. 25:1' .
Am?“

5 3:... v
“M”; .91 . u

. OJ 2. 1... \
A ,vs. 3.
31 u .. 2:

\

.
I I .4! V
Ln mu.» .49. ..
a. few...

1..

 

i ;.
rum

 

“43“”)th . 3.13;. ‘o |., ,1
a A: a mfifififiafi?

rrtfu

31‘ ,to 2.! J
‘ . awaken“.

32.1.2322: 311.7
‘94.}... ,. 7...?
«‘4... r 3.71:1 1,. ,

 

.1

,Z; LIBRARY
Michigan State
University

This is to certify that the
thesis entitled

QUANTIFYING AGGREGATION OF THE PARASITES OF
THE LYME DISEASE SYSTEM IN MENOMINEE COUNTY,
MICHIGAN

presented by
PAMELA L. ROY

has been accepted towards fulfillment
of the requirements for the

MS. degree in Fisheries and Wildlife

 

 

W/W J $449

 

fl/M'ajor Profess(br’s Signature

July 10, 2008

 

Date

MSU is an afiinnative-action, equal-opportunity employer

—.--—--.-.---gn-.-n--v—.-.-»— —.- - —

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K'IPro;IAcc&Pres/ClRC/DateDueiindd

 

QUANTIFYING AGGREGATION OF THE PARASITES OF THE LYME DISEASE
SYSTEM IN MENOMINEE COUNTY, MICHIGAN

By

Pamela L. Roy

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Department of Fisheries and Wildlife

2008

ABSTRACT

QUANTIFYING AGGREGATION OF THE PARASITES OF THE LYME DISEASE
SYSTEM IN NIENOMINEE COUNTY, MICHIGAN

By

Pamela L. Roy

Parasite aggregation is a phenomenon observed in nature across multiple systems
and refers to the tendency of most parasites to gather in or on a small number of hosts.
An aggregated parasite distribution may help a system maintain itself, and such may be
the case for Lyme disease transmission dynamics, establishment and maintenance.
Surveys of vegetation, questing ticks, and Peromyscus app. mice in Menominee County,
Michigan were conducted over the summer transmission period. Investigations for
parasite aggregation in Borrelia burgdorferi in tick and mammal hosts and vector ticks
(Ixodes scapularis) on small mammal hosts revealed aggregation for all populations.
Vegetation was compared with the spatial distribution of the pathogen, vector and
mammal host. Presence of coniferous trees was spatially correlated with questing larvae,
all questing nymphs, all mice and total ticks on mice. Implications of this aggregation

and spatial correlation on the Lyme disease cycle are discussed.

To overcoming imposter syndrome

iii

ACKNOWLEDGEMENTS
Jean Tsao

Sasha Kravchenko
Ned Walker
Ryan Henke
John Biando

Michael Erdman

Vemette Carlson

Karen Hubbard

Sarah Hamer

Graham Hickling

Michelle Rosen & Jonathon Mclain
Blair Bullard & Bill Morgan
Dustin Salter
Dan Beaudo
Mary Grace Stobierski & Kim Signs
Gabe Hamer
Wendell & Judith Johnson

Mercedes Ramirez

Kristin Bott

The Roys

iv

TABLE OF CONTENTS

List of Tables ............................................................................... vi
List of Figures .............................................................................. viii
Chapter 1
Introduction
Parasite aggregation ................................................................. 1
The study system: Lyme disease ................................................... 9
Describing aggregation ............................................................ 27
Research questions ................................................................. 33
Chapter 2
Methods
Field methods ......................................................................... 36
Laboratory methods ................................................................ 45
Statistical analyses ................................................................. 57
Chapter 3
Results
Descriptive statistics ................................................................ 63
Aggregation results ................................................................. 91
Analysis of spatial autocorrelation ............................................. 125
Chapter 4
Conclusion
Summary ............................................................................ 138
Discussion ........................................................................... 140
Conclusion .......................................................................... 154
Appendix .................................................................................... 157
References ................................................................................... 1 76

LIST OF TABLES

Table 1.1. Counties in regions of the state of Michigan with elevated case rates (cases
per 100,000 population) over a cumulative, 16-year period from 1992 to 2007. Data
courtesy Michigan Department of Community Health. See Figure 1.5 for map ....... 17

Table 1.2. I. scapularis burdens on P. leucopus collected in 2004 and 2005 in
southwestern Michigan. These data are supportive of the aggregation theory of ticks on
small mammal hosts. Unpublished data courtesy Sarah Hamer, 2006 .................... 20

Table 2.1. Daily field tasks representing one grid’s schedule. This cycle of tasks
occurred once per trapping grid for three grids. The entire cycle of three grids was
conducted four times over the course of the season .......................................... 38

Table 2.2. Categories used for understory classification. Any non-tree vegetation was
considered understory ............................................................................ 39

Table 2.3. Table of sources of the cultures used to create the standard curves for qPCR.
The standard curves were used to determine the number of organisms present in a
sample. All cultures were grown from adult I. scapularis collected from the field in April
2006. Listed is the location where that tick was collected, its assigned code, the resulting
standard curves, the number of spirochetes at each point in the curve and the strain of
each culture. All cultures grown and standardized by Sarah Hamer ...................... 46

Table 2.4. Variables compared using regression and a cross-correlogram. Vegetation
actually refers to three variables: understory, deciduous trees and coniferous trees, and
three separate analyses were run. Values of the coefficient of correlation, r, and the
distance at which the association was still significant are listed. Comparisons in bold
were significant for at least one of the runs. See Table 3.7 and 3.8 for results .......... 61

Table 3.1. Genera of trees at each grid. A total of 240 data points was taken at each
grid .................................................................................................. 64

Table 3.2. Percent of total small mammal captures among all three grids by species at
the field site in Menominee County, June-August 2006. A total of 230 individuals were
captured ............................................................................................ 70

Table 3.3. Population estimates and standard error for each 1.15 ha grid at each time
period. Three grids were sampled three times each over the trapping season from June to
August, 2006, in Menominee County, MI. The 3 grids were enumerated with Roman
numerals III, IV, and V. Population estimates were derived using Program CAPTURE
(United States Geological Survey, Patuxent, Rhode Island) ............................... 71

vi

Table 3.4. Variance to mean ratios, Lloyd’s mean crowding (LMC) index and the
negative binomial statistic k for number of B. burgdorferi organisms in the listed
populations of l. scapularis. Variance to mean ratios greater than one indicate
aggregation. Values of k less than one indicate aggregation. LMC values indicate how
many other parasites the average parasite shares a host with. Spirochetes in questing
larvae show very little aggregation because there are so few B. burgdorferi organisms in
the population, since generally, the ticks hatch uninfected. The spirochetes from the ticks
that had been feeding on mice show a high degree of aggregation, most likely as a
reflection of the distribution of the spirochetes among their hosts ........................... 92

Table 3.5. Determining spatial autocorrelation ofticks species and Iifestage, using
Moran’s Index ...................................................................................... 104

Table 3.6. Null hypothesis testing for aggregated spirochetes in the following
populations. For each listed population, the index of dispersion, n —— l and Chi—square
value with (n — 1) degrees of freedom are listed. Populations with indices of dispersion >
the Chi-square value are considered to deviate from a random distribution ............... 124

Table 3.7. Comparisons made using the cross-correlograms. The distance at which the
pairs were spatially correlated is listed for each trap period. Units are in decimal degrees.
Distance values listed all had an r value greater than the value of significance, 0.1069,
although individual r values are not listed here. If a 0 is listed, factors were not
correlated at any distance. Numbers smaller than 0.0002 (the minimum lag distance) and
greater than or equal to 0.0036 (the maximum lag distance) were not considered to be
significant. Variables with (*) were log-transformed for the analysis .................... 127

Table 3.8. Distances at which pairs of data were spatially correlated, for significant

associations only. Units listed are meters. Figure column listed figure where cross-
correlogram can be seen ......................................................................... 136

vii

LIST OF FIGURES

Figure 1.1. The two-year life cycle of Ixodes scapularis. Ticks start out as eggs in the
spring, and take two bloodmeals (males) or three bloodmeals (females) from different
hosts in their lifetimes. Figure modified from one produced by the American Lyme
Disease Foundation ............................................................................... 1 1

Figure 1.2. Schematic model of LD transmission dynamics ............................... 12

Figure 1.3. County map of the state of Michigan, including both the Upper and Lower
Peninsulas. Shading indicates Menominee County. Inset shows the study site
(N 45°9'46.2", W 87°42' 48.0"). Images in this thesis are presented in color ............ 13

Figure 1.4. A model of habitat suitability for I. scapularis establishment (Guerra et al.
2002) applied to Michigan predicts the potential future distribution of the tick, with the
darkest areas indicating highest risk for establishment upon introduction. Courtesy Erik
Foster, 2004, in thesis .............................................................................. 16

Figure 1.5. A map displaying the differences in infection prevalence of human Lyme
disease cases in Michigan. Case rates per 100,000 population are additive over a 16-year
period. See Table 1.1 for names of counties included in shaded regions. Data courtesy
Michigan Department of Community Health .................................................... 18

Figure 1.6. Ixodes scapularis on Peromyscus leucopus trapped at Van Buren State Park,
2004. Solid line shows the curve approximating the NBD. Unpublished data courtesy
Sarah Hamer, 2006 .................................................................................. 20

Figure 1.7. A visual representation of aggregation on multiple levels. In a host
population (mice, level 1), aggregation of ectoparasites (ticks, level 2), occurs. Further,
aggregation of endoparasites (spirochetes, level 3) occurs within the tick population. . . .21

Figure 1.8. A visual representation of the hypothesis that the upper 20% of parasite
loaded ticks harbor 80% of the spirochetes (false data set). Images in this thesis are
presented in color .................................................................................. 25

Figure 2.1. Eight by ten trapping grid with 12 In between trap lines (total 1.15 ha). Grey
rectangles at grid vertices each represent 2 Sherman long-folding traps used to minimize
the effect of trap saturation ........................................................................ 3 7

Figure 2.2. Scheme for vegetation sampling. The grey boxes represent the Sherman
traps present at each point where Easting and Northing coordinates were taken. The
12m2 box around it represents the area around each point considered for vegetation
sampling. Dotted lines represent the trap lines on the grid ................................... 39

viii

Figure 2.3. Scheme for drag-collecting ticks from grids. As indicated, dragging
transects were conducted between trapping lines. At each stop point, indicated by the
stop signs every 24 meters, drag cloths were lifted and checked on both sides for ticks.
Images in this thesis presented in color .......................................................... 41

Figure 2.4. Intensive drag-sampling scheme for questing ticks, for one of the 20 above-
labeled 24 m2 grids (Fig. 2.3), chosen at random for each grid, at each time step. Arrows
indicate transects dragged between trapping lines and stop signs indicate where drag
cloths were checked for ticks. Images in this thesis presented in color ..................... 42

Figure 2.5. The amplification plot for the standard curve trial for B. burgdorferi for five
dilutions ranging from 10'1 to 103. (SDS 2.3). Nine samples for each dilution had been
prepared, and the points for each order of magnitude clump together on the graph,
showing their close Ct values and thus tight reproducibility. Ct value refers to the cycle
number at which the PCR product with the labeled probe was detectable above a preset
threshold .............................................................................................. 47

Figure 2.6. An example of a standard curve used to determine numbers of spirochetes in
tick and tissue biopsy samples. Ct value refers to the cycle number at which the PCR
product with the labeled probe was detectable above a preset threshold. The regression
equation calculated by Microsoft Excel was used to determine the number of spirochetes
present in the sample. The Ct value of the sample was substituted for x, and y, or the
number of spirochetes was obtained algebraically ............................................. 50

Figure 2.7. Creating mammal DNA samples to optimize detection of B. burgdorferi
by PCR. Flowchart detailing the steps to create the 4 subsamples from each original
SOul elution of DNA from mammal ear biopsy punches. The 10 samples used for
optimization trials were those from older mice infested with ticks that had tested positive
for B. burgdorferi ................................................................................... 52

Figure 2.8. Generalized plate map for the ELISA assay. Several of the wells were left
empty in order to later calculate an average baseline (wells labeled ‘E’). The positive
controls (denoted ++ and +), negative control (denoted -) and samples (all other numbers)
were arranged in the remaining well according to the diagram...... .......56

Figure 2.9. Example of overlaying the rasterized surface (calculated from the drag
sample points, pink circles) with the trap points (yellow squares) from one grid. The
concentric colored circles in the bottom left are radiating from one point with a high
number of total ticks collected from dragging. The solid green around the rest of the
figure indicates zero values at the remaining trap points. Figures in this thesis presented
in color ............................................................................................... 59

Figure 3.0A. Kriged surface of understory vegetation layer. Data were interpolated

using ordinary kriging between 12 data points. Darker colors indicate heavier understory
cover. Dots represent trap points. Figures in this thesis presented in color ................. 65

ix

Figure 3.0B. Kriged surface of deciduous tree layer. Data were interpolated using
ordinary kriging between 12 data points. Darker colors indicate heavier deciduous tree
cover. Dots represent trap points. Figures in this thesis presented in color ................. 65

Figure 3.0C. Kriged surface of coniferous tree layer. Data were interpolated using
ordinary kriging between 12 data points. Darker colors indicate heavier coniferous tree
cover. Dots represent trap points. Figures in this thesis presented in color ................ 66

Figure 3.1. Questing Ixodes scapularis ticks collected over time at the study site
(Menominee County) in 2006. Data are pooled from all three trapping grids. Collections
are represented as proportions of the total collected (per stage) on that date rather than
raw numbers. Sampling appears to have begun at the end of the spring adult peak. No
obvious peak nymphal activity was detected. Larvae numbers peaked in mid-July. . . ..67

Figure 3.2. Population estimates for mice each trap period, by 1.15 ha grid.
Estimates calculated using program CAPTURE (United States Geological Survey,
Pamxent, Rhode Island). Error bars were calculated using the average 95% confidence
intervals for the grid ................................................................................ 72

Figure 3.3A. Questing Ixodes scapularis ticks testing positive and negative for
Borrelia burgdorferi. Approximately 3% of larvae, 22% of nymphs and 57% of adults
tested positive for B. burgdorferi. Larval ticks were not tested individually, but were
pooled with other larvae from the same animal or same drag transect ...................... 73

Figure 3.3B. On-host Ixodes scapularis ticks testing positive and negative for

Borrelia burgdorferi. Approximately 40% of larvae and 46% of nymphs tested positive
for B. burgdorferi. Larval ticks were not tested individually, but were pooled with other
larvae from the same animal or same drag transect .......................................... 74

Figure 3.4. Electrophoresis gel result from the ear biopsy DNA treatment trial. The
fragment size expected was a 987 bp amplicon of the intergenic spacer region (IGS)
between 165 (rrs) and a 23s (rrlA) rRNA genes in B. burgdorferi, (Bunikis et al. 2004b).
Sample 29, desalted and a 1:10 dilution (faint), and sample 38, both undiluted desalted
and a desalted 1:10 dilution show amplification. Standard used Invitrogen E-Gel low-
range quantitative DNA ladder .................................................................... 76

Figure 3.5. Frequency histogram for number of total ticks on mice over the course
of the trapping season. Numbers for each mouse are cumulative over the whole
season ................................................................................................. 78

Figure 3.6. Comparison of average tick burdens per mouse by trap period
(number) and age class (juvenile or adult) ................................................... 79

Figure 3.7. Comparison of total I. scapularis burdens (larvae and nymphs) between
juvenile (age=l) and adult (age=3) age groups for animals captured in trap period 1.
Only tick burdens for animals captured the first time are included in this dataset. Repeat
captures are omitted to avoid lack of independence and repeated measures issues.
Boxplot shows no significant difference between age groups ............................... 80

Figure 3.8. Comparison of total I. scapularis burdens (larvae and nymphs) between
juvenile (age=1) and adult (age=3) age groups for animals captured in trap period 2.
Only tick burdens for animals captured the first time are included in this dataset. Repeat
captures are omitted to avoid lack of independence and repeated measures issues.
Boxplot shows no significant difference between age groups ............................... 82

Figure 3.9. Comparison of total I. scapularis burdens (larvae and nymphs) between
juvenile (age=1) and adult (age=3) age groups for animals captured in trap period 3.
Only tick burdens for animals captured the first time are included in this dataset. Repeat
captures are omitted to avoid lack of independence and repeated measures issues.
Boxplot shows no significant difference between age groups ................................ 83

Figure 3.10. Frequency histogram showing the average adjusted optical density (OD)
values for the Peromyscus spp. serum samples collected in Menominee County from June
- August 2006. Two peaks in the data are evident; one in the negative range at 0.2 (*),
and the other in the positive range at 2.3(**). The cutoff value of 0.8784 is represented
by the dashed line. The cutoff was calculated to be 3 SD above the negative control
average. Values above the cutoff were considered positives, or serum samples with
antibodies to Borrelia burgdorferi, and values less than the value of the line are
considered negatives, samples without antibody response ................................... 85

Figure 3.11. Comparison of the average adjusted optical density (OD) values per
sample with the endpoint dilution at which the sample was still positive. Graph
demonstrates relationship between average adjusted optical density (OD) value for a
sample and the final dilution at which it was positive. Negative values are not shown on
the graph. The graph displays the trend that typically, higher average adjusted OD values
correlated with higher endpoint titers ............................................................ 86

Figure 3.12. Comparison of the average adjusted optical density (OD) values for
each population with the age of the population. Graph demonstrates the average OD
values of the serum samples of juvenile mice (age category 1) with adult mice (age
category 3) .......................................................................................... 87

Figure 3.13. Scatterplot comparison of tick burdens (log-transformed, on x-axis) and
average OD values (on y-axis) for trap period 1. Juvenile mice (a=1, black circles) and
adult mice (a=3, red circles) of the Peromyscus spp. were trapped at the field site in
Menominee County, 2006. Images in this thesis are presented in color ................... 88

xi

Figure 3.14. Scatterplot comparison of tick burdens (log-transformed, on x-axis) and
average OD values (on y-axis) for trap period 2. Juvenile mice (a=1, black circles) and
adult mice (a=3, red circles) of the Peromyscus app. were trapped at the field site in
Menominee County, 2006. Images in this thesis are presented in color .................. 89

Figure 3.15. Scatterplot comparison of tick burdens (log-transformed, on x-axi s) and
average OD values (on y-axis) for trap period 2. Juvenile mice (a=1, black circles) and
adult mice (a=3, red circles) of the Peromyscus spp. were trapped at the field site in
Menominee County, 2006. Images in this thesis are presented in color .................. 90

Figure 3.16. Variance to mean ratios for I. scapularis ticks on mice by trap period
and grid. The dotted line represents the cutoff for an aggregated population, or a
variance to mean ratio of 1. Overall, the larvae have higher ratios, indicating more highly
aggregated populations (black symbols). Trap period 1 shows the most highly
aggregated populations of ticks on mice ........................................................ 93

gure 3.17. Comparison of mean number of ticks found on mice per trap period by
grid, to the variance to mean ratio for the same population. Ticks were divided by
lifestage. Generally, as the number of ticks in the population increases, so does the
variance to mean ratio for that population, indicating that degree of aggregation may
partly be a function of n. This graph illustrates how smaller sample sizes may not reflect
the true degree of aggregation, since highly parasitized individuals (of which there are
few) will be missed ............................................................................... 94

Figure 3.18. Comparison of mean number of spirochetes found in mice per trap
period by grid, to the variance to mean ratio for the same population. Generally, as
the number of individuals in the population increases, so does the variance to mean ratio
for that population, indicating that degree of aggregation may partly be a function of n.
This graph illustrates how smaller sample sizes may not reflect the true degree of
aggregation, since highly parasitized individuals (of which there are few) will be missed.
The lowest point is the data point for grid IV, trap period 3. The highest point is grid IV,
trap period 1. There appear to be only 8 points because two of the data points overlap
completely .......................................................................................... 95

Figure 3.19. Gaussian spatially-correlated mouse chart. Variograrn for distance,
showing the approximate Gaussian curve for spatial correlation ............................ 97

Figure 3.20. Maps displaying areas of mouse capture, by trap period (1, 2 and 3). Dots
represent each trap point and the size of the dot indicates the O, 1 or 2 captures for that
trap period ........................................................................................... 98

Figure 3.21. Frequency histogram for number of larval ticks on mice at first

capture. The vzm ratio and k statistic indicate a high degree of aggregation. The LMC
index indicates that the average parasite endures a high degree of crowding ............ 100

xii

Figure 3.22. Frequency histogram for number of nymphal ticks on mice at first
capture. The vzm ratio and k statistic indicate a high degree of aggregation. The LMC
index indicates that the average parasite endures a high degree of crowding ............ 101

Figure 3.23. Frequency histogram for number of total ticks on mice at first capture.
The vzm ratio and k statistic indicate a high degree of aggregation. The LMC index
indicates that the average parasite endures a high degree of crowding .................... 102

Figure 3.24. Simultaneous parasitism of larval and nymphal I. scapularis on
individual mice. Graph shows the number of larvae and the number of nymphs that an
individual mouse had at first capture. When multiple mice had the same configuration of
larvae and nymphs, the graph still just displays one symbol. Overall, mice had higher
larval burdens than nymphal burdens. Not all mice that had larvae had nymphs, but
nearly all mice that had nymphs also had larvae feeding .................................... 103

Figure 3.25. Kernel density of all species of larvae from mice (A) a somewhat clustered
pattern, and of I. scapularis larvae from mice (B) also a clustered pattern. Descriptions
of clustering based on Moran’s Index analysis. Map shows all three grids together.
Darker areas indicate more ticks. Dots show GPS-located trap locations. Figures in this
thesis presented in color ......................................................................... 104

Figure 3.26. Kernel density of all nymphs from mice (A) a random pattern, and of I.
scapularis nymphs from mice (B) a clustered pattern, centered on grid IV. Descriptions
of clustering based on Moran’s Index analysis. Map shows all three grids. Darker areas
indicate more ticks. Dots show GPS-located trap locations. Figures in this thesis
presented in color ................................................................................. 105

Figure 3.27. Frequency histogram of number of mice with numbers of spirochetes.
This population is highly aggregated, as indicated by the large vzm ratio, small k. LMC
index indicates that the average parasite in this population experiences a lot of

crowding .......................................................................................... 106

Figure 3.28. Frequency histogram for spirochete loads in all ticks recovered from hosts.
The number of spirochetes is presented on a log scale. For each group, summary
statistics are presented: k parameter for the negative binomial distribution, variance to
mean ratio (V:M), and Lloyd’s mean crowding index (LMC). The k statistic indicates
that the larvae are more highly aggregated. Both sample populations have a V:M ratio to
the same order of magnitude, 10"4, indicating aggregation. The LMC index for nymphs
is higher, showing that that the average spirochete in a nymph on a host feels more
crowded than the average spirochete in a larva feeding on a host ........................ 107

Figure 3.29A. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent

trap points. Size of dots represent the number of infected mice captured at those trap
points. Increasing dot sizes represent 0, 1 and 2, respectively ........................... 109

xiii

Figure 3.29B. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the average adjusted OD values for the mice captured
at those trap points. The smallest dots represent mice with no developed antibody
response to B. burgdorferi, and the large dots represent those that do have antibodies
against the pathogen ............................................................................... 1 1.0

Figure 3.29C. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of spirochetes in the mice captured at
those trap points. Increasing dot sizes represent 0, 150, 500, 1500 and 5500,

respectively ......................................................................................... 1 1 1

Figure 3.29B. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of larvae on the mice captured at those
trap points. Increasing dot sizes represent 0, 5, 10, 20 and 35, respectively ............. 112

Figure 3.29E. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of infected larvae on the mice captured at
those trap points. Increasing dot sizes represent 0, 1, 2 and 3, respectively ............ 113

Figure 3.29F. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of spirochetes in larvae on the mice
captured at those trap points. Increasing dot sizes represent 0, 100, 1000, 9999 and
35,000 respectively ................................................................................ 114

Figure 3.29C. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of nymphs on the mice captured at those
trap points. Increasing dot sizes represent 0, 1, 3, 6 and 20 respectively ............... 115

Figure 3.29131. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of infected nymphs on the mice captured
at those trap points. Increasing dot sizes represent 0, 1 and 2 respectively ............. 116

Figure 3.291. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of spirochetes in nymphs on the mice
captured at those trap points. Increasing dot sizes represent 0, 100, 500, 1000 and 37,000
respectively ....................................................................................... 1 1 7

Figure 3.29J. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the total ticks on the mice captured at those trap
points. Increasing dot sizes represent 0, 3, 8, 15 and 54 respectively ................... 118

Figure 3.29K. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent

trap points. Size of dots represents the questing larvae at those trap points. Increasing
dot sizes represent 0, 2, 5, 16 and 79 respectively ............................................ 119

xiv

Figure 3.29L. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the questing nymphs at those trap points. Small and
large dot sizes represent 0 and 1 respectively ................................................. 120

Figure 3.29M. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots
represent trap points. Size of dots represents the infected questing nymphs at those trap
points. Increasing dot sizes represent 0, 1 and 2 respectively ............................. 121

Figure 3.29N. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the total questing ticks at those trap points.
Increasing dot sizes represent 0, 5, 17 and 79 respectively .................................. 122

Figure 3.290. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the total questing ticks at those trap points.
Increasing dot sizes represent 0, 250, 1500 and 5500 respectively ........................ 123

Figure 3.30A. Significant cross-correlogram for time 1. The x-axis shows the distance
in decimal degrees at which the pair is spatially correlated. The y-axis shows the r value,
the coefficient of correlation. The significant r = 0.1069 is represented by the dashed
line. Graph compares the spatial correlation between the number of infected mice and
conifers. They are spatially correlated between 0.0004 and 0.0027 decimal degrees... 128

Figure 3.30B. Significant cross-correlogram for time 1. The x-axis shows the distance
in decimal degrees at which the pair is spatially correlated. The y-axis shows the r value,
the coefficient of correlation. The significant r = 0.1069 is represented by the dashed
line. Graph compares the spatial correlation between the number of infected questing
nymphs and conifers. They are spatially correlated between 0.0021 and 0.0025 decimal
degrees .............................................................................................. 129

Figure 3.31A. Significant cross-correlogram for time 2. The x-axis shows the distance
in decimal degrees at which the pairs are spatially correlated. The y-axis shows the r
value, the coefficient of correlation. The significant r = 0.1069 is represented by the
dashed line. Graph compares the spatial correlation between deciduous trees and
questing larvae. They are spatially correlated between 0.0018 and 0.0020 decimal
degrees ............................................................................................. 130

Figure 3.31B. Significant cross-correlogram for time 2. The x-axis shows the distance
in decimal degrees ,at which the pairs are spatially correlated. The y-axis shows the r
value, the coefficient of correlation. The significant r = 0.1069 is represented by the
dashed line. Graph compares the spatial correlation between conifers and questing
larvae. They are spatially correlated between 0.0018 and 0.0024 decimal degrees... 131

XV

Figure 3.32A. Significant cross-correlogram for time 3. The x-axis shows the distance
in decimal degrees at which the pairs are spatially correlated. The y-axis shows the r
value, the coefficient of correlation. The significant r = 0.1069 is represented by the
dashed line. Graph compares the spatial correlation between total ticks and conifers.
They are spatially correlated between 0.0028 and 0.0034 decimal degrees ............ 132

Figure 3.32B. Significant cross-correlogram for time 3. The x-axis shows the distance
in decimal degrees at which the pairs are spatially correlated. The y-axis shows the r
value, the coefficient of correlation. The significant r = 0.1069 is represented by the
dashed line. Graph compares the spatial correlation between number of mice and
conifers. They are spatially correlated between 0.0032 and 0.0034 decimal degrees... 133

Figure 3.32C. Significant cross-correlogram for time 3. The x-axis shows the distance
in decimal degrees at which the pairs are spatially correlated. The y-axis shows the r
value, the coefficient of correlation. The significant r = 0.1069 is represented by the
dashed line. Graph compares the spatial correlation between questing nymphs and
conifers. They are spatially correlated at 0.0002 decimal degrees ......................... 134

Figure 3.33. Cross-correlogram for spatial correlation between deciduous trees and total

ticks on mice at time 3. Significant r = 0.1069, and all values are below that level. Thus,
there is no spatial correlation at any distance ................................................ 135

xvi

CHAPTER 1

INTRODUCTION

Parasite aggregation
The theory of parasite aggregation

This study’s theory of interest, aggregated distributions of parasite populations
among hosts, has a great mass of supporting ecological evidence from across diverse
systems (Shaw and Dobson 1995). Ecologi sts have long observed that parasites are
aggregated, rather than randomly distributed, among hosts (Crofton 1971, Anderson and
May 1978). Furthermore, there is evidence of a consistent finding that 20% of hosts
harbor 80% of the parasites (Woolhouse et al. 1997), and in some human/helminth
systems, that only 15% of available hosts feed 80% of parasites (Gregory and Woolhouse
1993). This aggregated distribution may have important implications for host-parasite
dynamics and thus for control and management of such systems.

Parasitism is an ecological association between species in which a parasite lives in
or on the body of a host (Anderson and May 1978). A more rigorous definition of
parasitism is where the parasite gains a benefit to the detriment of the host. Parasites
distributed across a host population can be measured by prevalence, or the number of
hosts infected, and by load, the number of parasites per host. Parasite aggregation is a
phenomenon often observed in nature, among both macro- (e.g. -ticks, lice) and micro-
(e. g‘.- bacteria, protozoa) parasites (Anderson and May 1978, May and Anderson 1978,
Shaw and Dobson 1995, Woolhouse et al. 1997). Aggregation refers to the tendency of

parasites to group among hosts in an uneven manner, where some hosts will have much

larger burdens of parasites than others, and the variance to mean ratio of these parasite
burdens is much greater than one. In aggregated systems, the prevalence may also be low
because many of the hosts will have zero parasites.

If aggregation did not occur, one would assume that a host population infected
with a particular parasite would have a parasite burden that follows a random distribution.
If parasites were distributed among hosts due only to random chance, we would assume
the Poisson distribution, where variance is equal to the mean. In nature, it is likely that
some potential hosts never encounter a particular parasite; some will encounter them over
and over again or may encounter many at once. Most will fall somewhere between those
extremes. This is considered the statistical distribution that serves as a null hypothesis to
the aggregation hypothesis. Any distribution that follows along a Poisson has a variance
to mean ratio of one; when that ratio rises above one, distribution is considered to be
greater than random and hence referred to as over-dispersed or aggregated (Shaw and
Dobson 1995, Elston et a1. 2001).

Multiple studies have documented the phenomenon seen in host-parasite systems
where the 20% of hosts with the highest parasite loads account for 80% of the parasites.
It should be noted that this finding is species-specific and thus refers to single-species
populations.

There has been at least one study documenting the difference in aggregation
among endemic foci, where it is demonstrated that there is a direct relationship between
pathogen prevalence and intensity of aggregation (Guyatt et al. 1994). The study found
that an increase in prevalence of parasite infections was related to an increase in intensity

of infections (prevalence of heavy infections). Thus, there can be heterogeneity in

aggregation patterns within an endemic area. The authors hypothesize that this difference
may be due to past geographic variation in transmission patterns, such as the age of the
endemic foci. Newly infected regions showed a pattern of more extreme aggregation.
This difference is hypothesized to be due to the lack of a ‘leveling’ effect that a
population with more infection experience and thus, increased immunity might have
(Guyatt et al. 1994).

Examples from the literature span many systems, and include helminthes (Pal and
Lewis 2004, Cattdori et al. 2004, Duerr et al. 2004, Churcher 2005), ticks (Randolph et
al. 1999a), bacteria (Randolph et al. 1996, Wang et a1. 2003), schistosomes (Guyatt et al.
1994, Woolhouse et al. 1998), trypanosomes (Lord et al. 1999) and viruses (Randolph et
al. 1996). A range of hosts for these aggregated organisms has also been studied,
including humans (Duerr et al. 2004), rodents (Randolph et al. 1999b) and birds (Elston
et al. 2001, Kirby et al. 2004). These are in addition to the numerous systems reviewed
by Crofton (1971) and Shaw and Dobson (1995).

The seminal texts on parasite aggregation are attributed to Crofton (1971) and
Anderson and May (1978). One of the central assumptions in Anderson and May's host-
parasite models (1978) is that each parasite increases the probability of host mortality. In
order to classify a species as a parasite, there are three conditions that these authors
require: use of the host as habitat, nutritional dependence on the host, and causing harm
to the host. This harm is often referred to as ‘mortality’ in the paper, but harm may be
described as any decrease in fitness, such as decrease in fecundity or immune fiinction.
Many organisms such as mites, fleas and ticks, appear to do little harm to their hosts

unless present in great numbers. Parasites range in the amount of harm they cause; some

very nearly kill their hosts and are parasitoid-like, while others act more like symbionts.
According to Anderson and May, all of these are considered parasites.

Anderson and May couch their descriptions of parasite-host interactions in a
predator-prey framework, a theory familiar to ecologists. In this light, the host is ‘prey’
to the parasite’s ‘predator’. It is ofien difficult to detect the detrimental effect that the
predator exerts on the prey since these effects may be subtle or often unobservable.
Sometimes the size of parasite and host populations is reliant on their interactions and
mimics the classic density-dependent sigmoid functions of predator-prey population
curves, and sometimes not. Although the phenomenon of parasite aggregation has been

well-documented and described, its mechanisms have not.

Parasite aggregation and tick-borne disease

Ticks are parasites. Laid in a clutch of hundreds or thousands, they begin life
aggregated. Once they emerge, they lay in wait for a host to walk by and hitch a ride,
since they cannot move very far on their own. They rely on vertebrate hosts for the
multiple bloodmeals they will take in their lives. They also rely on their hosts not
detecting and grooming them off for the duration of their meal. Once they have fed to
repletion, they drop off the host, wherever the host happens to be at the time. Since they
can walk no more than a few feet on their own (Daniels and Fish 1990), their distribution
is determined by host factors, which can include the host’s home range and availability of
good habitat for that host (Madhav et al. 2004).

In order for a microorganism to exploit a tick vector for transmission between

hosts, it is imperative that it be able to survive between life stages. That is, when a tick

molts to its next instar, a parasitic microorganism must be maintained through these
development processes. Further, in order for it to be sustained in susceptible vertebrate
populations, it must find a host in which it will survive until the next stage of the vector
will again become active. For tick-home pathogens, a multi-layered aggregation with
aggregation occurring at the vertebrate host level (ticks and microorganisms) and the tick
level (microorganisms).

Aggregation has been demonstrated for both ticks on rodent hosts as well as for
spirochetal bacteria among host-seeking ticks. For both parasites, a similar level of
aggregation has been found: 20% of the hosts harbor 80% of the parasites (Wang et al.
2003). Randolph et al. (1999b) shows that 80% of the ticks are found on only 20% of the
hosts. This pattern of 80/20 aggregation is the phenomenon of interest.

As numbers of available hosts increase, so do the numbers of parasitic ticks found
on those species. For example, Kirby et al. (2004) found that as numbers of red deer, the
favored host of the hard tick Ixodes ricinus, increase, the tick also flourishes.
Specifically, its infestations of red grouse chicks have increased in both prevalence
(number of chicks infested) and intensity (number of ticks per infested chick).

There are some disease systems for which aggregation is shown to have an
extremely beneficial, and sometimes necessary, effect for parasite maintenance in the
system. As reviewed by Randolph et al. (1996), in certain pathogen-tick-host systems,
the aggregation of feeding ticks on a host is a necessary means of propagating the spread
of the infection. These virus systems include Crimean-Congo hemorrhagic fever,
Kyansur forest disease, Louping illness, Thogoto, and tick—home encephalitis (TBE).

Ixodid (hard) ticks are responsible for two of these: Louping illness and TBE. Birds and

mammals can pose as the non-viremic hosts, showing infection only in the tissues where
the ticks are feeding, and not systemically (Randolph et al. 1996). Indeed, this clustering
of ticks allows for efficient transmission of the pathogen from infected tick to non-

systemically infected host to na'r've tick.

Potential mechanisms of aggregated parasite distributions

In the 30 years since Anderson and May (1978), the phenomenon of parasite
aggregation has been well-documented (Dye 1992, Shaw and Dobson 1995, Woolhouse
et al. 1997, Pugliese et al. 1998, , Lord et al. 1999, Elston et al. 2001, Duerr et al. 2003).
Known well to exist, the causal mechanisms of this observable fact are less understood.

There are several causes that have been hypothesized to contribute to parasite
aggregation. Heterogeneity in dispersal of parasites among hosts may be due to genetic,
physiological or behavioral differences in the hosts and may also vary in time and in
space (Shaw and Dobson 1995). Even small differences in susceptibility between hosts
can produce non-random, aggregated distributions of parasites (Anderson and May
1978). Other of these factors which might influence whether parasites become
aggregated on hosts include quality of host habitat, transmission process (for
microparasites which live inside the host) or the host’s migratory behavior (Shaw and
Dobson 1995).

Factors to be taken into account when considering pathogen transmission and
potential aggregation by a vector include vector competence, abundance and distribution,
as well as host choice, incubation period of the pathogen, and probability of transmitting

an infection (Dye 1992). Here, infection is defined as the colonization of a host organism

by a foreign species where the infecting organism, or pathogen, may interfere with the
normal functioning of the host. Transmission itself is also influenced by outside factors.
For example, in the Lyme disease system, density of feeding Ixodes scapularis larval
ticks on hosts influences transmission of Borrelia burgdorferi, the Lyme disease
pathogen. As density of the feeding larvae increases, the transmission efficacy of

B. burgdorferi increases as well, possibly because of immunosuppressive factors in the
tick’s saliva, allowing for unrestrained grth of B. burgdorferi spirochetes (Levin et al.
1997). However, higher densities of feeding larvae decrease chances of larval survival
and a smaller percentage of those ticks will survive to become nymphs, possibly because
not all ticks are able to feed to repletion, which may be due to grooming by the host
(Levin and Fish 1998).

Additional sources of aggregation may include seasonality of hosts or parasites,
aggregation of parasitic life stages, or host effects which may include behavior,
physiology or immunology (Shaw et al. 1998). It is clear to see how many of these
factors could heavily influence LD dynamics, with its strong patterns of seasonality, and
aggregation of the different life stages of its vector (Piesman and Spielman 1979, Wilson
and Spielman 1983, Levine et al. 1985).

One factor that may contribute to the abundance of the tick that vectors the Lyme
disease pathogen, I. scapularis, and hence, its aggregation, is drought (Schauber et al.
2005). Drought increases favorability of habitat for ticks, possibly by suppressing growth
of entomopathogenic soil fungus. In wet years, fewer I. scapularis survive and LD
incidence is lower the following year. Due to their lack of mobility, larval ticks will not

travel far from where they were hatched (Daniels and Fish 1990). In a year where more

of them survive, we might expect to see increased aggregation because of their higher
numbers at the larval stage. However, this may not necessarily be the case for other life
stages, which are more highly dispersed (Madhav 2004).

There may be some evidence of an age-intensity curve in regards to parasite
distribution among hosts. In some systems, it may be that peak intensity of infection
occurs towards the end of an organism’s life (Duerr et al. 2003, Cattadori 2005). Duerr et
al. (2003) have identified six factors which may contribute to this age-intensity
phenomenon- i) age-dependent exposure, ii) parasite-induced host mortality, iii)
heterogeneity within the host population, iv) clumped infection, v) density-dependent
parasite mortality and vi) density-dependent parasite establishment.

Several studies have documented the differences in infestation by parasites
between sexes. Hypothesized to be a result of immune differences, males of some taxa,
especially mammals, tend to carry the higher parasite load (in Hudson, 2002).

Once a system’s hosts and parasites have been described and characterized, age
and sex of host may give clues as to the parasite burden of an individual. If age and sex
distributions of parasites have been well-characterized, falling into certain categories of
those characteristics may indicate likelihood of a higher or lower parasite burden.
Having such a metric may be epidemiologically useful for targeting interventions that

would prevent spread or maintenance of a disease.

The study system: Lyme disease

Lyme disease (LD) is tick-borne, spirochetal disease endemic to much of the
United States, Europe and even parts of Asia (Kurtenbach et al. 2006). In the US, Lyme
disease is caused by Borrelia burgdorferi and transmitted by Ixodes scapularis; it is
characterized in humans by a distinctive erythema migrans (EM) skin lesion and systemic
symptoms. LD is classified as a zoonotic vector-borne disease, and as such, utilizes
wildlife reservoirs in addition to susceptible human and canine hosts. In 2003, LD
affected approximately 21,000 pe0ple in the US, (CDC 2005) and is the most prevalent
vector-bome disease in the US. Though not a fatal disease, it usually manifests with an
erythema mi grans (bulls’ eye) rash at the site of the bite or disseminated, arthritic joints,
flu-like symptoms, and rarely, neurological symptoms such as Bell’s palsy or 2nd or 3rd
degree heart block.

Additionally, the tick and vertebrate reservoir species have distinct roles in
propagating the disease. B. burgdorferi persists in ecosystems by alternating between the
midguts of ticks and the tissues of wildlife reservoirs. Because B. burgdorferi is not
spread person-to-person, it is a disease that relies on ecosystems to flourish by a year-
round maintenance in wildlife reservoirs and the vector tick. The pathogen is maintained
in the bloodstreams of mammals for 10 to 14 days (Hanincova et al. 2008), and thereafter
found only in skin and internal organs. Peromyscus leucopus, the major reservoir for
B. burgdorferi, are thought to be infected for life (Hanincova et al. 2008). The pathogen

is also maintained in infected ticks for their entire lifespan (VanBuskirk and Ostfeld

1995), provided that they do not feed on a host with bactericidal properties, such as the

western fence lizard (Sceloporus occidentalis) (Lane 1989) or the white-tailed deer
(Odocoileus virginianus) (Telford, Mather et al. 1988).

I. scapularis is a generalist parasite with three post-egg life stages, and a lifespan
of two years in the northern U.S. Ticks start out as uninfected larvae (Patrican 1997),
which typically feed on birds and small mammals in the northern U.S. Newly emerged
larvae have the opportunity to pick up the spirochete in their first bloodmeal from an
infected host. They then molt into nymphs, and seek their second bloodmeal. It is at this
life stage, because of their small size and potential for infectivity, nymphal ticks pose the
greatest threat for infection to humans (F alco 1988). This is also the first life stage
capable of transmitting the pathogen back to the wildlife population and completing the
transmission cycle. After nymphs obtain their blood meal and molt into adults, they
parasitize a large mammalian host, often a white-tailed deer, to feed and mate, dropping
off to lay eggs in the leaf litter of the forest (Figure 1.1).

This thesis explores whether pathogen distribution among the vector and host
populations of the LD system is aggregated, considering infection prevalence (proportion
of ticks and mice infected with B. burgdorferi) and vector load (number of ticks on
mice). This project is conducted in a Lyme disease endemic area of Michigan. It
measures the degree of aggregation among the hosts, vectors and pathogens in the
system, and the degree to which this aggregation is influenced by ecological factors. The
ecology of the system is considered and examined together with each step of the vector
life cycle (Figure 1.2) in order to form a more complete picture of the ecology of the

Lyme disease system of Menominee County.

10

 

WINTER

 

dormant
nymph

 

 

Figure 1.1. The two-year life cycle of Ixodes scapularis. Ticks start out as eggs in the
spring, and take two bloodmeals (males) or three bloodmeals (females) from different
hosts in their lifetimes. Figure modified from one produced by the American Lyme
Disease Foundation.

11

 

 

 

 

   

 

  

 

 

    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

      

 

   

 

    
 

 

 

 

  
   
 

 

 

Human/canine Lyme
disease
incidence
. 1.. Aggregation of
DenSIty of infected Aggregation of B bur dorferi in
nymphs ‘— questing ticks ' i ti ks
f ques "g C Habitat
Nymphal infection 1 4 factors
prevalence Aggregation of Aggregation of
1 ticks on mice 8. burgdorferi in
ticks on mice
Proportion of /
infected

 

 

 

Host immune
response

 

 

  

 

reservoirs \
T Aggregation of

Relative .
abundance of B- burgdorferi
in mice

reservoir and
non-reservoir
host species

 

 

 

 

 

i1

 

Figure 1.2. Schematic model of LD transmission dynamics.

The study of the LD system in Michigan is a subset of the national disease
system; the study of this subset may bring to the surface implications of importance for
the system in other geographic locations. In order to investigate the system as it operates
in an endemic area, sampling took place in Menominee Township in Menominee County,

Michigan (Figure 1.3).

12

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1.3. County map of the state of Michigan, including both the Upper and Lower
Peninsulas. Shading indicates Menominee County. Inset shows the study site
(N 45°9'46.2", W 87°42' 48.0"). Images in this thesis are presented in color.

13

Michigan ’s history of Borrelia burgdorferi and Ixodes scapularis

It has been over 30 years since LD was first discovered in Connecticut; now
people get ill with LD in many states across the country, although there still appears to be
just two major foci for the disease, one in New England and one in the Upper Midwest
(CDC 2008). The spatial and temporal trends of disease distribution in the US has
generally been a radial spread from the sites where it was first recognized, provided there
was no geographic barrier to the movement and establishment of I. scapularis and its
hosts, such as a coastline (Steere et al. 2004).

The history of Lyme disease in Michigan begins with a paper published in the
Vector Control Bulletin of the North Central States (Strand et al. 1992) documenting the
presence of B. burgdorferi-infected I. dammini (now I. scapularis) in Menominee
County. The Midwest’s first report of I. scapularis and erythema chronicum mi grans
rash in a human associated with the bite of that tick came from Wisconsin in 1970
(Jackson and DeFoliart 1970, Scrimenti 1970). Since then, additional populations of
I. scapularis infected with Borrelia burgdorferi have been found in Wisconsin (Anderson
et al. 1987), Minnesota (Drew 1988) and Indiana (Pinger et al. 1996). Menominee
County was the first county in Michigan to report an established population of
I. scapularis (Strand et al. 1992) and B. burgdorferi (Walker et al. 1994).

Thus, Menominee County was the first county in the state of Michigan to be
declared endemic, and several studies have found B. burgdorferi-positive I. scapularis
there since (Walker et al. 1994, Walker et al. 1998, Friedrich 2003, Foster 2004, Hamer
et al. 2007). Situated along a border with Wisconsin to the west, and with the shore of

Lake Michigan to its south, it may be that the geography of the county is such that the

14

populations of I. scapularis and B. burgdorferi have not had an opportunity to spread
radially. Large numbers of birds migrate along the flyway over the area, a peninsula
located between two Great Lakes. As such, the area may be used as a rest area for many
birds (Diehl et al. 2003) and it is possible that the ticks may have arrived as passengers of
these birds from the southern US.

More recently, unpublished work done by researchers at Michigan State
University has shown the development of a new endemic focus in Michigan. A
population of infected I. scapularis has been found in southwestern Michigan and
appears to be spreading north along the lakeshore (Foster 2004, Hamer et al. 2007,
Hamer 2008 (in progress)). A landscape model for Michigan (Foster 2004) based on data
from Wisconsin and Minnesota (Guerra et al. 2002) shows that much of Michigan is
suitable for both the tick and the pathogen to flourish (Figure 1.4). There is evidence of
established populations of I. scapularis, B. burgdorferi and human cases of disease in the
southwest portion of the state, and north along the Lake Michigan coast (Figure 1.4)
(Foster 2004, Hamer et al. 2007, Hamer 2008 (in progress)).

It is important to note that Menominee County is still the only county in Michigan
considered ‘endemic’ for LD. Lack of data from southwestern Michigan may be due to
constraints of legislature, physician willingness and Michigan Department of Community
Health (MDCH) resources. In order for MDCH to officially declare a county ‘endemic’,
multiple, culture-confirmed (typically from EM skin biopsy) cases of LD must be
reported. This requires physicians to biopsy patients and understandably, most

physicians are more interested in treating patients than aiding in the collection of this

15

data, which is not necessary for diagnosis. To our knowledge, no cases have been

culture-confirmed from this area (MDCH, personal communication).

f

  
 

[:1 County outline

Habitat Suitability
[”7 o - 25%

  
  

, 25-50%

- 50-75%

Kilometers

Figure 1.4. A model of habitat suitability for I. scapularis establishment (Guerra et al.
2002) applied to Michigan predicts the potential future distribution of the tick, with the
darkest areas indicating highest risk for establishment upon introduction. Courtesy Erik
Foster, 2004, in thesis.

In 1990, Lyme disease became a reportable disease in Michigan. Reliable human
case data is available electronically from the Michigan Department of Community Health
from 1992 to the present. Human case incidence from a 9-county region in the Upper
Peninsula, surrounding Menominee County summed over the 16-year period from 1992
to 2007 is approximately 67.3 cases per 100,000 population. Human case incidence from
the same period for a 6-county region of southwestern Michigan is just 5.1 cases per
100,000 population. The case rate for all the remaining counties combined is only 1.4 per

100,000 population for the same time frame (Table 1.1, Figure 1.5).

Table 1.1. Counties in regions of the state of Michigan with elevated case rates (cases
per 100,000 population) over a cumulative, 16-year period from 1992 to 2007. Data
courtesy Michigan Department of Community Health. See Figure 1.5 for map.

 

Regn Case rate Counties

 

Upper Peninsula 67.3 Alger, Baraga, Delta, Dickinson, Gogebic, Iron,
Marquette, Menominee, Ontonagon

 

 

Southwest 5.1 Allegan, Berrien, Kalamazoo, Muskegon, Ottawa,
Van Buren

 

 

 

 

l7

Case rates per 100,000 population 1992-2007

5.2?

    
  

/

 

 

 

 

 

 

 

 

Case rates by region \
1.4
- 5.1
- 67.3

 

 

 

 

 

 

 

 

Figure 1.5. A map displaying the differences in infection prevalence of human Lyme
disease cases in Michigan. Case rates per 100,000 population are additive over a 16-year
period. See Table 1.1 for names of counties included in shaded regions. Data courtesy
Michigan Department of Community Health.

In Menominee County, there exist two sympatric mouse species- Peromyscus
maniculatus gracilis (deer mouse) and Peromyscus leucopus (white-footed mouse)
(Strand et al. 1992, Friedrich 2003). The existence of these sympatric species most likely
has some impact on the disease system dynamics there. These species probably act
together as the most reservoir competent species; much like P. leucopus alone does in the
Northeastern US (LoGiudice et al. 2003). One researcher found no significant difference
in the infection prevalence of P. leucopus (34.2%) and P. maniculatus gracilis (27.7%) in
Menominee County (Friedrich 2003). Thus, for the purposes of my analysis, these two
species will be lumped together and referred to as Peromyscus spp.

There is some preliminary evidence to support the theory of aggregation of
I. scapularis among P. leucopus in Michigan (Table 1.2). Data for sites in Lower
Michigan demonstrate aggregation and, without detailed statistical analysis, appear to fit
the negative binomial distribution. At Van Buren State Park, in 2004, of 67 white-footed
mice trapped, 10% had no I. scapularis infestation, but the range of infestation included
up to 52 ticks on one individual (Figure 1.6). However, at Duck Lake that same year,
there were 178 P. leucopus trapped, with over 80% free of I. scapularis, and the upper
limit of tick load only 4 ticks (Hamer, 2006). The data show that that the populations
with more individuals infested also show a higher upper limit for tick burden (Table 1.2),

perhaps indicating a positive relationship between infection prevalence and intensity.

19

Table 1.2. I. scapularis burdens on P. leucopus collected in 2004 and 2005 in
southwestern Michigan. These data are supportive of the aggregation theory of ticks on
small mammal hosts. Unpublished data courtesy Sarah Hamer, 2006.

Y # P. %
2004 67 1

2005 15
2004 >80%
2005 88

 

   
  

 

 

 

12-
5:, Nymph
I Larva
g 8 _ E] No I. scapularis
“6
3
.Q
S
2 4-
0' . ,! . , 4

I — I . , ‘ . . . _
0 8 1 6 24 32 40 48
Num ber of I. scapularis
Figure 1.6. Ixodes scapularis on Peromyscus leucopus trapped at Van Buren State Park

2004. Solid line shows the curve approximating the NBD. Unpublished data courtesy
Sarah Hamer, 2006.

,

20

Aggregation & Lyme disease

There are several levels on which aggregation may occur in the LD system. First,
there is the distribution of ticks on small mammal (mouse) hosts. Second, the distribution
of B. burgdorferi in I. scapularis hosts. Third is the grouping of B. burgdorferi in the

mouse host (Figure 1.7).

 

Figure 1.7. A visual representation of aggregation on multiple levels. In a host
population (mice, level 1), aggregation of ectoparasites (ticks, level 2), occurs. Further,
aggregation of endoparasites (spirochetes, level 3) occurs within the tick population.

21

There are several ecological aspects of the LD system that may affect the spread
of the pathogen. The aspects of the disease system that this inquiry focuses on are the
distribution of pathogen load in the I. scapularis nymphs and adults, the distribution of
pathogen load in and I. scapularis ticks on Peromyscus spp. The maintenance cycle, and
by extension, the gradual spread of LD, requires that infected nymphs transmit the
pathogen to reservoir hosts and larval ticks to become infected from reservoir hosts.

The probability of a host becoming infected is a positive function of the
proportion of ticks that are infectious. Aggregation of the pathogen in hosts may
influence the proportion of ticks that are infections by infecting many of the naive larval
ticks that feed on them, which then molt into infected nymphs. The highly infected ticks
are able to deliver a high dose of pathogen to their next host. In this way, the cycle of the
pathogen moving from host to tick and to a new host is the basis for the spread of the LD
pathogen, and aggregation may play a role in this spread.

There is a high degree of aggregation of B. burgdorferi load among field collected
host-seeking nymphs in LD endemic woods of the Northeastern US (Wang et al. 2003),
and additional laboratory studies (Levin and Fish 2000) have demonstrated that the lower
20% of ticks positive for B. burgdorferi may not be capable of transmitting the disease
because their spirochete loads are too low to transmit an infectious dose. At a field site in
southern Connecticut, where the nymphal infection prevalence is ~ 32% and nymphal
densities are high, nearly 100% of mice become infected by summer’s end (Bunikis et al.
2004). There, pathogen aggregation in nymphs may not have a large ecological impact
on host infection dynamics, unless high pathogen loads in hosts allow for a more efficient

transmission back to ticks. If almost all the mouse hosts are infected, a tick’s chances of

22

becoming infected by feeding on any individual mouse in that population is high.
However, recent figures from Menominee County indicated infection prevalence for
Peromyscus spp. from 28-34% (Friedrich 2003).

It should be noted that in the Lyme disease system, ticks play multiple roles,
acting as a parasite, host and vector species. In the system in general, I. scapularis ticks
act as a vector of a parasite (B. burgdorferi). It is important to consider that the ticks are
simultaneously parasitized by B. burgdorferi as they parasitize a host for their blood
meal.

Intensity of infection or load of B. burgdorferi spirochetes in both the tick (Wang
et al. 2003) and host (Tsao, unpublished data, personal communication) has been
previously considered. The force of infection from tick to mammal and back most likely
plays an important role in this infection intensity, where force of infection is defined as
the number of new infections per number of susceptible hosts exposed per unit time.

For an endemic parasite, the force of infection can be calculated using age-
prevalence data (Anderson and May 1991). In some systems, it may be that peak
intensity of infection occurs towards the end of an organism’s life (Duerr et al. 2003,
Cattadori 2005). I hypothesize that it might be the casein the Lyme disease system that
mouse host populations demonstrate higher infection intensities later in life, because they
have had more opportunities to become infected, or their immune systems begin to fail as
they age.

Further, there may be evidence that multiple pathogen strains may aggregate in
hosts (Lord et al. 1999). For LD in particular, there is some evidence that there will be

more strains present in an endemic area, and fewer in a newly invaded area, such as

23

Lower Michigan. There is evidence that genotypic variation has some impact on the
pathogenesis of disease (Wang et a1. 2001) and hence, its impact on humans.

It has been shown in a laboratory experiment that approximately 20% of infected
nymphs are not transmitting B. burgdorferi to hosts (Levin and Fish 2000). I hypothesize
that this could be a parasite factor; there could be a threshold number of parasites in a
host that signals the parasite to move out of that host. It is possible that this 20% has
spirochete burdens too low to have successful colonization and therefore transmission.

Combining Woolhouse et al.’s hypothesis (1997) that only 20% of the hosts are
responsible for 80% of the infections with the estimated transmission data from the
aforementioned laboratory experiment (Levin and Fish 2000), it is possible that of the
infected ticks, the lower 20% are not contributing to LD spread, the middle 60% are
responsible for 20% of the incidence, and the remaining 20% of ticks with the highest
parasite load are contributing to 80% of the LD incident infections (Figure 1.8). Thus,
the proportion of ticks that are infectious is a subset of all infected ticks. The probability
of a host becoming infected may be a positive function of the number of parasitizin g
ficks

The intensity of infection, or the degree to which an organism is infected, is the
aspect of interest in regard to the ticks’ infection with B. burgdorferi. Intensity of
infection refers to the number of pathogenic organisms found inside a host. Force of
infection refers to the rate at which susceptible individuals become infected by a
pathogen, and can be used to compare the rate of transmission between different groups
of the population for the same disease, or even between different diseases. The force of

infection from tick to mammal species and then back to tick may also play an important

’24

Hypothesis of Porosite
Infection Ability

 

 

 

 

 

Figure 1.8. A visual representation of the hypothesis that the upper 20% of parasite
loaded ticks harbor 80% of the spirochetes (false data set). Images in this thesis are
presented in color.

25

role in regards to infection intensity. The intensity of the infection can affect
transmission efficiency in both directions. If a tick or host has a low number of
spirochetes, (and thus intensity of infection is low) transmission will not be efficient since
the few spirochetes transmitted may not be able to survive and replicate in the new host.

Studies have shown that the transmission of B. burgdorferi to the tick may be a
function of the density of feeding ticks on the host. In a laboratory study, density of
feeding larvae on infected mouse hosts was compared with the success of the acquisition
of the spirochete in the feeding larvae (Levin et al. 1997). At low densities (25 larvae per
mouse), up to 90% (23 ticks per mouse) of larvae successfully fed and molted; of those,
an average of 17% were infected. At high densities (250 larvae per mouse), only 16%
(40 ticks per mouse) of ticks successfully fed and molted, but on average, 32% were
infected. On the hosts with higher densities of feeding larvae, although a lower
percentage of ticks successfiilly molted, raw numbers of survivors were still higher, and
percentage of infected ticks was double that seen in the larvae fed at low densities.

When ticks are aggregated on hosts, their survivorship may be reduced, but the
overall number of infected ticks is still higher than if the ticks were spread out on hosts at
low densities. Thus, aggregation may help play a role in the maintenance of the LD
cycle. Another study that also demonstrated lower feeding success by I. scapularis with
increased densities of feeding larvae also showed that aggregation of the hosts reduced
density of feeding ticks since grooming behaviors increased when mice went from
solitary to co-nesting (Levin and Fish 1998). Although survivorship for I. scapularis is
less favorable under crowded conditions, it is more favorable for B. burgdorferi, since co-

feeding increases the chances of spirochete transmission to other vectors.

26

Describing Aggregation
How I. scapularis ticks become infected and B. burgdorferi are maintained among ticks

In the two year life cycle of I. scapularis, the tick will take three bloodmeals. The
first occurs in the summer of first year of life, as a larva, during which it can acquire the
B. burgdorferi pathogen, and after which, it molts into a nymph and over winters until the
following spring. That spring, it takes its nymphal bloodmeal, which is another
opportunity for it to acquire B. burgdorferi, and then molts into an adult, at which time
the females take their final bloodmeal that fall or winter. The way the tick finds its host
is slightly different each time.

As a larva, the tick seeks a host from the spot it hatched out of its egg (where its
egg and all its siblings’ eggs were laid), and usually seeks a small rodent or bird host to
feed on. However, I. scapularis are generalist feeders, and will opportunistically feed on
whatever host it can.

As a nymph, the tick is questing close to the spot where it had dropped off its first
host as an engorged larvae. It is also looking for a small—sized host to feed on. Finally,
as an adult, it seeks out its bloodmeal from the spot it dropped off its nymphal meal host,
but this time, often ends up on a large mammal (deer, bear, human, dog) to feed. Even
though it is possible that the tick might not become infected until its adult stage, at that
point, it is a dead-end host since it takes no further blood meals. It is also important to
note that transovarial transmission is low and therefore the amount of infected larvae is
negligible (Patrican 1997).

Although the tick has two chances to become infected and then pass on that

infection, there are still many ticks that do not acquire the B. burgdorferi spirochete in

27

either of their first two bloodmeals. Because there are so many uninfecteds, or ticks with
a zero spirochete burden, there are specific distributions that are used to describe the
distribution of spirochetes across tick or host populations.

Another way to think about this is the probability that a host becomes infected
with B. burgdorferi. That probability is a function of the probability of contact with a
tick, times the prevalence of infection (estimated as prevalence of B. burgdorferi in the
tick population), times the probability to transmit the parasite, which itself is a function of

spirochete load.

The negative binomial

The negative binomial distribution (NBD) has been in use for some time as a
descriptor of biological data (Bliss and Fisher 1953). In a review of 49 published
wildlife/parasite systems, Shaw et al. (1998) demonstrated that 48 of these 49 systems are
well-described by the NBD. Since then, more host/parasite systems that display strong
aggregation employ the NBD in order to statistically describe the system (Rosa and
Pugliese 2002, Wang et al. 2003). The NBD is a discrete probability distribution and
generalization of the Poisson that describes greater variance (Elliott 1977, in Guyatt et al.
1994)

The NBD of a population is described by the ‘k’ parameter, which can be

. 2 . 2 .
approxrmated by the formula k—hat = x2/ (5 — x), where x15 the sample mean and 5 IS

the sample variance. k is a value between 0 and infinity, and the smaller k gets, the more
highly aggregated the population. In a completely aggregated system, k = 0, and all the

parasites are on one host. Generally, when k S 5, the distribution is best approximated by

28

the NBD (Anderson and May 1978, Hudson et al. 2002). The ‘k’ parameter describes the
number of failures encountered before finding a success. When an outcome for the NBD
is termed a failure, it refers to it as a ‘0’- since the possibilities of outcomes in the
binomial distribution are limited to ‘0’ or ‘1’. Using the NBD to describe infection in
these host/parasite systems, the infected hosts are scored a ‘ l ’, ascribed the characteristic
‘success’, and the uninfected hosts scored ‘0’, or ‘failures’.

In these host/parasite systems, the ‘successes’ and ‘failures’ or ‘infecteds’ and
‘uninfecteds’, will lie on either side of some determined cut point. The cut point it
determined as a parasite burden above which an infection may be successfully
transmitted, and below which it cannot. Hosts with a number of parasites below that cut
point have an intensity of infection too low to transmit the pathogen successfully, and
therefore, those hosts are considered uninfected.

This distribution is useful when there are binomial outcomes, and the probability
of success can be described as a proportion, it. The ‘k’ parameter is used to describe the
degree of dispersion, with an inverse relationship to degree of aggregation (Paterson and
Lello 2003, Pal and Lewis 2004). As the parameter ‘k’ decreases, it indicates a higher
degree of aggregation (Hudson 2002). However, as the ‘k’ parameter approaches
infinity, the negative binomial converges with the Poisson (Elliott 1977, in Guyatt et al.
1994, Hudson et al.). Rosa and Pugliese (2002) point out that k is not a parameter

corresponding to a particular biological process, but instead, a population statistic.

29

Variance to mean ratio

Variance to mean ratios can be used to determine if aggregation occurs, where
values greater than one indicate aggregation (Lord et al. 1999, Elston et al. 2001).
Variance is generally assumed to be greater between groups than within groups (Elston et
al. 2001). However, when population sizes are small, the mean tends to be
underestimated. This is likely due to incomplete sampling of the population and thus
missing the individuals with the highest parasite loads (Gregory and Woolhouse 1993).
When the variance is equal to the mean, the distribution fits the random, or Poisson

distribution.

Mixed models
The generalized linear model (GLM) is also useful for describing distributions
found in nature. This is because mixed models incorporate an error term into the GLM to

account for random effects (Paterson and Lello 2003). The general model form is:

y,- = xiii + 8,, where there is a slope parameter ([3) and an error term (8). The standard

GLM will have variance equal to the mean, with Poisson distributed errors.

Here, the mixed models will be referred to as GLMM (generalized linear mixed
model). The GLMM can be used to analyze parasitological data that does not conform to
a normal distribution, and also controls for correlations between data points that arise
from grouped observations. A GLMM regression equation takes on this general form:

y,=X,—B+Z,b,- +e,-, where the newly introduced term, Z, is the random component of the

model. The properties of this model are as follows: var(yi)=E(u,) + var(ui). Here the

30

variance is far greater than the mean (Paterson and Lello 2003). The analogous GLMM
with Poisson-distributed errors is described by var(y,)=E(ui).

These GLMM have commonly been used when describing macroparasite
distributions among hosts, and employ a restricted maximum likelihood (REML) method
of analysis (Kirby et al. 2004). Wilson et al. (1996) argue that the negative binomial and
the GLM may not be suffi cient and that using log-transformed data for aggregation
analysis may sometimes be the better option. This is because the aggregation data often

have a large number of zeroes, and log-transforming changes the zero values.

Lloyd ’s mean crowding index
Lloyd’s mean crowding (LMC) index is the variance to mean ratio expressed

from the point of view of the parasite. The general form of the index is:

2 . . . .
m* = m + (s /m - 1), where m* IS the index parameter, m IS the sample mean, and s is

the sample variance (Lloyd 1967). If the distribution of parasites among hosts is normal,
the variance to mean ratio equals 1, and the index (m*) will be equal to the mean,
indicating that on average, as many parasitic organisms are experiencing crowding as
those not experiencing crowding. If the distribution is over-dispersed, then the variance
to mean ratio will be greater than 1, and hence, the crowding index will be greater than 1,
indicating that more parasitic organisms than not are experiencing crowding.

LMC measures the number of other parasites experienced by the average parasite
in the sample population. When aggregation occurs, and most of the parasites are

crowded in or on only a few hosts and only a few parasites are not crowded in or on their

31

host. The number of the index cannot be directly translated into an interpretation of the

number of other parasites a parasite is experiencing.

Statistically modeled systems

Statistical models provide powerful tools for describing how and why biological
processes influence parasite infection and aggregation, including exposure of hosts, and
virulence and movement of parasites. A deterministic model proposed by Crofton (1971)
does not take into account stochastic variables inherent in the study system. One basic
characteristic of these deterrninistically modeled systems is the assumption that the
parasite reproduces at a much higher rate than the host (Crofton 1971).

According to Paterson and Lello (2003 ), the problem with some statistical
parasitological models is pseudoreplication, whereupon spatial or temporal correlations
make observations non-independent. That is an issue for this study in this system, where
repeated sampling of the same mammals caught at different times forgoes assumptions of
independent capture events. Rosa and Pugliese (2002) argue that models that address
aggregation have thus far only focused on parasite abundance, but that focus should
perhaps shift to include factors affecting parasite dynamics and evolution. This study
attempts to address some of these concerns by taking into account host spatial structure
and immune response.

Other researchers have already attempted to model the Lyme disease system in a
process-based, rather than statistical way (VanBuskirk and Ostfeld 1998). Some of these
models even take into account the that there is not a constant species-specific rate of

infectivity, recently the issue of reservoir competence, or the degree to which a host. is

32

capable of harboring and propagating infection in its body and is subsequently able to
infect a feeding vector has been introduced (Schauber and Ostfeld 2002). The specific
infectivity of a host can be represented by multiplying the proportion infected (1) by the
reservoir competence of that species (Mather 1993, Schauber and Ostfeld 2002).
Certainly, in a disease system involving B. burgdorferi, we do not see any self-cure
immunity in any of the species considered reservoir hosts or disease-susceptible hosts.
Studies have documented the transmission dynamics for other disease systems
(Woolhouse et al. 1998, Randolph 1999a), or to create a predictive model of disease
transmission or disease risk, such as that for Babesia microti (Mather et al. 1996). A
study by Pugliese et al. (1998) developed a mechanistic epidemiological model for

macroparasite aggregation.

Research Questions

Through the study of the ecology of many vector-borne and zoonotic disease
systems, the scientific community has illuminated some of the events contributing to the
spread of these diseases. For instance, disease ecologists have linked the spread of West
Nile virus across the US to bird migration patterns (Reed et al. 2003, Owen et al. 2006).

By understanding the mechanisms for the transmission and maintenance of
vector-borne and zoonotic diseases, we have the opportunity to develop preventive
measures against these diseases. For example, if it is known that bird migration is

contributing to virus transmission to local birds, humans or horses via mosquitoes, then it

33

is prudent to control mosquito populations is high risk areas during high risk times of the
year.

My overall hypothesis is that the infected I. scapularis questing nymphs in
Menominee, Michigan have over-dispersed loads of B. burgdorferi. Further, adult life
stages of ticks will show the greatest spirochete loads, and adult reservoir-competent
hosts will show higher spirochete and tick loads than juveniles. The LD transmission
cycle in Menominee County may be influenced by changes in infection prevalence of
hosts, spirochete load, spirochete distribution in I. scapularis ticks and small mammal
hosts, and that the age-structure of those tick and reservoir host populations impact
spirochete loads and distributions in these same populations.

The null hypothesis is that a host population demonstrates a random distribution
of parasite burden. An even distribution would not be a realistic model for parasite
distribution, as it would assume that every host in the system has the same number of
parasites. Realistically, we know that some hosts will never become infected, and that
‘0’ burden cannot be accounted for in an even distribution, unless all hosts had ‘0’
parasite burden, which would nullify the disease system. Characteristics that distinguish
an aggregated distribution from a random one include a variance to mean ratio greater
than 1, the estimate of the NBD parameter k that is low, prevalence of infection and mean
parasite burden, dispersion pattern of parasites among hosts, as well as patterns in
aggregation, as defined by Shaw and Dobson (1995).

Through this research I have posed the following overall questions:

0 Is aggregation occurring in the parasite populations of Menominee

County’s endemic Lyme disease system?

34

o If so, do habitat factors contribute to this aggregation?
o Are any of these factors spatially correlated, and what are the possible
explanations for this spatial correlation?
These broader questions can be broken down into more focused, specific questions
that ask about the nature of aggregation in the system, and then the contributing factors.
Description of aggregation in the system
0 Does aggregation of parasites occur in this system and to what degree?
0 Does aggregation occur at the pathogen level, the tick level, or both?
0 Is there aggregation of ticks that are off-host?
o Is there aggregation of ticks that are on-host?
o Is there aggregation of hosts?
0 Are aggregation patterns different among different life stages of ticks?
Processes that contribute to aggregation in the system
0 What measured factors might have led to aggregation on any level?
0 What possible influence might these factors have?

I addressed these questions by performing, at three trapping grids, three times
each over the summer transmission period, a vegetation survey, a survey of questing
ticks, and a survey of Peromyscus spp mice. This study quantifies the distribution of
B. burgdorferi in larval, nymphal and adult I. scapularis ticks and mammal hosts as well
as the distribution of ticks on their small mammal hosts. The study determines the
proportion of ticks and hosts are carrying spirochetes and what proportion of hosts are
carrying ticks. Distributions of the spirochete, tick and host populations were examined

on the landscape. Implications of aggregation on the LD cycle are discussed.

35

CHAPTER 2

METHODS

Field methods
Study site, sampling design and field schedule

The study site is in southwestern Menominee County, in Michigan’s Upper
Peninsula. In this area, abundant Borrelia burgdorferi-infected tick populations exist
(Strand et al., 1992). Menominee County was chosen as the study area for this reason;
the particular site (N 45°9'46.2", W 87°42' 48.0") was chosen based on the oak-
dominated forest and sandy soil, 1. scapularis’ preferred habitat (Guerra et al., Figure
1.4). Figure 1.6 shows the study area where three, 1.15 ha grids were sampled for
mammal, tick and B. burgdorferi populations.

The three grids were situated 12 m apart. Three grids were chosen in order to
have replication for the area. The grids were placed according to where they would fit on
the chosen piece of private property. The grids were set up with an 8 by 10 array of trap
points spaced 12 m apart with an overall trapping area of 1.15 ha (Figure 2.1). The grids
were set to be this size to maximize mouse capture by using all available traps. Each
point had two Sherman traps spaced approximately 0.5 m apart. The traps were placed as
close to the trap point as possible while keeping them in areas where small mammals

were likely to be (i.e. — not on bare, uncanopied ground).

36

 

 

 

 

 

 

 

 

 

 

Figure 2.1. Eight by ten trapping grid with 12 m between trap lines (total 1.15 ha). Grey
rectangles at grid vertices each represent 2 Sherman long-folding traps used to minimize
the effect of trap saturation.

The fieldwork took place from the end of May to mid-August 2006 because this is
thought to be the time of year when both the ticks and their hosts are most active in this
area of Northern Michigan (Walker et al. 1994, Friedrich 2003). During this time, small
mammal traps were employed to trap rodents, questing ticks were collected by dragging,
and vegetation at each trap point was characterized. The procedure for each grid
(replicated per each of three grids, for three 12-night cycles) was as follows, with each of
the following tasks accomplished. The 160 traps were set and checked for 4 continuous

nights, for a total of 5760 trap nights. Whole grid dragging and fine-scale dragging

occurred on one dry day per trap period so as to maximize ability to capture questing

37

ticks. Vegetation sampling occurred once per grid for the whole season. Daily tasks are

outlined in detail in Table 2.1. Details for each of these tasks are explained below.

Table 2.1. Daily field tasks representing one grid’s schedule. This cycle of tasks
occurred once per trapping grid for three grids. The entire cycle of three grids was
conducted four times over the course of the season.

 

 

 

 

 

 

 

Task 1 Task 2 Task 3 Task 4
Day 1 Set up flags and two Drag for Sample Set traps open in the
Sherman traps at each ticks vegetation evening
trap point
Day 2 Traps checked Process Finish vegetation Set traps open in the
animals sample/ tick drag evening
Day 3 Traps checked Process Finish vegetation Set traps open in the
animals sample/ tick drag evening
Day 4 Traps checked Process Finish vegetation Set traps open in the
animals sample/ tick drag evening
Day 5 Traps checked Process Remove traps Begin Day 1, task 1
animals from grid on new grid

 

 

 

 

 

Vegetation sampling

Vegetation was surveyed on each grid because it is thought that habitat has an
influence on where organisms move and prefer to live, and vegetation is an important
part of habitat. In this case, we sampled both understory and trees since those can both
influence animals’ habits. Vegetation sampling occurred only once at each of the grids.
Vegetation data were collected at each of the 80 locations corresponding to the trap
points. Northing and the Easting for each trap point were recorded. Understory was
characterized for the 12 m square (24 m2) surrounding each trap (Figure 2.2), and
recorded as per the code in Table 2.2. Understory was considered to be any non-tree

vegetation, such as shrubs and herbaceous growth. From the flag, the genera of the three

38

 

closest mature trees were recorded, defined as having at least a 10” diameter at breast

height.

6m 6m
A

 

 

 

 

 

K

Figure 2.2. Scheme for vegetation sampling. The grey boxes represent the Sherman
traps present at each point where Easting and Northing coordinates were taken. The
12m2 box around it represents the area around each point considered for vegetation
sampling. Dotted lines represent the trap lines on the grid. Images in this thesis are
presented in color.

 

 

Table 2.2. Categories used for understory classification. Any non-tree vegetation was
considered understory.

No bare

Moderate 34% to 66%
67% 100%

 

39

Questing tick collection

Questing ticks were collected using a standard 1 m2 white corduroy cloth dragged
over the forest floor (F alco and Fish 1992), along predetermined transects (Figure 2.3).
These transects ran between the trap lines which had been previously set. We sampled
for questing ticks using two dragging schemes - ‘whole grid' and 'intensive sampling'.
We did this to later calculate if any ticks were missed with the coarser-scale sweep of the
area. Stops were made every 24 m for whole-grid transect dragging, and every 8 m for
intensive-sample dragging (see below for more details). A common distance to drag
between stops is 20 m (Diuk-Wasser et al. 2006), but we used 24 m since it lent itself to
easy counting with the 12 m spacing of traps. At each stop, if ticks were observed on
the cloth, they were removed and placed in labeled vials containing 70% ethanol. These
were stored at room temperature until identification could take place. Ticks of all life
stages were collected.

Dragging was completed following the arrows in Figure 2.3. For 'whole grid'
sample, dragging began 6m before the first flag, and ended 6 m past the last flag
indicated in Figure 2.3, and thus the total length of the transect was 120 m. The drag
cloth was checked at the place indicated in the figure by a flag, i.e., every 24 m. The
dragging transects ran between the trap transects.

At each grid, one of the little squares was randomly chosen for the 'intensive-
sample during each rotation (Figure 2.4). These intensive sample surveys were
conducted before the whole grid survey. For this survey, the cloth was checked every 8

m. In cases where there were large numbers of larvae on the cloth, they were lifted off

40

the cloth with clear packing tape and the tape was subsequently stuck to an acetate

transparency for later identification.

 

 

 

 

 

 

 

 

 

 

Figure 2.3. Scheme for drag-collecting ticks from grids. As indicated, dragging
transects were conducted between trapping lines. At each stop point, indicated by the
stop signs every 24 meters, drag cloths were lifted and checked on both sides for ticks.
Images in this thesis presented in color.

41

 

 

12m<
24m

 

 

 

V
12m

 

 

Figure 2.4. Intensive drag—sampling scheme for questing ticks, for one of the 20 above-
labeled 24 m2 grids (Fig. 2.3), chosen at random for each grid, at each time step. Arrows
indicate transects dragged between trapping lines and stop signs indicate where drag
cloths were checked for ticks. Images in this thesis presented in color.

Due to the nature of this preservation, these ticks could not be assayed for the presence of
B. burgdorferi. However, questing larvae are not likely to be infected since transovarial
transmission is highly inefficient (Piesman et al. 1986, Patrican 1997).

For the purpose of comparing tick densities across habitat types, I also sampled
another study site in the county for host—seeking ticks. This site is found in the Escanaba
River State Forest is on the eastern side of Menominee County. This habitat was

characterized by cedar swamp and a high water table. Dragging for ticks at this site

42

occurred once every two weeks during the season. 500 m were dragged each time and
used for comparison with the southwestern Menominee County site.
Small mammal trapping and processing

Trapping effort was limited to one grid per day; traps were set between 17:00 and
19:00 in the evening and checked each day at 7:00. Traps were baited with a mixture of
crimped oats and raw peanuts. Upon checking, if a trap was found to have an animal, it
was brought back to the processing station. Empty traps were closed at that time.
Processing was done while the animal was alert and not sedated. Data were collected as
follows: trap number and location of capture, recapture status, mass, age, sex,
reproductive status, species, ear length measurement (for Peromyscus spp. only) and
number of larval and nymphal ticks feeding. Ticks were removed and placed in labeled
vials containing 70% ethanol. Approximately 2 mm punch biopsy of ear skin was taken
and similarly preserved. Each animal was given a uniquely numbered ear tag (Monel
1005-1, National Band and Tag, Newport, KY). To obtain a blood sample, each mouse
was placed head first in a 50 ml conical tube with breathing holes drilled into it. Blood
was obtained by cutting off the tip of the tail with a sterile scalpel. Blood was collected
in Microvette tubes (Sarstedt, Numbrecht Rommelsdorf, Germany) and preserved on ice
until they could be centrifuged to separate serum from cells that afternoon. The bleeding
was stopped by the use of styptic powder and application of pressure.

After blood collection, mice were observed for 10-15 minutes. If the mice were
alert and healthy, they were released at the site of capture. Mice that appeared cold or

were still bleeding were treated by being placed on an activated charcoal hand warmer or

43

had more styptic powder applied to their tails, respectively. Mammal mortality due to
handling was low, for a total of three deaths of the 757 captures (0.4%).

Animals were treated in accordance with the approved Michigan State University
All University Committee on Animal Use and Care protocol (12/03-1 52-00). All
workers in the field were outfitted with the proper personal protective equipment. These
items consisted of a white cotton tick suit, latex gloves and a respirator for small mammal
handling. All workers were certified by the Michigan State University Office of
Radiation, Chemical and Biological Safety on their N95 respirators prior to animal

handling.

44

Laboratory methods
Tick identification

All ticks collected from the vegetation (questing) and off hosts (feeding) were
identified by life stage and species. To identify, ticks were placed in individual small
Petri dish under a dissection scope, and using a dichotomous key (Sonenshine 1979),
were determined to be larva, nymph, adult female or adult male. Upon identification,
nymphal and adult ticks were individually placed into new, DNAse/RNAse-free 1.5 ml
microcentrifuge tubes for storage at -80°C until DNA extraction could be completed.
Larvae were pooled by host or by drag location and date, due to their small size and low
probability of carrying the B. burgdorferi spirochete. Each pool contained between 1 and
100 larvae.

Each new tube was given a unique identifier, a number followed by a letter to
indicate species and life stage. ‘A’ coded for I. scapularis larvae, ‘B’ for I. scapularis
nymphs, ‘C’ for I. scapularis adult females and ‘D’ for I. scapularis adult males. For
example, 45B indicated the 45th tick identified in the series, and that it was an I.
scapularis nymph. A master spreadsheet correlated each tick back to the individual

mammal off of which it had been collected and capture date.

DNA extraction

DNA was extracted using DNeasy kits (Qiagen, Valencia, CA). Preparation of
the samples involved removing each tick from its tube, placing it on a Kimwipe
(Kimberly-Clark, Neenah, WI) inside a fume hood and allowing the ethanol to evaporate.

Ticks were then placed individually in a new, DNAse/RNAse-free 1.5 ml

45

microcentrifuge tubes, frozen in liquid nitrogen, and crushed with a sterile pestle in order
to crush the exoskeleton and allow the contents of the midgut to be exposed to the first
buffer solution in the kit.

The following modifications were made to the Qiagen protocol: ticks were lysed
overnight in lysis buffer, AE (elution) buffer was heated to 70°C before adding to the spin
column for final elution and each sample was eluted into a 50 pl final volume. Samples
were stored at -80°C until PCR could be performed. Each batch was extracted with both

a positive (B. burgdorferi culture) (Table 2.3) and negative (sterile water) control.

Table 2.3. Table of sources of the cultures used to create the standard curves for qPCR.
The standard curves were used to determine the number of organisms present in a
sample. All cultures were grown from adult 1. scapularis collected from the field in April
2006. Listed is the location where that tick was collected, its assigned code, the resulting
standard curves, the number of spirochetes at each point in the curve and the strain of
each culture. All cultures grown and standardized by Sarah Hamer.

 

 

 

 

 

 

 

 

 

Location Code Curves created Number of spirochetes Strain
collected
Castle Rock c131 A, B, C 227 * 104, 103, 102, 101, 100, 10’I C88
State Park, IL
Tippecanoe c017 D, E, F 2.82 * 104, 103, 102, 10‘, 10°, 10" 297
State Park, IN
Governor Dodge c336 G, H, 1 2,79 * 104, 103, 102, 101, 100, 10’1 156b
State Park, WI

 

Checking positive controls for DNA and quantitative reliability

Each tick sample was to be assayed for B. burgdorferi using a quantitative PCR.
For each run of this assay, standard curves of known quantities of B. burgdorferi
spirochetes were needed. To generate these standard curves, B. burgdorferi spirochetes

were first cultured and counted using a Petrof-Hauser chamber, by S. Hamer (Table 2.3).

46

 

Nine 5-point dilution series were created, each containing between 2 and 3 organisms
times the following orders of magnitude: 104, 103, 102, 101 and 10°. These cultures were
then extracted using the aforementioned Qiagen DNeasy kit and modifications. These
nine series were then tested for reproducibility using the qPCR protocol and found to

perform reliably (Figure 2.5).

1.000
1.000
1.000 E+l ,
i

1.000 5

1.000

 

o 5 IO 15 20 25 30 35 40
Circle

Figure 2.5. The amplification plot for the standard curve trial for B. burgdorferi for five
dilutions ranging from 10'1 to 103. (SDS 2.3). Nine samples for each dilution had been
prepared, and the points for each order of magnitude clump together on the graph,
showing their close Ct values and thus tight reproducibility. Ct value refers to the cycle
number at which the PCR product with the labeled probe was detectable above a preset
threshold.

47

Quantitative (qPCR) on extracted tick DNA for B. burgdorferi

Quantitative PCR (qPCR) is a method for detecting the exact number of copies of
a piece of DNA in a sample. By amplifying a portion of the genome for which there is
only one copy, the resultant copy number can reveal the number of B. burgdorferi
organisms in the sample (Wang et al. 2003). The amount in the sample can then be
extrapolated to the whole sample (tissue, tick) by multiplying that number by the volume
of the whole DNA extraction. For example, if one used 1 ul of a 50 ul sample, you
would multiply the final number by 50 to estimate the number of copies of the DNA in
the entire sample.

There are several other less-precise methods which can detect the spirochete’s
DNA, although not the copy number; those published in the literature, such as molecular
typing of B. burgdorferi PCR-RF LP (polymerase chain reaction restriction fragment
length polymorphism) analysis (Liveris et al. 1995). These methods also used a standard
dilution; the products were run out on a gel, and then the samples were compared to that
dilution series by comparing band intensity by eye. These methods do not have the
power to provide an estimate for the spirochete load in the sample with the same
precision as qPCR.

DNA from ticks was checked for the presence of the 16s intergenic spacer region
in B. burgdorferi, of which there is only one copy, via quantitative real-time PCR at
Michigan State University’s Research Technology Support Facility’s Genomics C ore.
The primer used was chosen from Tsao et al. 2004 (16s forward, 23 bases: 5’-GCT GTA
AAC GAT GCA CAC TTG GT- 3’ and 165 reverse, 22 bases: 5’ —GGC GGC ACA CTT

AAC ACG TTA G -3’). The process used a TaqMan major groove binding (MGB)

48

protein and the 6FAM fluorescent label, known as the probe (22 bases: 5’- 6FAM TTC
GGT ACT AAC TTT TAG TTA A MGB- 3’). This probe is employed in order to
quantify the PCR product without gel electrophoresis (Morrison et al. 1999).

All reactions were run on the Applied Biosystems (ABI) 7900. Master mix for 15
ul reaction was as follows: 7.5 pl 2x ABI TaqMan PCR master mix, 1 ul probe at 200
nm/L, 1 it] each forward and reverse 16s primers at 900 nm/L, 1.5 ul PCR-grade water,

and 3ul template DNA. Reactions were run in 96-well PCR plates. Each sample was run

in triplicate, along with a serial dilution of 100 - 105 B. burgdorferi spirochetes to create a

standard curve, positive and negative extraction controls and positive and negative PCR
controls (Table 2.3). Only one positive control (serial dilution) and set of three negative
controls (water substituted for DNA) were run per day for all the plates. The number of
plates run per day varied from one to four.

Each run was recorded on ABI’s software (SDS 2.3). The data were derived as a
Ct value, or the cycle number at which the amplicon crosses the threshold of detectibility.
This value was used to determine the number of spirochetes per sample. The three points
of each sample were averaged and that value was placed on a standard curve created from
the serial dilutions (Figure 2.6). The number of spirochetes in the sample was derived
from that particular day’s standard curves. That number was multiplied by 50 to
calculate the number of spirochetes in the tick, because only 1 ul of the 50 ul of extracted
DNA were used.

If the numbers for the replicates were not close in value (meaning there was a
difference of more than :1: 1.0) or if one or two of the values were recorded by the robot as

undetermined, the plot of the amplification curve was examined to determine if the value

49

was real or an artifact. Real amplification manifests as a smooth exponential curve,
whereas contamination and other interference will result in broken, jagged, or straight
lines. Samples determined to not show real amplification were recorded as zero
spirochetes. Samples where all three points came up as undetermined were also
determined to contain zero spirochetes. Undetermined samples were those that the
machine was not able to assess a Ct value on. If there appeared to be a discrepancy in the
data, and one of the replicates appeared to amplify and the other two did not, that sample
was re-run. If two of the replicates amplified and one did not, then the two that provided

a positive result were averaged for that sample’s Ct value.

 

100000 1

l

\ I

10000 _ LL
\\ y = 2E+096 0.6374x: ;

\ R2 = 0.9935 Ll

. \\
\.

15 20 25 30 35
Ct value

 

 

 

 

.3
O
O
O

 

 

_\
O
O

 

.3
O

 

 

Number of spirochetes in the sample
(log scale)

 

 

.0
_s

Figure 2.6. An example of a standard curve used to determine numbers of spirochetes in
tick and tissue biopsy samples. Ct value refers to the cycle number at which the PCR
product with the labeled probe was detectable above a preset threshold. The regression
equation calculated by Microsoft Excel was used to determine the number of spirochetes
present in the sample. The Ct value of the sample was substituted for x, and y, or the
number of spirochetes was obtained algebraically.

50

Mammal tissue DNA extraction and optimization

Ear tissue biopsy punches from all mammals captured underwent DNA extraction
using DNeasy kits (Qiagen, Valencia CA). Samples were prepared for extraction by
removing them from their original ethanol-filled tube to a Kimwipe to dry under the hood
before getting placed into a new, clean microcentrifuge tube. Modifications made to the
Qiagen protocol were the same as for the tick extractions. Ear punch DNA was stored at
4°C until the desalting procedure could be performed shortly thereafter. In order to
ensure that the amount of mammalian DNA in the ear punch biopsy extractions would
not overwhelm the primers for B. burgdorferi, an Optimization trial to determine DNA
sample concentration was performed; additionally we tested whether desalting the
extracted DNA would be needed. The total eluted 50 pl volume of the DNA was first
divided into two 25 pl aliquots. One was used for desalting. Desalting the samples
required the use of flow-through columns to remove impurities (Pure Biotech LLC,
Middlesex, NJ). The sample was placed on hydrated gel matrix inside a spin column and
spun in the microcentrifuge. The spinning allowed the eluted DNA to flow through, and
trapped any interfering salts in the gel. The remaining volume was smaller in all cases,
and varied from approximately 10 pl to 24 pl.

The desalted and virgin aliquots of each sample were then further divided by
removing 5 pl of the volume of DNA, adding it to a new tube, and diluting 1:10 by the
addition of 45 pl of sterile PCR-grade water. Thus, in total, four subsamples (Figure 2.7)
of the original elution were made (i.e., salted 1X, desalted 1X, salted 0.1X, and desalted
0.1X), and the PCR trials could begin in order to find the optimal combination of DNA

techniques.

51

A non-random sample (n = 10) of mammal DNA was chosen to undergo the trial
to determine the optimal treatment of DNA for detection of B. burgdorferi. The samples
were chosen because of the likelihood that the animal had been bitten by an infected tick.
Therefore, the samples were from older mice from which we had pulled an engorged tick

which tested positive for B. burgdorferi using qPCR.

 

 
  
      
     
    

 

 

 

 

 

 

 

 

 

    

 

 

 

     

 

 

 

 

 

 

 

 

 

Original
50p] sample
elution
25p] aliquot 25p] aliquot
desalted NOT
¢ desalted
10-24pl
remaining after _
procedure 5pl aliquot 20pl left
added to 45 pl unadulterated
water to create ,
1:10 dilution 2];
(50pl total)
5p] aliquot 5-l9pl
added to 45 pl remaining left 2 A
water to create undiluted
1:10 dilution
(50 pl total) 1B
1A

Figure 2.7. Creating mammal DNA samples to optimize detection of B. burgdorferi
by PCR. Flowchart detailing the steps to create the 4 subsamples from each original
50p] elution of DNA from mammal ear biopsy punches. The 10 samples used for
optimization trials were those from older mice infested with ticks that had tested positive
for B. burgdorferi.

52

The PCR trial conducted was a nested reaction for amplification of the intergenic
spacer region (IGS) between 165 (rrs) and a 23s (rrlA) rRNA genes in B. burgdorferi,
(Bunikis et al. 2004b). PCR with gel electrophoresis was used for this trial so that the
product could be visualized.

‘ The outer reaction was run in an Applied Biosystems GeneAmp 9700 96-well
therrnocycler in a 50 pl reaction volume. Each well contained 43 pl PCR Superrnix
(Invitrogen), 1 pl each outer-forward (IGS F: 5’GTA TGT TTA GTG AGG GGG
GTG3’) and outer-reverse (IGS R: 5’GGA TCA TAG CTC AGG TGG TTA G3’)
primers (0.16pmoles) and 5pl template DNA. The positive control was extraction from
culture c017-p8 (courtesy S. Hamer), and the negative control was water. Reaction
times were as follows. The initial denature ran for l min at 80°C and then 3 minutes at
94°C, followed by 35 cycles (30 seconds at 94°C, 30 seconds at 56°C, 60 seconds at
74°C). The final extension ran for 7 minutes at 72°C. Samples were stored at 4°C until
inner reaction could be completed.

The inner PCR followed the same protocol for mastermix preparation detailed for
the outer PCR reaction, but instead used the inner forward (IGS F n: 5’AGG GGG TGA
AGT CGT AAC AAG3’) and inner reverse (IGS Rn: S’GTC TGA TAA ACC TGA GGT
CGG A3’) primers. The template DNA used was 5uL of the product from the previous
PCR. The inner program was the same as above but with the following modifications:

the initial denature ran for 1 min at 80°C and 90 sec at 94°C, followed by 40 cycles of the

previous denature-anneal-extend cycle. Samples were stored at 4°C until gel

electrophoresis in 2% agarose gel.

53

Assaying mammal tissue biopsies for Borrelia burgdorferi

Quantitative PCR for biopsies was conducted as per the tick samples. The primer
targets the 16s intergenic spacer region (Bunikis et al. 2004b). The product is detected
using the TaqMan MGB (major groove binding) protein and the 6F AM fluorescent
labeled probe. Primer sequence: 5’6FAM-TTC-GGT-ACT-AAC-TTT-TAG-TTA-A-
MGB (Applied Biosystems). All reactions were run on the Applied Biosystems (AB1)
7900, at the Research and Technology Support Facility on site at Michigan State
University. Master mix for 15u1 reaction: 7.5 pl 2x ABI TaqMan PCR master mix, 1 pl
probe at 200 nm/L, lpl each forward and reverse 16s primers at 900nm/L, 1.5 pl PCR-
grade water, and 3 pl template DNA. Each sample was run in triplicate, along with a
serial dilution of spirochetes to create a standard curve, and positive and negative
extraction controls and positive and negative PCR controls.

Data were derived in the same manner as the tick qPCR assays, although the
number of spirochetes was calculated by multiplying the standard curve-derived value by
500 (times 10 for the 1:10 dilution, times 50 for the entire 50 pl elution). Rules for re-
testing and determination of real or artificial amplification were followed as described for

ficks

Enzyme-linked immunosorbent assay (ELISA) of Peromyscus spp. serum samples

Serum samples from Peromyscus spp. were tested for the presence of antibodies
to a B. burgdorferi antigen. Of the 238 small mammals captured during the field season,
167 were Peromyscus spp., and of those, 150 (90% of the total Peromyscus spp.

captured) were tested with ELISA. It was this study’s intent to look at Peromyscus spp.

54

only in order to control for confounding factors such as species differences. Furthermore,
because this test requires the use of a species-specific anti-conjugant, only Peromyscus
spp. could be tested. Sera from field-collected Peromyscus leuc0pus from Lyme endemic
site in Connecticut were used as positive and negative controls (Tsao 2004 and Bunikis
2004)

Medium binding 96-well plates were used for this assay. Plates were first blocked
with 50pl 1% milk-PBS overnight at room temperature (22°C). 1% milk-PBS made with
dried milk powder. The following day, after rinsing wells twice with PBS-Tween 20,
serum samples were diluted 1:100 in 1% milk-PBS and added to the wells. Samples were
run in duplicate. The plates were then incubated at room temperature for one hour, and
washed six times with PBS-Tween 20. The diluted (1:1000) goat anti-Peromyscus
leucopus IgG horseradish peroxidase conjugated antiserum (60 pl) (Kirkegaard & Perry,
Gaithersburg, Maryland) was added to the wells in bright light and incubated one hour at
room temperature. Following that, plates were washed eight times with PBS-Tween 20.
100 pl 3.3’, 5.5’-tetramethylbenzidine (TMB) substrate was then added, and after three
minutes at room temperature, the reaction was stopped with 50 pl 1N HCl .

Immediately after reaction termination, samples were read in a Molecular Devices
Kinetic Microplate Reader (Sunnyvale, CA). Samples with optical density (OD) readings
> 3 standard deviations above the negative value mean were considered positive. All OD
values were normalized by subtracting the empty well average (see plate map, Figure
2.8). Samples that yielded a positive result were subsequently run in an endpoint titer
assay to determine the dilution at which the sample would remain positive, in order to

assess the relative strength of antibody response.

55

The endpoint titer assay was performed using the same ELISA protocol as for the
screening assay, but the samples tested were run at three dilutions: 1:200, 1:400 and
1:800. The endpoint titer for a specific sample was determined to be the last dilution at

which the OD reading was still > 3 times above the negative value mean.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PLATE MAP
Peromcus spp. sawles A

1 2 3 4 5 6 7 8 9 1 0 1 1 1 2
A 6 6 7 7 1 0 1 0 12 12 1 5 1 5 1 6 1 6
B 1 8 18 1 9 1 9 20 20 21 21 22 22 5 5
C 24 24 27 27 28 28 29 29 30 30 31 31
D 32 32 43 43 42 42 44 44 45 45 46 46
E 47 47 48 48 49 49 36 36 37 37 34 34
F 35 35 4O 40 41 41 38 38 39 39 54 54
G 51 51 52 52 218 218 101 101 55 55 56 56
H 57 57 E E 470++ 470++ 449+ 449+ 298- 298- E E

 

 

 

 

 

Figure 2.8. Generalized plate map for the ELISA assay. Several of the wells were left
empty in order to later calculate an average baseline (wells labeled ‘E’). The positive
controls (denoted ++ and +), negative control (denoted -) and samples (all other numbers)
were arranged in the remaining well according to the diagram.

Statistical analyses

All data analysis was canied out using SAS (SAS Institute, Cary, NC) and
ArcGIS (ESRI, Redlands, CA).

For measurements of variance to mean ratios, LMC index and k values, basic
algebra was employed to obtain these numbers for the populations. The variance to mean
ratios were compared against the null hypothesis that the populations followed a random

distribution, and therefore were compared to Chi-square values accordingly. This

56

 

method, derived from Elliot (1977) is called the Index of Dispersion (ID), and uses the
following formula:
2
ID=(s /x) * (n— l)
where s2 is the sample variance, x is the sample mean and n is the sample size.

This Index is then compared to the Chi-square value for the appropriate (n — 1) degrees of
freedom. A value greater than that Chi-square value indicates that the population
deviates from the random distribution.

It has been shown that as sample size decreases, values of sample mean parasite
burden, variance, and level of aggregation are underestimated because there is low power
for detection of the few hosts with large parasite burdens (Gregory and Woolhouse 1993).
The sample mean will underestimate the true mean as sample size decreases and/or
aggregation increases (Pacala and Dobson 1988). Therefore, if host sample sizes are
unequal in regard to age classes, there may also be an artifact of significant patterns in
age classes where sample sizes are smaller and aggregation more severe, such as in the
oldest age classes (Gregory and Woolhouse 1993).

Age classes matter for both ticks and mice; the numbers of adults will invariably
differ from the numbers of juveniles or nymphs based on factors such as mortality rates,
behavior, and seasonality. Thus, sample size for statistical power was considered for this
study so as to avoid this potential problem. When considering groupings for statistical
analysis of measures of aggregation, as often as possible, the data were broken down by
trap period and by grid. When sample sizes were low, like for nymphal ticks picked up

on the drag, they were grouped by trap period and not by grid, since activity of these ticks

57

is seasonal and their activity is more likely to vary by season than by place in the same
region.

Maps displaying kernel density were compiled for comparison of clustering
patterns for nymphs off mice of all species and nymphs off mice that were just I.
scapularis. Similar maps for larvae were constructed as well. Data tables with latitude
and longitude coordinates were entered into ArcMap. Clusters were examined visually
on the maps.

Questing ticks were not sampled at the trap points. In order to derive the numbers
for the questing ticks at each trap point, the data were plotted in ArcMap (ESRI,
Redlands, CA) and interpolated by ordinary kriging, using default parameters. This
created a continuous surface over the grid. The trap points were overlaid with this
surface (Figure 2.9) and the value of the rasterized surface was extracted at each trap
point to get an approximate value for each. That value is the estimated number of ticks

for that point that would have been collected by dragging over that point.

58

 

Figure 2.9. Example of overlaying the rasterized surface (calculated from the drag
sample points, pink circles) with the trap points (yellow squares) fiom one grid. The
concentric colored circles in the bottom lefi are radiating from one point with a high
number of total ticks collected from dragging. The solid green around the rest of the
figure indicates zero values at the remaining trap points. Figures in this thesis presented
in color.

Further, the data used for the mouse captures was the first capture of that
individual in the trapping period only. In order to avoid the problems associated with
repeat captures and lack of independence, only these captures were used in the spatial
analysis. In certain cases, this made the sample size for a grid at a particular time period
very small, and the analysis reflects this lack of power, since the grids were lumped by

trap period for all other analyses.

59

Data were then analyzed using cross-correlogram and cross-variogram techniques
to determine at what distances variables were spatially correlated. Using a cross
variogram, lag distances were measured, whereby the units were a function of the 12 m
spacing, and thus one unit was considered equal to 12 m. Lag distances are the distance
between points where the maximum lag distance is the distance between the two points
farthest apart; the minimum lag distance is the distance between the two closest points.

Vegetation (understory, coniferous and deciduous trees) were compared to mouse
locations, larvae on mice, nymphs on mice and questing ticks. Quantified loads of
B. burgdorferi were compared to drag ticks, mice and larvae on mice, nymphs on mice
and OD values. For a complete list of comparisons, see Table 2.4.

All spatial analyses were done using SAS (SAS Institute, Cary, NC). First, all of
the variable comparisons listed in Table 2.4 were examined using regression and a cross-
correlogram of the residuals at each of three trap periods, for a total of three analyses per
comparison. Data sets were examined individually and heavily skewed sets were log-
transformed. Since captures and ticks picked up on dragging were low at some of these
‘snapshots’, grids were combined at each time period and examined by time period,

because sample sizes were too low to show significant correlograms.

6O

Table 2.4. Variables compared using regression and a cross-correlogram. Vegetation
actually refers to three variables: understory, deciduous trees and coniferous trees, and
three separate analyses were run. Values of the coefficient of correlation, r, and the
distance at which the association was still significant are listed. Comparisons in bold
were significant for at least one of the runs. See Table 3.7 and 3.8 for results.

 

 

 

 

 

 

 

 

 

 

Factor 1 Factor 2
Vegetation number of mice
Vegetation number of infected mice
Vegetation total ticks on mice
Vegetation questing larvae
Vegetation questing nymphs
Vegetation infected questing larvae
Vegetation infectedjuesting nymphs

Questing larvae number of infected mice
Questing nymphs number of infected mice

 

Total questing ticks number of infected mice

 

Questing larvae total ticks on mice

 

Questing nymphs total ticks on mice

 

Total questing ticks total ticks on mice

 

Infected questing larvae infected larvae on mice

 

Infected questing nymphs

infected pymphs on mice

 

Spirochetes in questingnymphs

infected larvae on mice

 

 

Spirochetes in questing larvae

 

infected nymphs on mice

 

 

Significance was determined by calculating r, the coefficient of correlation and
determining the distance at which the two were still related. Ifthe significant cutoff
distance, determined by r, was smaller than one unit (the minimum lag distance possible,
or the distance between 2 trap points, in meters or in decimal degrees) then the
association was not considered significant. Values of r for each comparison are listed in
Table 2.4.

The value of r was calculated using the formula:

.=,*(rn—.a)/(r—1-.A2)

61

where t is looked up in a t distribution table by significance value and sample size, and
then r is solved algebraically. For these comparisons, 0.05 was chosen as the level of
significance. Sample sizes for all comparisons were based on 240 observations. Thus,
the significant r value used for a cutoff was 0.1069.

Since distances at which the variables are spatially correlated are given in decimal
degrees, these distances must be converted to meters in order to understand them in terms
of real distances. This was done using the conversion factor that one degree of latitude at
45°Nis equal to 111.113 km, or 111,113 in.

Associations that were significant by this process and made sense biologically
were then examined by spatial regression in order to make predictions about one variable

based on another. These statistical models will be based on the formula:
Y = flo + fl1X(Si)
where Y is the dependent variable, ,80 is the intercept, 61X is the independent variable and

s,- is the residual, or error term.

Datasets were examined at each time period with a semi-variogram. These

variograms can be found in the Appendix, Figure A] to Figure A19.

62

CHAPTER 3
RESULTS
Descriptive statistics
Vegetation sampling

All three grids were sampled for vegetation at each of 80 trap points on the grid.
Each point was assessed for amount of understory; a categorical descriptor was assigned
(see Table 2.2 for description). Understory coverage was compared using chi-square test.
Grids were found to have significant differences in understory composition (x2 statistic
682.4, df = 24, p-value < 0.0001). These differences were further investigated using the
chi-square test to do pair-wise comparisons. Grids III and IV were found to have
significantly different understory composition (x2 statistic 468.0, df = 21, p-value <
0.0001). Grids III and V were found to have significantly different understory
composition (x2 statistic 468.0, df = 15, p-value < 0.0001). Grids IV and V were found
to have significantly different understory composition (x2 statistic 450.2, df = 18, p-value
< 0.0001). In sum, all three grids were significantly different from one another in
understory coverage. When taken together, grid 111 had the greatest amount of
understory, an average category classification of 1.84 at each trap point, followed by grid
V, with 1.83 at each point, and then grid IV, with 0.99 at each point.

Plots were also sampled by recording the genera of the three closest mature trees
to each trap point and compared using the chi square test. Overall, there were 14 genera
of trees present at the grids. Grids were found to be significantly different in tree
composition (x2 statistic 5167.1, df = 228, p-value < 0.0001). Table 3.1 shows a list of

trees and their numbers at each grid. Overall, grid HI has about twice as many oaks as

63

the other grids. Grid IV has more than twice as many hemlocks than the other grids, and

grid V has twelve times more aspens than the others.

Table 3.1. Genera of trees at each grid. A total of 240 data points was taken at each

 

 

 

grid.

. Tree (genus) Grid III Grid IV Grid V
Oak 75 38 28
Maple 82 75 82
Hemlock 15 54 21
Basswood 0 2 1
Ash 12 31 36
Pine 25 14 12
Aspen 3 0 36
Cedar 4 14 18
Birch 13 8 4
Filbert O 1 O
Spruce 1 O 0
Elm 1 0 0
No tree 9 3 2
Total 240 240 240

 

 

 

These differences were further investigated using the chi-square test to do pair-
wise comparisons. Grids IH and IV were found to be significantly different from each
other (x2 statistic 3953.2, df = 180, p-value < 0.0001). Grids III and V were found to be
significantly different from each other (x2 statistic 3680.5, df = 154, p-value < 0.0001 ).
Grids IV and V were found to be significantly different from each other (12 statistic
3620.8, df = 150, p-value < 0.0001). Kriged surfaces of all three vegetation samples for

all three grids are shown in Figure 3.0A, Figure 3.0B and Figure 3.0C.

64

 

Figure 3.0A. Kriged surface of understory vegetation layer. Data were interpolated
using ordinary kriging between 12 data points. Darker colors indicate heavier understory
cover. Dots represent trap points. Figures in this thesis presented in color.

 

Figure 3.08. Kriged surface of deciduous tree layer. Data were interpolated using
ordinary kriging between 12 data points. Darker colors indicate heavier deciduous tree
cover. Dots represent trap points. Figures in this thesis presented in color.

65

 

Figure 3.0C. Kriged surface of coniferous tree layer. Data were interpolated using
ordinary kriging between 12 data points. Darker colors indicate heavier coniferous tree
cover. Dots represent trap points. Figures in this thesis presented in color.

Drag-collecting questing ticks

Each of the three grids was dragged for questing ticks during each of the three
trapping sessions for a total of 9 collections. The three grids were chosen as replicates
for each trap period. The phenology for the Ixodes scapularis ticks collected over the
course of the season, by life stage, is shown in Figure 3.1. For this graph, in order to get
an understanding of the change in activity over time, data are pooled from all three
trapping grids. Drag collections began on June 13 and ended August 5, with period 1
running from June 13 to June 25, period 2 from July 7 to July 19, and period 3 from July
23 to August 5. Collections are represented as proportions of the total collected (per

stage) on that date rather than raw numbers. Sampling appears to have begun at the end

66

of the spring adult peak, no obvious peak nymphal activity was detected, and larval

numbers peaked in mid-July.

 

 

 

 

 

 

 

 

60
—Adults
50 4
— Nymphs
_ - - Larvae
:3 4o —
.9
0.—
0
id 30 r
:
no
2
a 20 —
10 4 / ’
\\ /
0‘ /
0 N r F \ __

 

Mr r

11-Jun 18-Jun 25-0un 2-Jul 9-Jul 16-Jul 23qu 30-Jul
Date

Figure 3.1. Questing Ixodes scapularis ticks collected over time at the study site
(Menominee County) in 2006. Data are pooled from all three trapping grids. Collections
are represented as proportions of the total collected (per stage) on that date rather than
raw numbers. Sampling appears to have begun at the end of the spring adult peak. No
obvious peak nymphal activity was detected. Larvae numbers peaked in mi d-July.

The drag collected questing I. scapularis ticks were sampled two ways, in order to
get a relative measure of how sensitive the sampling methods were. Drag-collecting can
be a sensitive method to detect questing ticks in the environment (Fish 1995), but there is
a need to balance covering a large area with the frequency of stops. If one is st0pping
frequently, there is more time and effort being spent, but stops farther apart may allow for

ticks to fall off the cloth again before they can be seen. In part, this whole grid/subset

67

comparison was done to get a feel for the balance between effort and detection ability. If
we find that our large-scale effort is missing some ticks, we can note that in our analysis,
and use this information to inform our next study. If we find that the detection methods
were equal, then we can be satisfied with the amount of effort put into the drag-sampling
scheme.

The ticks were collected first on one small sub-square of each grid (Figure 2.3),

equal to l/20’h, or 5% of the total area and stops were made every 8 m, over a total of

72 m2, to check the cloth for ticks. Second, sampling over the entire grid was conducted,
with stops every 24 m, over a total of 480 m2. Although the small subsquare itself

2 . 2
represents 5% of the total area, the amount dragged, 72 m , IS actually 15% of 480 m .

However, in order to accurately compare them, we need to take into account the

entire sampled area. Ifthe entire area is considered the whole-grid drag plus the

subsquare, (480 m2 + 72 m2), the whole-grid drag accounts for 87%, and the subsquare is

13% of that total. Therefore, we would expect the whole-grid collected ticks to account
for 87% of the total ticks recovered, and the subset of ticks to account for 13%. Of
course, this assumes spatial homogeneity of ticks over the landscape.

Specifically, the percent of I. scapularis larvae detected overall in the intensive
sample was 9.1% of the total, instead of the 13% that we would expect. Of all ticks
recovered, 8.4% of larvae, 32% of nymphs and 11.6% of adults were recovered from
these intensive samples. The results are mixed- for total ticks and larvae, the intensive

sample underestimated. The estimation for the adult ticks was about right, and the

68

nymphal ticks were overestimated. These results do not strongly indicate that many
ticks, besides nymphs, were missed on the large-scale dragging effort.

On average for the whole season, drag collection yielded about 8 I. scapularis
nymphs per 1000 m2. At the peak of their activity, that number was 17 per 1000 m2. For
all I. scapularis lifestages, there was an average of 78 per 1000 m2, but that number
varied from 2 to 89 at each of the sampling times. Variation through time is shown in

Figure 3.1.

Mammals

Most of the small mammals captured at the site were Peromyscus spp., accounting
for 68.3% of the 230 total individual animals captured. Since there were so few
Peromyscus maniculatus gracilis caught, they were added to the numbers of the
Peromyscus leucopus since they essentially function as a single species in this system
(Friedrich 2003). Because of the number of mice caught and their role in the Lyme
disease system as a highly reservoir-competent host, these mouse species are considered
the main study population. Other small mammals caught, with their relative percentages

of the total number of captures, are listed in Table 3.2.

69

Table 3.2. Percent of total small mammal captures among all three grids by species at
the field site in Menominee County, June-August 2006. A total of 230 individuals were
captured.

 

 

 

 

 

 

 

 

 

 

 

 

 

Species Common name # caught % of total
Peromyscus leucopus White-footed mouse 151 65.7
Peromyscus maniculatus Deer mouse 6 2. 6
gracilis
T amias striatus Eastern chipmunk 32 13.9
C lethrionomus gapperi Southern red-backed vole 29 12.6
Blarina brevicauda Short-tailed shrew 5 2.2
Glaucomys sabrinus Northern flying squirrel 4 1.7
Dedelphis virginiana (juvenilg Opossum 2 0.9
Napeozapus insignis Woodland jumping mouse 1 0.4

 

Traps were set for a total of 5,760 trap nights. The 705 individual captures of all
mammals provide a capture success rate of 12.2%. Of the 705 captures, 579 of those
(82.0%) were Peromyscus spp.

1 Population estimates for Permyscus spp. at the study site were estimated for each
grid at each time period. Population abundances were derived using the interactive
software program CAPTURE (United States Geological Survey, Patuxent, Rhode Island).
Assumptions of the test for abundance included a closed population. Estimates for each
of the nine intervals are listed in Table 3.3. These population estimates are also shown in
graphical form in Figure 3.2. Population estimates listed by trap period and by grid
yields a range from 19 to 56. Including the 95% confidence intervals in a range estimate
expands to between 16 and 77 individuals per grid per trap period. For the graph, error
bars were calculated using the average 95% confidence intervals for that grid.

Over time, the population sizes appear to shrink and then grow. Sample sizes for

grid IV were more consistent within the grid and did not track with the other grids sample

70

 

sizes at time period 3. Sample sizes for grids HI and V are consistent throughout the
season. There was no significant difference in numbers among grids. Approximately
34% of mice were captured on more than one grid, suggesting a lack of independence of

grids.

Table 3.3. Population estimates and standard error for each 1.15 ha grid at each time
period. Three grids were sampled three times each over the trapping season from June to
August, 2006, in Menominee County, MI. The three grids were enumerated with Roman
numerals IH, IV, and V. Population estimates were derived using Program CAPTURE
(United States Geological Survey, Patuxent, Rhode Island).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Trap Period Grid Population estimate 95% confidence interval
1 III 19, se=3.72 16-32
1 IV 27, so = 3.33 25 - 40
1 V 42, se = 5.82 36 - 59
2 III 25, se = 2.34 25 - 41
2 IV 27, se = 3.48 24 - 40
2 V 27, se = 2.62 27 - 41
3 1H 56, se = 7.38 47 - 77
3 IV 29,se=3.13 27-41
3 V 51,se=7.30 42-71

 

71

 

 

 

70

 

lGridllI EiGn‘le IGridV] I

 

 

 

 

 

 

 

 

 

 

 

 

Population estimate

 

 

 

10—

 

 

ol

 

Figure 3.2. Population estimates for mice each trap period, by 1.15 ha grid.
Estimates calculated using program CAPTURE (United States Geological Survey,
Patuxent, Rhode Island). Error bars were calculated using the average 95% confidence
intervals for the grid.

Quanti'jying spirochete load in ticks

A total of 929 tick samples were assayed using the quantitative PCR method.
These tick samples included flat questing ticks, engorged and partially engorged ticks
removed from hosts, all three lifestages (larvae, nymphs and adults) and both
Dermacentor variabilis and Ixodes scapularis species. In total, 30.8% were positive.

This number represents of a mix of species, life stages and on and off host ticks.

72

The breakdown of percent positive ticks is presented in Figure 3.3A and
Figure 3.3B. In these, I. scapularis positives are shown by life stage and location (on or
off host). In these data, larval ticks were not tested individually, but were pooled together
with other larvae from the same animal or same drag transect because most were
expected to be negative. Generally, more of the on-host ticks were positive than the off-
host ticks. Infection prevalence of questing I. scapularis increases with life stage, and

adults had the highest rate at 57%.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100
\§
Positive \
80 I Negative \ \\
40 t
20 a
0 - .

Larvae Nymphs Adults

Figure 3.3A. Questing Ixodes scapularis ticks testing positive and negative for
Borrelia burgdorferi. Approximately 3% of larvae, 22% of nymphs and 57% of adults
tested positive for B. burgdorferi. Larval ticks were not tested individually, but were
pooled with other larvae from the same animal or same drag transect.

73

300

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   
 

7/
250 ./ Positive
I Negative
200
150 -
100 -
50 ~ 7
0 - .

 

 

 

 

Larvae Nymphs

Figure 3.3B. On-host Ixodes scapularis ticks testing positive and negative for
Borrelia burgdorferi. Approximately 40% of larvae and 46% of nymphs tested positive
for B. burgdorferi. Larval ticks were not tested individually, but were pooled with other
larvae from the same animal or same drag transect.

74

Quantifiring spirochete load in mouse tissue

Before PCR of the mouse tissue could begin, the optimal treatment and dilution
for the eluted DNA had to be determined. One trial was conducted using 10 samples.
Using the results of the trial to determine the best dilution for the mammal ear bi0psy
DNA (Figure 2.7) by PCR amplification, the desalted 1:10 dilution of the original DNA
was selected. The gel picture shows amplification most clearly in the lanes for this
combination of treatments (Figure 3.4). The samples that had been desalted and diluted
1:10 were selected for use in qPCR. The fragment size expected was a 987 base pair
amplicon. Sample 29, desalted and a 1:10 dilution, and sample 38, both undiluted
desalted and a desalted 1:10 dilution show amplification. The molecular ladder used was
the Invitrogen E-Gel low-range quantitative DNA ladder.

Ear biopsies were taken on mice at their first time of capture during each trap
period and checked for infection by qPCR. Overall infection prevalence for mice at their
first capture was 33%. For trap period 1, 50% of mice trapped first during that trap
period were infected, trap period 2, 26% of mice were infected, and trap period 3, 16% of
mice were infected.

Spirochete burden values were compared for juvenile (young of the year) mice
and adult mice at each trap period, and adult mice were expected to have higher burdens
because of their increases chances of exposure. At trap period 1, spirochete burdens were
found to be significantly different, (Student’s t-test, one tailed, unequal variance,

p = 0.005, df = 50) as adults had a higher spirochete load, with an average of 405
spirochetes and juveniles had an average of just 18. For trap period 2, spirochete burdens

between age classes were also found to be significantly different, (Student’s t-test, one

75

tailed, unequal variance, p = 0.017, df = 45) with adults demonstrating a higher average
burden (110 spirochetes) than juveniles (18 spirochetes). For time 3, the age classes were
still significantly different (Student’s t-test, one tailed, unequal variance, p = 0.015,

df = 36). Adults had an average of 2805 spirochetes and juveniles had zero.

.4. -.

29D1:10

Positive
Molecular
ladder

Negative con trols

 

-
v
a.

 

Figure 3.4. Electrophoresis gel result from the ear biopsy DNA treatment trial. The
fragment size expected was a 987 bp amplicon of the intergenic spacer region (IGS)
between 165 (rrs) and a 23s (rrlA) rRNA genes in B. burgdorferi, (Bunikis et al. 2004b).
Sample 29, desalted and a 1:10 dilution (faint), and sample 38, both undiluted desalted
and a desalted 1:10 dilution show amplification. Standard used Invitrogen E-Gel low—
range quantitative DNA ladder.

76

Ticks on mammals

By individual mouse, 74% (165 of 224) of mice became parasitized over the
course of the trap season. If each mouse capture is counted separately, of 705 capture
events, 396 were parasitized, so per capture, 56.2% of mice were parasitized. The lower
per-capture parasitism rate indicates that at least some mice had tick burdens of zero at
some captures and burdens of greater than zero at others. A bar chart showing total tick
burdens over the course of the trapping season is displayed in Figure 3.5. Mice were
numbered chronologically as they were caught, and so mice caught early on in the season
are represented by lower numbers. Mice caught earlier in the season also showed overall
higher total tick burdens as compared to those caught later on. This difference is due to
the seasonal activity of the ticks (Figure 3.1) as well as the increased likelihood that a
mouse caught early in the season would be caught again and multiple captures added up
to larger tick burdens.

Data used for the following analyses including the animal’s first capture only
during each of the trapping periods. Recapture data were eliminated from the analysis to
avoid issues surrounding lack of independence and repeated measures. Since only first
captures were used, sample sizes were low at each grid (~15 individuals) and therefore,
were lumped together with the other grids’ samples from that same trap period in order to

increase statistical power.

77

 

100

 

 

 

 

 

 

 

 

 

 

 

 

 

i
INymphs per mouse l
80 | Larvae per mouse 11*
in 60 i
x
.2
'3’
o
'— 40
2° . ,2'
l . I [til I I
' ': r. ill3il': .l i! ,
.i' 1:11.?" I. I . hiiiil' : 5' 11.3,”; ’1' L ,
0 a " 11' liiwi..'..‘il'i:l 'liallli‘f. . i . r 1.931" 11.1.1 I.’ i.
1 21 41 61 81 101 121 141 161 181 201 221

Animal number

Figure 3.5. Frequency histogram for number of total ticks on mice over the course
of the trapping season. Numbers for each mouse are cumulative over the whole season.

Burdens of I. scapularis ticks parasitizing mice were compared on mice of

different age classes. Mice in the field were aged by size and stage of pelage (small, all

grey fur = 1, or juvenile, medium, grey to brown fur = 2, or subadult, and large, all brown

fur = 3, or adult). Since all mice categorized 1 and 2 were young of the year, and their

sample sizes were relatively small compared to the adult mice, they were lumped together

in the juvenile category, denoted in some of the figures by the numeral 1. Thus, mice

were divided by age class and categorized as a juvenile or an adult, where the adults

displayed brown pelage and were sexually mature. Comparison of tick burdens by age

class and trap period is shown in Figure 3.6.

78

Nymphs

l Larvae

Number

 

Juvenile Adult Juvenile Adult Juvenile Adult

1 1 2 2 3 3 i

Figure 3.6. Comparison of average tick burdens per mouse by trap period
(number) and age class (juvenile or adult).

For the first trap period, larval burdens were found to differ by age (Student’s t-
test, two tailed, unequal variance, p=0.029, df = 50). However, nymphal burdens were
not significantly different based on age (Student’s t—test, two tailed, unequal variance,
p=0.097, df = 46). Adult mice were likely to feed more nymphal ticks than juvenile
mice. On average, adults fed almost 10 larvae and 2 nymphs, and juveniles fed almost 6
larvae and less than one nymph in the first round of captures. Comparison of the total
number of larval and nymphal I. scapularis ticks is illustrated in Figure 3.7, and shows no
significant difference in overall tick burden between age groups. Juveniles had an

average of more than 6 ticks per animal and adults a burden of 12 ticks per animal.

79

50

45

40

35

30

20

15

10

Figure 3.7. Comparison of total I. scapularis burdens (larvae and nymphs) between
juvenile (age=1) and adult (age=3) age groups for animals captured in trap period 1.
Only tick burdens for animals captured the first time are included in this dataset. Repeat
captures are omitted to avoid lack of independence and repeated measures issues.
Boxplot shows no significant difference between age groups.

80

For the second trap period, larval burdens were found to differ by age (Student’s
t-test, two tailed, unequal variance, p=0.04, df = 36). Nymphal burdens were also
significantly different based on age (Student’s t-test, two tailed, unequal variance,
p=0.001). Adult mice were likely to feed both more larvae and nymphs than juvenile
mice. On average, adults fed more than 3 larvae and 1 nymph, and juveniles fed 2 larvae
and less than one nymph. Comparison of the total number of ticks is illustrated in Figure
3.8, and shows no significant difference in overall tick burden between age groups.
Juvenile mice fed an average of nearly 3 ticks in total and adults an average of almost 5
ticks in total for the trap period.

For the third trap period, larval burdens were not found to differ by age (Student’s
t-test, one tailed, unequal variance, p=0.068), nor were nymphal burdens (Student’s t-test,
two tailed, unequal variance, p=0.256). On average, adults fed 1.5 larvae and less than
one nymph, and juveniles fed less than one larva and less than one nymph. Comparison
of the total number of ticks is illustrated in Figure 3.9 and shows no significant difference
in overall tick burden between age groups. Juvenile mice fed an average of less than 2
ticks per animal, and adults also fed an average of less than 2 ticks.

Since the season changed over the course of the study, the mice that were
juveniles in the beginning had matured by the end of the study. All the hosts were aged
at each time and differences among age classes compared. However, by the end of the
season (the third and last rotation), almost all of the mice were part of the most mature
age class, and thus, comparisons between age classes were probably not as meaningful

due to lack of sample size of the younger age class.

81

I6

14

O

12

10

OO
+-----+-—-—-—-+-—-—--—+----+--—-—-—+-—-—-—-+-——-—--+---—-—~+—

Figure 3.8. Comparison of total I. scapularis burdens (larvae and nymphs) between
juvenile (age=1) and adult (age=3) age groups for animals captured in trap period 2.
Only tick burdens for animals captured the first time are included in this dataset. Repeat
captures are omitted to avoid lack of independence and repeated measures issues.
Boxplot shows no significant difference between age groups.

82

I
8 + 0
I
l
l
7 +
l
I
l
6 +
l
I
l
5 + 0
l
l
l
4 + |
l 'l
I l
l l
3 + |
l l
I l
l l
2 + + ----- + j
I I l l
| l + j + ----- +
l I I l l
1 + at _____ t *_,+-_r
l l I l |
I l l I l
l l l | l
0 + + ----- + + ----- +
............ +-----------+-----
1 3

Figure 3.9. Comparison of total I. scapularis burdens (larvae and nymphs) between
juvenile (age=1) and adult (age=3) age groups for animals captured in trap period 3.
Only tick burdens for animals captured the first time are included in this dataset. Repeat
captures are omitted to avoid lack of independence and repeated measures issues.
Boxplot shows no significant difference between age groups.

83

Mouse infection status as measured by serology (ELISA)
A one-way AN OVA was conducted to determine effect of plate on sample OD.

Using a null hypothesis that there is no effect of plate, analysis determined that there is

indeed no effect of plate (F =0.00, F < 0.5, 3,383 (critical value = 2.6), p= 0.9996). Since

Fcalculated < Ftables we fail to reject the null hypothesis and say that there is no detectable

difference in the values of the positive and negative controls between plates.

Of the total number (238) of animals caught, 167 were Peromyscus spp. Of these,
150 had serum samples suitable for testing. Thus, serology was completed on 90% of the
captured mice. For mice tested, 50% were seropositive at the time of testing. Of those,
75 were gave a positive result at the first 1:100 dilution, and 75 gave a negative result.
Sera from field-collected Peromyscus leucopus from Lyme endemic site in Connecticut
were used as positive and negative controls (Tsao 2004 and Bunikis 2004). A cutoff
value for negative and positive values was determined by averaging the values from the
negative controls in the plates, and adding 3 standard deviations to that number. The
resultant value was 0.8784. All samples whose optical density (OD) values fell above
that cutoff were considered positive. Two peaks, a negative and a positive peak, can be
seen in the frequency histogram of the OD values (Figure 3.10).

Samples that yielded a positive result were then re-tested to determine an endpoint
titer for positivity. Each sample that had previously been positive at a dilution of 1:100,
were then run at 1:200, 1:400 and 1:800. These results, compared with the original
average adjusted OD values, are shown in Figure 3.11. Negative values are not shown
on this graph. The graph displays the trend that typically, higher average adjusted OD

values correlated with higher endpoint titers.

84

 

25 .

 

 

 

* I l

' l

20 I 5

' l

. l

I l

g 15 _ ,

c l
g l

8 ' *4. l

t: 10 — i %

 

01
I
I—I1_

 

0— |"|
0 0.3 0.6

Figure 3.10. Frequency histogram showing the average adjusted optical density (OD)
values for the Peromyscus spp. serum samples collected in Menominee County from June
- August 2006. Two peaks in the data are evident; one in the negative range at 0.2 (*),
and the other in the positive range at 2.3(**). The cutoff value of 0.8784 is represented
by the dashed line. The cutoff was calculated to be 3 SD above the negative control
average. Values above the cutoff were considered positives, or serum samples with
antibodies to Borrelia burgdorferi, and values less than the value of the line are
considered negatives, samples without antibody response.

0.9 1 2 1 5 1.8 2.1 2.4

Average adjusted OD value

85

 

 

 

 

 

 

 

 

 

 

 

 

 

1:800 e 4% J
I
t: i
.9
*5 l
‘5 i
a; l
8. l
- 4v 4—H—o—o :
a
g l
“J
1:200 * f % ‘
1:100 ‘W i
0 I I I I i

0.500 1.000 1.500 2.000 2.500 3.000

Average adjusted OD

Figure 3.11. Comparison of the average adjusted optical density (OD) values per
sample with the endpoint dilution at which the sample was still positive. Graph
demonstrates relationship between average adjusted optical density (OD) value for a
sample and the final dilution at which it was positive. Negative values are not shown on
the graph. The graph displays the trend that typically, higher average adjusted OD values
correlated with higher endpoint titers.

There was no clear pattern of OD values over time or by grid. Average adjusted
OD values were compared for juvenile (young of the year) mice and adult mice, and were
found to be significantly different, (Student’s t-test, one tailed, unequal variance,
p<0.0001) as adults had a higher average OD value of 1.7 and juveniles had an average of
just 1.1 (Figure 3.12). Comparison of the relationship between tick burden (log-
transforrned) and average adjusted OD value between age groups for trap period 1 is
shown in a scatterplot in Figure 3.13. The scatterplot for trap period 2 (Figure 3.14) and
trap period 3 (Figure 3.15) are also shown. At all three time periods, the adult mice

generally had higher OD values and total tick burdens then their juvenile counterparts.

86

l
2.5 + I

I l E

l l l

I l I
2.25 + I + ----- +

| l 5 l

l l 3 I

l l l l
2 + E 1i _____ a

| I l l

| l i l

| l i l
1.75 + I l l

l r ----- + + l

I I I i l

l l l l I
1.5 + i l + ----- +

l l l l

l l l l

I I l l
1.25 + l l l

I t _____ t i

l I | l

l I + l I
1 + l l l

| l l l

I l l l

| i I l
0.75 + l l

I l l I

l 1 l l

I l l l
0.5 + I l l

I l l i

| i l

l l l 0
0.25 + I I

I + ..... + O

I I 0

l l
O + o

............ +-----------+-----
1 3

Figure 3.12. Comparison of the average adjusted optical density (OD) values for
each population with the age of the population. Graph demonstrates the average OD
values of the serum samples of juvenile mice (age category 1) with adult mice (age
category 3).

87

 

 

 

3
o o o
o o
2 O 8 o
0 ° 0
O C
0
° 0
00
O
l o 0 ° 0 0
OD ° 0 D o
o
o o
o o
o o
0
o O
o
O
0 o
0 o g C’00
O O
7 f1 Vir if vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
-l 0 l 2 3 4
Log of tick burdens

Figure 3.13. Scatterplot comparison of tick burdens (log-transformed, on x-axis) and
average OD values (on y-axis) for trap period 1. Juvenile mice (a=1, black circles) and
adult mice (a=3, red circles) of the Peromyscus spp. were trapped at the field site in
Menominee County, 2006. Images in this thesis are presented in color.

88

 

 

 

 

 

3 .
2 O O
O
O o o O
o o o O
0
0D1
0
O
O
O . O 0 O
O n
O
0 o
O
o O
S E 0 g 0 0 o o
o ’3
r T f r vvvvvvvvv I rrrrrrrrr I T fir I
-l 0 l 2 3
Log of tick burdens

Figure 3.14. Scatterplot comparison of tick burdens (log-transformed, on x-axis) and
average OD values (y-axis) for trap period 2. Juvenile mice (a=1, black circles) and adult
mice (a=3, red circles) of the Peromyscus spp. were trapped at the field site in
Menominee County, 2006. Images in this thesis are presented in color.

89

 

00

OD

0C)

 

 

 

Log of tick burdens

Figure 3.15. Scatterplot comparison of tick burdens (log-transformed, on x-axis) and
average OD values (on y-axis) for trap period 3. Juvenile mice (a=l, black circles) and
adult mice (a=3, red circles) of the Peromyscus spp. were trapped at the field site in
Menominee County, 2006. Images in this thesis are presented in color.

For analysis of the relationship between the number of ticks (data for tick burdens
on mammals were log-transformed in order to meet normality requirements for data
analysis) and the average adjusted optical density (OD) values, it does not matter whether
the three times are looked at separately or together. For trap periods 1 and 2, for age 3
the two measures were positively related, and for trap period 3 these factors were not
related. At age 1, there is no relationship between tick burden and OD value at any time

interval.

90

Aggregation results

Aggregation was assessed using three measures: the variance to mean ratio, the k
statistic of the negative binomial distribution and Lloyd’s mean crowding index (LMC).
Measures of aggregation were calculated for the following populations: spirochetes in
questing ticks, spirochetes in bloodfeeding ticks, spirochetes in mice and bloodfeeding
ticks on mice. The ticks were grouped by lifestage. Only larval ticks from animals were
separated by trap period and grid. Nymphal ticks from animals were separated by trap
period only because sample sizes were small, and questing ticks were not separated by
trap period or by grid since sample sizes were too low to have sufficient power.

Spirochetes in questing larvae show very little aggregation because there are so
few B. burgdorferi organisms in the population, since generally, the ticks hatch
uninfected. Loads in the larval ticks that were positive were very low; the one positive
pool had a total of 9 spirochetes. The spirochetes from the ticks that had been feeding on
mice show a high degree of aggregation, most likely as a reflection of the distribution of
the spirochetes among their hosts. In general, measures of aggregation for all examined
populations indicated higher degrees of aggregation when grids were lumped by trap

period than when examined alone.

Variance to mean ratios

The variance to mean ratio indicates an aggregated population when greater than
one. Table 3.4 shows the variance to mean ratios for the listed populations. Variance to
mean ratios for populations of I. scapularis ticks on mice by lifestage, trap period and

grid are presented in Figure 3.16. Overall, the larvae have higher ratios, indicating more

91

highly aggregated p0pu1ations. In general, trap period 1 shows the most highly

aggregated populations of ticks on mice. Comparisons of variance to mean ratios and

sample sizes for ticks on mice and spirochetes in mice are shown in Figure 3.17 and

Figure 3.18, respectively.

Table 3.4. Variance to mean ratios, Lloyd’s mean crowding (LMC) index and the
negative binomial statistic k for number of B. burgdorferi organisms in the listed
populations of I. scapularis. Variance to mean ratios greater than one indicate
aggregation. Values of k less than one indicate aggregation. LMC values indicate how
many other parasites the average parasite shares a host with. Spirochetes in questing
larvae show very little aggregation because there are so few B. burgdorferi organisms in
the population, since generally, the ticks hatch uninfected. The spirochetes from the ticks
that had been feeding on mice show a high degree of aggregation, most likely as a
reflection of the distribution of the spirochetes among their hosts.

 

 

 

 

 

 

 

 

 

 

 

Population Variance to k Lloyd’s mean
mean ratio crowding index
All larval ticks on mice 14680 0.0657 15660
Larval ticks, trap period 1, grid HI 4970 0.1507 5725
Larval ticks, trap period 1, grid IV 13524 0.1749 15894
Larval ticks, trap period 1, grid V 1152 0.2037 1392
Larval ticks, trap period 2, grid III 4848 0.0640 5174
Larval ticks, trap period 2, grid IV 23252 0.1018 25629
Larval ticks, trap period 2, grid V 7115 0.0881 7753
Larval ticks, trap period 3, grid III 6605 0.0514 6964
Larval ticks, trap period 3, grid IV 16763 0.0356 17387
Larval ticks, trap period 3, grid V 5393 0.0603 5735
All nymphal ticks on mice 22895 0.1313 25908
Nymphal ticks, trap period 1 20720 0.1214 23244
Nymphal ticks, trap period 2 27893 0.2223 34097
Nymphal ticks, trap period 3 2668 0.3261 3541
Questing larvae 8.8 0.0594 -7
Questing nymphs 5984 0.0630 6376
Questing adults 3600 0.3689 4930
Mice, trap period 1, grid III 2731 0.1585 3163
Mice, trap period 1, grid IV 403 0.6924 680
Mice, trap period 1, grid V 119 0.3468 159
Mice, trap period 2, grid H1 585 0.1377 664
Mice, trap period 2, grid IV 230 0.2511 286
Mice, trap period 2, grid V 266 0.2753 337
Mice, trap period 3, grid HI 53462 0.0775 57605
Mice, trap period 3, grid IV 82 0.1012 89
Mice, trap period 3, grid V 859 0.2511 1073

 

92

 

 

10

 

 

 

 

 

 

 

 

 

 

o Grid m larvae I ,
8 0 Grid Ill nymphs _
.g o 0 Grid IV larvae I
E 9 Grid IV nymphs I
g 6 a Grid v larvae ”—‘I
8 I :2. Grid v nymphs I
8 4 I
c E
E A o ‘ t
" o ‘ 2 I
> ,
2 ‘"
a 0 I
“'""‘"""""'""""‘""Z """""""" 6' """"""""
0 . a. .

1 2 3

Trap period

Figure 3.16. Variance to mean ratios for I. scapularis ticks on mice by trap period
and grid. The dotted line represents the cutoff for an aggregated population, or a
variance to mean ratio of 1. Overall, the larvae have higher ratios, indicating more highly
aggregated populations (black symbols). Trap period 1 shows the most highly
aggregated populations of ticks on mice.

93

 

 

 

 

 

 

 

 

 

 

 

 

12.00
10.00 0
,-
3; 8.00 i a
E R2=0.8609 / o
3 / R2=0.6358
° 6.00 .1
.2 /
E /
g 4.00 ’
I I ’f n
I 90 W90 9 Larvae l
2.00 liar” L
~' I Nymphs I
. l
0.00 fl T - f fl 1 l T T

 

0 20 40 60 80 100 120 140 160 180 200 220
Mean number of ticks per population

Figure 3.17. Comparison of mean number of ticks found on mice per trap period by
grid, to the variance to mean ratio for the same population. Ticks were divided by
lifestage. Generally, as the number of ticks in the population increases, so does the
variance to mean ratio for that population, indicating that degree of aggregation may
partly be a function of n. This graph illustrates how smaller sample sizes may not reflect
the true degree of aggregation, since highly parasitized individuals (of which there are
few) will be missed.

Figure 3.17 shows the comparison of number of ticks found on mice per trap
period by grid to the variance to mean ratio for the same population. In this figure, ticks
were divided by lifestage. Generally, as the number of ticks in the population increases,
so does the variance to mean ratio for that population, indicating that degree of
aggregation may partly be a function of n. This graph illustrates how smaller sample
sizes may not reflect the true degree of aggregation, since highly parasitized individuals

(of which there are few) will be missed.

94

 

3000

2500

 

2000

1500 /
1000
/
500 '

fir f

 

 

 

Variance to mean ratio

 

 

 

j T

0 1000 2000 3000 4000 5000 6000
Mean number of spirochetes per mouse in the population

Figure 3.18. Comparison of mean number of spirochetes found in mice per trap
period by grid, to the variance to mean ratio for the same population. Generally, as
the number of individuals in the population increases, so does the variance to mean ratio
for that population, indicating that degree of aggregation may partly be a function of n.
This graph illustrates how smaller sample sizes may not reflect the true degree of
aggregation, since highly parasitized individuals (of which there are few) will be missed.
The lowest point is the data point for grid IV, trap period 3. The highest point is grid IV,
trap period 1. There appear to be only 8 points because two of the data points overlap
completely.

Figure 3.18 demonstrates comparison of number of spirochetes found in mice per
trap period by grid to the variance to mean ratio for the same population. As the number
of individuals in the population increases, so does the variance to mean ratio for that
population, indicating that degree of aggregation may partly be a fimction of n. This
graph illustrates how smaller sample sizes, especially when loads are lower, may not

reflect the true degree of aggregation, since highly parasitized individuals (of which there

are few) will be missed.

95

Negative binomial distribution
Values of k, a measure of aggregation, were calculated using the following

approximation of k, from Ludwig and Reynolds (1988), which is as follows:

2 . .
k = x2/ (5 —— x), where x = the sample mean and s = the sample variance. k rs a value

between 0 and infinity, and the smaller k gets, (but typically less than 1) the more highly
aggregated the population. Table 3.4 shows the k parameter for the listed populations.

All of the values of k were below 1, indicating a negative binomial distribution.

Lloyd’s mean crowding index

The calculation for crowding is a straightforward algebraic equation:

2 . .
m* = x + (s /x - 1), where m* = the adjusted mean (or LMC index), x = sample mean,

5 = the sample variance. Table 3.4 shows the LMC index for the listed populations. The

LMC index is high for all populations, indicating crowded conditions for all.

Spatial clustering of mice

ANCOVA for spatially correlated relationships shows that the Gaussian model
fits the data the best (AIC= 119.8), indicating that mouse captures were spatially
correlated (Figure 3.19). The captures were plotted on the grids at which they were
captured in ArcMap (ESRI, Redlands, CA) and these are presented, by trap period, in
Figure 3.20. Trap period 1 has the most captures, and trap period 3 the least. The very

smallest dots on the maps indicate zeroes.

96

gnu-Hoo-‘fim<

 

L l I

r l ; ' ‘
0.0000 0.0005 0.0010 0.0015 0.0020 0.0025 0.0030

Distance (decimal degrees)

Figure 3.19. Gaussian spatially-correlated mouse chart. Variogram for distance,
showing the approximate Gaussian curve for spatial correlation.

97

o
o
. . o o o . o
..... . . . e 0.. . 0.... o o
. . .
0
. O Q
o . .9..0.0o 0 .° .. o ’ .0.
. o 0 . o
C . Q
. o
o o
- . - ... 0 .~. I
o .0 ° 9
o 0 o
O ,
'0
O
o
‘ . O. .
o o. ' o.
O . ' .
0 ° ' ° ‘°
0 o
’0 0 ~. 2
o
o
. 9
°' C
O ..
o .. o
o
0 o
o ..o..‘ '9
0... ’. ‘ o
O. Q. .
O. .
0 o o

O O o O 0
o .0 . . . 9.
o
.0
o
‘ o
o . o. ..'0.
o . ° 0 .
o. z o 0 °. .
‘. o o
. I...
o
. o
O
o .9 o 3
o
. e o .0
O ..
o
.0... g .
'°' 0 °
. a" ’ .. .0 .
O . Q . o O
0 . o. '. o .
. . o o. .
o ‘ 0 o .
o o . .. o . o
9 o
O . O O . O O. O. . O . .
. o o
. o .. o o. .0 . o .. . .
. . . . . . . ... . O . . . O
. o
. ° '0 0.. ‘0. ’g o . ...
. -- ° 0 ° . ° ’ o
o . .
o
o
o . . . o e °° .‘ .. 0.
o 0

Figure 3.20. Maps displaying areas of mouse capture, by trap period (1, 2 and 3). Dots
represent each trap point and the size of the dot indicates the 0, 1 or 2 captures for that

trap period.

98

Looking at the all three grids and trap periods together because mouse densities
on grids did not differ over time, there was a spatial correlation in average adjusted OD
data up to 0.0006 distance units, or 66.7 m. The distance units used were decimal
degrees; one degree of latitude at the 45th parallel is equal to 111.133 km. This means
that mice up to 66.7 m apart have related values for their average adjusted CBS. The OD
value is a measure of the level of antibody response that the mouse has had to Borrelia
burgdorferi. Therefore, we assume that mice with similar OD values have had similar
levels of exposure to B. burgdorferi, not taking individual immunological differences into
account. Therefore, mice within 66.7 m of each other are likely to have similar exposure

levels to B. burgdorferi, and at distances greater than that, are not likely to be similar.

Aggregation and spatial clustering of ticks

Aggregation of ticks on mice occurred at our field site. Frequency histograms
show aggregated distributions of larvae on mice (Figure 3.21), nymphs on mice (Figure
3.22) and total ticks on mice (Figure 3.23) for first capture of individuals at each time
period. For first time captures, Figure 3.24 describes how the same mouse can be
simultaneously parasitized by both larvae and nymphs. Overall, mice had higher larval
burdens than nymphal burdens. Graph does not differentiate if there was more than one
mouse at a data point.

There were two lifestages each of two species of ticks on the mice. Larvae and
nymphs of both I. scapularis and Dermacentor variabilis were present. For all three of

the above histograms, the variance to mean ratio and k statistic indicate a high degree of

99

aggregation, and the LMC index indicates that the average parasite endures a high degree

 

 

 

 

 

 

ofcrowding.
100
90
N=224
80 V:M=14,680
k=0.066
w 70 LMC=15,660
E
E 60
{I
E
“5 50
2*
E 40
5
30
20
10
0 r IITTII III-l-II-IIIIIIlT--IIIIrlllllTIl‘lI I II-

 

 

15913172125293337414549535761

Number of larval ticks per mouse

Figure 3.21. Frequency histogram for number of larval ticks on mice at first
capture. The v: m ratio and k statistic indicate a high degree of aggregation. The LMC
index indicates that the average parasite endures a high degree of crowding.

100

 

160 -

 

 

 

 

 

 

N=224
V:M=22,895
k=0.1313
w 1204 LMC=25,908
E
E
II
E
.5 80-
h
8
E
3
Z
40-
O‘Ir . fIrlikliIrI'-IlIfllfilIlTrIlll-ll:.

 

14 7 1013 1619 22 25 28 3134 37 40
Number of nymphal ticks per mouse

Figure 3.22. Frequency histogram for number of nymphal ticks on mice at first
capture. The v: m ratio and k statistic indicate a high degree of aggregation. The LMC
index indicates that the average parasite endures a high degree of crowding.

101

 

 

 

 

 

 

 

 

 

60 N=224
V:M=19.6
k=0.376

.02 LMC=25.6

E

'E 40

(6

Ir.—

0

a;

'2

3 20

z
I 0 ll 1. ll l l l-llllllllill-llillrlljllITlllll-lllllllillllfl

0 5 1015202530 354045 5055606570

I Number of ticks per mouse

 

Figure 3.23. Frequency histogram for number of total ticks on mice at first capture.
The v: m ratio and k statistic indicate a high degree of aggregation. The LMC index
indicates that the average parasite endures a high degree of crowding.

102

 

 

35 0

 

30

 

25

 

20

 

 

Larvae

15

 

10 4

 

% 4
e

 

 

0 5 10 15 20
Nymphs

Figure 3.24. Simultaneous parasitism of larval and nymphal I. scapularis on
individual mice. Graph shows the number of larvae and the number of nymphs that an
individual mouse had at first capture. When multiple mice had the same configuration of
larvae and nymphs, the graph still just displays one symbol. Overall, mice had higher
larval burdens than nymphal burdens. Not all mice that had larvae had nymphs, but
nearly all mice that had nymphs also had larvae feeding.

Ticks can be considered to be clustered, or aggregated in space both on and of
hosts. Since we have exact location data for hosts and the data of their tick burdens at
that time, we can create a map of where the ticks are, and thus, see how they aggregate in
space. Tick clustering by lifestage and species was analyzed using kernel density (for a
visual of where the most ticks were) and Moran’s Index to get a degree of clustering.
Table 3.5 and Figure 3.25 show ticks cluster spatially based on both lifestage and species.
Nymphal ticks were mapped using kernel density, and demonstrate a clustered pattern by

lifestage and species as well (Figure 3.26).

103

Table 3.5. Determining spatial autocorrelation of ticks species and lifestage, using
Moran’s Index.

Pattern

      

Figure 3.25. Kernel density of all species of larvae from mice (A) a somewhat clustered
pattern, and of I. scapularis larvae from mice (B) also a clustered pattern. Descriptions
of clustering based on Moran’s Index analysis. Map shows all three grids together.
Darker areas indicate more ticks. Dots show GPS-located trap locations. Figures in this
thesis presented in color.

104

 

Figure 3.26. Kernel density of all nymphs from mice (A) a random pattern, and of I.
scapularis nymphs from mice (B) a clustered pattern, centered on grid IV. Descriptions
of clustering based on Moran’s Index analysis. Map shows all three grids. Darker areas
indicate more ticks. Dots show GPS-located trap locations. Figures in this thesis
presented in color.
Aggregation and spatial clustering of Borrelia burgdorferi

Aggregation of spirochetes in mice occurs (Figure 3.27), as does aggregation of
spirochetes in all ticks found on mice (Figure 3.28). These histograms show, on a log
scale, how most of the sample population contained zero B. burgdofleri organisms, few
to none in the 10-100 range, with a peak in the 10005 for both populations. Although

possible, quantitative PCR was not perfectly accurate at picking up spirochete loads less

than 10 and thus, this bin in nearly empty for both histograms. Both histograms show

105

how the populations are highly aggregated, as indicated by the large variance to mean
ratio and small k. The LMC index indicates that the average parasite in this population

experiences a lot of crowding.

90

80 N = 32

70 V:M = 40,088
k = 0.014

60 LMC = 40,635

Frequency
'8 8 8 8

.3
0

<1%

 

O

0 10 100 1000 10000 100000
Spirochetes (log scale)

Figure 3.27. Frequency histogram of number of mice with numbers of spirochetes.

This population is highly aggregated, as indicated by the large vzm ratio, small k. LMC
index indicates that the average parasite in this population experiences a lot of crowding.

106

 

  
  
   
   
   

 

 

140 —

I nymphs E larvae
‘20 ‘ k=0.1313 k=0.0657 ‘”—'—~
100 - V:M = 22895 V:M =14680 ____

 

LMC = 25908 LMC = 15660

 

 

 

 

 

Number ticks

 

 

 

 

 

1 1% 3%

 

0 10 100 1000 10.000 100,000
Number spirochetes (log scale)

Figure 3.28. Frequency histogram for spirochete loads in all ticks recovered from hosts.
The number of spirochetes is presented on a log scale. For each group, summary
statistics are presented: k parameter for the negative binomial distribution, variance to
mean ratio (v: m), and Lloyd’s mean crowding index (LMC). The k statistic indicates
that the larvae are more highly aggregated. Both sample populations have a v: m ratio to
the same order of magnitude, 10"4, indicating aggregation. The LMC index for nymphs
is higher, showing that that the average spirochete in a nymph on a host feels more
crowded than the average spirochete in a larva feeding on a host.

Further, the method of spirochete detection in mammal tissues is not completely
accurate or reliable due to the nature of where in the bacteria may be in the animal.
Sometimes, even an infected animal will show up negative because the tissue sample that
we have taken is the not area where the bacteria are in the skin, or the mouse was only
recently infected and the spirochetes are still circulating in the blood and have not made it
into the skin. Therefore, this method of detection is not the best for determining which
mice have past or current B. burgdorferi infection. Thus, this measure, combined with

the ELISA to capture antibodies to a past infection with B. burgdorferi gives the best

107

measure of infected mice. This combination of metrics was used to count ‘infected’
mice, which were used as a factor in the spatial analysis. If one of the metrics gave a
positive result, then the mouse was counted as infected.

Maps for datasets were created at each time period. The maps were created using
ArcMap software (ESRI, Redlands, CA). In these maps, the trap points are represented
by dots, and the larger the dots are, the more of that particular organism was found at that
particular point. Scales for each of the maps are listed in the figure legends. These maps
are shown in Figures 3.29A through 3.290. These maps are visuals displaying how the
bacteria, ticks and mammals were clustered in space. Over time, the places where these

organisms were picked up changed.

108

S .. C. .
e O
. 9, . '
' 1
O
f.
'5
o- '-
. O
L-o -
' O ' " O
f g o
. o
. a. .
e . .. . . . .
O J'
. . .. O . 6 c . o
o. a n . .
. I
I .5. . .. .
o
o .. I. .
O ' . O
. i
O
‘ o
u
.0. Q.-
Q
' 2
. 0".
..
O . I.
o ." ’
i- . .‘
‘ " '0 a. '
C '. ..
e- 9 ’ '
e I
l ‘ .- o
.. . 'I . I o . ‘
. . .
. a. ‘0 .. ‘ . o
o ' ,' - .
O ..
a. .3
9 . . .
o
.0
o
\
.0
6
o
O
' "’ 3
O
'c
.0 . go
.0 '.
o o
‘1. . .' .
. o
. .
o l o
o . '-
. . ..
° e
. o
O a
f‘ .
. 9. ' .0
o
' . .V. o ' . . .
O o
. o O
. o l‘
o o o o
I c
0

Figure 3.29A. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represent the number of infected mice captured at those trap

points. Increasing dot sizes represent 0, 1 and 2, respectively.

109

. - .. . ,9
It" '
' . '0.
' _ ,. or
' U
.' ' ‘
I C. .
to ~
.: . . e o .
. ' ‘ . .v
.. . a. .‘ .
. . . . ‘ . . a
o .9 o . . .
' . . . .- . D
a. .... u'
0 .
'0
~ 0
. . D
U
I’_
' o S
O ' t . .
o 9“ e
.0 o
o .' - '.
. . '.u . . . . a
o_ ' . . , - o
o. '
. 0. .' 0‘. .
.0
.'o
' 'U
0?
§
0. ° .
2. ' a.
' I
. 3
' . 0' I.
o . .
‘C ‘ ...‘
.-
0..
'0
. ‘-
I}

Figure 3.29B. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the average adjusted OD values for the mice captured
at those trap points. The smallest dots represent mice with no developed antibody
response to B. burgdorferi, and the large dots represent those that do have antibodies
against the pathogen.

110

.0‘. t
En. -'
.
fl .' . O .
C ' ‘-
e
o . a
. c. ' .
. . o .. .
O .
. C . . . o . O C
O
-' .’ o .
o
o '. .
o
O
'5
O
' 2
I"
f .
3
E. o
. ..
.-
. c
.' o ’,
v
.. ’
a
o 9 ,
a} '
. I
O
\
C
' " I 3
..
q I
‘0 1.
. o
.-
'.
a . .
. U
I
. . O O
o
c .' '
o X
C

Figure 3.29C. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of spirochetes in the mice captured at
those trap points. Increasing dot sizes represent 0, 150, 500, 1500 and 5500, respectively.

111

. .. ,3
.‘ o
b . o
.- . 1
' '0
Q t
. ' " 39. ' °'
.- .. . . - a
O. . .. '
'o o 0 . .1 . o
. e .‘o . , .
._ .‘Os , .
O '. .3. o
'o .
O . r.
c ',.
o- ,. ‘
' .. . '0 e o
s . ‘ - -'5 .
" '.. :0 . .
o .' o
O U ..
o ' . 'o'. o
.0
. . N O.
a".

Figure 3.29B. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of larvae on the mice captured at those
trap points. Increasing dot sizes represent 0, 5, 10, 20 and 35, respectively.

112

o- t
._-o . 0'
o. . .
o . 'o
‘C
. .
. Q 0
C o
C . . . . 0..
‘- . o
O
. ..$ a
C
K
C
' 2
0 a
c ' 9.
. O
' '.o c .
U ' . .
.
. .
a. .. .
. .
o .' " .
. C
..
g '-
I . .
\
a . ‘
’ o
I I. 3
C ‘ .
' . O. .
' C
e 0 . o
I . 0 .
.0 I
o. '
'C
U. ..
3 $
. . .

Figure 3.29E. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of infected larvae on the mice captured at
those trap points. Increasing dot sizes represent 0, 1, 2 and 3, respectively.

113

. .' .0
. O
'.
I o.
. 1
0‘ ‘
':_. 0'
‘ O O. .
O . .0 . ‘0
. .. . . O
. Q o " .-
. O
. Q 0 o .
O .-.. . .
.. . ' .'
O
s.
0
fl
. 2
.C_ r
. Q- , 9
o- ' .
'0
. O . o
. e
. . I O .
'. .. .0 O
. o. 5..
C . . .
' o
a .. .
~ I
O . .
‘ o
o O
.. ‘. .
' . a. .0.
. o ' 0 o 0
° 0
I O
c .‘.
.0
a}

Figure 3.29F. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of spirochetes in larvae on the mice
captured at those trap points. Increasing dot sizes represent 0, 100, 1000, 9999 and
35,000 respectively.

114

.0.
' 1
‘ ,._
u '- . ‘ .'
.' c. .
o . '0 '
. .
. ‘ o o
. . . . I .0
e e .
Q '. . '. o
O
K C
, C
C
0' 2
° C
. o
I
E- .
fl . 0.. .
. O
o_ ' - _'Z .
..' 0
O ' ‘°
C
I I 8 .
‘
a. .00
$
' s
.0
I. .I'

Figure 3.29C. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of nymphs on the mice captured at those
trap points. Increasing dot sizes represent 0, 1, 3, 6 and 20 respectively.

115

Figure 3.29B. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of infected nymphs on the mice captured
at those trap points. Increasing dot sizes represent 0, 1 and 2 respectively.

116

I
...
C
.0. o
C

I ‘5 .

OI
‘ C
..
.. 2
' no ‘5
‘ ‘ .‘.
.: .,
C.‘
\
O
t
o’

8

C
..
I
0,.
O
..

c.‘

C

Figure 3.291. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the number of spirochetes in nymphs on the mice
captured at those trap points. Increasing dot sizes represent 0, 100, 500, 1000 and 37,000
respectively.

117

.o
o . . l' .
‘ .O
, C
. 1
. -'-'
a
u '.
. o. ' " ' ‘
c
. A. o o
O '- 9. ‘ I. .
.. o O ‘ o
o c .. .
o o
, C Q o ‘. .. ..
. a , ’0' '. . ,
0.. ’ . . . °
I o. ' a C . a
o
a
. .. | .
. .
O
\ Q
.
.
c 2
v . .‘.
. O
I ..
O '1.» .
0 ' ‘ c
' . . ". .. 0 ..
o . ‘- ,
. .
o ' ' o
O . I
.. ' ‘.‘o . o 0
0 U .
0.. 0.. . * I.
C ' ,° .
.0
0 ' M‘ .
D
‘ 0.
O ' 3
‘ O
o . '
. . . ‘0. O
' . I D
o ' . '0
o . .' .
o n
. o C
. .0 ‘0'. .
. . .
c o O . o .
. I . .
o I ‘

Figure 3.29J. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the total ticks on the mice captured at those trap
points. Increasing dot sizes represent 0, 3, 8, 15 and 54 respectively.

118

”k .. ..
.- . . 0.. .
~ ‘3- .°
‘00.. '~ " 2."... o
o p 2
. . .0 ...‘o. .
. g. ;' .‘. . .. o
'. 4'0"- .5 r
. «‘9'. '
.-- " .-g ..
' 3

Figure 3.29K. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the questing larvae at those trap points. Increasing
dot sizes represent 0, 2, 5, 16 and 79 respectively.

119

I . '
. 0 II ‘
. I . D.
I. ‘ ' ‘
U
I I I '.. .
U
I
" O
C t
. . I I
‘ I
I 0' I . -
. . ‘ ‘ .g I I
I I
. I.“ .I
I ' I I
9 fl 0 .
I I I ’I
O . . .
I I . I . '
O
I . 4 .

‘
fl I
. . I . I I "
I ‘ I . I
.
'I I '.‘ .
I
I
.I
I
f ’6
. I I
Q ‘0

I . .
9 . I 9
II I. . I I
-. .
. O . I O. I . .
. I O I O I '
I ‘ '
I I‘ . O ' I
. I . I I I
I I I ...
o I I
I O . O
I . A... I 'I . . . I ‘. I ..I
. I
9 I .
. I
. I I
. 0‘ I
. - - r .s -
‘I I
I . 9
I
I
I
I
C ' .
- . ‘ D
0'
I I .
. g . I.
I
.u . I .
I I
’I I I
.v‘
I
I
g I
‘
. I
O
' I
I ‘.
I I
' I
9 0.. ‘ I
I. I
U . ‘
.I. I I .
. t 9. O
‘ I I
I I. . I
I
. I I 'I
I
O I
I I
I
‘ I. ’ I I
" I ‘ . I
. I. I
' ' C I ' I . I .
I I
0 .9 O. .3 I . I .
- . . . . .a . I . I I
I‘0
I . Q 'g . . .
. 0". I I I . . , .
I 9'.
' I
. I
I I I
. II I
. I Q 0 I .‘ '
I
I O Q. C I
I
I
I
I

Figure 3.29L. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the questing nymphs at those trap points. Small and
large dot sizes represent 0 and 1 respectively.

120

I I
I
. I I. ‘ "
I
I ' I 0. ‘ ,
I.“ 'I. .
' I o
O I O
t J . O. . Q
I I I
I . .
.. I ‘ C 0
I. I I
I
. I . I . .
U I I
. I . ‘ I I.9 . .
. I I. 'I ' I . o I . I . I .
I
I . . ’ . " . .D
I I I
I
9 . I.. . Q . I . ' . . I
. I . O . . O
_ I
. ' . C. O
' . I I I I . $ .
. U I I
I .‘ I
I
I
‘
I
I
I . O '
, .0 C ..
. . I ‘ I
" ‘ ' “ 2
I I
'I I I ....
I
l
,. I
'I
I
.I ’0
I
O
' I
I
. . . IQ ‘ 9
0 g 0. ‘ .
'l. I I ..
, I I
I I . .
I v I. . '
I I I I
I . . ' .
I I I
I I . I
.. I
. I . O ' ‘ o
I . . ' I . . I
I . .
. ' .I I I - .
I I I I
I I I ‘ . . ' .I'.
I O
I O O I
- ' I " OI . .. ° ‘ ' . I. . I
I I
I . .
I
.
I I I
. ° . ° . s .
I I . D.
-
O
I
I
I
I
I
5 . .
I
. . II ..
. ' I . I
I ' .I
I I
‘ I I '.. '
I
.
I
I
O
I '9.
I
Q ..
I 0‘ ‘ 'I
I
I U I . ‘ . I
I“ ' ' I
. 0‘ Q Q
I I
I I. . I. . .
I I I I
C - '
. . I . ' I I
I. O I
O . g
I ' . I. I ’ .
' . g . O ' O
I
I I . I
I I. . . I . g I
. I I . I I I
o l I I .
I I I
I I
‘ I '. .. ‘ I. . . ,. 0 . p C
. I
O I
. I
. I I
I I I
‘ I I I ' ‘
I ‘0 I
I ' I
I
I
D
I

Figure 3.29M. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots

represent trap points. Size of dots represents the infected questing nymphs at those trap
points. Increasing dot sizes represent 0, 1 and 2 respectively.

121

O
0 ° '. .s
I
.I . . . .. .
~ . . 'I
.
I
'- 1
O
t ." . ,
0
j ..o. . .0
. O . ¥ . .
. n
' . .. .. o
o
C
O . .
o . .
O. 1
9 . .. n
. . C. . . f .. .
' o ' . . o g o o
O ' .0.
o. o’OQ' 0 o.-
o 00 - ‘ ' O
o .O“ .0
. ..
l. .
i
' o
‘
a
“ 'b
. o
. o
. 2
O
I
'0
. .n ‘
I ... . “ . O
. . O I * '
o O o 0. o
o i ' '.
O . a
o " . ' 0
O
' .03
I . '
é. . a l 0.. . '
. ' 'bfi 0 9° ‘
O . . .
C
. § - o o ..‘
. 0' _
o. ' v:
I ..
°. .
Q ."
. 3
4’ ’-
-‘ ‘
. E. . I
I 4. O
O O. .
. 9
o
. ’. .
o
o . D
. ‘
O . '9‘

Figure 3.29N. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the total questing ticks at those trap points.
Increasing dot sizes represent 0, 5, 17 and 79 respectively.

122

. ..o .0...
D .-
.' o o t. _
.I .
C . . '.
U
- 1
.‘Q . '
'0
. Q ..o‘ . O
. C”. . . ... 4
. . .. - . I.. . .
. . cl .. o 5 . . . . .
C . 9’
. . I ' = o. °“o ' ' '
r -. o o .
U" ' . . ' '0" ' o
. . .‘o. ' 'h . a a . .0
. ° ‘..
. . .. o
o .’
c . ..
t.
‘ .
‘ 5
' 0' 2
. 0' U" ‘
I ...
3.0 o .‘
. C .
‘ '. .'o 00
. 0 . . ' a
. O. . o . .
. ‘ o O o. .l l.“’ q...' . O . Q
C . . Q
.. ..: D~. . '. .‘. .U . '0.
. ~ I. ..‘ . O
S
v . ‘
o E... '
.0 . . .0
v " ° . .
o 4' o ... ‘. O
‘I 0 q.’
. . o a. ‘...
I“ . '

Figure 3.290. Maps for trap periods 1, 2 and 3 displaying all three grids. Dots represent
trap points. Size of dots represents the total questing ticks at those trap points.
Increasing dot sizes represent 0, 250, 1500 and 5500 respectively.

123

To ensure that these aggregated distributions did indeed deviate from a random,

or Poisson distribution, the index of dispersion (ID) for the datasets were compared with

the Chi-square values for datasets with the same sample size. ID values that exceed the

Chi-square value are considered to deviate from the random distribution. These values
all exceeded the Chi-square value for the appropriate n — 1 sample size, and are listed in

Table 3.6.

Table 3.6. Null hypothesis testing for aggregated spirochetes in the following
populations. For each listed population, the index of dispersion, n — 1 and Chi-square
value with (n — 1) degrees of freedom are listed. Populations with indices of dispersion >
the Chi-square value are considered to deviate from a random distribution.

 

 

 

 

 

 

 

 

 

 

Population Index of dispersion n - l Chi-square value
All larval ticks on mice 3816800 260 149.48
Larval ticks, trap period 1, grid III 69580 14 36.12
Larval ticks, trap period 1, grid IV 581532 43 77.42
Larval ticks, trap period 1, grid V 38016 33 63.87
Larval ticks, trap period 2, grid 111 135744 28 56.89
Larval ticks, trap period 2, grid IV 627804 27 55.48
Larval ticks, trap period 2, grid V 192105 27 55.48
Larval ticks, trap period 3, grid III 211360 32 62.49
Larval ticks, trap period 3, grid IV 486127 29 58.3
Larval ticks, trap period 3, grid V 102467 19 43.82
All nymphal ticks on mice 1007380 44 78.75
Nymphal ticks, trap period 1 518000 25 52.62
Nymphal ticks, trap period 2 251037 9 27.88
Nymphal ticks, trap period 3 21344 8 26.13
Questing larvae 158.4 18 42.31
Questing nymphs 209440 35 66.62
‘Questing adults 136800 38 70.71
IVIice, trap period 1, grid HI 35503 13 34.53
Mice, trap period 1, grid IV 8060 20 45.32
Moe, trap period 1, grid V 1785 15 37.7
Mice, trap period 2, grid 111 11700 20 45.32
Mice, trap period 2, grid IV 2990 13 34.53
Mice, trap period 2, grid V 3458 13 34.53
Mice, trap period 3, grid 111 641544 12 32.91
Mice, trap period 3, grid IV 738 9 27.88
Mice, trap period 3, grid V 8590 10 29.59

 

 

 

 

124

 

Analysis of spatial correlation
Spatial analysis

Trap points were located with a GPS unit with an error of approximately 30m.
The traps were spaced at 12 m intervals, so distances cannot be accurately measured on
the map outputs. In the future, traps could be spaced further apart, although this may not
be biologically meaningful, or a more accurate GPS unit is needed. As a result of this,
and because drag tick sample locations were not georeferenced with a GPS unit, false
coordinates were used to derive the number of drag ticks at each trap point for the spatial
analysis portion of the data analysis.

The three grids were rotated through in order to serve as replicates. However,
sample sizes at the individual grid level were too small for meaningful comparisons to be
made. Therefore, possible associations were examined by grouping the grids by time
period in order to increase the numbers of the positive responses per analysis and serving
to increase power. Indeed, the trends within time periods are more similar than the trends
within a single grid over time, suggesting that seasonality is one of the strongest factors
influencing the activity and thus, the aggregation of the organisms in the system.

For the spatial autocorrelation procedure, I used SAS (SAS Institute, Cary, NC).
Latitude and longitude coordinates were used, along with the first capture of an
individual mouse at that time period. Only data related to I. scapularis was used. Other
mammal species, repeat captures, other species of tick were not used. Measured factors

were compared in a pair-wi se fashion to determine which of the pairs were correlated
spatially. These comparisons along with the distances at which they were spatially

correlated with significant coefficient of correlation are listed in Table 3.7.

125

Of the 27 pair-wise comparisons made for each time period, for the purpose of
‘screening’ variables to see which were spatially related, a total of 17 were significant at
any distance. For Figures 3.30 to 3.32, the x-axis shows the distance in decimal degrees
at which the pairs are spatially correlated. The y-axis shows the r value, the coefficient
of correlation. The significant r = 0.1069 is represented by the dashed line. See Methods
for the calculation.

Cross-correlograms for significant associations for trap period 1 were number of
infected mice and conifers (Figure 3.30A) and infected questing nymphs and conifers
(Figure 3.30B). For trap period 2, there were significant associations between deciduous
trees and questing larvae (Figure 3.31A), and conifers and questing larvae (Figure
3313). For trap period 3, significant associations were total ticks and conifers (Figure
3 .32A), number of mice and conifers (Figure 3.32B) and questing nymphs and conifers

(Figure 3.32C).

126

Table 3.7. Comparisons made using the cross-correlograms. The distance at which the
pairs were spatially correlated is listed for each trap period. Units are in decimal degrees.
Distance values listed all had an r value greater than the value of significance, 0.1069,

although individual r values are not listed here. If a 0 is listed, factors were not

correlated at any distance. Numbers smaller than 0.0002 (the minimum lag distance) and
greater than or equal to 0.0036 (the maximum lag distance) were not considered to be
significant. Variables with C‘) were log-transformed for the analysis.

 

 

 

 

 

 

 

 

 

 

 

Factor 1 Factor 2 TP 1 TP 2 TP 3
Understory 0 0 0
Number of mice Deciduous trees 0 0 0
0.0032-
Coniferous trees 0 0 0.0034
Understory 0 0 0
Number of Deciduous trees 0 0 0
0.0004-
infected mice* Coniferous trees 0.0027 0 0
Questing larvae 0 0 0
Questing nymphs 0 0 0
Infected quesitng nymphs 0 O 0
Understory 0 0 0
Deciduous trees 0 0 0
0.0028-
Total ticks on Coniferous trees 0 0 0.0034
mice* Questing larvae 0 0 0
Questing nymphs O 0 0
Infected quesitng nymphs 0 0 0
Understory 0 0 0
0.0018-
Questing larvae* Deciduous trees 0 0.0020 0
0.0018-
Coniferous trees 0 0.0024 0
Understory 0 0 0
Questing
nymphs* Deciduous trees 0 0 0
Coniferous trees 0 0 0.0002
Understory 0 0 O
Infected questing Deciduous trees 0 0 0
0.0021-
nymphs* Coniferous trees 0.0025 0 0
Infected nymphs on mice 0 0 0
Spirochetes in Infected larvae on mice 0 0 0
questing nymphs“ Infected nymphs on mice 0 0 0

 

127

 

 

 

0.4 ‘i" *
*
0.2 ~-
* * *
__.___*_____________*. ________
0.0 --
*
-032 '1'-
*
-o.4 -»
l l l l I
0.000 0.001 0.002 0.003 0.004

Figure 3.30A. Significant cross-correlogram for time 1. The x-axis shows the distance
in decimal degrees at which the pair is spatially correlated. The y-axis shows the r value,
the coefficient of correlation. The significant r = 0.1069 is represented by the dashed
line. Graph compares the spatial correlation between the number of infected mice and
conifers. They are spatially correlated between 0.0004 and 0.0027 decimal degrees.

128

0.2 -~

0.0 -— * *

'001 "b

-0.2 a

 

l | l | l
l | l | I
0.000 0.001 0.002 0.003 0.004

 

Figure 3.30B. Significant cross-correlogram for time 1. The x-axis shows the distance
in decimal degrees at which the pair is spatially correlated. The y-axis shows the r value,
the coefficient of correlation. The significant r = 0.1069 is represented by the dashed
line. Graph compares the spatial correlation between the number of infected questing
nymphs and conifers. They are spatially correlated between 0.0021 and 0.0025 decimal
degrees.

129

 

 

0.2 --
*
a:
_4. ________________________
0.1 -*-
*
1:
* * *
*-
* * *
0.0 ~- *
t a»
*
a»
*
-0.1 -- *
l I l I I
0.000 0.R1 0.%2 0.003 0.004

Figure 3.31A. Significant cross-correlogram for time 2. The x-axis shows the distance
in decimal degrees at which the pairs are spatially correlated. The y-axis shows the r
value, the coefficient of correlation. The significant r = 0.1069 is represented by the
dashed line. Graph compares the spatial correlation between deciduous trees and
questing larvae. They are spatially correlated between 0.0018 and 0.0020 decimal
degrees.

130

 

 

0.2 l
*
*
*
_m__________g _____________
0 1 w
*
'k *
'k ‘k
0.0 w
'k 'k *
'k 'k
'k
*
'k
-0.1 w
-0.2 w *
l l l l 1
0.000 0.001 0.002 0.003 0.004

Figure 3.31B. Significant cross-correlogram for time 2. The x-axis shows the distance in
decimal degrees at which the pairs are spatially correlated. The y-axis shows the r value,
the coefficient of correlation. The significant r = 0.1069 is represented by the dashed
line. Graph compares the spatial correlation between conifers and questing larvae. They
are spatially correlated between 0.0018 and 0.0024 decimal degrees.

131

 

 

*
0.2 --
* *

001 '"'"

*

* * *
*
* *
0.0 -- * *
* *
* *
* *
*

'0-1 ""

l l l l 1

0.000 0.001 0.002 0.003 0.004

Figure 3.32A. Significant cross-correlogram for time 3. The x-axis shows the distance
in decimal degrees at which the pairs are spatially correlated. The y-axis shows the r
value, the coefficient of correlation. The significant r = 0.1069 is represented by the
dashed line. Graph compares the spatial correlation between total ticks and conifers.
They are spatially correlated between 0.0028 and 0.0034 decimal degrees.

132

0.20 l
0.15 + *

__._l. ___________________________
0.10 l
0.05 + * *
0.00 l

-0.05 l * * *

-0.10 l

 

l l l
O . 000 0101 0 . 002 0 . 003 0 . 004

Figure 3.32B. Significant cross-correlogram for time 3. The x-axis shows the distance
in decimal degrees at which the pairs are spatially correlated. The y-axis shows the r
value, the coefficient of correlation. The significant r = 0.1069 is represented by the
dashed line. Graph compares the spatial correlation between number of mice and
conifers. They are spatially correlated between 0.0032 and 0.0034 decimal degrees.

133

 

 

0.15 -- *
*
0.10 --
* *
* *
0.05 ~-
*
*
*
0.00 -- *
l l l l 1
0.000 0.001 0.002 0.003 0.004

Figure 3.32C. Significant cross-correlogram for time 3. The x-axis shows the distance
in decimal degrees at which the pairs are spatially correlated. The y-axis shows the r
value, the coefficient of correlation. The significant r = 0.1069 is represented by the
dashed line. Graph compares the spatial correlation between questing nymphs and
conifers. They are spatially correlated at 0.0002 decimal degrees.

134

0'1 "-

0.1 ""

 

| l l
| l I
0.000 0.001 0.002 0.003 0.004

Distance Between Pairs

Figure 3.33. Cross-correlogram for spatial correlation between deciduous trees and total
ticks on mice at time 3. Significant r = 0.1069, and all values are below that level. Thus,
there is no spatial correlation at any distance.

The remaining comparisons were not spatially correlated. An example of a non-
significant cross correlogram is shown in Figure 3.33, with all values below the

significant r value of 0. 1069. Thus, there is no spatial correlation at any distance. Non-

135

significant correlograms included those with individual values that were above the
significant r value, but lacked a trend in the data points and demonstrated a scattered
‘cloud’ of points where there were points at the same distances where some were above
the level of significance and some were below. Distances at which comparisons were

significant are listed in meters in Table 3.8.

Table 3.8. Distances at which pairs of data were spatially correlated, for significant
associations only. Units listed are meters. Figure column listed figure where cross-
correlogram can be seen.

 

 

 

 

 

 

 

 

 

 

 

 

Factor 1 Factor 2 Figure Distance (m)
Number of mice conifers 3.32B 352 — 377
Number of infected mice conifers 3.30A 44 - 300
Total ticks on mice conifers 3.32A 311 — 377
Questing larvae deciduous trees 3.31A 200 — 222
conifers 3 .3 1B 200 — 266
Questing nymphs conifers 3.32C 22
Infected questing nymphs conifers 3.30B 231 — 275

 

Distances at which correlations were significant ranged from 22 m to 377 m.
Conifers were involved in 6 of the associations and deciduous trees in one. Questing
larvae and questing nymphs were correlated to vegetation variables as were ticks on
mice. Mice themselves were not spatially correlated to any other factor.

Significant associations were further examined using a cross-variogram, for the
purpose of model building and the ability to make predictions of a dependent variable
based on the value of another. This method will allow a model to be developed and thus,
be able to get a prediction when only one variable of interest is available later on. All
pairs with significant cross-correlograms (Table 3.7, Table 3.8, Figures 3.30A to 3.32C)

were further examined by cross-variogram.

136

Only four of these showed significant spatial autocorrelation according to the
cross-variograms. (Figure 3.33A thru 3.33D). No significant spatial regression equations

were produced from the datasets.

137

CHAPTER 4

DISCUSSION
Summary

The aim of this project was to examine the phenomenon of aggregation in the

parasites of the endemic Lyme disease system in Menominee County, and speculate how
this aggregation and spread of the parasites through time and space might contribute to
the maintenance of such a disease cycle. Furthermore, I was able to look at the question
of aggregation in a spatially explicit manner. Through this research I have gained

insights and answers to the following questions.

0 Is aggregation occurring in the parasite populations of Menominee County’s endemic
Lyme disease system?
All parasite population datasets did indeed show aggregation. Spirochetes in
ticks, ticks on mice and spirochetes in mice all handin exceed the cutoff value that

indicates aggregation for each of the metrics with which they were measured.

0 If aggregation is occurring, do habitat factors contribute to this aggregation?

Multiple factors were examined as potential contributors to the aggregation of
the system in addition to being potentially aggregated themselves. These were deciduous
trees, coniferous trees, understory coverage, the number of mice, the number of mice
infected with B. burgdorferi, the antibody response of mice to B. burgdorferi, the number
of spirochetes in the mice, the number of larvae on the mice, the number of infected

larvae on the mice, the number of spirochetes in the infected larvae on the mice, the

138

number of nymphs on the mice, the number of infected nymphs on the mice, the number
of spirochetes in the infected nymphs on the mice, the total number of ticks 0n the mice,
the number of questing larvae, the number of questing nymphs, the number of infected
questing nymphs, the number of spirochetes in the infected questing nymphs and the total
number of questing ticks. These factors were examined spatially to‘determine
relationships in space between these variables. However, coniferous trees were most

often significantly spatially autocorrelated with other populations.

0 Are any of these factors spatially correlated, and what are the possible explanations
for this spatial correlation?

The examined pairs of data are listed in Table 3.7, and significant
correlograms which indicate spatial correlation between variables are shown in Figure
3.30A through Figure 3.32C. In six of these seven significant associations, conifers were
spatially correlated to questing larvae, questing nymphs, infected questing nymphs, total
ticks on mice, number of mice and number of infected mice.

Since both larval and nymphal ticks are dispersed by their hosts, it appears as
though the hosts have a preference for the coniferous trees, producing these positive
associations with these other populations. It makes sense biologically, and for the
transmission of the pathogen, that both the vector species and the host species frequent
the same places. This data show that the infected and uninfected tick vectors and the
mouse hosts are both found in association with coniferous trees.

Crucial to the Lyme disease cycle is the interaction between infected ticks and

naive hosts, to create more infected hosts, and the interaction between infected hosts and

139

naive vectors, to create more infected vectors. This indicates that the spatial nature of

this ecological disease process is critical for its perpetuation.

Discussion

The three grids were rotated through in order to serve as replicates. Indeed, the
trends within time periods are more similar than the trends within a single grid,
suggesting that seasonality is one of the strongest factors influencing the activity and
thus, the ecology and aggregation of the organisms in the system.

Ecological interactions are often dictated by abiotic factors. To say a system
seasonally driven means that a combination of sunlight availability, temperature, air
moisture, precipitation and other factors are cues to the organisms in the system to begin
activity. That activity consists of different things for different classes of organism.
Trees, for example, will begin to grow buds and leaf out as response to sunlight and
temperature cues. Mice come out of hibernation and mate when the seasons change from
winter to spring. Ticks, in response to the right temperature and moisture conditions also
become active in the spring and begin to host-seek. The activities change as the seasons
progress and the abiotic conditions alter.

For ticks, those levels of activity can be tracked by phenology curves. As shown
with the phenology curve for questing ticks at the study site (Figure 3.1), each lifestage of
the tick has a slightly different time at which its activity peaks. The measured activity in
this case is host-seeking behavior. A drop in activity occurs as more and more of those

ticks find hosts and take a bloodmeal or die off. This removes ticks from the cohort of

140

host-seekers, and eventually, as all the ticks find a bloodmeal, there are no more left to
detect in the environment as the measured activity of those ticks has stopped occurrin g.

Each parasite dataset shows aggregation. Spirochetes in ticks, ticks on mice and
spirochetes in mice all demonstrate aggregation according to the variance to mean ratios,
the ‘k’ statistic and Lloyd’s mean crowding index for those populations (Table 3.4). This
holds true whether the groups are lumped by time period or separated by grid.

However, I found LMC no more useful to use than any other measure of
aggregation such as the variance to mean ratio or the ‘k’ parameter of the NBD. In fact, it
tracks fairly closely with the variance to mean ratio in value for all sample populations in
this project (Table 3.4). It is perhaps important to have the ‘parasite’s perspective’.
Looking at this measure for highly aggregated populations may eventually reveal a limit
or a threshold at which the parasites no longer tolerate the crowding and at which their
fitness is decreased.

I had predicted an upper asymptotic threshold for B. burgdorferi spirochete load
in I. scapularis. The upper asymptotic threshold was predicted because the physical
space inside the tick for these bacteria is limited. This number is hypothesized by Wang
(2003) to be above 200,000 but the exact figure is not known. In this dataset, the highest
spirochete burden was 103,429 spirochetes, and was found in a questing, adult male
I. scapularis.

For most of the analyses, the dataset consisted of only I. scapularis species ticks.
However, in order to understand what efi‘ect multi-species parasitism has on the small
mammal population and whether aggregation patterns change when multiple species are

examined, I looked at both I. scapularis and Dermacentor variabilis nymphs together for

141

a kernel density and Moran’s Index analysis. When both species of nymphal tick are
analyzed together, the pattern of clustering is not evident (Table 3.4, Figure 3.25, Figure
3.26).

The study was conducted as such a scale that no differences were detected
between areas of the property in terms of disease risk. Further, areas of aggregation
identified were just that- the areas identified. Certainly there were areas where
aggregation was occurring, but not identified because those places were not sampled on
that date or that particular area may have been missed. No clear area of constant tick and
mouse host aggregation was identified.

The data do indicate that on a larger scale, areas that are preferred by hosts such
as white-tailed deer, mice and ticks will be the areas where patterns of aggregation are
seen. If an area is preferred, populations of these organisms will be higher. These data
show that the higher the population size, the better the chance for aggregation (Figure
3.17 and Figure 3.18). If hosts are frequenting an area, there may be many engorged
larvae there that fell off their host in that spot, contributing to next year’s infected
nymphs. Ifthe area remains favorable, the hosts will visit it again the following season
and then have the opportunity to pick up those infected nymphs. Aggregation effectively
increases the chance of a vector infecting more vectors. If a tick with a high spirochete
burden feeds on a mouse that is also feeding many larvae, or will feed many larvae, that
tick has a better chance of spreading the spirochetes than one with a low load. If only a
few spirochetes are passed to a host, it may be that none can survive since there are too

few to override the host’s immune system.

142

Infection appears to be additive by life stage (Figure 3.3A), where very few, if
any larvae are infected, then up to a third of nymphs are infected and over half of adults
are infected. The previously described aggregation scenario could contribute to this as
each tick picks up more spirochetes with each bloodmeal, and since they have three
chances to become infected by the time they are adults.

Perhaps this is a manifestation of parasite dilution where one species of parasite
subsumes space that the other might use. If a host is located by an I. scapularis nymph,
but is already loaded with feeding D. variabilis ticks, there is less space for the
I. scapularis tick to feed, and vice versa. In this case, if D. variabilis ticks, (which are
non-competent vectors for B. burgdorferi (Piesman and Happ 1997),) are taking feeding
space away from potentially infected I. scapularis ticks, they may serve to dilute the
effect of disease transmission from the competent vector. However, the effect of feeding
D. variabilis ticks did not seem to prevent the aggregation and feeding of I. scapularis
ticks, since that population was strongly aggregated. Since there was no control area
where D. variabilis ticks were not present and feeding, it is not possible at this time to
determine what specific effects those ticks had on the aggregation response of I.
scapularis.

There are multiple potential effects that aggregation may have on B. burgdorferi
transmission dynamics. First, there is the establishment of the pathogen in a foreign
geographic area. It is possible that aggregation may help increase the rate of spread in a
new area, whereby only a few hosts need to be highly infected in order to efficiently
infect more naive vectors. The second potential effect is the maintenance of the disease

cycle in that newly established area. Although a definitive link cannot be proven, this

143

system appears to be a deeply entrenched ecologically-based disease system in tandem
with multiple levels of highly aggregated parasites.

The examined factor that emerged as having the most spatial autocorrelati on was
the most stationary, the vegetation. Spatial spread of plants is determined by exogenous
processes, such as water availability and soil type (F ortin and Dale 2005). The response
of the spatial structure of trees is influenced by exogenous variables whose spatial
autocorrelation is what drives the spatial structure of the tree species. Thus, trees are not
considered to have an independent spatial structure, and instead it is considered that their
spatial structure is a reflection of the other factors their growth relies on, such as bedrock,
soil and water. These assumptions were also held for the understory vegetation. The
vegetation exhibited no movement in the time frame of the project and thus the nature of
these variables (understory, deciduous trees and coniferous trees) is entirely stationary.
One measurement was taken for the vegetation during the first time period and then used
in the analysis for that grid for the remaining time periods. Vegetation may be
aggregated by type due to the exogenous factors of an area and that aggregation may
exert some pressure on the spatial structure of the hosts that require a particular species or
habitat type for food and shelter.

The next least mobile of the examined variables were the questing larval
blacklegged ticks. Since larval ticks hatch in a clutch of eggs, and a clutch may number
in the hundreds or thousands (Main et al. 1982), they begin their lives aggregated. Where
the eggs are placed is a function of where the mother tick has laid them. The mother tick
lays her eggs in the spot she drops off her final host- likely a white-tailed deer

(Odocoileus virginianus), where she has blood-fed and presumably mated with a male.

144

Engorged with blood and now carrying fertilization for her eggs, she drops off the deer
wherever the deer happens to be. Due to her engorged state, she does not have the use of
her legs, and must wait to digest the meal and produce the eggs she lays in that very spot.
Thus, to travel back on the chain of events, the eggs are laid where the deer come to rest.
Deer, like other mammals, prefer and rely on certain habitat types and cover- and this
habitat preference is largely based on vegetation (MDNR 1999).

Once the eggs hatch, the larval ticks cannot move far from that spot when they
begin to host-seek, and rarely move farther than 3 m away (Daniels and Fish 1990). Any
small host that encounters this clutch of newly hatched ticks will likely be parasitized by
multiple ticks. Another host that does not encounter a larval clutch may escape larval
parasitism altogether. This provides the set-up for a highly aggregated system. If a
mouse that encounters the ticks is already infected with B. burgdorferi from a previous
parasitism event with an infected nymph, then most of the naive larvae that feed on that
mouse will also become infected (LoGiudice et al. 2003). If the larvae feed on an
uninfected host, then they will also remain uninfected, and maintain that status until they
become nymphs. Since blacklegged ticks feed only once per lifestage, and only feed on
our host of interest, the mouse, at most twice in a lifetime (as a larvae and then again as a
nymph), if a larval tick does not pick up the B. burgdorferi pathogen on its first go-round,
it will not be able to spread that infection at its nymphal feeding.

Even if a tick becomes infected as a nymph by feeding on a mouse, it is
essentially no longer playing a role in the maintenance of Lyme disease in the
environment. Since many adult ticks feed on white-tailed deer, and the deer are capable

of clearing the pathogen from their system, that tick is not contributing to any more

145

infected hosts. However, the infected nymphs play an important role when we look at
cases of Lyme disease in humans. Infected nymphs are implicated in most cases of Lyme
disease, due to their small size and ability to be easily overlooked for 2 to 3 days while
feeding on a human. Thus, although the infected nymphs are not contributing back into
the LD cycle when they feed on a human (who are dead-end hosts for the pathogen, since
infected humans will not usually be fed on by additional uninfected ticks while
spirochetemic) the pathogen will cause cases of LD in the human population of the
endemic area.

Nymphs are any more mobile than larvae. Nymphs of the Ixodes ricinus species
have been documented to move only about 1 m (Gray 1985). Although they may be able
to climb up on vegetation in order to obtain a better position to find a host, nymphs rarely
move more than 2 111 (F alco 1987). It is assumed that they do not go far from the place
where they dropped off as an engorged larva after a bloodmeal. They are dispersed by
hosts and less aggregated in the environment than larvae (Madhav et al. 2004).

Adult ticks do not feed on mouse hosts, and although they were picked up on drag
cloths, they do not contribute to the maintenance of the cycle in the small mammal hosts,
except that they are able to reproduce and then generate the next cohort of blacklegged
ticks. Even if the mother tick is infected with B. burgdorferi, her offspring will not be, as
B. burgdorferi is not passed transovarially.

These results support the current knowledge of how larval and nymphal I.
scapularis ticks are clustered and dispersed. Larvae and nymphs both showed high levels
of aggregation, and were both considered clustered by Moran’s Index criteria (Table 3.5).

Further, larvae and nymphs did have a similar spatial structure, as they were often both

146

associated with the same hosts (Figure 3.5). Again, this is critical for the transmission
and maintenance of the disease cycle because it allows the interaction of both naive and
infected hosts and vectors.

Table 3.5 and Figure 3.25 show ticks cluster spatially based on both lifestage and
species. The larval clustering may be due to mammals moving through a recently hatched
clutch of eggs. Larval ticks do not have the ability to move very far on their own, and
emerge together in a synchronous fashion. Nymphal ticks were mapped using kernel
density, and demonstrate a clustered pattern by lifestage and species as well (Figure
3 .26).

In order to determine spatial correlation between measured variables in order to
infer that these variables are indeed related spatially, 27 comparisons were made for each
of 3 trap periods, for a total of 81. These comparisons are listed in Table 3.7. Distances
at which pairs were significantly correlated are listed in Table 3.8.

The correlograms for significant associations are shown in Figure 3.30A through
Figure 3.32C. Conifers were the most associated factor, making up half of six pairs of
data. Coniferous trees were spatially correlated with number of mice, number of infected
mice, total ticks on mice, questing larvae, questing nymphs and infected questing
nymphs. Each of these pairs were significant at only one time period.

The data show that larvae were spatially correlated with conifers upwards of
dozens of meters, but typically only travel a maximum of 3 m (Daniels and Fish 1990).
This suggests that the larvae are where they are not because of their own preference, but
because of where the mother tick laid the eggs where the tick dropped off the deer. Deer

do prefer habitat with plentiful edible cedar browse and cover, and prefer to bed down for

147

the night in areas where cover is thickest (MDNR 1999). Thus, their preferences may
belie where the larval ticks end up questing.

The closest association was between questing nymphs and conifers; these are
spatially correlated at 22 m, at time 3. Questing larvae were also associated with
conifers; this may not be surprising given the preference of white—tailed deer in Michigan
to prefer swampy and coniferous habitat (MDNR 1999). This makes sense in light of the
earlier assumption that white-tailed deer are actually responsible for the dispersal of
larvae, and larvae quest from the same spot they hatch.

Deciduous trees were spatially correlated in only one of the pairs, in this case,
with questing larvae. Since the cover of deciduous trees was fairly even, and most points
had at least one deciduous tree, it is reasonable that there was no ‘effect’ detected. The
‘variable’ of deciduous trees was almost a ‘constant’ in this study because of its ubiquity.
Of 240 trap points, only 19, or only 7.9% of trap points did not have a deciduous tree
within 6 m of the point. In other words, 92.1% of trap points had at least 1 deciduous tree
with 6 m.

The study system is an endemic Lyme disease system in Menominee County in
Michigan’s Upper Peninsula. There is thought to have been an established transmission
cycle and blacklegged tick population there since the 1980’s, the evidence for which was
the number of Lyme disease cases in humans reported from the area at that time. The
first confirmed case in Michigan was from Menominee County in 1985 (MG Stobierski,
personal communication). Studies confirmed the presence of the vector population later

on (Strand et al. 1992).

148

In Menominee County, the density of infected nymphs (epidemiologically the

. . . . 2 .
most important life stage of the tick) rs about 6 per 1000 m (county-Wide average),

similar to other endemic areas in the Northeast, such as Westchester County, where the

. . . . . 2 .
densrty of infected nymphs rs approxrmately 8 infected nymphs per 1000 m (Druk-

Wasser et al. 2006). The LD case rate of Menominee County, about 50 people per
100,000 per year, is similar to Westchester County New York which also has about 50
cases per 100,000 per year (New York State Department of Health 2005).

Although much of what we have learned about this disease cycle agrees with what
is understood about the cycle in other endemic areas of the US, there are still some
notable differences. First, the habitat structure in Menominee County is different. In the
northeast, the preferred habitat type for the hosts of I. scapularis is oak or maple-
dominated forest with sandy soil (Ostfeld et al. 1996) and this habitat type is widespread
in that area. Furthermore, in the northeast, increased understory has been associated with
the small mammal hosts of I. scapularis (Prusinski 2006).

In Menominee County, we primarily observed lowland cedar swamp, interspersed
by farmland. The study site was most similar to the northeast sites in that it was maple
and oak dominated with no standing water and sandy soil. However, the majority of the
county is either cedar swamp or farmland. Since I. scapularis cannot survive in standing
water, which is common in swampy lowland areas, these areas do not seem to be suitable
habitat. However, drag-sampling to assess the presence of blacklegged ticks in this
habitat type was still successful and I. scapularis specimens of all life stages were

recovered.

149

Secondly, the phenology of the tick’s life stage in Menominee is different from
what is observed in the northeastern US. In the northeast, the adults have a peak of host-
seeking activity in the spring and again in the fall, but the fall peak tends to be larger.
Further, the nymphs are out feeding in spring and early summer, followed by the larvae
throughout the summer (Fish 1995). This set up seems to be ideal for allowing infected
nymphs to transmit the pathogen to as many hosts as possible before the naive larvae
come out to feed, increasing a larva’s chances of picking up B. burgdorferi.

Conifers were the factor most frequently positively associated with other
populations. The distances at which these pairs were correlated are listed in Table 3.8.
Presumably, nymphal ticks are dispersed on the landscape by vertebrate hosts (Madhav
et al. 2004). Nymphal ticks are not constrained by hatch site as are the larvae, and hence
their clustering may have more to do with their fitness and host factors such as habitat
preference.

Even though conifers were positively associated with the presence of the tick and
its mouse hosts, this statement cannot be applied to all areas of Menominee County.
Given that Menominee County is largely swampy, it is surprising to see that these ticks
thrive here. Typically associated with higher and drier areas in the Midwest (Guerra
et al. 2002), the ticks are clearly surviving and having an impact in Menominee. So, what
is it about this place that makes it so ideal for this tick and pathogen? Perhaps it is the
interspersion of farmland and forest creating forest fragments where deer and mice thrive
without the presence of many of their predators. This is one of the reasons posited for the
thriving populations of I. scapularis and B. burgdorferi in northeastern US habitats

(Allen et al. 2000, Ostfeld and Keesing 2000). These forest fragments may be too small

150

for the predators of these animals to live, so the populations of white-tailed deer and
white-footed mice are allowed to grow relatively unchecked. These hosts are ideal for
the survival of I. scapularis ticks and B. burgdorferi, so the disease cycle is maintained.

The study site was a mix of hardwood and conifers, and is located near a river but
with sandy soil. It may be the mix of oak and maple in the drier areas combined with
hemlocks and cedar in the wetter areas to provide both the habitat that the deer prefer
with plenty of cedar browse, with the habitat the ticks prefer, with dry, sandy soils.

In Menominee, the lowland cedar swamps certainly do house ticks, but because of
the frequent standing water in these areas, they are not ideal habitat for I. scapularis.
These areas may actually be limiting where the ticks can survive, and perhaps help
explain why Lyme disease has not spread far beyond Menominee County’s borders in the
past 30 years.

Closer examination of the data reveals additional deviations from Lyme disease
cycle patterns in the northeastern US. In Michigan, a different pattern in the phenol ogy
of the life stages has been observed in this data and others (Walker and Hamer, personal
communications). First, the bi-modal adult peak does exist, but the spring peak seems
larger than the fall peak. Additionally, the nymphs and larvae are out feeding at around
the same time; there is not a large nymphal peak preceding the larval peak (Figure 3 . 1).

The data were analyzed to see if it fit the 20/80 distribution described in the
literature (W oolhouse et al. 1997) (i.e., if 20% of the ticks account for 80% of the
infections). In order to see if this could hold true for this dataset, the number of questing
ticks in the upper 20% of spirochete burden was calculated. Ticks off hosts were not

used because of the confounding factor that we are unable to know if a tick was infected

151

on this host or a previous host. The number that comprised 80% of infected mouse hosts
was also calculated. In this scenario, we only use nymphs, since they are the only life
stage capable of infecting a mouse host.

Our drag-sample size of nymphs from the 3 grids was low, only 36 individuals,
and the upper 20% infected was only 7 nymphs. It is not then possible for that small
number of nymphs to infect 80% of our mouse population, which are approximately 121
animals if each tick takes only one bloodmeal. Therefore, these data do not support the
20/80 hypothesis for ticks and infected mouse hosts. However, 22% of the nymphs do
have spirochetes and spirochetes are aggregated in questing nymphs. Even though the
data do not necessarily support this theory, it is important to note that we certainly did not
detect all the questing nymphs. However, we can use this data to make a prediction.
Based on the rate of infectivity of questing nymphs (22%), if a mouse had 5 nymphs
feeding on it, we could predict that at least one of those ticks was positive and thus, the
mouse would also become infected with B. burgdorferi.

Other studies documenting aggregation of B. burgdorferi and I. scapularis have
not measured aggregation of B. burgdorferi in the vertebrate host. To our knowledge,
this is the first study that does. That said, it is important to acknowledge that that
particular quantification aspect poses potential problems. Since quantification was done
on an ear tissue biopsy sample from the animal, it represents only a small portion of the
total animal, but is not a representation of the total spirochete load in the animal.

Secondly, spirochete loads may not be evenly distributed throughout the
vertebrate body; it is assumed that since most ticks will feed around the head and neck

area where the blood vessels are close to the surface of the skin that taking a

152

representative sample from that area would be most appropriate; however, no ‘control’
samples were taken from other areas of the body and quantified by spirochete load.
Further, if a mouse is newly infected, the spirochetes will still be circulating in the
bloodstream and may not yet have had time to migrate to the skin (Hanincova 2008).

Additionally, it may be that the ELISA test was not the most sensitive or specific
test that could have been run. The antigen used was sonicated whole-cell B31 Borrelia
burgdorferi, specifically, a strain that produces a lot of OspC. The strain in circulates in
all Lyme endemic areas, however, it may not be the prevalent circulating strain here in
the Midwest. It may be that the antibodies produced by the mice here are not as specific
to the OspC in this strain as they are to the strain or strains that infected them. This calls
the sensitivity of the ELISA test into question, and it should be acknowledged that the
positive antibody response prevalence could have been underestimated as a result.

In order to get around these limitations, mouse infection status was determined
based on a combination of infected larvae pulled off of them, the number of spirochetes
measured in their tissues and their antibody response to prior infection with
B. burgdorferi.

Not all potential factors were examined with respect to the aggregated populations
in the system. There may be some factors not examined in this study, such as
immunological variables or grooming behavior, which may play a role in either
promoting or deterring tick presence. Including these factors in future studies may

provide more insight to this disease system’s dynamics.

153

Conclusion

There is a delicate balance for parasites between aggregating and evenly
dispersing. Ifthey are too evenly dispersed, and each host gets only a few parasitic
organisms, chances of survival of that parasite are not good if that number of parasites is
overwhelmed by the host immune system. On the other extreme, if the parasites are
highly overdispersed, they will be crowded into only a few hosts; that parasite burden
may be too much for those hosts and the hosts instead may succumb to the infection. If
the host is killed off too quickly to allow transmission of the parasite to a new host, the
parasite has also lost out on an opportunity to propagate. However, this may not be the
case for the LD system. Since the wildlife reservoirs never show evidence of disease
even though they are infected, morbidity may not play a role. Instead, it could be the
parasites that suffer, and too many in one host may limit their fitness and ability to find
their way into a new tick vector.

It is assumed that the parasites in this system have achieved a balance. All of the
parasite populations are aggregated, but not so aggregated that they have killed off their
hosts before they can be passed on to other hosts. Further, the evidence for a balanced
system is in the case rates of human disease which have stayed fairly constant in the past
20 years.

Several clear conclusions have come of this research. First, Menominee County
has a unique, isolated endemic Lyme disease system. Aggregations of ticks and
spirochetes are spatially associated and driven by habitat, especially proximity to

coniferous trees. Some of these associations are surprising based on what we know about

154

other endemic LD systems, where oak-dominated habitats are the most suitable
I. scapularis habitat (Glass et al. 1994).

Spatial work like this may allow predictions based on habitat type and
aggregations of spirochetes, ticks and reservoir hosts to help predict risk. Ability to
predict disease risk can lead to an increase in prevention and thus be used to target public
health messages. For many emerging diseases, as well as invasive species, rarely are
there baseline data against which to measure rates of spread. These data represent a
current snapshot of Lyme disease cycle dynamics in Menominee County and provide data
for future comparisons.

In order to target public health messages, it is important to be able to identify
areas and populations of highest risk and then focus the efforts there. The predictions
here that conifers positively predict the presence of infected questing nymphs in
Menominee County could be an important one. Nymphs are the most epidemiologically
important lifestage (Barbour and Fish 1993 ), and the habitat heterogeneity of the endemic
county indicates that not all of the area is high-risk. Even infected mouse hosts are ’
positively associated with these conifers, putting people at risk by exposing more naive
vectors to the pathogen. Therefore, the predictions made from the data may be
information that can be used by agencies and citizens in the state to measure risk to
companion animals, domestic animals and humans.

This research sheds light on areas of Menominee County where people should be
especially vigilant about tick prophylaxis. Wooded areas of mixed oak-dominated sandy
upland and low cedar swamp are perhaps the most dangerous for encountering an

infected questing I. scapularis nymph. Aggregation, in combination with the spatial

155

distribution of the pathogen, vector and host organisms is the driver of this disease cycle
in Menominee County. The ecological factors contributing to this system provide a
deeply entrenched disease cycle that allows for its perpetuation. Avoidance of these
high-ri sk habitats and other prophylactic behaviors such as tucking pants into socks, the
use of acaricides and checking oneself for ticks regularly are the norm among
Menominee County citizens. Without drastic changes to the ecosystem, it will not be
possible to disrupt the disease cycle. Until our understanding of Lyme disease system
dynamics increases even further, we should continue to emphasize vigilance in hi gh-risk

areas.

156

0.92 ' °

Variogram 0_ 82 . °
TP 1 0 l
072 i

0.62

0.52 .

0.42 .
0.92

 

002 o l
Variogram ' t

TP2 072 . i

0.62 .
0.52 °

0.42 Na. ---
0.92 °

 

0.82 .

Variogram 0.72 . ‘

TP3 '

0.62 .

0.52 °

 

 

0.42
0 100 200 300 400

Distance in meters

Figure A.1. Variograms for deciduous tree data for all trap periods. The variograms
are the same because data was taken once and used for all 3 trap periods.

157

Variogram
TP 1

Variogram
TP 2

Variogram

TP3

1.8

1.5

1.2

0.9

0.6

0.3
1.8

1.5

1.2

0.9

0.6

1.5

1.2

0.9

0.6

0.3

 

 

 

 

 

200 300 400

Distance in meters

0 100

Figure A.2. Variograms for coniferous tree data for all trap periods. The
variograms are the same because data was taken once and used for all 3 trap periods.

158

1.05

Variogram 0.9 ' ‘
TP 1 °
0.75

0.6 -
0.45

0.3
1.05

 

0.9 .
Variogram .

TP 2 0.75

0.6 .

0.45

 

 

0 .3 _._..__ ,- .___--. --
1 .05
O .9 ‘ °

Variogram 0.75

TP3 . °

0.6

0.45

 

0.3

 

0 100 200 300 400

Distance in meters

Figure A.3. Variograms for understory data for all trap periods. The variograms are
the same because data was taken once and used for all 3 trap periods.

159

0.31

Variogram 0.275

TP 1
0.25

0.225

0.2

0.175
0.205

0.19
Variogram

TP2

0.16

0.135

012
0.31

0.275

Variogram

TP3

0.25

0.225

0.2

0.175

0.175

 

 

 

100 200 300 400

Distance in meters

Figure A.4. Variograms for number of mice data for all trap periods.

160

0.15

Variogram 0.142
TP 1

0.128
0.115
0.105

0.09
0.125

0.11
Variogram

TP 2 0.095

0.08
0.065

0.05
0.054

0.50

Variogram
TP 3

0.46

0.42

0.38

0.034

 

 

 

 

F

 

 

100

200 300 400

Distance in meters

Figure A.5. Variograms for number of infected mice data for all trap periods.

161

Variogram
TP 1

Variogram
TP 2

Variogram

TP3

Figure A.6.

0.54

0.50

r
.. Vfiw.....m-__.c.— -_..._‘ a v..._..——.-'

0.46 ' . g
(142 °

0.38

 

0.34 -,._ . , - . ,
0.31 '

0.28
0.25 ° .

0.22 ' ° '
0.19

0.16 .
0.60 E

040 g.
(L36 5 .
(124 §

0422;

 

(10 5 °
0 100 200 300 400

Distance in meters

 

Variograms for average adjusted OD values for all trap periods.

162

Variogram
TP 1

Variogram
TP 2

Variogram
TP 3

Figure A.7.

0402 . .
r1095 , . . .
(1088
(1081 .
r1074 .

l

r

_ l

- fiu-v-y—v- _ ..--~»r . .. »‘ >0 r r—n _. --...‘ o-r -~o« . -. “mwfl. v“ o --..-...-._.,_- f --v-..‘...._ -- w J

I

. l

 

048 . g
042 - ° .
(136

0.30

 

 

 

0.24 ~44_ - - (M __‘

0.75
0.6
0.45 .

0:3 . . . . .

 

 

0 100 200 300 400

Distance in meters

Variograms for spirochetes in mice for all trap periods.

163

0.156

Variogram 0.149 . ' '
TF’1

 

0.142 .

0.135

0.128

 

0.121
0.137

0.131
Variogram

TP2 0,25 . . l

 

0.119 . 5
0.113

0.23 -

0.19

Variogram
TP 3

0.15

0.11 '

0.07

 

0.03

 

0 100 200 300 400

Distance in meters

Figure A.8. Variograms for larvae on mice for all trap periods.

164

 

0.120
Variogram 0105 . . . _ t
TP1 -
0.090

0.075 ~
0.060

0.450 f
0.120 1

 

0.105 ' '
Variogram =. .

TP 2 0.090 ‘

0.075

0.060

 

 

0.450 " . ' ' . -
0.97

 

0.93

Variogram
TP 3

0.89

0.85 '

0.81

 

0.77

 

0 100 200 300 400

Distance in meters

Figure A.9. Variograms for infected larvae on mice for all trap periods.

165

 

6.0

Variogram 5.0 . . .
TP 1
4.0

3.0

1.0
6.2

5.8
Variogram

TP2 , ° -

5.0 '

4.6

 

 

I u 2 3. ._’..'.'-..—. -q—w—--u~.»H —— w -o v -o ‘ - -.~.-—-~ o—ov—v—o-v-o‘ ow—Ho- cw. -- r-9+ o C—v-V-C-vv-w

3.7; .

3.3

Variogram .
TP 3

2.5 .

2.1

 

1.7 ° 1
0 100 200 300 400

 

Distance in meters

Figure A.10. Variograms for spirochetes in larvae on mice for all trap periods.

166

0.50

Variogram Q45
TP 1

0.40
0.35
0.30

0.25
0.32

0.29
Variogram

TP 2 0.26

0.23
0.20

0.17
0.52

0.44

Variogram
TP 3

0.36

0.28

0.20

0.12

 

 

 

 

e
g e
o
0 e
9 e
e
o
e . e
0
e
e
o
e
'v ~v-v—>>v~-v‘<1>-'U4"v‘ v --v m V'T'C“y‘ r—v-v - 4 'v wyv n V
I
e
e
e
. e
e
e
O
O
O
o
0
o
e
e
e
e
l
a
e
I . . . .
..........................................................................................

Distance in meters

Figure A.11. Variograms for nymphs on mice for all trap periods.

167

Variogram
TP 1

Variogram
TP 2

Variogram
TP 3

0.028

0.024

0.020

0.016

0.012

0.016

0.013

0.0.10

0.007

0.004

0.001
0.022

0.018

0.014

0.010

0.006

0.002

 

 

 

 

 

0 100

200 300 400

Distance in meters

Figure A.12. Variograms for infected nymphs on mice for all trap periods.

168

 

0.54

Variogram Q47
TP 1 . ,
0.39

0.31 . .

0.23 , 1

0.15 .
0.015 . '

 

0.012 ' . ° ' . -
Variogram

TP 2 0.009

0.006
0.003

0.000
0.92 '

0.82 °

Variogram
TP 3

0.72

0.62

0.52

 

0.42

 

0 100 200 300 400

Distance in meters

Figure A.13. Variograms for spirochetes in nymphs on mice for all trap periods.

169

1.9

Variogram 1.75
TP 1

1.6
1.45
1.3

1.15
1.04

0.98
Variogram

TPZ 092

0.86
0.80

0.74
0.95

0.70

Variogram
TP 3

0.55

0.40

0.25

0.10

 

 

 

 

100 200 300

Distance in meters

Figure A.14. Variograms for total ticks on mice for all trap periods.

170

400

0.11 o ' .

Variogram 0.09
TP 1

0.07
0.05

0.03

 

. ,
0.25

0.21
Variogram

TP2 0.17

0.13
0.9

0.12 '

0.10

Variogram 0.08

TP 3 °
0.06

0.04 . .

 

0.02

 

0 100 200 300 400

Distance in meters

Figure A.15. Variograms for questing larvae for all trap periods.

171

0.12

Variogram 0.10
TP 1

0.08
0.06
0.04

0.02
0.05

0.04
Variogram

TP 2 0.03

0.02

0.01

0.45

0.36

Variogram
TP 3

0.27

0.18

0.09

0.00

 

 

 

 

 

i
100 200 - 300 400

Distance in meters

Figure A.16. Variograms for questing nymphs for all trap periods.

172

 

 

0.11 '
Variogram 009
TP1
0.07 °
0.05 . '
0.03 - . ' . i
0.01 _ , - , - , , ..-... 2-3
0.15 - '
0.12 - , . .
Variogram
TP2 0.09 .
0.06
0.03
0.00 .. .
Variogram
TP3 000 -e e e e e e o

 

 

0 100 200 300 400

Distance in meters

Figure A.17. Variograms for infected questing nymphs for all trap periods.

173

Variogram 1.0
TP 1

0.8
0.6
0.4

0.2
2.3‘ --

 

Variogram
TP 2

 

1.4
1.1

0.8
0.10

0.08

Variogram 0.06

TP3 , .

0.04 . .

0.02

 

0.00

 

0 100 200 300 400

Distance in meters

Figure A.18. Variograms for all questing ticks for all trap periods.

174

 

13 °

Variogram 12
TP1 ' ' .

11

10

 

Variogram
TP 2

(DO

Variogram

TP3

 

 

0 ""2 - e
0 1 00 200 300 400

Distance in meters

Figure A.19. Variograms for spirochetes in questing ticks for all trap periods.

175

REFERENCES

Anderson JF, Duray PH, et al. 1987. Prevalence of Borrelia burgdorferi in white-footed
mice and Ixodes dammini at Fort McCoy, Wisconsin. Journal of Clinical
Microbiology 25(8): 1495-1497.

Anderson RM and May RM. 1978. Regulation and stability of host-parasite population
interactions: I. Regulatory processes. Journal of Animal Ecology 47: 219-247.

Barbour AG and Fish D. 1993. The biological and social phenomenon of Lyme disease.
Science. 260: 1610-1616.

Begon M, Bennett M, et al. 2002. A clarification of transmission terms in host-
microparasite models: numbers, densities and areas. Epidemiology of Infections
129: 147-153.

Bliss CI and Fisher RA. 1953. Fitting the negative binomial distribution to biological
data. Biometrics 9: 176-199.

Bunikis J, Tsao J, et al. 2004. Population of Peromyscus leucopus mice: a longitudinal
study in an area where Lyme borreliosis is highly endemic. Journal of Infectious
Diseases 189: 1515-1523.

Bunikis J, Garpmo U, et al. 2004. Sequence typing reveals extensive strain diversity of
the Lyme borreliosis agents Borrelia burgdorferi in North America and Borrelia
afzelii in Europe. Microbiology 150: 1741 — 1755.

Cattadori IM, Boag B, et al. 2005. Peak shift and epidemiology in a seasonal host-
nematode system. Proceedings of the Royal Society B-Biological Sciences
272(1568): 1163-1169.

Centers for Disease Control. www.cdcgov. 2005.

Churcher TS, Ferguson NM, et al. 2005. Density dependence and overdispersion in the
transmission of helminth parasites. Parasitology 131: 121-132.

Crofton HID. 1971. A model of host-parasite relationships. Parasitology 63: 343-3 64.

Diehl RH, Larkin RP, et a1. 2003. Radar observations of bird migration over the Great
Lakes. Auk 120(2): 278-290.

Diuk-Wasser MA, Gatewood AG, et al. 2006. Spatiotemporal patterns of host-seeking

Ixodes scapularis nymphs (Acari : Ixodidae) in the United States. Journal of
Medical Entomology 43(2): 166-176.

176

Drew ML, Loken KI, et al. 1988. Ixodes dammim' occurrence and prevalence of infection
with Borrelia spp in Minnesota. Journal of Wildlife Diseases 24(4): 708-710.

Duerr HP, Dietz K, et al. 2003. On the interpretation of age-intensity profiles and
dispersion patterns in parasitological surveys. Parasitology 126:87-101.

Dye C. 1992. The analysis of parasite transmission by bloodsucking insects. Annual
Review of Entomology 37: 1-19.

Elliot IM, 1977. Statistical analysis of samples of benthic invertebrates. Ambleside,
Freshwater Biological Association.

Elston DA, Moss R, et al. 2001. Analysis of aggregation, a worked example: numbers of
ticks on red grouse chicks. Parasitology 122: 563-569.

F alco RC. 1987 . Abundance of the deer tick, Ixodes dammim' (Acari: Ixodidae), in a
Lyme endemic area of southeastern New York state. PhD dissertation, Fordham
University.

Falco RC and Fish D. 1988. Ticks parasitizing humans in a Lyme disease endemic area
of southern New York State. American Journal of Epidemiology 128(5): 1146-
1152.

Falco RC and Fish D. 1992. A comparison of methods for sampling the deer tick, Ixodes
dammim', in a Lyme disease endemic area. Experimental & Applied Acarology
14(2): 165-173.

Fish D. 1995. Environmental risk and prevention of Lyme disease. American Journal of
Medicine 98: 82-S9.

Foster E. 2004. Ixodes scapularis (Acari: Ixodidae) and Borrelia burgdorferi in
southwest Michigan: population ecology and verification of a geographic risk
model. Entomology. East Lansing, MI, Michigan State University: 87.

Friedrich T. 2003. Ecology of sympatric species of Peromyscus as hosts for Ixodes
scapularis and Borrelia burgdorferi in northern Michigan forests. Ecology and
Evolutionary Biology. Ann Arbor, MI, University of Michigan: 123.

Gaba S, Ginot V, et al. 2005. Modelling macroparasite aggregation using a nematode-
sheep system: the Weibull distribution as an alternative to the negative binomial
distribution? Parasitology 131: 393-401.

Glass GE, Amerasinghe FP, et al. 1994. Predicting Ixodes scapularis abundance on

white-tailed deer using geographic information systems. American Journal of
Tropical Medicine and Hygiene 51: 538-544.

177

Gordis L. 1996. Epidemiology. Philadelphia, W.B. Saunders Company.

Gray J S. 1985. A carbon dioxide trap for prolonged sampling of Ixodes ricinus
populations. Experimental Applied Acarology 1: 35-44.

Gregory RD and Woolhouse ME]. 1993. Quantification of parasite aggregation: a
simulation study. Acta Tropica 54: 131-139.

Guerra M, Walker E, et al. 2002. Predicting the risk of Lyme disease: habitat suitability
for Ixodes scapularis in the North Central United States. Emerging Infectious
Diseases 8(3): 289 - 297.

Guyatt HL, Smith T, et al. 1994. Aggregation in schistosomiasis - comparison of the
relationships between prevalence and intensity in different endemic areas.
Parasitology 109: 45-55.

Hamer SA, Roy PL, et al. 2007. Zoonotic Pathogens in Ixodes scapularis, Michigan.
Emerging Infectious Diseases 13(7): 1131-1133.

Hamer S. 2008. Invasion of blacklegged ticks in Michigan. [Dissertation- in progress].

Hanincova K, Kurtenbach K, et al. 2006. Epidemic spread of Lyme borreliosis,
northeastern United States. Emerging Infectious Diseases 12(4): 604-611.

Hanincova K, Ogden NH, et al. 2008. Fitness variation of Borrelia burgdorferi sensu
stricto strains in mice. Applied and Environmental Microbiology 74(1): 153-157.

Hudson PJ, editor. 2002. The Ecology of Wildlife Diseases. lst ed. New York, NY:
Oxford University Press. 197p.

Jackson J0 and Defoliart GR. 1970. Ixodes scapularis Say in northern Wisconsin.
Journal of Medical Entomology 7(1): 124-125.

Kirby AD, Smith AA, et al. 2004. Rising burden of immature sheep ticks (Ixodes ricinus)
on red grouse (LagOpus lagopus scoticus) chicks in the Scotti sh uplands. Medical
and Veterinary Entomology 18(1): 67-70.

Kurtenbach K, Hanincova K et al. 2006. Fundamental processes in one evolutionary
ecology of Lyme borreliosis. Nature Reviews in Microbiology 4(9): 660-669.

Lane RS and Loye JE. (1989). Lyme disease in California: Interrelationship of Ixodes

pacificus (Acari: Ixodidae), the western fence lizard (Scelopoms occidentalis),
and Borrelia burgdorferi. Journal of Medical Entomology 26(4): 272-278.

178

Leung B. 1998. Aggregated parasite distributions on hosts in a homogeneous
environment: examining the Poisson null model. International Journal for
Parasitology 28(11): 1709-1712.

Levin M, Papero M, et al. 1997. Feeding density influences acquisition of Borrelia
burgdorferi in larval Ixodes scapularis (Acari: Ixodidae). Journal of Medical
Entomology 34(5): 569-572.

Levin ML and Fish D. 1998. Density-dependent factors regulating feeding success of
Ixodes scapularis larvae (Acari : Ixodidae). Journal of Parasitology 84(1): 36-43.

Levin ML and Fish D. 2000. Acquisition and coinfection and simultaneous transmission
of Borrelia burgdorferi and Ehrlichia phagocytophila by Ixodes scapularis ticks.
Infection and Immunity. 68(4): 2183-2186.

Levine J, Wilson ML, Spielman A. 1985. Mice as reservoirs of the Lyme disease
spirochete. American Journal of Tropical Medicine and Hygiene 34: 355-360.

Liveris D, Gazumyan A, et al. 1995. Molecular typing of Borrelia burgdorferi sensu lato
by PCR-restriction fragment length polymorphism analysis. Journal of Clinical
Microbiology 33(3): 589-595.

Lloyd M. 1967. Mean crowding. Journal of Animal Ecology 36(1): 1-30.

LoGiudice K, Ostfeld R, et al. 2003. The ecology of infectious disease: effects of host
diversity and community composition on Lyme disease risk. Proceedings of the
National Academy of Science USA 100(2): 567-571.

Lord CC, Barnard B, et al. 1999. Aggregation and distribution of strains in
microparasites. Philosophical Transactions of the Royal Society of London Series
B-Biological Sciences 354(1384): 799-807.

Madhav N, Brownstein J S, et al. 2004. A dispersal model for the range expansion of
blacklegged tick (Acari: Ixodidae). Journal of Medical Entomology 41(5): 842-
852.

Main AJ, Carey AB, et al. 1982. Immature Ixodes dammini (Acari: Ixodidae) on small
mammals in Connecticut, USA. Journal of Medical Entomology 19: 655-664.

Mather TN. 1993. The dynamics of spirochete transmission between ticks and
vertebrates. Pages 43-60 in HS Ginsberg, editor. Ecology and environmental
management of Lyme disease. Rutgers University Press, Piscataway, NJ.

Mather TN, Nicholson MC, et al. 1996. Entomological correlates of Babesia micronoti

prevalence in an area where Ixodes scapularis (Acari: Ixodidae) is endemic.
Journal of Medical Entomology 33(5):866—870.

179

May RM and Anderson RM. 1978. Regulation and stability of host-parasite population
interactions: H. Destabilizing processes. Journal of Animal Ecology 47: 249-267.

Michigan Department of Natural Resources (MDNR). 1999. White-tailed deer: species
management.
http://www.michigandnr.com@ublications/pdfs/huntiggwildlifehabitat/Landowner
s_Guide/Sgaecies_Mgmt/Deer.htm.

 

Morrison TB, Ma Y, et al. 1999. Rapid and sensitive quantification of Borrelia
burgdorferi-infected mouse tissues by continuous fluorescent monitoring of PCR.
Journal of Clinical Microbiology 37(4): 987-992.

New York State Department of Health. 2005.
http://www.health.statenv.us/statistics/chac/general/lvmehtm.

Owen J, Moore F, et al. 2006. Migrating birds as dispersal vehicles for West Nile virus.
Ecohealth 3(2): 79-85.

Ostfeld RS, Jones CG, et al. 1996. Of mice and mast: ecological connections in eastern
deciduous forests. Bioscience 46: 323-330.

Ostfeld RS and Keesing F. 2000. Biodiversity and disease risk: the case of Lyme disease.
Conservation Biology 14(3): 722-728.

Pacala SW and Dobson AP. 1988. The relationship between the number of parsites/host
and host age: population dynamic causes and maximum likelihood estimation.
Parasitology 96: 197-210.

Pal P and Lewis JW. 2004. Parasite aggregations in host populations using a reformulated
negative binomial model. Journal of Helminthology 78(1): 57-61.

Paterson S and Lello J. 2003. Mixed models: getting the best use of parasitologi cal data.
Trends in Parasitology 19(8): 370-375.

Patrican L. 1997. Absence of Lyme disease spirochetes in larval progeny of naturally
infected Ixodes scapularis (Acari: Ixodidae) fed on dogs. Journal of Medical
Entomology 34(1): 52-55.

Piesman J and Spielman A. 1979. Host association and seasonal abundance of immature
Ixodes dammini in southeastern Massachusetts. Annals of the Entomological
Society of America 72: 829-832.

Piesman J, Donahue JG et al. 1986. Transovarially acquired Lyme disease spirochetes
(Borrelia burgdorferi) in field-collected larval Ixodes dammini (Acari: Ixodidae).
Journal of Medical Entomology 23(2): 219.

180

Piesman J and Happ CM. 1997. Ability of the Lyme disease spirochete Borrelia
burgdorferi to infect rodents and three species of human-biting ticks (blacklegged
tick, American dog tick, lone star tick) (Acari: Ixodidae). J oumal of Medical
Entomology 34(4): 451-456.

Pinger R, Timmons RL, et al. 1996. Spread of Ixodes scapularis (Acari: Ixodidae) in
Indiana: Collections of adults in 1991-1994 and description of a Borrelia
burgdorferi-infected population. Journal of Medical Entomology 33(5): 852-855.

Prusinski MA, Chen HY, et al. 2006. Habitat structure associated with Borrelia
burgdorferi prevalence in small mammals in New York State. Environmental
Entomology 35(2): 308-319.

Pugliese A, Rosa R, et al. 1998. Analysis of a model for macroparasitic infection with
variable aggregation and clumped infections. Journal of Mathematical Biology
36(5): 419-447.

Randolph SE, Gem L, et al. 1996. Co-feeding ticks: Epidemiological significance for
tick-borne pathogen transmission. Parasitology Today 12(12): 472-479.

Randolph SE. 1999a. Epidemiological uses of a population model for the tick
Rhipicephalus appendiculatus. Tropical Medicine & International Health 4(9):
A34-A42.

Randolph SE, Miklisova D, et al. 1999b. Incidence from coincidence: patterns of tick
infestations on rodents facilitate transmission of tick-borne encephalitis virus.
Parasitology 118: 177-186.

Reed KD, Meece JK, et al. 2003. Birds, migration, and emerging zoonoses: West Nile
virus, Lyme disease, influenza A and enteropathogens. Clinical Medicine &
Research 1(1): 5-12.

Rosa R and Pugliese A. 2002. Aggregation, stability, and oscillations in different models
for host-macroparasite interactions. Theoretical Population Biology 61(3): 319-
334.

Schauber EM and Ostfeld RS. 2002. Modeling the effects of reservoir competence decay
and demographic turnover in Lyme disease ecology. Ecological Applications
12(4): 1142-1162.

Schauber EM, Ostfeld RS, Evans AS. 2005. What is the best predictor of annual Lyme

disease incidence: Weather, mice, or acorns? Ecological Applications 15(2): 575-
586.

181

Scrimenti R. 1970. Erythema chronicum migrans. Archives of Dermatology 102(1): 104-
105.

Shaw DJ and Dobson AP. 1995. Patterns of macroparasite abundance and aggregation in
wildlife populations: a quantitative review. Parasitology 111: S111-133.

Shaw DJ, Grenfell BT, Dobson AP. 1998. Patterns of macroparasite aggregation in
wildlife host populations. Parasitology 117: 597-609.

Sonenshine DE. 1979. Ticks of Virginia (Acari: Metastigmata). Insects of Virginia No.
13. Research Division Bulletin 139, Virginia Polytechnic Institute and State
University, Blacksburg VA.

Steere AC, Coburn J, et al. 2004. The emergence of Lyme disease. Journal of Clinical
Investigation 113(8): 1093-1101.

Strand RM, Walker ED, et al. 1992. Field studies on Ixodes dammini in the Upper
Peninsula of Michigan. Vector Control Bulletin of the North Central States 1:
1 1 1-1 18.

Telford SR, TN Mather, et al. 1988. Incompetence of deer as reservoirs of the Lyme-
disease spirochete. American Journal of Tropical Medicine and Hygiene 39(1):
105-109.

Van Buskirk J and Ostfeld RS. 1995. Controlling Lyme—disease by modifying the density
and species composition of tick hosts. Ecological Applications 5(4): 1133-1140.

Van Buskirk J and Ostfeld RS. 1998. Habitat heterogeneity, dispersal, and local risk of
exposure to Lyme disease. Ecological Applications 8(2): 365-378.

Walker ED, Smith TW, et al. 1994. Prevalence of Borrelia burgdorferi in host-seeking
ticks (Acari: Ixodidae) from a Lyme disease endemic area in Northern Michigan.
Journal of Medical Entomology 31(4): 524-528.

Walker ED, Stobierski MG, et al. 1998. Geographic distribution of ticks (Acari21xodidae)
in Michigan, with emphasis on Ixodes scapularis and Borrelia burgdorferi.
Journal of Medical Entomology 35(5): 872-882.

Wang G, Ojaimi C, et al. Impact of genotypic variation of Borrelia burgdorferi sensu
stricto on kinetics of dissemination and severity of disease in C3H/HeJ mice.
Infection and Immunity 69(7): 4303-4312.

Wang G, Liveris D, et al. 2003. Real-time PCR for simultaneous detection and
quantification of Borrelia burgdorferi in field—collected Ixodes scapularis ticks
from the northeastern United States. Applied and Environmental Microbiology
69(8): 4561-4565.

182

Wilson K, Grenfell BT, et al. 1996. Analysis of aggregated parasite distributions: A
comparison of methods. Functional Ecology 10(5): 592-601.

Wilson, ML and Spielman A. 1983. Seasonal activity of immature Ixodes dammini
(Acari: Ixodidae). Journal of Medical Entomology 22: 408-414.

Woolhouse MEJ, Dye C, et al. 1997. Heterogeneities in the transmission of infectious
agents: implications for the design of control programs. Population Biology
94(1): 338-342.

Woolhouse MEI, Etard JF, et al. 1998. Heterogeneities in schistosome transmission
dynamics and control. Parasitology 117: 475-482.

183

   

1"13111111131111“;in1111,111