lower
peninsula, is infected with bovine tuberculosis (Mycobacterium bovis) (TB) (Figure 1).
This is the only occurrence of self-sustaining bovine TB in a free-ranging cervid
population in North America. There are no well-established control programs for TB in
wild deer, and at the time of its discovery in MI, much about the transmission of bovine
TB was unknown. TB has also been discovered in cattle herds (dairy and beef) and other
wildlife (bears (Ursus americanus), bobcats (Lynx rufus), coyotes (Cam's latrans), elk
(Cervus elaphus), opossums (Didelphus virginiana), raccoons (Procyon lotor), and red
foxes ( VulpeS vulpes)) within the state. As a result of the presence of TB in livestock, the
state’s TB-status was downgraded from “accredited-free” to “modiﬁed accredited-free”
(USDA 2000). The loss of “accredited-free” status has resulted in movement restrictions
and mandatory TB testing for livestock. The costs of the imposed regulations on the beef
and dairy industry, TB testing of wildlife and livestock, and compensation for
depopulated herds has made Michigan’s TB problem a multimillion dollar issue. There is
a collaborative research effort among wildlife biologists, veterinarians, microbiologists,
epidemiologists, and managers to understand and control bovine tuberculosis in free-
ranging wildlife and domestic animals. Research indicates that white-tailed deer are the

reservoir host for TB in MI while other wildlife are hypothesized to be only spillover or

 

O I I I L.
"'"'I'""I'""I"""I""'I
! I ! !‘ L. . ""T """
-.-..J L._.-L...,...L._,.... ---.rl !

L._..L.... i i ,' .---{.

.J " o o .—.Lou1- L ’._.

t—H - -
I I I I I ...-L...o—I---

i i i i. I !
—-1—-—-.r-A0'-’—OT-JIL-~.rLo—o‘Foéc'Lrg’unul-d-o

“I'm-iMIéH hﬂ'i"'r‘i"“~ N

“3:: +

Figure 1. Map of the lower peninsula of Michigan highlighting the 5 county region where
the deer population is infected with bovine tuberculosis (DMU452).

dead end hosts (Schmitt et a1. 1997). One of the goals in studying TB in MI, as stated by
the United States Department of Agriculture (USDA), is to determine the long-term,
baseline prevalence of TB and the rate of TB transmission in white-tailed deer.

During the early 20th century, bovine TB was found in high prevalence within
cattle populations throughout MI. As the result of an intensive eradication program, MI
livestock gained USDA TB-free accreditation in 1979 (MDNR Activity Report 1997).
However, in 1975, a TB-infected deer was harvested in Alcona county and a second TB—
positive deer was identiﬁed in Alpena county in 1994 (Schmitt et al. 1997). In 1995, the
Michigan Department of Natural Resources (MDNR) screened livestock and hunter-
harvested deer for TB in the counties surrounding the region where the ﬁrst 2 TB-positive
deer were harvested. While no infected livestock were found to be infected at that time,
an additional 18 TB-infected deer were identiﬁed (Table 1). The 5 counties from which
TB-infected deer were harvested in the northeast lower peninsula of M1 were designated
the main TB management area, named Deer Management Unit 452 (DMU452), and
included Alcona, Alpena, Montrnorency, Oscoda, and Presque Isle counties. The area in
the northeast lower peninsula of MI designated DMU452 has changed over time. For the
purposes of this dissertation, DMU452 will refer to the 5 county area that DMU452
encompassed at the beginning of this study. From 1995 to the present, hunters were asked
to voluntarily submit the heads of deer harvested from throughout MI to MDNR check
stations. Deer heads were sent to the MI Animal Health Diagnostic Laboratory where the
cranial lymph nodes were visually inspected for signs of infection with TB. Suspected
infections were cultured and serially tested with a genetic probe for M. bovis serving as

the deﬁnitive test. The MDNR’S survey resulted in the identiﬁcation of an additional 336

Table 1. Number of harvested deer screened for TB and number of TB-positive deer in
Michigan from 1995-2000.

 

 

 

Year Deer TB-positive
Screened Deer
1995 403 18
1996 4967 56
1997 3720 73
1998 9057 78
1999 19500 58
2000 25858 53

 

 

 

 

TB-positive deer in MI from 1995-2000 (Table 1) (O’Brien et al. 2002). In addition to
increased surveillance of wildlife for TB, management is aimed at eradication of TB from

the deer population in order to eliminate the spillover of TB into livestock.

Models of disease transmission

Interactions between host and pathogen populations are driven by transmission
dynamics (Begon et al. 1999). Host-pathogen models are used to design management
strategies for disease control (McCallum et al. 2001). Traditionally, in the majority of
both empirical and theoretical studies, disease transmission dynamics have been assumed
to be density dependent (mass action) (Anderson and May 1992, Grenfell and Dobson
1995, Begon et a1. 1999, McCallum et al. 2001). In density dependent models of disease
transmission, contacts among infected and susceptible individuals are assumed to be
random and to increase in proportion to population Size (Begon et a1. 1999). Culling in
order to reduce population density is a common management strategy for controlling
disease based on density dependent models (McCallum et al. 2001). This policy will fail,
however, if disease transmission is not host-density dependent (McCallum et al. 2001).

Alternatively, disease transmission dynamics may be frequency dependent where
contacts among infected and susceptible individuals are independent of population size
(Begon et al. 1999). In addition, incidence of infection is often patchily distributed, where
infected individuals are likely to be closer to other infected individuals than expected
under random mixing, the typically assumed model of disease transmission (McCallum et

al. 2001). The manner in which transmission dynamics are modeled will have profound

effects on the success of management actions aimed at eliminating disease transmission
(Dobson and Meagher 1996).

Most zoonotic pathogens are endemic, always present at some low level, in
wildlife populations (Childs et al. 1998, Begon et a1. 1999). Most empirical studies of
disease transmission in wildlife, however, have been conducted opportunistically in
response to epidemic outbreaks which generally occur over short time periods (Begon et
al. 1999). Endemic diseases are rarely easily eradiated, and ecological changes can
promote their ampliﬁcation to epidemic levels (increased incidence) (Childs et a1. 1998).
AS such, it is important to understand the key mechanisms of disease transmission to
devise effective control measures (Childs et al. 1998).

Recently, numerous studies have focused on determining the mechanisms of
disease transmission in natural populations. Boone et al. (1998) evaluated the ecological
correlates of hantavirus in deer mice. They found a complex non-linear relationship
between density and disease prevalence such that above a certain threshold, density of
deer mice no longer inﬂuenced hantavirus prevalence (Boone et al. 1998). In addition,
there are often differences in the frequency of infection among demographic segments of
wildlife populations (i.e. age, sex) (Mills and Childs 1998). Patterns of cowpox infection
in rodents vary as a function of age and sex (Hazel et al. 2000). The difference between
male and female tick dispersal and host speciﬁcity may be important in understanding
Lyme disease transmission (Meeus et al. 2002). Higher rates of Puumala virus infection
among adult male bank voles in comparison to the rest of the population was used to infer
that aggressive interaction among males was a major mechanism of disease transmission

(Escutenaire et al. 2002).

The management actions taken to eliminate TB from free-ranging deer have been
built upon a density dependent model of epidemic disease transmission. TB management
strategies focus on actions aimed at minimizing social contact and crowding of deer at
artiﬁcial feeding sites and on decreasing population numbers. One major action taken to
decrease the transmission of TB among deer was to ban both supplemental feeding and
baiting in DMU452. Supplemental feeding is deﬁned as the placement of food designed
to help deer survive harsh winters. Baiting is deﬁned as putting food out for the speciﬁc
purpose of attracting deer for hunting. The basis for the bans was the result of an in-depth
risk assessment conducted on the TB problem in MI by the USDA that recommended
decreasing interactions among individuals and increasing hunting to decrease the
population size (MDNR TB Activity Report 1997). Artiﬁcial feeding brings large
numbers of deer into contact for prolonged periods of time, in contrast to the normal
feeding practices of deer where they remain spread out in space foraging over greater
distances in order to obtain necessary resources. At artiﬁcial feed sites, the probability of
encountering food contaminated with M. bovis is greater than in a naturally feeding deer
population due to the large numbers of deer that feed at supplemental sites (MDNR
Activity Report 1997). In the absence of artiﬁcial feeding sites, it was hypothesized that
deer would disperse in space in search of food and would not congregate at feeding sites.
It was believed that the elimination of large aggregations of deer would decrease TB
transmission.

The second major management action was to increase the antlerless (doe) deer
harvest in DMU452 in order to decrease overall deer population density. Increased

hunting, especially of female deer, was intended to reduce deer numbers in DMU452 to

levels that could be supported by natural habitat. In addition, hunters from DMU452 were
asked to voluntarily submit the heads of their harvested deer to the MDNR to check for
signs of TB infection. It was hypothesized that the combination of both management
strategies would reduce the transmission of bovine TB. When transmission decreases
below a certain threshold, more deer infected with bovine TB will die each year than
would become infected, and prevalence rate will decline.

The goal of management is to decrease disease prevalence to less than 1 percent
by the fall of 2003, and to have the disease eliminated in the wild deer herd by the fall of
201 O (MDNR TB Activity Report 1997). The bans on supplemental feeding and baiting
along with increased hunting are likely to result in decreased transmission of TB among
the deer population as a whole. However, because of the behavior and social structure of
white-tailed deer populations that results in nonrandom interactions among individuals
(Hawkins and Klimstra 1970, Marchinton and Hirth 1984), the feed ban and reduction in
population density alone are not likely to eradicate TB. An understanding of the role of
deer ecology in disease transmission is necessary to evaluate transmission dynamics
between deer and M. bovis and whether current management actions will eliminate
disease transmission.

It has become increasingly recognized that wildlife ecology plays a key role in
disease transmission (Boone et al. 1998, Mills and Childs 1998, Begon et al. 1999, Hazel
et al. 2000, Escutenaire et al. 2002, Meeus et al. 2002). In order to eliminate disease from
wildlife populations it is necessary to understand host species’ ecology. The preferred
habitats of white-tailed deer are primarily deciduous and coniferous forests that provide

cover (Marchinton and Hirth 1984). Males generally have larger ranges than females and

10

both sexes tend to use the same ranges across years (Marchinton and Hirth 1984) likely
restricting the range over which an infected deer could transmit TB. Deer do not tend to
migrate long distances, but do often make seasonal movements in search of food
(Marchinton and Hirth 1984). The availability of preferred habitat and food types are
likely to inﬂuence the distance and direction of deer movement, which in turn is likely to
play a key role in the spread of TB geographically.

In addition to the inﬂuence of habitat preference and movement ecology, social
structure and behavior are likely to play important roles in the transmission of disease in
wildlife populations (Mills and Childs 1998, Begon et al. 1999). The social structure of
white-tailed deer is likely to inﬂuence which animals are most likely to become infected
with TB in the absence of artiﬁcial feeding. White-tailed deer have a matriarchal social
structure (Hawkins and Klimstra 1970). Females live in groups usually consisting of an
adult doe, her yearlings and fawns, and often her older offspring as well as offspring of
female relatives. In a study of white-tailed deer in southem Illinois, HaWkins and
Klimstra (1970) found that 87% of yearling females remained philopatric and were
observed with other group members in greater than 60% of all observations. In contrast,
80% of yearling males were found to disperse (Hawkins and Klimstra 1970). For those
males that did not disperse, they moved an average of 3 miles from their mothers' home
ranges (Hawkins and Klimstra 1970). Adult and young males are occasionally found in
groups of 2-5 during the non-breeding season. However, during the breeding season
(September through December in MI) adult males are solitary. The strong site ﬁdelity and
frequency of contact among related females and male fawns suggests that matriarchal

structure is likely a strong contributor to disease transmission. Speciﬁcally, the

11

probability of infection with TB is likely non-random such that deer that are closely

related to an infected deer have a higher risk of TB infection than do nonrelatives.

Utility of molecular markers for studying disease in wildlife populations

The importance of a species’ ecology in disease transmission is well documented.
Many recent studies were conducted over multiple years in intensive ﬁeld situations in
which rates of infection differed as a function of age and sex (Boone et al. 1998, Begon et
a1. 1999, Hazel et al. 2000, Escutenaire et al. 2002). Differences in infection rate across
segments of the population (i.e. age, sex) were used to infer how behavior inﬂuenced
disease transmission. Long-term, ﬁeld intensive studies are very difﬁcult, time intensive,
and expensive. Multi-year studies are often not practical for wildlife diseases that have
reached epidemic levels and demand immediate management actions.

In order to evaluate the role of social structure in TB transmission, it is necessary
to determine relationships among both TB-infected and non-infected deer. It would be
impossible to directly observe the movement, behavior, and relationships of the entire
white-tailed deer population in DMU452. In the absence of known pedigree relationships,
molecular markers can be used to estimate genealogical relationships among animals
(Avise 1994, Queller and Goodnight 1989, Lynch and Ritland 1999). Microsatellite
markers are especially valuable for determining relationships among animals. Loci that
possess large numbers of alleles as well as rare alleles are useful in determining
genealogical relationships among individuals in close proximity (Parker et al. 1998).

When individuals share an allele at a highly polymorphic locus there is an increased

12

probability that these alleles are identical by descent. The probability that alleles are
identical by descent increases when individuals share alleles that are rare in the
population. Multiple genetic markers can be used to estimate the degree of coancestry
among individuals. Coancestry among individuals can be used to recreate pedigrees when
direct observation of behavior is not possible and pedigrees are not known. Levels of
coancestry among infected individuals when compared to coancestry among non-infected
deer can be used to make inferences about mechanisms of disease transmission and the
probability of infection within wildlife populations.

Analyses of wildlife spatial genetic structure can be used to infer the impact of
management actions on movement patterns and breeding behaviors that would be
difﬁcult to determine using other methods. Population genetic structuring and gene ﬂow
can be measured indirectly by assaying geographic variation in the frequency of heritable
genetic markers (Scribner et al. 1997). In order to characterize the pattern of genetic
variability over large geographic areas, moderately variable microsatellite loci are
desirable to characterize individuals within a population and distinguish them individuals
from other populations (Parker et al. 1998). The change in deer spatial structure as
measured by spatial variation in gene frequencies can be incorporated into management
strategies aimed at delineating hotspots of TB-infected deer, targeting management
actions, and predicting the transmission and distribution of bovine tuberculosis.

In this dissertation, genealogical relationships among deer are estimated to infer
the role of white—tailed deer social structure and behavior in disease transmission in
DMU452, and the impacts of management on deer spatial genetic structure. Results are

incorporated into a simple model that highlights the role of deer ecology in disease

13

transmission dynamics. This is the ﬁrst study of its kind to use genetics to infer the role
that wildlife species’ ecology plays, speciﬁcally in terms of behavior and social structure,

in the transmission dynamics between host and pathogen populations.

Sample collection and general methods

Muscle tissue samples from the neck region of approximately 9000, 17,000, and
22,000 hunter-harvested deer submitted to the MDNR in 1998, 1999, and 2000
respectively, were collected. Deer were individually identiﬁed with unique tag numbers
that corresponded to an MDNR database containing the sex, age, harvest location, and
TB status of each animal. Each sample was placed in a 1.5 14L tube ﬁlled with a
preservative tissue storage buffer or 95% ethanol, and labeled with the deer’s tag number.
Samples were stored frozen at either -20 or -70 °C. DNA was extracted from the muscle
tissue of animals used in the analyses in the following chapters (~5000) using the
QIAGEN DNEasy extraction kit and quantiﬁed using ﬂuorometry or spectrophotometry.
Extracted DNA was used for microsatellite ampliﬁcation. Forty-three ﬂuorescently
labeled microsatellite markers were screened and over 30 loci were optimized using a
subset of deer tissue from DMU452 (~ 100 animals) (Appendix I, II). Microsatellite
regions were ampliﬁed using the polymerase chain reaction, separated using
polyacrylamide gel electrophoresis, and visualized using a Hitachi FMBIO-II laser
scanner. The markers and techniques described above were used to address the research

questions presented in the following chapters.

14

CHAPTER 2
GENEALOGICAL RELATIONSHIPS INFLUENCE
THE PROBABILITY OF INFECTION WITH BOVINE TUBERCULOSIS
IN FREE-RANGING WHITE-TAILED DEER

Zoonoses are of increasing importance to wildlife conservation and human health
(Childs et a1. 1998). An understanding of the underlying mechanisms of disease
transmission is necessary to control wildlife disease. Traditionally, in the majority of both
empirical and theoretical studies, disease transmission dynamics have been assumed to be
density dependent (Anderson and May 1992, Grenfell and Dobson 1995, Begon et al.
1999), where contacts among infected and susceptible individuals are assumed to be
random and to increase in proportion to population size (Begon et al. 1999).
Alternatively, disease transmission dynamics may be frequency dependent where the
number of contacts among infected and susceptible individuals is ﬁxed and independent
of population size (Begon et al. 1999). The manner in which transmission dynamics are
modeled will have profound effects on the success of management actions aimed at
eliminating disease transmission. ’

The white-tailed deer (Odocoileus virginianus) population in Michigan (MI),
primarily within a 5 county area of the northeast lower peninsula of MI (DMU452) is
infected with bovine tuberculosis (Mycobacterium bovis) (TB) (Figure 2, Table 2). Direct
contact between animals in which ﬂuids are exchanged, or the ingestion of food
previously contacted by a TB-positive animal are hypothesized to be the primary routes
of TB transmission (Schmitt et al. 1997). Management actions taken by the Michigan
Department of Natural Resources (MDNR) to eradicate TB are based upon a density

dependent model of epidemic disease transmission. One of the primary management

15

 

I
o o .. I

—'~.:+--o-:Fo-‘n-:Fo—c—:Fc-‘J" >
I I I I i.. . "”7“”

-o-. as... o u‘ . 0 -.."J
‘I I" L 1' "L T " , I

O
L.

LO-ILD.. ‘ ! L.T._GI{.L‘

EH

I l I i i '!

-‘—.-0.r.L‘T.';O~ierOf.FO-O’LrQ’—'L.b

”i . i I l I
g---.-}MI¢HIeAII.-L~T-L--. N
I . ! . I I

"'ii +

Figure 2. Map of the lower peninsula of Michigan highlighting the 5 county region where
the deer population is infected with bovine tuberculosis (DMU452).

16

Table 2. Number of harvested deer screened for TB and number of TB-positive deer in
Michigan from 1995-2000.

 

 

 

Year Deer TB-positive
Screened Deer
1995 403 18
1996 4967 56
1997 3720 73
1998 9057 78
1999 19500 58
2000 25 858 53

 

 

 

 

l7

actions taken by the MDNR was to ban artiﬁcial feeding in 1999 in order to minimize
contacts and crowding among deer. In addition, antlerless (doe) harvest levels were
increased in DMU452 in order to decrease deer population size. Culling in order to
reduce population density is a common management strategy for controlling disease
based on density dependent models of disease transmission (McCallum et al. 2001). This
policy will fail, however, if disease transmission is not host-density dependent
(McCallum et al. 2001). The ban on artiﬁcial feeding and increased hunting do not
account for aspects of deer ecology and social structure that may play important density
independent roles in the transmission of disease. The ban on artiﬁcial feeding and the
reduction of population density alone, therefore, may not eradicate TB from the deer
population.

In order to understand and eliminate disease in a wildlife population it is
necessary to understand characteristics of the species’ ecology that might play important
roles in the transmission and maintenance of disease (Boone et al. 1998, Mills and Childs
1998, Begon et al. 1999, Hazel et al. 2000, Escutenaire et al. 2002, Meeus et al. 2002).
Complex social systems and behavior are likely to play important roles in the
transmission of disease in wildlife populations. Aspects of white-tailed deer social
structure and behavior are likely to play important roles in TB transmission. Deer have a
matriarchal social structure in which females live in groups usually consisting of an adult
female, her yearlings and fawns and often her older offspring, her sisters and their
offspring (Hawkins and Klimstra 1970). Hawkins and Klimstra (1970) found that 87% of
yearling females remained philopatric and were observed with other group members in

greater than 60% of all observations. In contrast to the strong association among female

18

deer, 80% of yearling males were found to disperse (Hawkins and Klimstra 1970). For
those males that did not disperse, their main areas of use were on average 3 miles distant
from their mothers' home ranges (Hawkins and Klimstra 1970). Adult and young males
are occasionally found in transitory groups of 2-5 during the non-breeding season
(Marchinton and Hirth 1984). However, during the breeding season, which runs from
September through December in MI, adult males are solitary (Marchinton and Hirth
1984). The strong site ﬁdelity and frequency of contact among related females indicates
that the matriarchal structure is likely to facilitate disease transmission within matrilines
relative to the probability of transmission across matrilines. Consideration of the
inﬂuences of movement and social structure on TB transmission is important when
devising management strategies to eradicate TB from the deer pOpulation.

The high degree of social structure present in most white-tailed deer populations
suggests that another likely risk factor for TB transmission that has not been accounted
for in current MDNR management actions is genetic relatedness to TB-infected deer. If
genetic relatedness plays a role in the probability of infection with TB, those deer at
greatest risk for TB infection will be members of family groups in which one or more
members is already infected. In the absence of artiﬁcial feeding, deer will be forced to
spread out over space in search of food and will not crowd together at feeding sites.
Following the feeding ban, deer coming into closest, most frequent contact will be close
relatives. Thus, it was hypothesized that those deer most likely to become infected with
TB would be close relatives of TB-infected deer. It was hypothesized that artiﬁcial
feeding served to exaggerate TB transmission by bringing many different family groups

into contact at feed sites. Transmission at artiﬁcial feeding sites facilitated the movement

19

of TB from within a few lineages to across many (and likely uninfected) genealogical
lineages. In the absence of artiﬁcial feeding, congregations of large numbers of unrelated
deer at feeding sites should decrease and TB transmission among non-relatives should be
less ﬁ'equent. However, if genetic relationships play an important role in the probability
of TB infection, elimination of artiﬁcial feeding alone will not eliminate disease
transmission among deer.

Determination of all mechanisms of TB transmission is important for devising
management actions that are likely to successfully control the disease (Childs et al.
1998). If genealogical relationships prove to be important risk factor for TB infection, the
disease is likely to be endemic, persisting within some matrilines. As a result, additional
management strategies beyond the ban on artiﬁcial feeding and current level of
population density reduction would be necessary to suppress transmission. In order to
determine the importance of deer ecology, speciﬁcally genealogical relationships, in TB
transmission, relationships among both TB-infected and non-infected deer in DMU452
needed to be determined. It would be impossible to directly observe the movement,
behavior, and relationships of the entire white-tailed deer population in DMU452. In the
absence of known pedigree relationships, molecular markers can be used to estimate
genealogical relationships among animals. If TB transmission is non-random with respect
to genealogical relationships, and if social structure plays an important role in the
probability of infection with TB, it was hypothesized that, on average, TB-positive deer
would be more highly related than non-infected deer. In particular, if the white-tailed

deer’s ecology itself plays an important role in the transmission and maintenance of TB,

20

we expected that TB-positive deer would be more highly related than non-infected deer

independent of the presence or absence of artiﬁcial feeding.

Methods

Muscle tissue samples were collected from the neck region of hunter-harvested
deer submitted from throughout the state to the MDNR during October-December of
1998, 1999, and 2000. Sample storage and DNA extraction were carried out as described
in Chapter 1. All TB-positive deer harvested in DMU452 during 1998-2000 and whose
harvest location was identiﬁed to 26-ka were selected for genotyping. The number of
harvested TB-positive deer each year was only a small fraction of the total number of
deer harvested and screened for TB (Table 2). In order to more accurately characterize
allele frequencies in the deer population in DMU452, numerous non-infected deer were
also genotyped (Table 3). Non-infected samples were chosen based on research on the
movement patterns of deer in DMU452 that indicated the average maxirnum movement
of deer was approximately 12 kilometers (Gamer 2001). For each TB-positive deer, a
subset of non-infected deer within the 12 kilometer radius were also identiﬁed and
selected for genotyping.

All samples were genotyped at 11 polymorphic microsatellite loci (BI/12,
BM4107, Cervid l, Cervid 2, ETH152, OarFCBl93, RT 20, RT 23, RT 24, RT 27, and
SRCRSP-IO; Appendix I, II) (Buchanan and Crawford 1993, Bhebhe et a1. 1994,
DeWoody et al. 1995, Talbot et al. 1996, Wilson et al. 1997). Precise estimation of
relatedness requires negligible genotyping errors, no null alleles, selectively neutral

alleles, a rarity of mutations, random mating, and independent/ unlinked loci (Van de

21

Table 3. Sampling area and number of deer used in the estimation of genealogical
relationships among TB-positive deer (TB+) and among non-infected deer (TB-) from

1998-2000.

 

 

Year Sampling Area TB+ TB-
1998 30.1 x 43.2 mi 67 547
1999 21.6x 33.0 mi 46 311
2000 51.6 x 51.9 mi 35 253

 

 

 

 

 

22

Casteele et al. 2001). Exact tests as described by Guo and Thompson (1992) as
implemented in the program GENEPOP (Raymond and Rousset 1995) were used to
ensure that loci were independent and that observed genotype frequencies accurately
reﬂected expectations under Hardy-Weinberg equilibrium. SUbsequent analyses were
canied out for TB-positive and non-infected animals for each of the three years. Allele
frequencies, allelic diversity, and heterozygosity were calculated using GENEPOP
(Raymond and Rousset 1995).

Molecular marker data were used to estimate the degree of relatedness among
individuals to statistically infer genealogical relationships among animals. The pairwise
coefﬁcient of relatedness is the probability that 2 alleles at a locus from 2 different
individuals are identical by descent. For diploid systems, the coefﬁcient of relatedness is
expected to be equal to 2 times the coefﬁcient of coancestry, and is estimated by the
fraction of alleles that 2 individuals Share that are identical by descent. Marker-based
relatedness estimates are typically characterized by large errors of inference (Ritland
1996, Lynch and Ritland 1999). The performance of different estimators varies with the
shape of the allele frequency distributions, the number of alleles per locus, and the
characteristics of the population (i.e. age, sex composition) under study (Van de Casteele
et al. 2001). Because of the differential performance of relatedness estimators, relatedness
was calculated using 3 different estimators: the regression-based estimator developed by
Queller and Goodnight (1989) implemented in the program KINSHIP, the allele sharing
method described by Bowcock et al. (1994), and the regression-based method-of-
moments estimator developed by Lynch and Ritland (1999) implemented in the program

DELRIOUS (Stone and Bjorklund 2001).

23

KINSHIP 1.3 was used to calculate estimates of relatedness (Qery values)
between all pairs TB-positive deer and estimates of relatedness between all pairs of non-
infected deer for 1998, 1999, and 2000. The coefﬁcient of relatedness (Qery) derived by

Queller and Goodnight (1989) is represented in the following equation:

ESE-p")

QGr

xy = giipx ‘P‘i

For each pairwise estimate, the coefﬁcient of relationship is summed over all loci, k and
all alleles at each locus, a. The frequency of an allele in individual x (0.5 or 1) is
represented by p, while py is the frequency of that same allele in individual y (0, 0.5, or
1). The mean frequency of a given allele in the entire population is represented by p'. The
statistical bias present in p. is corrected for by omitting the 2 individuals being compared.
The resulting rxy calculation is an estimate of the proportion of alleles 2 individuals share
that are identical by descent (IBD). Estimates of relatedness between 2 individuals range
from 1 (all alleles IBD) to —-1. In a randomly mating population, mean relatedness is
expected to be zero. Negative estimates of relatedness indicate that 2 individuals Share
fewer alleles IBD than expected based on the population mean. Standard errors
associated with estimates of pairwise r.y values were estimated through jackkniﬁng
across loci using RELATEDNESS 4.2c (Queller and Goodnight 1989).

Pairwise relatedness estimates between all TB-positive deer and between all non-
infected deer were also calculated using the similarity measure described by Bowcock et

al. (1994):

24

 

The similarity between the multi-locus genotype of 2 individuals, P5, is evaluated based
on the sum of the number of alleles 2 individuals share, P5, across all loci, L, divided by 2
times the number of loci screened. Similarity values range from 0 (no alleles shared) to 1
(all alleles shared).

Relatedness estimates were also calculated using Lynch and Ritland’s (1999)
statistic as implemented in the program DELRIOUS (Stone and Bjorklund 2001). The

Lynch and Ritland (1999) estimator is:

LRr = pa(Sbc +de)+pb(Sac +Sad)-4Papb
xy (1+SabXPa +Pb)-4Papb

 

For each pairwise estimate of relatedness, individual x possesses alleles a and b and the
second individual, y, possesses alleles c and d. If individual x is homozygous, Sub = 1 and
Sal, = 0 if it is heterozygous. If allele 0 from individual x is the same as allele c from
individual y then, Sac = 1, while Sac = 0 if alleles a and c are different. There are 6
different S values corresponding to the 6 possible combinations of the 4 alleles. The
frequencies of alleles a and b in the population are represented by pa and pb respectively.
The estimators are calculated on the basis of conditional probabilities involving allele
ﬁequencies and Similarity measures. For each calculation, one individual serves as the

reference and the probability of its locus-speciﬁc genotype in the second individual is

25

conditioned on the reference individual (Lynch and Ritland 1999). For multiple loci, the
estimators are multiplied by locus-speciﬁc weights to account for sampling variance
differences among loci (Stone and Bjorklund 2001). Lynch and Ritland's LRrxy values

range from -1 (completely unrelated) to 1 (genetically identiCal).

Relatedness Among TB-positive and Non-infected Deer Across All of DMU452

TB-positive deer harvested in DMU452 in 1998, 1999, and 2000 whose harvest
location was identiﬁed to 26-10112 were selected (Table 3). In order to control for
geographic separation among individuals, a non-infected deer was randomly selected
from the same or adjacent 26-ka area from which each TB-positive deer was harvested.
Relatedness between all pairs of TB-positive deer and between all pairs of non-infected
deer was calculated using the 3 estimators described above.

Pairwise similarity data are interdependent and analysis using standard parametric
or nonparametric tests is not appropriate (Dietz 1983). Permutation tests were used to
evaluate signiﬁcance of differences between relatedness estimates for TB-positive and
non-infected deer for each year (Wayne et al. 1991, Lehman et al. 1992). A SAS-based
program (A. Saxton, University of Tennessee, unpublished software) was used to ﬁrst
compute the difference between the means of the 2 groups. The data from the 2 groups
were pooled, and these data were randomly subsampled 500 times. For each
subsarnpling, the difference between the observed and theoretical mean calculated from
the pooled data was calculated and the one-tailed probability of the two means being
different was tested (Ratnayeke et al. 2002). Based upon the hypothesis that the

genealogical model played an important role in TB transmission among deer, it was

26

predicted that mean pairwise estimates of relatedness would be greater for TB-positive
deer than for non-infected deer.

The relationship between genetic differentiation and geographic distance was
calculated for all pairs of TB-positive deer and non-infected‘deer and compared in order
to more explicitly examine the role of genealogical relationships in disease transmission.
Cavalli-Sforza Edwards chord distances (1967) were calculated for all pairs of TB-
positive deer and all pairs of non-infected individuals for each year using BIOSYS
(Swoford and Selander 1981). X,Y (UTM) coordinates were used to calculate Euclidean
geographic distances between individuals using the program PASSAGE (Rosenberg
2000). Mantel tests (Mantel 1967) of correlations between matrices of genetic and
geographic distance between pairs of TB-positive deer and pairs of non-infected deer for
each year were performed using the program PASSAGE (Rosenberg 2000). Signiﬁcant
relationships between genetic and geographic distance were evaluated through
randomization tests in which the order of elements within one matrix Was randomly
perrnutated 999 times (Smouse et al. 1986). If there is a strong (non-random) association
between the 2 matrices, the iterations will be more extreme than the observed association.
A signiﬁcant relationship between the degree to which TB-positive deer were genetically
differentiated and their geographic distance from one another relative to that among non-
infected deer would indicate that the probability of infection with TB is not random, but

rather a function of genealogical relationships among infected individuals.

27

Relatedness Among TB-positive and Non-infected Deer at Distances Consistent With
Deer Movement in Northeast Michigan

Mean relatedness between TB-positive deer was compared to mean relatedness
between non-infected deer at distances consistent with deer movement distances in
DMU452 using the Bowcock et al. (1994) and Queller and Goodnight (1989) estimators
described above for each year. Mean relatedness between pairs of TB-positive deer
separated by several geographic distances was calculated. There were many more non-
infected deer than TB-positive deer (Table 3). Mean relatedness among non-infected deer
was calculated by randomly selecting a number of non-infected deer equivalent to the
number of TB-positive deer at each distance class. This process was repeated 5,000
times. The probability that TB-positive deer were more closely related than non-infected
deer at a given distance class was calculated as the number of times the mean relatedness
among TB-positive deer was greater than the mean relatedness among non-infected deer
divided by 5,000 (Manly 1997). We expected that if the probability of infection with TB
was not random, but rather a function of genealogical relationships, that for deer
separated by distances in which all deer could conceivably come into contact, mean
relatedness among TB-positive deer would be higher than relatedness among non-

infected deer.

Results
Hardy Weinberg tests after Bonferroni correction indicated an excess of
homozygotes at one of 11 loci (p< 0.005). One pair of 55 pairwise comparisons of loci

were in linkage disequilibrium (p< 0.0009). Both excess homozygosity, oﬁen indicative

28

of null alleles, and gametic disequilibrium reduce the ability to resolve relationships
among individuals. The power to accurately resolve pedigree relationships among
individuals, however, increases with the number of loci used. Therefore, data from all
loci were used for estimation of relatedness between individirals. I

Heterozygosity and allelic diversity were high for both TB-positive and non-
infected groups of deer for all years, both prior to and following the ban on artiﬁcial
feeding (Table 4). Allelic diversity at each locus was high ranging from 9-22 alleles per
locus (mean=15.4i 1.1). Loci that possess large numbers of alleles are useful in
determining genealogical relationships among individuals in close proximity (Parker et
al. 1998). When individuals share an allele at a highly polymorphic locus there is an

increased probability that the alleles are identical by descent.

Relatedness Among TB-positive and Non-infected Deer Across All of DMU452

Mean relatedness was Signiﬁcantly higher among TB-positive deer than among
non-infected deer over the whole of DMU452 both before (1998) and after management
changes (2000) as measured by both the Queller and Goodnight (1989) and Bowcock et
al. (1994) estimators (Table 5). The distribution of relatedness values between pairs of
TB-positive deer was skewed towards higher relatedness than was the distribution
between pairs of non-infected deer both prior to and following management changes as
measured by Queller and Goodnight’s (1989) coefﬁcient of relatedness (Qery) and
Bowcock et al.’s (1994) similarity index (Table 5) (Figure 3, 4). No Signiﬁcant
differences were detected for any year using Lynch and Ritland’s (1999) coefﬁcient

(Table 5) (Figure 5).

29

Table 4. Mean heterozygosity and allelic diversity (iSE) among TB-positive deer (+) and
amongIon-infected deer (-) for 1998-2000.

 

 

 

 

 

Year TB Status Heterozygosity Allelic Diversity
1998 + 0.788 (0.026) 12.2 (1.0)
1998 - 0.835 (0.024) 12.ﬂ1.1)
1999 + 0.805 (0.031) 11.6 (0.9)
1999 - 0.837 (0.028) 11.9 (0.9)
2000 + 0.843 (0.032) 11.0 (0.8)
2000 - 0.821 (0.023) 11.5 (0.8)

 

 

 

 

 

30

Table 5. Mean pairwise estimates of relatedness among TB—positive deer (+) and among
non-infected deer (-) during 1998-2000 as measured by Queller and Goodnight’s (1989)
Qery (iSE), Bowcock et al.’s (1994) Similarity, and Lynch and Ritland’s (1999) LRrxy.
Associated p-values are based on permutation tests of differences in mean relatedness
between TB-positive deer and non-infected deer for each measure for each year.

 

 

 

 

 

Year Qery Similarity LRrxy
1998 + 0.025 (0.010) 0.238 -0.020
1998 - -0.015 (0.009) 0.229 -0.016

-value ‘ <0.001 <0.001 0.744
1999 + 0.002 (0.015) 0.236 -0.025
1999 - 0.001 (0.017) 0.248 -0.023

-value 0.446 1 0.652
2000 + 0.014 (0.025) 0.255 -0.034
2000 - -0.009 (0.018) 0.236 -0.033

p-value <0.001 <0.001 0.544

 

 

 

 

31

 

0.35

 

 

-1998-

0'30 I m 1998+

 

 

 

0.25 -
0.20 .
0.15 ‘

Frequency

0.10 .

A

0.05- . I

0.00 . . —. - « ' ' - i

oessoaosozoJQOQIozoSO4osoe
Qery

     
 

 

 

 

 

 

 

 

0.35

 

 

 

 

0.25 <
0.20 .
0.15 .

Frequency

0.10 .
0.05 <

 

 

 

 

 

 

0.00 v r ’1— ’ ‘
-0.6-0.5-0.4-0.3-0.2-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6

Qer

 

0.35
0.30 ‘ — 2000-

 

 

 

. .d ‘

 

0.25 .
0.20 -
0.15 -

Frequency

0.10 j
0.05 -

 

 

 

 

 

 

 

 

0.00

-0.6-0.5-0.4~O.3-0.2-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6
Qer

Figure 3. Distribution of Queller and Goodnight’s estimates of relatedness (Qery)
among TB-positive deer and among non-infected deer in the year prior to the ban on
artiﬁcial feeding (a), in the year in which artiﬁcial feeding was banned (b), and in the
year following the ban on artiﬁcial feeding (c)

32

 

0.45
0.40 'l
0.35 .
0.30 -
0.25 'l
0.20 .
0.15 .
0.10 .
0.05 .
0.00 -

Frequency

 

0.45

0.0

0.1

 

 

 

 

 

 

— 1998-
12:: 1998+

 

 

 

 

 

IEL

 

0.2

Similarity

0.3

0.4

0.5 0.6

 

0.40 .
0.35 4
0.30 .
0.25 I
0.20 .
0.15 .
0.10 .
0.05 .

Frequency

 

0.00 -

0.45

0.0

0.1

 

 

f—l

 

 

 

 

— 1999-
:: 1999+

 

 

 

 

0.2

0.3

I

0.4

Similarity

n ,_
05

0.6

 

0.40 ~
0.35 -
0.30 -
025 ~
0.20 l
0.15 I
0.10 «

Frequency

 

0.05 ]
0.00

0.0

0.1

 

 

 

 

 

 

—2000-
c212000+

 

 

 

 

 

IL.

 

0.2

0.3

 

0.4

Similarity

0.5 0.6

Figure 4. Distribution of Bowcock et al.’s Similarity values among TB-positive deer and
among non-infected deer in the year prior to the ban on artiﬁcial feeding (a), in the year
in which artiﬁcial feeding was banned (b), and in the year following the ban on artiﬁcial

feeding (c).

33

 

0.55
0.50 ~ — 1998-
045 ~ I: 1998+
0.40 .
0.35 -
0.30 "I
0.25 I a
0.20 I
0.15 4
0.10 i

0.05 « I l
0.00 v . . T .
0605-0403-0201 0.0 0.1 0.2 0.3 0.4 0.5 0.
LR

 

 

 

 

Frequency

 

 

 

 

 

f

0.55 xy

0.50 'I - 1999-
0.45 i :1 1999+
0.40 i
0.35 4
0.30 I b
0.25 J
020 I
0.15 I
0.10 'l
0.05 . m I

0.00 . .

-0.6-0.5-0.4-0.3-0.2-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.

 

 

 

 

 

Frequency

 

 

 

 

 

LRrxy

 

0.55

0'50 ‘ - 2000-
0-45 ‘ I22: 2000+
0.40 .

035‘
0,30. 0
0.25«
0.20J
0.15~
0.10«
0.05«
0.00 I . . -.=a

06050403020100010203040506

LRrxy

 

 

 

 

Frequency

 

 

 

 

Figure 5. Distribution of Lynch and Ritland’s estimates of relatedness (LRrxy) among TB-
positive deer and among non-infected deer in the year prior to the ban on artiﬁcial
feeding (a), in the year in which artiﬁcial feeding was banned (b), and in the year
following the ban on artiﬁcial feeding (c).

34

There was a signiﬁcant correlation between the degree to which TB-positive deer
were genetically differentiated and their geographic distance from one another following
the ban on artiﬁcial feeding (2000) and a marginal relationship when artiﬁcial feeding
was occurring (1998) (Table 6). TB-infected deer that were Close to one another in space
were more genetically Similar than were TB-positive deer separated by larger geographic
distances. There was not a signiﬁcant relationship between genetic differentiation and
geographic distance for non-infected deer before or after management actions (Table 6).
Non-infected deer that were close to one another in space were no more similar
genetically than non-infected deer separated by larger distances. The lack of genetic
Similarity between non-infected deer that were geographically proximal relative to
genetic similarity between non-infected deer separated by larger geographic distances
indicates that the non-infected deer were random samples of the deer population. The
increasing degree of genetic differentiation between TB-positive deer with increasing
geographic distance between them, however, indicates that TB-positive deer were not a
random sample of the entire deer population. Speciﬁcally, TB-infected deer that were
geographically close to one another in space were highly genetically similar to one
another relative to TB-infected deer ﬁrrther apart in space indicating that the risk of

infection with TB is not random, but rather is a function of genealogical relationships.

Relatedness Among TB-positive and Non-infected Deer at Distances Consistent With
Deer Movement in Northeast Michigan
Mean relatedness among TB-positive deer was higher than mean relatedness

among non-infected deer harvested in the same 26-ka both prior to (1998) and

35

Table 6. Results of Mantel tests of correlations between genetic and geographic distance
including results of permutation tests of signiﬁcance and associated p-values for TB-
positive deer (+) and for non-infected deer (—) during 1998-2000.

 

 

 

 

 

 

# of Iterations out of 999
Year Correlation < Observed > Observed p-value
1998 + 0.079 937 62 . 0.063
1998 - -0.054 177 822 0.823
1999 + 0.009 550 449 0.450
1999 - -0.030 290 709 0.710
2000 + 0.144 957 42 0.043
2000 - -0.061 216 783 0.784

 

 

 

 

 

 

36

coincident with the ban on artiﬁcial feeding and increased hunting (1999, 2000) for both
the Queller and Goodnight (1989) and Bowcock et al.’s (1994) measures (Figure 6, 7)
(Table 7, 8). In all years, there was a general pattern of higher relatedness at shorter
distances and decreasing relatedness with increasing separation distance (Figure 6, 7). In
addition in 2000, mean relatedness among TB-positive deer was consistently higher than
mean relatedness among non-infected deer up to distances consistent with the average
maximum distance of deer movement (<20km) in northeast M1 for both Queller and
Goodnight’s (1989) Qery and Bowcock et al.’s (1994) similarity (Figure 6, 7) (Table 7,
8). At distances beyond which deer generally move in DMU452, relatedness between
TB-positive and between non-infected deer declined and did not differ (Figure 6, 7). The
higher relatedness among TB-positive deer relative to relatedness among non-infected
deer separated by distances in which all individuals could conceivably come into contact

indicates that all deer were not equally likely to become infected with TB.

Discussion

Estimates of genetic relatedness among TB-positive deer were higher than were
comparable estimates among non-infected deer both when artiﬁcial feeding was
occurring (1998) and after it had been banned (2000) as measured by 2 of 3 estimators.
Results were consistent both independent of the geographic proximity of individuals and
at distances consistent with what is known about deer movement in northeast MI. The
greater genetic similarity among TB-positive deer relative to non-infected deer indicates
that the probability of TB infection is non-random, and that genealogical relationships

play an important role in the maintenance of TB. The role that genealogical relationships

37

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.08
0 1998+
o 1998-
IF‘ 006 JL 3
0.02 / . y
‘ 8 0 oz: - i
000‘ 0 O O 3
o o O
-0021 0
-004 . . L - - A . . . . ﬁﬁ.
051015202530354045505560
Distance (km)
012
* o 1999+
o 1999-
l.
S 0.10/V * /t/ b
8 0.02 ”I /
o 0 °
0.00 «o 0 0 o o o
. . o 0 o
-0.02« ° 0
4104 . . . T . . . - . . ., . .
051015202530354045505560
Distance (km)
052
F o 2000+
o 2000-
C
? 0.50// /;
8 0.044 * /
0.02 ‘ . . . .
0.00 . 000 o o 0 2 o
-0.02« O
-004i ﬂ

 

 

0 51015202530354045505560

Distance (km)

Figure 6. Queller and Goodnight estimates of relatedness (Qery) among TB-positive
deer and among non-infected deer at different distances of geographic separation in the
year prior to the ban on artiﬁcial feeding (a), in the year in which artiﬁcial feeding was
banned (b), and in the year following the ban on artiﬁcial feeding (c). Asterisks indicate
signiﬁcant differences in mean pairwise relatedness between TB-positive and non-
infected deer (p<0.05).

38

Similarity

Similarity

Similarity

Figure 7. Bowcock et al. Similarity values among TB-positive deer and among non-
infected deer at different distances of geographic separation in the year prior to the ban
on artiﬁcial feeding (a), in the year in which artiﬁcial feeding was banned (b), and in the
year following the ban on artiﬁcial feeding (c). Asterisks indicate signiﬁcant differences
in mean pairwise relatedness between TB-positive and non-infected deer (p<0.05).

030

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o 1998+
028? o 1998-
0.26 4
O 090 3 .
0.24 « O O O
O
0.22 < Bi
0.20.......,.....
051015202530354045505560
Distance (km)
0.30

L o 1999+
0.28« o 1999-
0.26~ O

000 o
0.24 p . O O 8 2
O
. O
0.22 4 o
o_2o.............
051015202530354045505560
Distance (km)
0.60 h
o 2000+
0.58« o 2000-
0.56 -

/ /
0.30/ * * /
0.28- 0
0.26 «D o
0.24« 000 O 8 o O 0

O
022«
0.20 -

 

 

 

Distance (km)

39

0 51015202530354045505560

 

 

 

 

 

 

ﬁo Eoo- too- So Boo. oooo- So too. Eoo- oo
So 2o? 99o o~.o oaoo- Boo- moo oooo. too on
. 2o Boo- :3. So o oooo- moo oooo- oooo- ow
8o ~oo.o oooo moo «So. «So So «So. «So on
moo mooo- oroo woo Boo. Boo woo Nooo «So om
So Boo- foo moo Fooo Fooo 5o oooo oooo- 9
Bo mooo mmoo 3o oooo oooo 5o oooo Soo A
to oooo moo to mooo Boo. woo Boo too o
oo oooo Nooo moo oooo «So. So ~8o Nooo o
So moo o—oo moo Boo 83 ..o :oo Boo o
8.9-... - ooo~ + ooo~ 3.9-.. - 82 + 8% mega - 82 + 82 8:900

 

.AExM cozmomacm oEqumoIMmdo moocﬁwﬁ 2.90:6 am
3 Loop 0>zmmmc-m._. 5053 new At Loon 02:399. 505.09 829a 8663mm ucm >500 cams. .x. 030...

 

 

40

 

 

 

 

 

mvd mNNd mmNd vmd nmmd med mad w _.N.o vNNd cm
26 ENG mvwd mod mvwd mmwd 3.6 mmmd mvmd om
med NVNd Nde sad mvmd 3N6 m6 mmmd 5N6 ow
. mod mvwd mmwd mud mewd mvmd v.0 3N6 Nvmd on
5.0 VNd mend vmd mvwd mvwd cod vvmd mud ON
006 3N6 mud mvd nvmd 3N6 wad mvwd mmwd N_.
3.6 hvmd mvud mud mvmd vwd de mvwd wmwd x.
56 mud vomd mad mvmd NNd ﬁnd mvmd wvwd m
mvd mvwd ..mmd sud Nvmd mNNd mod mmNd nvwd m
36 ENG and mod mvwd 5N6 3.0 mNd omwd o
029d - OOON + OOON 0:_m>-a I mam? + mam? 03.0?ﬂ - mam? + ammo 00:90.5

 

.AExv 530.0000 oEdEmo0m Co 008906 600.50 “0 3 L000
00635-5: c0053 new AL .000 0>=_moq-m.r 0053 0029-0 02080000 0:0 btm__E_0 c005. .o 030..

 

 

 

41

play in the risk of infection with TB is increasingly important following the ban on
artiﬁcial feeding that dramatically reduced or eliminated the risk of infection resulting
from contacts at artiﬁcial feeding Sites. Regardless of management actions aimed at
decreasing population density, crowding, and contacts among deer at food piles, data
imply that matrilineal groups likely serve as reservoirs of TB infection within the deer
population. The ban on artiﬁcial feeding and reduction in total population density reduce
contacts among individuals and thus disease transmission across matrilines at feeding
sites, but these actions alone are unlikely to eradicate TB. Genetic data suggest that
management must also focus on aspects of social and kin structure facilitating infection.
Unlike results found using the Queller and Goodnight (1989) and Bowcock et al.
(1994) measures, analyses using the Lynch and Ritland (1999) measure did not indicate
signiﬁcant differences in relatedness between TB-positive and non-infected deer (Table
5) (Figure 5). The performance of different estimators varies with the shape of the allele
frequency distributions, the number of alleles per locus, and the population composition
(Van de Casteele et al. 2001). Van de Casteele et al. (2001) found that from known
pedigrees, Queller and Goodnight’s (1989) coefﬁcient of relatedness more accurately
estimated relationships than did Lynch and Ritland’s (1999) estimator for highly
polymorphic markers. Lynch and Ritland’s estimator also was not as accurate as the
Queller and Goodnight estimator for multiple locus analyses when the number of alleles
varies across loci (Van de Casteele et al. 2001). The 11 loci used in these analyses were
extremely polymorphic ranging from 9-22 alleles per locus (mean=15.4i~ 1.1) (Table 4).

Due to the characteristics of the loci used for these analyses, the Lynch and Ritland

42

(1999) coefﬁcient of relatedness was not likely the best estimator of relatedness for these
data.

Data from 1999 did not indicate a higher degree of relatedness among TB-positive
deer than among non-infected deer (Table 5) (Figure 3, 4). TB-positive deer harvested in
the same 2.6-krn2, however, were more highly related than were non-infected deer in
1999 (Table 7, 8) (Figure 6, 7). In addition to the ban on artiﬁcial feeding, antlerless
hunting quotas were increased in 1999. Increased harvest of females could have resulted
in a larger number of females or younger animals used in the 1999 analysis than in 1998
or 2000. A review of the samples used in the 1999 analysis, however, did not indicate a
bias towards females (36 females, 41 males). A one-way analysis of variance indicated
that the age of sampled deer did not differ as a function of year (F: 0.76, df=- 2, p= 0.47).
The only difference in the 1999 samples, unlike samples from 1998 and 2000, was that
non-infected deer were signiﬁcantly younger than TB-positive deer (t= -3.43, df= 75, p=
0.001). Given that TB-positive deer are consistently older than non-infected deer
(O’Brien et al. 2002), the younger age of non-infected deer samples in 1999 does not
explain why results do not agree with those from the years prior to and following 1999.

Recently, numerous studies have focused on determining how host ecology
inﬂuences the dynamics of disease transmission in natural populations (Boone et al.
1998, Mills and Childs 1998, Begon et a1. 1999, Hazel et al. 2000, Escutenaire et al.
2002, Meeus et al. 2002). There are often differences in the frequency of infection among
demographic segments of wildlife populations (Mills and Childs 1998). Patterns of
cowpox infection in rodents vary as a function of age and sex (Hazel et al. 2000). The

difference between male and female tick dispersal and host speciﬁcity may be important

43

in transmission of Lyme disease (Meeus et al. 2002). Differences among age groups and
sex in infection with Puumala virus were used to infer how behavior inﬂuenced disease
transmission among bank voles (Escutenaire et al. 2002). Boone et al. (1998) found that
adult male deer mice had a higher prevalence of hantavirus than did other segments of the
population infening that wounding during territorial disputes was a major cause of
disease transmission (Boone et al. 1998). In addition, they found a complex non-linear
relationship between density and disease prevalence (Boone et al. 1998). Above a certain
density, there was no predictive correlation between density and hantavirus prevalence
(Boone et a1. 1998).

Complex social behavior can also inﬂuence disease transmission in wildlife. The
rate of bovine TB transmission in brushtail possums increased markedly when highly
social possums were infected compared to transmission rates among randomly mixed
possums (Comer et al. 2002). Transmission risk depended on the spatial proximity of
individuals and rates of social interaction (Corner et al. 2002). The seaSonality in rabies
incidence in jackal populations can be explained by the seasonality of their social
behavior (Loveridge and Macdonald 2001 ). Increased contact among neighboring groups
of j ackals during mating and dispersal was correlated with increased incidence of rabies
while decreased home ranges and neighbor overlap following breeding were correlated
with decreased rabies prevalence (Loveridge and Macdonald 2001). Changes in the
population density of badgers in Great Britain, the main reservoir there for bovine
tuberculosis, were related to incidence of TB in cattle (Hutchings and White 2000).
Hutchings and White (2000) found that human culling of badger populations decreased

population densities sufﬁciently to disrupt their territorial behavior and communal latrine

use. Rather than eliminating disease transmission, culling resulted in a wider spatial
dissemination of badger urination and an increased probability of cattle contact with
urine-contaminated grass (Hutchings and White 2000). Clearly, knowledge of host
ecology can increase understanding of mechanisms of diseaSe transmission in wildlife.

It has been demonstrated based upon estimates of genealogical relationships that
the probability of infection with TB is not random. Rather, matrilineal groups likely serve
as reservoirs of TB infection that is facilitated by deer’s strong social structure. The close
association and high rate of contact that occurs among individuals within matrilines
seems to be the most likely mechanism by which genealogical relationships inﬂuence
disease transmission. There is also the possibility, however, that the higher relatedness
among TB-infected individuals relative to relatedness among non-infected individuals is
due to genetic susceptibility to TB infection. Additionally, the fact that groups of related
individuals share a common environment which in turn might pose a risk of TB infection
cannot be diScounted.

This is the ﬁrst study of its kind to use molecular markers and genealogical
relationships to infer wildlife ecology’s role in transmission dynamics between host and
pathogen populations. Data clearly demonstrate the utility of using molecular markers
and population genetic theory to aid managers in the development of timely and effective

strategies for the control of wildlife disease.

45

CHAPTER 3
WHITE-TAILED DEER SPATIAL GENETIC STRUCTURE PRIOR TO AND
COINCIDENT WITH CHANGES IN ARTIFICIAL FEEDING REGULATIONS

Spatial Structure in White-Tailed Deer

Humans can affect animal social and demographic structures that can affect levels
of genetic diversity within and among wildlife populations (Chesser 1983, McCullough
and Chesser 1987, VanDenBussche et al. 1987, Kilgo et a1. 1998, Trombulak et al. 2000).
Disturbances such as habitat loss and fragmentation, hunting, and artiﬁcial feeding can
change animal spatial dispersion, movements, and foraging behavior (Kilgo et al. 1998,
Trombulak et al. 2000). White-tailed deer (Odocoileus virginianus) often live in close
proximity to humans and are impacted by anthropogenic inﬂuences (Scribner et al. 1997).

Deer are large, highly mobile mammals and might be expected to show little
spatial structuring at a microgeographic scale due to their potential to disperse long
distances, large home ranges, and seasonal migration (McCullough 1979, Marchinton and
Hirth 1984). Characteristics of white-tailed deer ecology, including site ﬁdelity, female
philopatry, and a polygamous mating system, however, may explain the documentation
of spatial genetic structuring in many white-tailed deer populations (Cronin et al. 1991 ,
Scribner 1993, Scribner et al. 1997, Purdue et al. 2000).

White-tailed deer have a matriarchal social structure in which females live in
multi-generational female groups (Hawkins and Klimstra 1970, McCullough 1979,
Marchinton and Hirth 1984). Hawkins and Klimstra (1970) found that 87% of yearling

females remained philopatric and were observed with other group members in greater

46

than 60% of all observations. In contrast, 80% of yearling males were found to disperse
(Hawkins and Klimstra 1970). For those males that did not disperse, home ranges were
typically established an average of 3 miles away from their mothers' home range
(Hawkins and Klimstra 1970). Adult and young males are oCcasionally found in groups
of 2-5 during the non-breeding season (Hawkins and Klimstra 1970, Marchinton and
Hirth 1984, McCullough 1979). However, during the breeding season, which runs from
September through December, adult males are solitary and establish breeding hierarchies
(Hawkins and Klimstra 1970, Marchinton and Hirth 1984, McCullough 1979). Older,
dominant males are presumed to obtain a greater proportion of breeding opportunities
than younger males (Hawkins and Klimstra 1970, Marchinton and Hirth 1984,
McCullough 1979).

Female philopatry, male-biased dispersal, and non-random mating create genetic
substructure within populations (Chesser 1991a). A matrilineal social system in which
females are philopatric to natal areas over multiple years creates spatial genetic structure
over microgeographic scales (Chesser 1991a, Matthews & Porter 1993). Philopatry leads
to high coancestry among individuals within social lineages (0) relative to coancestry
among individuals from different social lineages (0t) (Chesser 1991a). Male breeding
hierarchies can either reinforce divergence or homogenize genetic diversity in a
matrilineally-structured social system (Chesser 199la,b, Matthews and Porter 1993). A
situation in which a single male mates with most or all females in a matriline both
increases coancestries within matrilines and the differentiation among matrilines (Chesser
1991a,b). Differentiation among matrilines decreases, however, when all males in a

population have relatively equitable mating opportunities, more than one male mates with

47

females within a matriline, or males mate with females in several matrilines (Chesser
1991a,b). Changes in local white-tailed deer breeding structure due to high rates of
population turnover, young age structure, sex-biased gene ﬂow, and method of harvest
can alter spatial genetic structure (Scribner et al. 1985, Scribner 1993, Scribner et al.

1997).

White-Tailed Deer in Michigan

In Michigan (MI), before human settlement, white-tailed deer were most abundant
in the southern part of the state where the habitat primarily consisted of open oak and
hickory forests ideal for deer (Langenau 1994, MDNR 1993). In the northern lower
peninsula of MI, forests of primarily pine and hardwood that did not provide optimal
habitat resulted in low carrying capacity for deer (MDNR 1993). However, during the
early 20th century, numerous hunting clubs were established in northeast MI. Due to high
hunting demand and shortage of natural browse, artiﬁcial feeding was widely used to
enhance deer population numbers. Artiﬁcially high deer densities resulting from artiﬁcial
feeding practices has the potential to have negative effects on the natural vegetation and
on the deer population itself in terms of increased transmission and prevalence of disease
(Chouinard and Filion 2001, Miller et al. 2003).

The white-tailed deer population in the northeast lower peninsula of MI (referred
to here as Deer Management Unit 452 (DMU452)) is infected with bovine tuberculosis
(Mycobacterium bovis) (TB). Higher prevalence of TB in this region relative to the rest

of MI is likely due to the region's long history of artiﬁcial feeding (Schmitt et al. 1997).

48

Artiﬁcial food sources have resulted in high deer densities (>35 per miz), crowding, and
high rates of contact among deer over food (Schmitt et al. 1997). In an effort to decrease
TB transmission among deer, artiﬁcial feeding was banned in DMU452 in 1999. In
addition to facilitating TB transmission, artiﬁcial feeding islikely to have affected deer
Spatial genetic structure by altering deer movement and breeding behavior. Knowledge of
deer spatial genetic structure as measured by spatial variation in gene frequencies can be
useful in predicting the distance of gene ﬂow and thus the transmission and distribution
of TB across DMU452.

In order to characterize the effect of artiﬁcial feeding on deer population genetic
structure and the potential for disease transmission, the degree of variation in allele
frequencies among groups of deer within 2 regions of DMU452 that have different
histories of artiﬁcial feeding was estimated (Figure 8). It would be impossible to directly
observe the movement, behavior, and relationships of the entire white-tailed deer
population in DMU452. In the absence of known pedigree relationships, the degree of
spatial variation in allele frequencies and gene ﬂow can be estimated indirectly by
assaying geographic variation in the frequency of heritable genetic markers (Scribner et
al. 1997).

Spatial autocorrelation techniques are often used to describe non-random patterns
in spatial distributions of alleles or genotypes (Sokal and Oden 1978, Cliff and Ord 1981,
Sokal and Jacquez 1991, Epperson 1993, Sokal et al. 1997, Koenig 1999). Speciﬁcally,
spatial autocorrelation is a measure of the degree of interdependence between variables
(Cliff and Ord 1981). Autocorrelation analysis has advantages in describing population

structure over standard techniques such as Fst, a descriptor of the overall degree of

49

Sampling Network:

ex. 23.3-ka -

~--

r--

ex. 2.6-km2

 

ﬁ-‘ﬁi—rwa‘;
”TIES-l
imi+£~t~
- ....... E] .............. N 900----
I 11.3 kilometers
South

------------------‘---’---- ......

Figure 8. Map of northeast MI (DMU452) and networks for the analysis of spatial genetic
structure among “subpopulations” of deer. In the North region is a depiction of the
network of 23.3-lcrn2 quadrats. Each point indicates the center of the 233-ka quadrat
and the box is an example of a single 23.3-krn2 quadrat. In the South region is an example

of the network of 2.6-lcrn2 quadrats.

50

variance among locations, because it explicitly describes patterns of genetic variation
over space (Barbujani 1987, Sork et al. 1999). Spatial autocorrelation techniques can be
used to characterize the geographic variation in biological populations and to make
inferences concerning ecological or demographic features 0f populations and the
landscape they inhabit that lead to non-random spatial associations (Oden and Sokal
1986, Sokal et al. 1997).

Random mixing of large numbers of deer at feeding sites contrasts with a natural
free-ranging population in which deer, especially females, interact primarily with their
relatives (Hawkins and Klimstra 1970). Artiﬁcial feeding disrupts the spatial segregation
of matrilines by bringing groups of deer into the same spatial location. Speciﬁcally, in the
absence of artiﬁcial feeding, individuals from several matrilines may coexist in a
common area, but the degree of social interaction and probability of contact among these
individuals increases when the population is artiﬁcially fed. It was hypothesized that the
greater the intensity of artiﬁcial feeding, the greater homogenization of population
genetic diversity. Due to intensive artiﬁcial feeding, deer in southern areas of DMU452
(South) were expected to exhibit weaker spatial genetic structuring relative to the less
heavily artiﬁcially fed northern DMU452 deer (North). Following the ban on artiﬁcial
feeding, it was hypothesized that there would be a stronger relationship between genetic
differentiation among groups of deer and their geographic distance from one another in

both the North and South regions (2000).

51

Methods

DMU452 was divided into 2 regions on the basis of differing histories of artiﬁcial
feeding intensity (Figure 8). One region referred to as the South region was located in the
region of most intensive artiﬁcial feeding (in terms of the nUmber and size of feeding
sites) and the highest incidence of TB (319 TB—positive deer identiﬁed from 1995-2000).
The South region was located primarily in Alpena and Montrnorency counties and
extended south into the north-most sections of Alcona and Oscoda counties and covered
2,323—krn2 (Figure 8). Deer throughout South have been intensively artiﬁcially fed
(hundreds of thousands of kilograms per year) for several decades (Hickling 2002, Miller
et al. 2003). The second region, referred to as the North region was located in the
northem-most part of DMU452 where incidence of TB was lower (5 TB-positive deer
identiﬁed from 1995-2000). Artiﬁcial feeding in this region has historically been lower
than in the South region. The North region was located 11.3 kilometers to the north of the
South region .in southern Presque Isle county and the very northem-most portions of
Alpena and Montrnorency counties and covered 1,994-km2 (Figure 8).

The North and South regions were divided into two different sized networks of
“subpopulations” (quadrats) (Figure 8). The ﬁrst network consisted of 37 (North region)
and 40 (South region) approximately 233-ka quadrats each separated by a 3.2 km
buffer (Figure 8). The second network consisted of 26-ka quadrats (Figure 8). The
number of 26-ka quadrats in each region ranged from 142-258 based on sample
availability in each year of the analysis (Table 9). Spatial locations of each quadrat for the

233-ka network. were deﬁned by the X,Y coordinate (UTM) of the centermost 26-ka

52

Table 9. Number of deer genotyped and number of quadrats in each network for the
North and South regions during 1998, 1999, and 2000.

 

 

 

 

Region Year # Individuals # 23.3-km2 # 26-1an

quadrats quadrats
North 1998 644 37 205
North 1999 616 37 _176
North 2000 351 37 142
South 1998 654 40 252
South 1999 663 40 258
South 2000 726 40 255

 

 

 

 

 

 

53

section of each 23.3-krn2 quadrat. The spatial location for quadrats within the 2.6-krn2
network was deﬁned by the UTM location of the center of each 26-ka quadrat.

Samples of hunter-harvested deer were collected as described in Chapter 1. All
deer samples from 1998, 1999, and 2000 (~3660) harvested within the quadrats described
above were selected for the following analyses. Because sampling intensity across the
network was a function of hunter harvests, sample sizes were not uniform across quadrats
of each network (Table 9).

DNA was extracted ﬁ'om deer samples using the QIAGEN DNEasy protocol.
Analyses were based on the spatial dispersion of alleles within and among quadrats of
deer for both regions at 3 moderately polymorphic loci (9-12 alleles per locus) (IGF-l ,
OBCAM, and RT-9, Appendices I, H) (Kirkpatrick 1992, Fries et al. 1993, Wilson et al.
1997). Exact tests described by Guo and Thompson (1992) as implemented in the
program GENEPOP (Raymond and Rousset 1995) were used to estimate levels of
gametic disequilibrium to ensure. that loci were independent and to determine whether
observed allele frequencies accurately reﬂected expectations under Hardy-Weinberg
equilibrium.

The program GENEPOP was used to estimate allele frequencies, observed and
expected heterozygosity, and allelic diversity for each locus for each region for each year
(Raymond and Rousset 1995). Allelic richness was calculated using the program
CONTRIB in order to adjust allelic diversity measures for sample size differences among
years and between regions (Petit et al. 1998). Due to limitations in the capacity of some
software, some data in the network of 2.6-krn2 quadrats were grouped in order to ﬁt the

dimensions of the software. Four individuals possessing unique or extremely rare alleles

54

(frequency <0.0001) were removed from subsequent analyses. Additionally, 4 rare alleles
(frequency <0.002) at the IGF-l locus were binned with the next largest size allele and
the allele was renamed. The degree of population genetic divergence within each region
for each year was assayed by measuring the variation in allele frequencies among all
quadrats of deer (Fst) (Weir and Cockerharn 1984), and was estimated for each network
using the program GENEPOP (Raymond and Rousset 1995).

Genotypes of deer within each quadrat were used to calculate allele frequencies
for each quadrat of deer using GENEPOP (Raymond and Rousset 1995). Allele
frequencies of deer within quadrats were used to estimate Cavalli-Sforza Edwards chord
distances (1967) between quadrats of deer for each network for each region and for each
year using the program BIOSYS (Swofford and Selander 1981) or the program PHYLIP
3.5 (Felsenstein 1993). UTM coordinates were used to calculate Euclidean geographic
distances between quadrats using the program PASSAGE (Rosenberg 2000). Mantel tests
were implemented in the program PASSAGE to estimate the relationship between
genetic distance and geographic distance matrices constructed from all pairwise
comparisons of quadrats within each region for each year (Rosenberg 2000). A Mantel
test examines the relationship between matrices whose values are measures between 2
points or sampling units (Mantel 1967, Fortin and Gurevitch 2001). Signiﬁcance of
correlations between genetic and geographic distance were evaluated through
randomization tests in which the elements of one matrix were randomly permuted 1000
times (Smouse et al. 1986). If there is a strong (non-random) association between the 2
matrices, the iterations will be more extreme (biased towards either larger or smaller

values) than the observed association.

55

In addition to characterizing the magnitude of spatial autocorrelation over an
entire region, spatial relationships among locations can be characterized at a series of
successively larger distance classes (Rosenberg 2000). Mantel tests (Mantel 1967) were
used to estimate the degree of genetic differentiation among quadrats of deer separated by
several geographic distances for each region for each year using the program PASSAGE
(Rosenberg 2000). The normalized Mantel correlation was estimated between a matrix of
genetic distances between pairs of quadrats of deer and a weight matrix in which only
pairs of quadrats of deer separated by geographic distances within the focal distance class
were included. Correlogram plots of the correlation coefﬁcient at each distance class
were constructed for each region for each year to summarize patterns of genetic variation
in space (Sokal and Oden 1978), and to identify geographic distances at which

autocorrelation is high or low (Diniz-Filho and Telles 2002).

Results

Heterozygosity and allelic richness were high in deer populations in both the
North and South regions. Regions did not differ from one another in heterozygosity (t=
0.737, df= 2, p= 0.538) or allelic richness (P -0.157, df= 2, p= 0.889) over time (Table
10).

Estimates of variance in allele frequencies among quadrats of deer (Fst) in the
network of 23.3-km2 quadrats were close to zero, and indicate that the size of quadrats
may have exceeded the scale of spatial genetic structure present in both regions during
each year (Diniz-Filho and Telles 2002). Fst values were effectively zero in South (-

0.001, -0.001, and -0.001) and relatively low in North (0.004, -0.003, and 0.004) for

56

Table 10. Heterozygosity and allelic richness (iSE) for deer genotyped in the North and
South regions during 1998, 1999, and 2000.

 

 

 

 

 

 

 

Region Year Observed Allelic
Heterozygosity Richness

North 1998 0.803 (0.023) 9.26 (0.06)
North 1999 0.771 (0.045) 8.99 (0.05)
North 2000 0.778 (0.032) 8.67 (0.03)
South 1998 0.760 (0.046) 9.16 (0.06)
South 1999 0.779 (0.041) 8.58 (0.06)
South 2000 0.778 (0.047) 9.33 (0.07)

 

57

 

1998, 1999, and 2000 respectively. The network of 2.6-ka quadrats may be a more
biologically appropriate scale at which to characterize deer spatial genetic structure in
DMU452.

The degree of genetic divergence among quadrats (Fst) as measured using the
network of 26-1an quadrats was effectively zero in the South region and relatively low
in the North region prior to the ban on artiﬁcial feeding (1998) (Table 11). The near zero
Fst values imply that allele frequencies among quadrats of deer were relatively
homogenous when artiﬁcial feeding was occurring. The degree of population structure as
measured by variation in allele frequencies among quadrats of deer (F st) was greater in
both regions following the ban on artiﬁcial feeding (2000) (Table 11). Higher estimates
of inter-quadrat variance in allele frequencies was observed among quadrats of deer in
the North region relative to the South region both prior to and following changes in
feeding regulations (Table 11). The greater variance in allele frequencies among quadrats
of deer in the North region relative to the South region may be due to the North region’s
less intensive artiﬁcial feeding history.

There was a signiﬁcant correlation between genetic and geographic distance
among quadrats in the South region in 2000 and a marginal relationship among quadrats
in the North region in 1998 as measured using the network of 23.3-km2 quadrats of deer
(Table 12). Correlograms depicted increased genetic differentiation between quadrats
with increased geographic distance in the North region in all years, and in the South
region in 2000 (Figure 9).

There was no signiﬁcant relationship between genetic and geographic distance in

either the North or South regions in 1998 as measured by the network of 26-ka

58

Table 11. Measures of allelic variance within and among quadrats for the network of 2.6-
km2 quadrats for the North and South regions during 1998, 1999, and 2000.

 

 

 

 

 

Region Year F is F st Fit
North 1998 0.020 0.005 0.025
North 1999 0.025 0.010 0.034
North 2000 0.013 0.027 0.039
South 1998 0.050 0.001 0.050
South 1999 -0. 166 -0.049 0223
South 2000 0.023 0.004 0.027

 

 

 

 

59

 

Table 12. Results of Mantel tests of correlations between genetic differentiation between
quadrats and their geographic distance including results of permutation tests of
signiﬁcance and associated p-values for the network of 233-10112 quadrats for the North
and South regions during 1998, 1999, and 2000.

 

 

 

 

 

# of iterations out of 1000
Region Year Correlation < Observed > Observed p-value
North 1998 0.1 15 909 91 0.092
North 1999 0.023 648 352 0.353
North 2000 0.061 797 203 0.204
South 1998 -0.168 6 994 0.994
South 1999 -0.081 115 885 0.885
South 2000 0.134 960 40 0.041

 

 

 

 

 

 

 

60

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.15
+ North 1998
0.10 1 0- South 1998
,5 0.05 « » 0
‘0
- a
g 0.00«
8 '0
-0.05 <
O .
-0.10 . ' 0'
5101520253035404550
Geographic Distance (km)
0.15
+ North 1999
0.10 < O - South 1999
,5 0.05 .
26.5
g 0.00 b
o
O
-0.05 .
-0.10 «
5101520253035404550
Geographic Distance (km)
0.15
—0— North 2000
010 ‘ >0 - South 2000
.5 0.054
E
C
g 0.00
o
O
-0.05 4
-0.10 <

 

 

 

5101520253035404550
Geographic Distance (km)

Figure 9. Correlation between quadrats of deer separated by different geographic
distances in the North and South regions for the network of 23.3-lcm2 quadrats in the year
prior to the ban on artiﬁcial feeding (1998) (a), the year in which feeding was banned
(1999) (b), and the year following the ban on feeding (2000).

61

quadrats of deer (Table 13). When artiﬁcial feeding was occurring, spatial genetic
structuring was not detected in either region. There was, however, signiﬁcant Spatial
autocorrelation in both regions in 2000 (Table 13). Following the ban on artiﬁcial
feeding, both regions showed evidence of signiﬁcant heterOgeneity among quadrats of
deer across space. The magnitude of spatial autocorrelation was higher in the North
region than in the South region in all years (Table 13). A stronger signature of spatial
genetic structuring in the North region relative to the South region can likely be explained
by the North region’s historically lower level of artiﬁcial feeding.

Correlograms of correlation coefﬁcients at several geographic distances indicated
the degree of genetic differentiation between quadrats of deer increased with increasing
spatial distance in both the North and the South regions in 2000 relative to 1998 using the
network of 2.6-km2 quadrats (Figure 10). In addition, in 2000, the magnitude of genetic
differentiation between quadrats of deer as a function of increasing geographic distance
was greater among in the North region than in the South region (Figure 10). The
geographic distance at which the autocorrelation becomes non-signiﬁcant can be deﬁned
as the intercept of the correlograrn (Diniz-Filho and Telles 2002), and lies between 13-18
km in each region for 2000. The intercept of a correlograrn deﬁnes the minimum
geographic distance at which samples are independent or the diameter of a genetic patch
(Diniz-Filho and Telles 2002). The intercepts of the correlograms for both regions are
consistent with telemetry data that indicates that the average maximum distance of deer

movement in DMU452 is approximately 16 km (Garner 2001).

62

Table 13. Results of Mantel tests of correlations between genetic differentiation between
quadrats and their geographic distance including results of permutation tests of
signiﬁcance and associated p—values for the network of 26-ka quadrats for the North
and South regions during 1998, 1999, and 2000.

 

 

 

 

 

 

 

 

 

 

# of iterations out of 1000
Region Year Correlation < Observed > Observed p-value
North 1998 0.023 805 195 0.196
North 1999 0.007 625 375 0.376
North 2000 0.087 996 4 0.005
South 1998 -0.008 313 687 0.687
South 1999 -0.009 366 634 0.634
South 2000 0.061 992 8 0.009

 

63

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.06
+ North 1998
0‘04 ‘ - 0» South 1998
r: 0.02 <
.9
g 0.00« a
8
0 -0.02 ~
-0.04 .
-0.06....-......
5101520253035404550556065
Geographic Distance (km)
0.06
+ North 1999
004 ~ 0- South 1999
c 0.02‘
.9
E; 0.00 b
‘6
0 -002.
-0.04 <
-0.06.....-.s...
5101520253035404550556065
Geographic Distance (km)
0.06
—0— North 2000
0-04 I '0” South 2000
r: 0.02 -
.g
(B
" 0.00< C
8
o
O 0.02 '1
-0.04 «
‘0.06 v 1 a r r ﬂ 1 v I v y
5101520253035404550556065

Geographic Distance (km)

Figure 10. Correlation between quadrats of deer separated by different geographic
distances in the North and South regions for the network of 2.6-krn2 quadrats in the year
prior to the ban on artiﬁcial feeding (1998) (a), the year in which feeding was banned
(1999) (b), and the year following the ban on feeding (2000).

Discussion

Prior to the ban on artiﬁcial feeding, there was no evidence of genetic divergence
among quadrats of deer (F st) or signiﬁcant correlation between genetic differentiation
and geographic distance between quadrats of deer in either the North or South regions.
The absence of Signiﬁcant spatial autocorrelation in 1998 may be explained by the
aggregation of multiple matrilines at artiﬁcial feeding sites. Following the ban on
artiﬁcial feeding, there was a demonstrable increase in the degree population genetic
structure as measured by variance among quadrats in allele frequencies (Fst) in both the
North and South regions. Additionally, there were signiﬁcant correlations between the
degree to which quadrats of deer were genetically differentiated from one another and
their geographic distance in both regions. The detection of signiﬁcant spatial genetic
structure in 2000 is likely the result of a greater dispersion of matrilines across space.
Increasing genetic differentiation between groups of deer with increasing geographic
distance in both regions suggests that there is likely limited gene ﬂow among quadrats of
deer as they become increasingly separated in space as expected in an isolation by
distance model (Purdue et al. 2000).

The degree to which quadrats of deer were genetically differentiated from one
another was greater in the North region relative to the South region for all time periods.
The greater spatial genetic structuring found in the North region relative to the South
region is likely due to the North region’s historically lower level of artiﬁcial feeding. The
magnitude at which artiﬁcial feeding was practiced in the South region likely served to
bring larger numbers of individuals together at artiﬁcial feeding sites than did artiﬁcial

feeding sites in the North region. The aggregation of a larger number of individuals ﬁ'om

65

a variety of different matrilines in the South region likely resulted in a greater degree of
homogenization of allele frequencies than occurred in the North region.

In addition, telemetry studies of deer located in the South region conducted by
Garner (2001) documented that when artiﬁcial feeding was‘occurring, deer moved
extremely short distances relative to movement distances documented in deer elsewhere
in the United States (Garner 2001). The lesser degree of Spatial genetic structuring in the
South region relative to the North region may be a function of the extremely short
movement distances for deer in the South region when food was abundant. Turnover of
the population and recruitment into the population would be expected to result in
increasing spatial genetic structuring in the South region in the years following the ban on
artiﬁcial feeding as more traditional movement behavior resumes in the region.

Data from 1999, the year in which artiﬁcial feeding was banned, for both regions
and both networks were somewhat unusual compared with results from 1998 and 2000
(Figure 9, 10) (Table 12, 13). It is possible that the lack of food at locations where deer
found it for several decades may have resulted in somewhat unusual movement in 1999.
Deer show considerable ﬁdelity to feeding sites, and it may take population turnover and
recruitment for all individuals to abandon sites where food piles no longer occur.

Social structuring in natural populations can impede the movement of genes
among breeding groups, serving to partition population genetic variation (Chesser
199la,b, Sugg et al. 1996). For animals in which females remain in natal areas near
relatives, genetic differences will accrue among spatially segregated matrilines (Chesser
1991a, Matthews & Porter 1993). Male breeding structure can either accentuate levels of

divergence or homogenize gene frequencies across group structured populations (Chesser

66

199la,b, Matthews & Porter 1993). Gene correlations among individuals within social
lineages (0) are enhanced when a Single or a few males breed within a lineage (Chesser
1991a). However, when numerous males mate within a social lineage or when males
mate with females from several lineages, coancestry among individuals within a lineage
(0) decreases relative to coancestry among individuals from different lineages (or) serving
to homogenize population gene frequencies (Chesser 19913). Future studies of animals
recruited into the deer population after the ban on artiﬁcial feeding may be useful in
addressing the impact of the artiﬁcial feeding ban and reduced density on the deer
population's mating system.

Numerous studies have documented the presence of spatial genetic structure in
white-tailed deer populations (Cronin at e1. 1991, Matthew and Porter 1993, Scribner
1993, Purdue et al. 2000). Few studies, however, have characterized the degree of
homogenization in allele frequencies among quadrats of deer across Space as was
documented here when artiﬁcial feeding was occurring. Several studies have documented
that anthropogenic actions such as the method and intensity of harvest and translocations
can alter spatial genetic structuring in deer populations (Leberg 1990, Scribner 1993,
Scribner et al. 1997, Leberg and Ellsworth 1999). Leberg and Ellsworth (1999) found that
animal translocations reduced the degree to which populations were genetically
differentiated as a function of their geographic distance from one another. Scribner et al.
(1997) found signiﬁcant temporal variation in microgeographic gene frequencies that
could likely explained by high rates of population turnover, young age structure, and
method of harvest affecting local breeding structure. The results presented here are the

ﬁrst to document the extreme homogenization of allele frequencies among groups of deer

67

when artiﬁcial feeding was occurring, and the rapid emergence of spatial heterogeneity in
allele frequencies among quadrats of deer following the ban on artiﬁcial feeding.
Analyses of the degree to which natural populations are spatially genetically
structured can be used to make inferences regarding the effects of management actions on
movement patterns and breeding behaviors that would be difﬁcult to determine using
traditional methods. Changes in deer spatial genetic structure coincident with changes in
management can also have important implications with regard to disease transmission in
wildlife populations. Disease transmission will be affected by the spatial distribution of
infected individuals (Fromont et al. 1998a,b, Tully et al. 1999, Fulford et al. 2002).
Clusters of genetically similar groups of deer in space indicate that there is spatial genetic
structuring in DMU452 that likely limits the movement of TB across genetically
differentiated groups. Knowledge of the magnitude of spatial genetic structure and the
distance of gene ﬂow as measured by spatial variation in allele frequencies can be
incorporated into management strategies aimed at delineating hotspots of TB, targeting
regions for management actions such as intensive density reductions, and predicting the

future transmission and distribution of bovine tuberculosis.

68

CHAPTER 4
A SIMPLE MODEL HIGHLIGHTING THE IMPORTANCE OF
SOCIAL STRUCTURE IN THE MAINTENANCE OF BOVINE TUBERCULOSIS
IN F REE-RANGING WHITE-TAILED DEER

Transmission dynamics drive interactions between host and pathogen populations
(Begon et al. 1999). Host-pathogen models are used to design management strategies for
disease control (McCallum et al. 2001). Traditionally, in the majority of both empirical
and theoretical studies, disease transmission dynamics have been assumed to be density
dependent (mass action) (Anderson and May 1992, Grenfell and Dobson 1995, Begon et
al. 1999, McCallum et al. 2001). In density dependent models of disease transmission
contacts among infected and susceptible individuals are assumed to be random and to
increase in proportion to population size (Begon et al. 1999). Culling in order to reduce
population density is a common management strategy for controlling disease based on
density dependent models (McCallum et al. 2001). This policy will fail, however, if
disease transmission is not solely a function of host density (McCallum et al. 2001).

Alternatively, disease transmission dynamics may be frequency dependent where
the number of contacts among infected and susceptible individuals are ﬁxed and
independent of population size (Begon et al. 1999). In addition, the incidence of infection
is often patchily distributed, where infected individuals are likely to be closer to other
infected individuals than expected under random mixing which is the typically assumed
model of disease transmission (McCallum et al. 2000). The manner in which transmission
dynamics are modeled will have profound effects on the success of management actions

aimed at eliminating disease transmission (Dobson and Meagher 1996).

69

Most zoonotic pathogens are endemic, always present at some low level, in
wildlife populations (Childs et al.1998, Begon et a1. 1999). Most empirical studies of
disease transmission in wildlife, however, have been conducted opportunistically in
response to epidemic outbreaks that generally occur over short time periods (Begon et al.
1999). Endemic diseases are rarely easily eradiated, and ecological changes can promote
their ampliﬁcation to epidemic levels (increased incidence) (Childs et al. 1998). As such,
it is important to understand the key mechanisms of disease transmission to devise
effective control measures (Childs et al. 1998).

The white-tailed deer (Odocoileus virginianus) population in the lower peninsula
of northeast Michigan (MI) (DMU452) is infected with bovine tuberculosis
(Mycobacterium bovis) (TB). The management actions to eliminate TB from free-ranging
deer are based upon a density dependent model of disease transmission. Management is
focused on minimizing contacts and crowding among deer at feeding sites and decreasing
population density. One major action taken in order to decrease the transmission of TB
was to ban artiﬁcial feeding. Artiﬁcial feeding results in large aggregations of deer for
prolonged periods of time. It is likely that the probability of encountering food
contaminated with M. bovis at artiﬁcial feeding sites is greater than in a deer population
foraging on natural foods. In the absence of feeding sites, it was hypothesized that deer
would disperse in search of food, and contacts and crowding among deer would decrease
thereby greatly reducing TB transmission.

The second major management action was to increase the antlerless (female) deer
harvest to decrease deer population density. Increased hunting, especially of female deer,

was intended to reduce deer numbers to levels that could be supported by natural habitat.

70

It was hypothesized that the combination of the management strategies described above
would reduce TB transmission. When transmission decreases to some threshold level,
more deer infected with bovine TB will die each year than would become infected, and
disease prevalence would be expected to decline. Management goals were to decrease
disease prevalence to less than one percent by the fall of 2003, and to have TB eliminated
from the wild deer herd by the fall of 201 O (MDNR TB Activity Report 1997).

The ban on artiﬁcial feeding along with increased hunting are likely to result in
decreased transmission of TB among deeriby eliminating the risk of infection due to
contacts among individuals at feed sites. However, the matrilineal social structure and
female philopatry of white-tailed deer populations results in frequent interactions among
related individuals and rarer interaction among non-relatives (Hawkins and Klimstra
1970, Marchinton and Hirth 1984). The artiﬁcial feeding ban and reduction in population
Size do not speciﬁcally address the non-random probability of infection with TB as a
function of white-tailed deer ecology. Speciﬁcally, deer that are closely related to an
infected deer have a higher risk of TB infection than do nonrelatives.

An understanding of the role of deer ecology in disease transmission is necessary
to evaluate whether transmission dynamics more closely resemble density dependent or
frequency dependent processes, and whether current management actions will eliminate
the disease. Numerous recent studies have demonstrated that wildlife (host) ecology
plays a key role in disease transmission (Boone et al. 1998, Mills and Childs 1998, Begon
et al. 1999, Hazel et al. 2000, Escutenaire et al. 2002, Meeus et al. 2002). As was
demonstrated in Chapter 2 of this dissertation, TB-positive deer were more closely

related than were non-infected deer both prior to and following management efforts to

71

decrease contacts over food and to decrease density in northeast MI. Clearly, the
probability of infection with TB is non-random such that deer that are closely related to
an infected deer have a higher risk of TB infection than do nonrelatives suggesting that
genealogical relationships play a key role in transmission 0f TB. In order to effectively
control TB, all mechanisms of TB transmission must be accounted for in management
actions.

McCarty and Miller (1998) were the ﬁrst to model TB transmission among deer
in northeast Michigan. Interactions among infected and susceptible individuals were
assumed to be random and to change as a function of population size (McCarty and
Miller 1998). In their model, all individuals had an equal probability of becoming
infected with TB. McCarty and Miller (1998) described 3 factors that determined the
probability that an individual would become infected: the number of infected individuals
in the population, the number of contacts the individual made, and the total population
size. The model predicted that TB would increase substantially over time regardless of
management efforts to reduce disease transmission (McCarty and Miller 1998).

The effectiveness of management actions to reduce population density and
artiﬁcial feeding on TB prevalence was modeled (Hickling 2002). Because of the mass
action model’s success in understanding TB transmission in New Zealand brushtail
possums (T richosurus vulpecula), a Similar model was used to examine TB transmission
among deer (Hickling 2002). A regression model in the region of highest TB prevalence
predicted that reduction in herd density by one-third along with reduction in artiﬁcial
feeding by 75% would decrease TB prevalence to 0.5% after 10 years (Hickling 2002).

The model indicated that under current management, complete eradication of TB was

72

unlikely. The model did predict, however, that further reductions in either deer density
(an additional 40-50%) or artiﬁcial feeding alone could potentially result in the
elimination of TB (Hickling 2002).

The McCarty and Miller (1998) and Hickling (2002) models predict dramatically
different trajectories for TB in deer in DMU452. Unfortunately, both models fail to
incorporate characteristics of deer ecology that inﬂuence the probability of infection with
TB. Results from Chapter 2 of this dissertation indicated that deer infected with TB were
more closely related than were non—infected individuals both prior to and following
efforts to reduce density and contacts among deer. TB-infected deer were more closely
related than were non-infected deer within distances consistent with deer movement in
the lower peninsula of northeast MI indicating that the probability of infection with TB is
not random. Thus, unlike what has previously been modeled, TB transmission, facilitated
through interactions among relatives within matrilines, is at least partially frequency
dependent. Here, a model is presented highlighting interactions among relatives (within

matrilines) as an important risk factor for TB infection.

Methods

STELLA modeling software (Costanza et al. 1998) was used to create a
conceptual and simple quantitative model of TB transmission to evaluate whether TB
would be eliminated from DMU452 over a 20 year period starting in 1998, the year prior
to management actions taken to eradicate TB (Figure 11). The main purpose of the model
was to explicitly incorporate TB transmission among relatives (non-random, frequency

dependent), and separate it from random transmission among non-relatives (density

73

Matriline
Size

Infectious .
I nteractio ns

#ofUnreIated .
Interactions .

M
Population

Year
Density

 

 

 

 

Transmission
Rate J
6 .2. TB t1
TB Deer TB t ,LK.
Per Mile ' i

I Year
T # TB DEER
:2

Initial
# TB

6
Unrelated _
Interactions @

Proportion
Screened

Figure 11. A conceptual model of TB transmission among deer in the northeast lower

peninsula of Michigan.

74

dependent). Separation of density dependent and independent components was deemed
appropriate based on results from Chapter 2 that indicated that TB-positive deer were
more closely related than non-infected deer, implying that interactions and disease
transmission within groups of related and non-related indiViduals must be different.
The initial population density was set at 50 deer per square mile based on estimates of
deer population density in 1998. Density was modeled to decline as predicted by the
MDNR (Schmitt et al. 1997, Hickling 2002) (Table 14). The number of interactions
among unrelated individuals was set at 15% of population density per square mile (Table
14). The number of interactions among unrelated individuals was set at 15% in order to
make interactions among non-relatives nearly the same as the number of interactions
among related individuals, size of matrilines, at the beginning of the model. The
probability of interaction among non-related individuals was expected to decline
following the ban on artiﬁcial feeding and as population density declined (Table 14).
Matriline Size was parameterized based on previous studies of matrilineal
behavior in which females have been shown to live in groups usually consisting of an
adult doe, her yearlings and fawns, and often her older offspring as well as her sisters and
their offspring (Hawkins and Klimstra 1970, Marchinton and Hirth 1984). Deer within
matrilines have been observed to come into frequent contact with one another. Hawkins
and Klimstra (1970) found that 87% of yearling females remained philopatric and were
observed with other group members in greater than 60% of all observations. The average
size of a deer matriline varies based on a number of characteristics including rate of

population turnover, type of habitat, and age structure of the population. Matriline size

75

Table 14. Initial values of parameters included in the model of TB in deer in northeast

 

 

Michigan.
Parameter Value
Year Counter (0,21)

Population Density per mi2

Unrelated Interactions

# of Unrelated Interactions
Matriline Size
Transmission Rate

# of Infectious Interactions

Proportion Screened

# of TB Deer Per Mile
TB

TB tl

Initial # of TB Deer

# of TB Deer

Year 0-1= 50, 2= 41, 3-14= 32, 15-17=18,18-20=12
Constant 15%

Population Density * Unrelated Interactions

Constant 7

Constant 15%

(# of Unrelated Interactions + Matriline Size) *
Transmission Rate

Constant 25%

# of Infectious Interactions * Proportion Screened
TBt (t-dt) + (Deer TB — TBtl) "' dt

TBt

58

If year = 0 then Initial # TB+ Deer else # of TB Deer Per
Mile * (TBt + TBtl)

 

 

 

76

 

was modeled here to be a constant 7 deer assumed to be coming into frequent contact
(Table 14).

The rate of disease transmission per contact is one of the most difﬁcult
components of the model to parameterize. A number of stridies have been carried out to
evaluate the efﬁcacy of TB transmission among captive animals and from artiﬁcial feed
(Whipple and Palmer 2000). Very little is known, however, about TB transmission
resulting from face to face contacts among free-ranging deer. For this model, the
probability of TB transmission among interacting deer was extrapolated ﬁom recent
studies to be 15% of all contacts (Table 14) (Pahner et al. 1999).

Total infectious interactions was the sum of the number of interactions among
unrelated individuals and matriline size multiplied by the transmission rate (Table 14).
Not all deer that are infected with TB in DMU452 are detected because only harvested
deer are screened for TB. A parameter called “Proportion Screened” was included in the
model to account for the fact that only harvested deer, approximately25% of the
population, are screened for TB (Table 14) (Hickling 2002). The number of TB-positive
deer detected per mile was a function of the number of infectious interactions multiplied
by the proportion of deer screened for TB (Table 14). The total number of TB-positive
deer was determined by multiplying the number of TB-positive deer per mile by the
number of TB-positive deer in the previous year, initially 78 (Table 14) (O’Brien 2002).

Validation of this model was not possible because there were no pro-existing data
regarding TB transmission in this or any other free-ranging cervid population in North
America. The model was veriﬁed by comparing its prediction of TB in the ﬁrst 4 years to

empirical data (1998, 1999, 2000, and 2001) using a paired t-test.

77

The impacts of a variety of management scenarios on TB transmission were
simulated by manipulating key model parameters. Speciﬁcally, the frequency of
interaction among unrelated individuals was varied (0-25%) to reﬂect rates of contact
among deer at different intensities of artiﬁcial feeding. Population density was i
manipulated to reﬂect different possible management goals. In addition, the matriline size
was varied (0-7) to evaluate the impact of related interactions on the ntunber of TB-

positive deer.

Results

The output of the model indicated that the number of TB-positive deer declined as
population density and interactions among unrelated deer declined (Table 15). Unlike the
goals set by management to decrease disease prevalence to less than one percent by the
fall of 2003, and to have the disease eliminated from wild deer by the fall of 2010
(MDNR TB Activity Report 1997), however, TB was not eliminated from the population
(Table 15). The prediction of the model was that approximately 7 TB-positive deer would
be identiﬁed in DMU452 in 2010. The ban on artiﬁcial feeding served to reduce
interactions among unrelated individuals, but did not have an equivalent effect on
interactions among related deer within matrilines. The continued higher incidence of TB
among genetically similar individuals is responsible for the maintenance of TB in deer.

The model’s prediction of the number of TB-positive deer in the ﬁrst 4 years did
not differ from the actual number of TB-positive deer detected in 1998-2001 (t= -0.36,
df= 3, p= 0.74) (Table 16). The similarity of the model predictions to empirical data for

the ﬁrst 4 years helps verify the conceptual basis for pararneterization of the model.

78

Table 15. Model predictions of population density, number of unrelated interactions, and
number of TB-positive deer in northeast Michigan over 20 years. Years in which speciﬁc
prevalence goals were set by mana ement are 2003 (0.5%) and 2010 (0%).

 

 

 

Year Population # Unrelated # TB Deer
Densiy Interactions
0 50 10 78

1998 50 10 79.56
1999 41 8.2 72.56
2000 32 6.4 58.34
2001 32 6.4 46.9
2002 32 6.4 37.71
2003 32 6.4 30.32
2004 32 6.4 24.38
2005 32 6.4 19.6
2006 32 6.4 15.76
2007 32 6.4 12.67
2008 32 6.4 10.19
2009 32 6.4 8.19
2010 32 6.4 6.58
2011 32 6.4 5.29
2012 18 3.6 3.37
2013 18 3.6 2.14
2014 18 3.6 1.36
2015 12 2.4 0.77
2016 12 2.4 0.43
2017 12 2.4 0.24

 

 

 

 

 

79

Table 16. Number of TB-positive deer in northeast Michigan as predicted by the model
and empirical data from 1998-2001.

 

 

 

 

 

Year Model Prediction Empirical Data
1 (1998) 79.56 78
2 (1999) 72.56 58
3 (2000) 58.34 53
4 (2001) 46.9 60

 

 

80

When parameters were manipulated to reﬂect alternative management scenarios,
the number of TB-positive deer predicted in 2010 varied substantially. Further, more
drastic reductions in population density and in interactions among unrelated individuals
beyond current management goals resulted in fewer TB-pOsitive deer (Table 17). It is
doubtful, however, as to whether such extreme reductions in density or interaction rate
among unrelated individuals could be accomplished. Reducing the size of matrilines, and
thus interactions among related individuals, had the most rapid and greatest impact on

reducing the number of TB-positive deer (Table 17).

Discussion

The model predicts that TB will decline following the ban on artiﬁcial feeding
and decreased population density, but that TB will not be eliminated under current
management actions. The model clearly illustrates that interactions among related
individuals within matrilines play a key role in the maintenance of disease. Over time,
interactions among unrelated individuals decline, but interactions among relatives within
matrilines are constant and a function of matriline size. These interactions among
relatives are responsible for the continued maintenance of TB in the deer population.

Previous attempts to model TB transmission among deer in DMU452 did not
explicitly account for transmission among related individuals (McCarty and Miller 1998,
Hickling 2002). The model developed by McCarty and Miller (1998) predicted that TB
would increase substantially over time regardless of management efforts to reduce
disease transmission. Empirical data of TB prevalence in DMU452 in the years following

the development of the McCarty and Miller (1998) model indicated that TB prevalence

81

Table 17. Number of TB-positive deer in northeast Michigan in 2010 when initial values

of parameters were varied.

 

 

 

Parameter Manipulation # of TB Deer
Management Goal 0
Model 6.6
Unrelated Interactions 10% of population density 0.17
Unrelated Interactions 5% of population density 0.17
Unrelated Interactions 0% of population density 0
Unrelated Interactions 25% of population density 114.1
Population Density 18 per mi2 3.3
Population Density 10 per mi2 2.8
Population Density 5 per mi2 2.5
Population Density 50 per mi2 100.9
Matriline Size 4 0.27
Matriline Size 2 0.02
Matriline Size 1 0

 

 

82

 

has not increased (Hickling 2002). A regression model in the area of highest TB
prevalence predicted that a one-third reduction in herd density along with a 75%
reduction in artiﬁcial feeding would result in a TB prevalence of 0.5% after 10 or more
years (Hickling 2002). The model predicted that further reductions in deer density (up to
40-50%) or artiﬁcial feeding could potentially result in an elimination of TB (Hickling
2002)

The density dependent Hickling model was based upon the success of density
reductions in controlling TB in brushtail possums. The social structure of possums,
however, is quite different from white-tailed deer. Brushtail possums are basically
solitary animals (Nowak 1999). At high population densities individuals share dens and
home ranges overlap, but at low population densities possums become solitary, territorial,
and maintain discrete home ranges (Nowak 1999). White-tailed deer are highly social
animals in which females are philopatric and live in matrilines while males live in buck
groups for part of the year (Hawkins and Klimstra 1970, Marchinton” and Hirth 1984). In
addition, data from Chapter 2 clearly demonstrates that deer that are related to TB-
infected deer are more likely to be infected with TB than are nonrelatives. Therefore, and
as illustrated in the model presented here, manipulation of artiﬁcial feeding and overall
population density are unlikely to result in the elimination of TB. Instead, TB is likely to
be maintained at low prevalence in the deer population due to transmission among
individuals within matrilines.

The model presented here was intended to highlight that the risk of TB infection
as a ﬁrnction of genetic relatedness to an infected deer plays an important role in the

transmission and maintenance of TB. The concepts presented in this model could be

83

incorporated into the Hickling model to increase its accuracy and predictability for use as
a management tool. The model presented here also highlights areas of further research
whose results might suggest management actions that will increase the likelihood that TB
will be eliminated. For example, the results from Chapter 2 that indicated that the
probability of TB infection is not solely density dependent suggests that there may be a
critical threshold of population density below which density no longer predicts TB
transmission. Below this threshold, TB infection might occur predominantly within a few
infected matrilines. In addition, the relative role of density dependence in TB
transmission will depend upon how many individuals within a matriline are infected with
TB at a given time. If no more than one individual within a matriline is infected in a
given time period then a reduction in total herd density will increase the probability that
TB will be eliminated from a matriline. If, however, numerous individuals within a
matriline are infected at the same time, then reducing total deer population density is not
likely to eliminate all infected matrilines. Rather, more site speciﬁc reductions targeted at
TB-infected matrilines might be more effective.

Characteristics of host ecology are increasingly recognized as playing key roles in host-
pathogen transmission dynamics (Boone et al. 1998, Mills and Childs 1998, Begon et al.
1999, Hazel et al. 2000, Escutenaire et al. 2002, Meeus et al. 2002). In order to eliminate
disease from wildlife populations it is necessary to understand the role that host ecology
may play in disease transmission. This model is based on data that indicated that even
prior to management actions aimed at eliminating TB, there was a non-random risk of
infection with TB such that deer that were closely related to an infected deer had a higher

risk of TB infection than did nonrelatives. As a result of the artiﬁcial feeding ban, lower

84

population densities, and female philopatry, unrelated individuals come into contact less
frequently, and the disparity in the risk of infection for relatives of TB-infected deer
versus nonrelatives will increase. The results of the model presented here highlight the

role that deer social structure plays in the transmission and maintenance of TB.

85

CONCLUSIONS

Genealogical Relationships and the Probability of TB Infection

In this dissertation, estimates of genealogical relationships were used to infer the
role that white-tailed deer ecology plays in the risk of infection with bovine tuberculosis.
TB—positive deer were more closely related than were non-infected deer over the entirety
of DMU452 as well as at distances consistent with deer movement in the northeast lower
peninsula of MI (Garner 2001). Relatedness among TB-positive deer was higher than
among non-infected deer both before and after management actions aimed at eradicating
TB indicating that artiﬁcial feeding is not the sole mechanism by which TB is
transmitted. Results indicate that the probability of infection is non-random such that deer
related to a TB-infected deer have a greater probability of infection than do nonrelatives.
Deer’s strong matrilineal structure together with the non-random probability of infection
implies that the risk of TB infection is not solely a function of density. The ban on
artiﬁcial feeding and the current level of density reduction alone are unlikely to eradicate
TB. Management must also focus on deer social groups that likely facilitate the
maintenance of disease. The high rate of contact that occurs among related individuals
within matrilines relative to rates of contact among individuals from different matrilines
is one hypothesis as to the mechanism by which TB is transmitted among relatives.
Additional mechanisms by which related groups facilitate the maintenance of TB that
must also be evaluated include the possibility of genetic susceptibility to TB and risk

factors in the environment to which all members of a matriline might be exposed.

86

Microgeographic Genetic Structure in Deer

Analyses of wildlife spatial genetic structure can be used to infer the impact of
managementactions on social structure, movement ecology, and breeding behaviors that
would be difﬁcult to determine using traditional methods. Molecular markers were used
to evaluate the impact of artiﬁcial feeding and high deer densities on deer spatial genetic
structure. Prior to the artiﬁcial feeding ban and increased levels of harvest, there was no
Si gniﬁcant relationship between genetic differentiation between groups of deer and their
geographic distance from one another. Artiﬁcial feeding and high deer densities
homogenized spatial genetic structure in DMU452 likely due to the aggregation of
multiple matrilines at'feeding sites. Following changes in management, there was
evidence of heterogeneity in allele frequencies among quadrats of deer as well as a
signiﬁcant relationship between genetic differentiation between quadrats of deer and their
geographic distance from one another. The magnitude of genetic differentiation between
groups of deer as a function of geographic distance is likely the result of white-tailed deer
matrilineal structure, female philopatry, and site ﬁdelity previously documented in
observational and telemetry studies (Hawkins and Klimstra 1970, Marchinton and Hirth
1984). Changes in deer spatial genetic structure coincident with changes in management
can have important implications with regard to disease transmission in wildlife
populations. Disease transmission will be affected by the spatial distribution of infected

individuals (Fromont et al. 1998a,b, Tully et al. 1999, Fulford et al. 2002). Clusters of

87

genetically similar groups of deer in space indicate that there is spatial genetic structuring

in DMU452, and likely limited movement of TB across genetically differentiated groups.

Future Research Directions

The results presented in this dissertation generate a number of questions regarding
TB transmission and spatial genetic structure in white-tailed deer that can be addressed
by further research. Results presented here indicate that the probability of TB infection is
non-random and not solely density dependent. One major issue arising ﬁ'om these results
is the relative role of density in the maintenance of TB. The role of density will depend
upon the size of matrilines in DMU452 and how many individuals within a matriline are
infected with TB at a given time. Characterization of the variability of the pathogen,
Mycobacterium bovis, in the deer population, speciﬁcally within and among related
groups of deer, would help to clarify whether contacts among individuals within
matrilines facilitate disease transmission. Investigation into genetic susceptibility for TB
infection would also contribute to an understanding of the risks of infection.

Characterization of spatial genetic structure in deer in DMU452 can also be
approached using mitochondrial DNA (mtDNA) that is maternally inherited. Because
females are typically philopatric, spatial genetic structure as measured using mtDNA
would be useful in more precisely deﬁning matriarchal structure in DMU452. Future
studies of animals recruited into the population after the ban on artiﬁcial feeding may be
useﬁrl in addressing the impact of the artiﬁcial feeding ban and reduced density on the

deer population's mating system. In addition, the relationship between genetic

88

differentiation between groups of deer and the characteristics of habitat separating them
may be useful in spatial explicit models aimed at predicting the transmission and

distribution of TB in northeast MI.

Contributions to Management

The results presented in this dissertation contribute to efforts aimed at
understanding the transmission and maintenance of bovine tuberculosis in white-tailed
deer. An incorporation of these results with the research from other scientists has the
potential to lead to a more complete understanding of the mechanisms of TB transmission
among white-tailed deer. Such understanding will aid managers in developing strategies
to eliminate TB. Speciﬁcally, the non-random probability of infection with TB as a
function of genetic similarity to TB-infected individuals and the spatial aggregation of
genetically similar groups of deer can be incorporated into spatially explicit models of the
transmission and distribution of TB. Knowledge of the degree of population genetic
structuring as measured by spatial variation in allele frequencies can be used with data
identifying hotspots of TB to target such regions for management actions such as
intensive density reductions in hotspot areas. Targeted density reductions in hotspots will

be more likely to affect matrilineal groups that serve as reservoirs of TB infection.

89

General Contributions

Zoonoses are of increasing importance to wildlife conservation and human health
(Childs et al. 1998). Transmission dynamics are driven by interactions between host and
pathogen populations (Begon et al. 1999). It has become increasingly recognized that
wildlife ecology plays a key role in disease transmission (Boone et al. 1998, Mills and
Childs 1998, Begon et al. 1999, Hazel et al. 2000, Escutenaire et al. 2002, Meeus et al.
2002). It has been demonstrated based upon estimates of genealogical relationships that a
signiﬁcant reservoir of TB is within related groups likely facilitated by deer’s strong
social structure. This is the ﬁrst study of its kind to use genetics to infer the role of
wildlife behavior and social structure in transmission dynamics between host and
pathOgen populations. It would have been impossible to directly observe the movement,
behavior, and relationships of the entire white-tailed deer population in DMU452. With
access to hunter-harvested deer, however, molecular markers enabled genealogical
relationships among individuals at both macro and microgeographic scales to be rapidly
estimated. Data clearly demonstrate the utility of using molecular markers and population
genetic theory to infer the role of host ecology in the maintenance of disease in wildlife.
Finally, a number of studies have documented that anthropogenic impacts such as the
method and intensity of harvest and translocations can alter spatial genetic structuring in
deer populations (Leberg 1990, Scribner 1993, Scribner et al. 1997, Leberg and Ellsworth
1999). This study, however, is the ﬁrst to document a relative homogenization of allele

frequencies across a deer population when artiﬁcial feeding was occurring.

90

APPENDICES

91

APPENDD( I: MICROSATELLITE LOCI SCREENED IN WHITE-TAILED DEER

Lwrs

BL 25

BL 42

BM 203

BM 888

BM 4107 a
BM 4208

BM 5004
BMC 1009
BMI 861 (Y)
Cervid 1 a
Cervid 2 a
Cervid 3
Cervid 14
CRFA

CSN3
ETH152 a
lGF1 b

INRA 008 (Y)
INRA 057 (Y)
INRA 062 (Y)
INRA 124 (Y)
INRA 126 (Y)
IRBP2
MAF35
MAF65
MAF70
OarFCB193 a
OarFCB304
OBCAM ”

RT 1

RT 5

RT 6

RT 7

RT9b
RT10

RT 20 a
RT23a

RT 24 a

RT 27 a

RT 30
SRCRSP-3
SRCRSP-S
SRCRsp-ioa

Ampliﬁed

yes
yes
yes
no
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
no
yes
yes
yes
yes
yes
no
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
no
yes
yes
yes
yes
yes
no
yes
yes

Talbot et al. 1996
Talbot et al. 1996
Talbot et al. 1996
Talbot et al. 1996
Talbot et al. 1996
Talbot et al. 1996
Talbot et al. 1996
Talbot et al. 1996
Bishop et al. 1994
DeWoody et al. 1995
DeWoody et al. 1995
DeWoody et al. 1995
DeWoody et al. 1995
Moore et al. 1992
Bishop et al. 1994
Talbot et al. 1996
Kirkpatrick 1992
Vaiman et al. 1994
Vaiman et al. 1994
Vaiman et al. 1994
Vaiman et al. 1994
Vaiman et al. 1994
Bishop et al. 1994
Swarbick et al. 1991
Buchanan et al. 1992
Buchanan and Crawford 1992
Buchanan and Crawford 1993
Buchanan and Crawford 1993
Fries et al. 1993
Wilson et al. 1997
Wilson et al. 1997
Wilson et al. 1997
Wilson et al. 1997
Wilson et al. 1997
Wilson et al. 1997
Wilson et al. 1997
Wilson et al. 1997
Wilson et al. 1997
Wilson et al. 1997
Wilson et al. 1997
Arevalo et al. 1994
Arevalo et al. 1994

FROM DMU452
Optimized # of Alleles Size Range Reference
yes 6 170-190
yes 1 1 236-252
yes 1 7 200-230
na na na '
yes 1 5 139-1 67
yes 15 140-180
yes 23 180-230
no na ~270
no: amps in ferns na 140-150
yes 21 161-199
yes 1 1 1 58-1 78
yes 1 3 160-210
no na ~190
yes na 230-260
yes 3 185-195
yes 14 200-232
yes 1 7 1 14-1 58
na na na
yes 1 ?
yes 1 ?
~yes 1 smeary
yes 1 181
yes 8 160-180
na na na
no na ~230
yes 18 100-140
yes 14 93-125
yes 2 120-140
yes 10 194-214
yes 10 21 0-240
no na ~160
no na ~85
yes 1 5 200-240
yes 1 1 105-1 29
na na na
yes 13 223-259
yes 22 148-190
yes 13 202-236
yes 1 5 1 39-1 67
yes 20 160-210
na na na
yes 4 144-1 80
yes 8 208-222

a: used in Chapter 2, ”= used in Chapter 3

92

Bhebhe et al. 1994

 

Gm 000 cm NA. Vm 3 ES _ va
am 000 on No Nm vo SE _ «5
on 000 on Nb on vm SE _ ea
cm 000 on No wv va SE _ ed
on 000 on NA. On Va 58 — vo
om SE ~ No mm g 58 _ 00
cm £8 _ NA. cm va SE — ea
BEND 08C. vcoaxm .0055. 0:00:09 REE
mN mN mN mN
: Nav— mQE N.3
vd m6 md md
@— _ w; _
o._ Harm 9o _ mNod _
Wm N _ mN
mN mN nN mN
”EU .2<OmO .752 no mONEm m 2300 326030
3.x .2<OmOJ_-mO_ ”x0323: 2.50.2
wN mN mN wN mN mN mN mN mN mN
52 N4; N.m_ Num— o.c_ NE 52 >6— mvN— 06—
md md md m6 md md md md md md
m m._ _ o o _ N mud 0 m6
wd _ ~ — _ _ _ mch mch Yo
N N m._ N _ mN N ~ _ m.—
mN mN mN mN mN mN mN ms: mN mN mN
TmO— REM— M a. m|o|~ mdmmOMm gem—)5 N 2300 _ 2300 g
42<UmO 2va .mmmOMm mOmSO .ZmONm—mhm .woszm

 

 

 

 

 

 

 

 

mum—ESE Mama Qua—(«Hrmtmg QmN_2:A—O «On— wmm-Omm NEEDS u= VAR—2952‘

I

 

 

 

 

eooo20

oe0<2 0250
8.00030

28

.200 .8320 .820
800030

.3920 .900

N EEO ._ 2260

-02 .0000 .2000 .2
000000 .m EEO E0
63

62668

82.680 000

<29
02:0
as

0002 2e 2
.0820 208:: o—
mHZv SE N
56.: 60:0

<75
ONEV
GS

802 26 2
SEE— 2825 o—
mHZc .28 N

50.: 60:0
22:86.6. :3

93

REFERENCES

94

REFERENCES

Anderson, R.M., and RM. May. 1992. Infectious diseases of humans: dynamics and
control. Oxford University Press, Oxford.

Arevalo, E., D.A. Holder, J .N. Derr, E. Bhebhe, R.A. Linn, F. Ruvuna, S.K. Davis, and
J .F . Taylor. 1994. Caprine microsatellite dinucleotide repeat polymorphisms at the SR-
CRSP-l, SR-CRSP-2, SR-CRSP-3, SR-CRSP-4, and SR-CRSP-S loci. Animal Genetics
25:202.

Avise, J .C. 1994. Molecular markers, natural history and evolution. Chapman Hall, New
York, 51 lpp.

Barbujani, G. 1987. Autocorrelation of gene frequencies under isolation-by-distance.
Genetics 117:772-782.

Begon, M., S.M. Hazel, D. Baxby, K. Bown, R. Cavanagh, J. Chantrey, T. Jones, and M.
Bennett. 1999. Transmission dynamics of a zoonotic pathogen within and between
wildlife host species. Proceedings of the Royal Society of London B 266:1939-1945.

Bhebhe, E., J. Kogi, D. Holder, E. Arevalo, J .N. Derr, R.A. Linn, F. Ruvuna, S.K. Davis,
and J .F . Taylor. 1994. Caprine microsatellite dinucleotide repeat polymorphisms at the
SR-CRSP-6, SR-CRSP-7, SR-CRSP-8, SR-CRSP-9 and SR-CRSP-IO loci. Animal
Genetics 25:203.

Bishop, M.D., S.M. Kappes, J .W. Keele, R.T. Stone, S.L.F. Sunden, G.A. Hawkins, S.S.
Toldo, R. Fries, M.D. Grosz, J. Y00, and CW. Beattie. 1994. A genetic linkage map for
cattle. Genetics 136:619-639.

Boone, J .D., E.W. Otteson, K.C. McGwire, P. Villard, J .E. Rowe, and SC. St. Jeor. 1998.
Ecology and demographics of hantavirus infections in rodent populations in the Walker

River Basin of Nevada and California. American Journal of Tropical Medicine 59:445-
451.

Bowcock, A.M., A. Ruiz-Linaares, J. Tomfohrde, E. Minch, J .R. Kidd, and LL. Cavalli-
Sforza. 1994. High resolution of human evolutionary trees with polymorphic
microsatellites. Nature 368:455-457.

Buchanan, F .C., P.A. Swarbrick, and A.M. Crawford. 1992. Ovine dinucleotide repeat
polymorphism at the MAP 65 locus. Animal Genetics 23:85.

Buchanan, RC, and A.M. Crawford. 1992. Ovine dinucleotide repeat polymorphism at
the MAF7O locus. Animal Genetics 23:185.

95

Buchanan, RC, and A.M. Crawford. 1993. Ovine microsatellites at the OarFCBl 1,
OarFCB128, OarFCBl93, OarFCB266 and OarFCB304 loci. Animal Genetics 24:145.

Cavalli-Sforza, LL, and A.W.F. Edwards. 1967. Phylogenetic analysis: models and
estimation procedures. Evolution 21 :550-5 70.

Chesser, R.K. 1983. Genetic variability within and among populations of the black-tailed
prairie dog. Evolution 37:320-331.

Chesser, R.K. 1991a. Inﬂuence of gene ﬂow and breeding tactics on gene diversity
within populations. Genetics 129:573-583.

Chesser, R.K. 1991b. Gene diversity and female philopatry. Genetics 127:437-447.

Childs, J ., R.E. Shope, D. Fish, F.X. Meslin, C.J. Peters, K. Johnson, E. Debess, D.
Dennis, and S. Jenkins. 1998. Emerging zoonoses. Emerging Infectious Diseases 4:452-
454.

Chouinard, A. and L. F ilion. 2001. Detrimental effects of white-tailed deer browsing on
balsam ﬁr growth and recruitment in a second growth stand on Anticosti Island, Quebec.
Ecoscience 82199-210.

Cliff, AD. and J .K. Ord. 1981. Spatial processes: models and applications. Pion Limited,
London, 266 pp.

Corner, L.A.L., D.U. Pfeiffer, G.W. de Lisle, R.S. Morris, and BM. Buddle. 2002.
Natural transmission of Mycobacterium bovis infection in captive brushtail possums
( T richosurus vulpecula). New Zealand Veterinary Journal 50:154-162.

Costanza, R., D. Duplisea, and U. Kautsky. 1998. Ecological modelling and economic
systems with STELLA- Introduction. Ecological Modelling 110:1-4.

Cronin, M.A., M.E. Nelson, and DP. Pac. 1991. Spatial heterogeneity of mitochondrial
DNA and allozymes among populations of white-tailed deer and mule deer. J oumal of
Heredity 82:118-127.

DeWoody, J .A., R.L. Honeycutt, and LC. Skow. 1995. Microsatellite markers in white-
tailed deer. Journal of Heredity 86:317-319.

Dietz, E]. 1983. Permutation tests for association between two distance matrices.
Systematic Zoology 32:21-26.

Diniz-Filho, J .A.F ., and MP. Telles. 2002. Spatial autocorrelation analysis and the

identiﬁcation of operational units for conservation in continuous populations.
Conservation Biology 16:924-935.

96

Dobson, A., and M. Meagher. 1996. The population dynamics of brucellosis in the
Yellowstone National Park. Ecology 77: 1026-1036.

Epperson, B.K. 1993. Recent advances in correlation studies of spatial patterns of genetic
variation. Evolutionary Biology 27295-155.

Escutenaire, S., P. Chalon, F. De J aegere, L. Karelle-Bui, G. Mees, B. Brochier, F.
Rozenfeld, and P-P. Pastoret. 2002. Behavioral, physiologic, and habitat inﬂuences on
the dynamics of Puumala virus infection in bank voles (Clethrionomys glarelous).
Emerging Infectious Diseases 82930-936.

Felsenstein, J. 1993. PHYLIP (phylogeny interface package) version 3.5. Distributed by
the author. Department of Genetics, University of Washington, Seattle.

Fortin, M.-J., and J. Gurevitch. 2001. Mantel tests: spatial structure in ﬁeld experiments.
In Design and analysis of ecological experiments. S.M. Scheiner & J. Gurevitch eds.,
Oxford University Press, New York, NY. Pp. 308-326.

Fries, R., A. Eggen, and J .E. Womack. 1993. The bovine gene map. Mammal Genome
4:405-428.

Fromont, E., D. Pontier, and M. Langlais. 1998a. Dynamics of a feline retrovirus (F eLV)
in host populations with variable spatial structure. Proceedings of the Royal Society of
London B 265:1097-1104.

Fromont, E., M. Artois, and D. Pontier. 1998b. Epidemiology of feline leukemia virus
(FeLV) and structure of domestic cat populations. Journal of Wildlife Management
62:978-988.

Fulford, G.R., M.G. Roberts, and J .A.P. Heesterbeek. 2002. The metapopulation
dynamics of an infectious disease: tuberculosis in possums. Theoretical Population
Biology 61 :15-29.

Garner, MS. 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. Ph.D. Dissertation. Michigan State University,
East Lansing, MI, 270pp.

Geist, V., B. O’Gara, and RS. Hoffmann. 2000. Taxonomy and the conservation of
biodiversity. In Ecology and Management of Large Mammals in North America. pp. 1-
26. (eds. S. Demarais and PR. Krausman). Upper Saddle River: Prentice Hall.

Grenfell, B.T., and Dobson, AP. 1995. Ecology of infectious diseases in natural
populations. Cambridge University Press, Cambridge.

97

Guo, S.W., and EA. Thompson. 1992. Performing the exact test of Hardy-Weinberg
proportions for multiple alleles. Biometrics 48:361-3 72.

Hawkins, R.E., and W.D. Klimstra. 1970. A preliminary study of the social organization
of white-tailed deer. Journal of Wildlife Management 34:407-419.

Hazel, S.M., M. Bennett, J. Chantrey, K. Bown, R. Vavanagh, T.R. Jones, D. Baxby, and
M. Begon. 2000. A longitudinal study of an endemic disease in its wildlife reservoir:
cowpox and rodents. Epidemiological Infection 124:551-562.

Hickling, GJ. 2002. Dynamics of bovine tuberculosis in wild white-tailed deer in
Michigan. Wildlife Division Report no. 3363. Michigan Department of Natural
Resources, 34pp.

Hutchings, M.R., and P.C.L. White. 2000. Mustelid scent-marking in managed
ecosystems: implications for population management. Mammal Review 30: 157-169.

Kilgo, J .C., R.F. Labisky, and DE. Fritzen. 1998. Inﬂuences of hunting behavior of
white-tailed deer: implications for conservation of the Florida panther. Conservation
Biology 12:1359-1364.

Kirkpatrick, B.W. 1992. Identiﬁcation of a conserved microsatellite site in the porcine
and bovine insulin-like grth factor-I gene 5’ ﬂank. Animal Genetics 23:543-548.

Koenig, W.D. 1999. Spatial autocorrelation of ecological phenomena. Trends in Ecology
and Evolution 14:22-26.

Langenau, E. 1994. Michigan’s management centennial. Michigan Out-of-Doors. 42-44.

Leberg, PL. 1990. Genetic considerations in the design of reintroduction programs.
Transactions of the North American Wildlife and Natural Resources Conference 55:609-
619.

Leberg, PL, and BL. Ellsworth. 1999. Further evaluation of the genetic consequences
of translocations on southeastern white-tailed deer populations. Journal of Wildlife
Management 63:327-334.

Lehman, N., P. Clarkson, L.D. Mech, T.J. Meier, and R.K. Wayne. 1992. A study of the
genetic relationships within and among wolf packs using DNA ﬁngerprinting and
mitochondrial DNA. Behavioral Ecology and Sociobiology 3:83-94.

Loveridge, A.J., and D.W. Macdonald. 2001. Seasonality in spatial organization and

dispersal of sympatn'c jackals (Cam's mesomelas and C. adustus): implications for rabies
management. Journal of Zoology 253:101-111.

98

Lynch, M., and K. Ritland. 1999. Estimation of pairwise relatedness with molecular
markers. Genetics 152: 1 753-1 766.

Manly, BR]. 1997. Randomization, bootstrap and Monte Carlo methods in biology.
Chapman and Hall.

Mantel, N. 1967. The detection of disease clustering and a generalized regression
approach. Cancer Research 27:487-497.

Marchinton, R.L., and DH. Hirth. 1984. Behavior In White-tailed deer ecology and
management. L.K. Halls, ed., Stackpole Books, Harrisburg, PA, Pp. 129-168.

Matthews, N.E., and W.F. Porter. 1993. Effect of social structure on genetic structure of
free-ranging white-tailed deer in the Adirondack mountains. Journal of Mammalogy
74:33-43.

McCallum, H., N. Barlow, and J. Hone. 2001. How should pathogen transmission be
modelled? Trends in Ecology and Evolution 16:295-300.

McCarty, W.W., and M.W. Miller. 1998. A versatile model of disease transmission
applied to forecasting bovine tuberculosis dynamics in white-tailed deer populations.
Journal of Wildlife Diseases 34:722-730.

McCullough D.R.1979. The George Reserve Deer Herd: Population Ecology of a K-
' Selected Species. Ann Arbor: University of Michigan Press.

McCullough, D., and R.K. Chesser. 1987. Genetic variation within and among
populations of the Mexican prairie dog. Journal of Mammalogy 68:555-560.

Meeus, T. de, L. Beati, C. Delaye, A. Aeschlimann, and F. Renaud. 2002. Sex-biased
genetic structure in the vector of Lyme disease, Ixodes rincus. Evolution 56:1802-1807.

Michigan Department of Natural Resources. 1997. TB activity report.

Michigan Department of Natural Resources. 1993. Wildlife Division report.

Miller, R.A., J .B. Kaneene, S.D. Fitzgerald, and S.M. Schmitt. 2003. Evaluation of the
inﬂuence of supplemental feeding of white-tailed deer (Odocoileus virginianus) on the
prevalence of bovine tuberculosis in the Michigan wild deer population. J oumal of

Wildlife Diseases 39:94-95.

Mills, J .N., and J .E. Childs. 1998. Ecologic studies of rodent reservoirs: their relevance
for human health. Emerging Infectious Diseases 4:529-537.

99

Moore, S.S., W. Barendse, K.T. Berger, S.M. Armitage, and D.J.S. Hetzel. 1992. Bovine
and ovine microsatellites from the EMBL and GENBANK databases. Animal Genetics
23:463-467.

Nowak, RM. 1999. Walker’s Mammals of the World Volume 1. Johns Hopkins
University Press, Baltimore, pp. 92-94.

O’Brien, D.J., S.M. Schmitt, J .S. F ierke, S.A. Hogle, S.R. Winterstein, T.M. Cooley,
W.E. Moritz, K.L. Diegel, S.D. Fitzgerald, D.E. Berry, and J .B. Kaneene. 2002.
Epidemiology of Mycobacterium bovis in free-ranging white-tailed deer, Michigan, USA,
1995-2000. Preventive Veterinary Medicine 54:47-63.

Oden, N.L., and RR. Sokal. 1986. Directional autocorrelation: an extension of spatial
correlograms to two dimensions. Systematic Zoology 35:608-617.

Palmer, M.V., D.L. Whipple, and SC. Olsen. 1999. Development of a model of natural
infection with Mycobacterium bovis in white-tailed deer. J oumal of Wildlife Diseases
35:450-457.

Parker, P.G., A.A. Snow, M.D. Schug, G.C. Booton, and RA. Fuerst. 1998. What
molecules can tell us about populations: choosing and using a molecular marker. Ecology
79:361-382.

Petit, R.J., A. El Mousadik, and O. Pons. 1998. Identifying populations for conservation
on the basis of genetic markers. Conservation Biology 12:844-855.

Purdue, J .R., M.H. Smith, and J .C . Patton. 2000. Female philopatry and extreme and
extreme spatial genetic heterogeneity in white-tailed deer. Journal of Mammalogy
8 1 : 1 79-1 85.

Queller, DC, and KP. Goodnight. 1989. Estimating relatedness using genetic markers.
Evolution 43:258-275.

Ratnayeke, S., G.A. Tuskan, and MR. Pelton. 2002. Genetic relatedness and female
spatial organization in a solitary carnivore, the raccoon, Procyon lotor. Molecular
Ecology 11:1115-1124.

Raymond, M., and F. Rousset. 1995. GENEPOP (Version 1.2). Population genetics
software for exact tests and ecumenicism. Journal of Heredity 86:248-249.

Ritland, K. 1996. Estimators for pairwise relatedness and individual inbreeding
coefficients. Genetical Research 67: 175-185.

Rosenberg, MS. 2000. PASSAGE. Pattern analysis, spatial statistics, and geographic
exegesis. Version 1.0. Department of Biology, Arizona State University, Tempe, AZ.

100

 

Schmitt, S.M., S.D. Fitzgerald, T.M. Cooley, C.S. Bruning-Fann, L. Sullivan, D. Berry,
T. Carlson, R.B. Minnis, J .B. Payeur, and J. Sikarskie. 1997. Bovine tuberculosis in free-
ranging white-tailed deer from Michigan. Journal of Wildlife Diseases 33:749-758.

Scribner, K.T., M.C. Wooten, M.H. Smith, and RE. Johns. 1985. Demographic and
genetic characteristics of white-tailed deer populations subjected to still or dog hunting.
In Game harvest management, S.L. Beason and S.F. Roberson, eds., Caesar Kleberg
Wildlife Research Institute, Texas, pp. 197-212.

Scribner, K.T. 1993. Conservation genetics of managed ungulate populations. Acta
Theriologica 82:89-101.

Scribner, K.T., M.H. Smith, and R.K. Chesser. 1997. Spatial and temporal variability of
microgeographic structure in white-tailed deer. Journal of Mammalogy 78:744-755.

Smouse, P.E., J .C. Long, and RR. Sokal. 1986. Multiple regression and correlation
extensions of the Mantel test of matrix correspondence. Systematic Zoology 35:627-632.

Sokal, RR, and N.L. Oden. 1978. Spatial autocorrelation in biology. 1. Methodology.
Biological Journal of the Linnean Society 10:199-228.

Sokal, RR, and GM. J acquez. 1991. Testing inferences about micro-evolutionary
processes by means of spatial autocorrelation analysis. Evolution 45:152-168.

Sokal, R. R., N. L. Oden, and B. A. Thomson. 1997. A simulation study of
microevolutionary inferences by spatial autocorrelation analysis. Biological Journal of
the Linnean Society 60: 73- 93.

Sork, V.L., J. Nason, D.R. Campbell, and J .F. Fernandez. 1999. Landscape approaches to
historical and contemporary gene ﬂow in plants. Trends in Ecology and Evolution
14:219-224.

Stone, J ., and M. Bjorklund. 2001. DELRIOUS: a computer program designed to analyze
molecular marker data and calculate delta and relatedness estimates with conﬁdence.
Molecular Ecology 1:209-212.

Sugg, D.W., R.K. Chesser, F .S. Dobson, and J .L. Hoogland. 1996. Population genetics
meets behavioral ecology. Trends in Ecology and Evolution 11:338-342.

Swarbrick, P.A., F .C. Buchanan, and A.M. Crawford. 1991. Ovine dinucleotide repeat
polymorphism at the MAF35 locus. Animal Genetics 22:369-370.

Swofford, D.L., and R.K. Selander. 1981. BIOSYS-l: a FORTRAN program for the
comprehensive analysis of electrophoretic data in population genetics. Journal of
Heredity 72:281-283.

101

Talbot, J ., J. Haigh, and Y. Plante. 1996. A parentage evaluation test in North American
elk (wapiti) using microsatellites of ovine and bovine origin. Animal Genetics 27:117-
1 19.

Trombulak, SC, and CA. Frissell. 2000. Review of ecological effects of roads on
terrestrial and aquatic communities. Conservation Biology 14:18-30.

Tully, O., P. Gargan, W.R. Poole, and KP. Whelan. 1999. Spatial and temporal variation
in the infestation of sea trout (Salmo trutta L.) by the caligid copepod Lepeophtheirus
salmonis (Kroyer) in relation to sources of infection in Ireland. Parasitology 119:41-51.

United States Department of Agriculture. 2000. Tuberculosis in cattle and bison; state
and zone designations. Federal Register 65:39780-39782.

Vaiman, D., D. Mercier, K. Moazami-Goudarzi, A. Eggen, R. Ciarnpolini, A. Leplingle,
R. Vehnala, J. Kaukinen, S.L. Varvio, P. Martin, H. Leveziel, and G. Guerin. 1994. A set
of 99 cattle microsatellites: characterization, synteny mapping, and polymorphism.
Mammal Genome 5:288-297.

Van de Casteele, T., P. Galbusera, and E. Matthysen. 2001. A comparison of
microsatellite-based pairwise relatedness estimators. Molecular Ecology 10:1539-1549.

VanDenBussche, R.A., M.J. Hamilton, R.K. Chesser, and KT. Scribner. 1987. Genetic
differentiation among cottontails from isolated playa basins. Genetica 75:153-157.

Wayne, R.K., D.A. Gilbert, N. Lehman, D. Girman, P.J.P. Gogan, D.A. Gilbert, K.
Hansen, R.O. Peterson, U.S. Seal, A. Eisenhawer, L.D. Mech, and R.J. Krumenaker.
1991. Conservation genetics of the endangered Isle Royale gray wolf. Conservation
Biology 5:41-51. '

Weir, B.S., and CC. Cockerharn. 1984. Estimating F-statistics for the analysis of
population structure. Evolution 38: 1358-1370.

Whipple, D.L, and M.V. Palmer. 2000. Survival of Mycobacterium bovis on feeds. In
Conference on bovine tuberculosis proceedings 2000. Lansing, MI, p. 17.

Wilson, G.A., C. Strobeck, L. Wu, and J .W. Cofﬁn. 1997. Characterization of

microsatellite loci in caribou, Rangifer tarandus, and their use in other artiodactyls.
Molecular Ecology 6:697 -699.

102

TA
1

‘ MICHK‘S‘AN‘S TE UNIVEFS'leLIB‘RAR‘|ES
‘4 (“HM ‘ “
3 1293 02469 2356