Genetic Structure of the White-Footed Mouse in the Context of the Emergence of Lyme Disease in Southern Québec

Genetic structure of the white-footed mouse in the context of the emergence of Lyme disease in southern Québec

Anita Rogic Department of Biology McGill University, Montreal

Submitted August, 2012

A thesis submitted to McGill University in partial fulfillment of the requirements of the degree of M.Sc.

TABLE OF CONTENTS

ABSTRACT……………………………………………………. 3

RESUMÉ……………………………………………………….. 4

ACKNOWLEDGEMENTS……………………………………. 5

LIST OF TABLES……………………………………………... 6

LIST OF FIGURES……………………………………………. 8

CONTRIBUTION OF AUTHORS……………………………. 9

GENERAL INTRODUCTION……………………………….. 10 References……………………………………………... 23

GENETIC STRUCTURE OF THE WHITE-FOOTED MOUSE IN THE CONTEXT OF THE EMERGENCE OF LYME DISEASE IN SOUTHERN QUÉBEC……………………………………….. 33

Introduction……………………………………………. 34

Materials & Methods…………………………………... 39

Results…………………………………………………. 45

Discussion…………………………………………….... 49

References……………………………………………… 58

TABLES……………………………………………………….. 72

FIGURES………………………………………………………. 80

GENERAL CONCLUSIONS………………………………..... 83 References……………………………………………... 85

APPENDIX……………………………………………………. 86

ABSTRACT

The white-footed mouse (Peromyscus leucopus) has expanded its northern limit into southern Québec over the last few decades. P. leucopus is a great disperser and colonizer and is of particular interest due to its link to human health: it is considered one of the primary reservoirs for the spirochete bacterium that causes Lyme disease in humans. There is no current information on whether the mice are establishing successful populations on the mountains and forest fragments found scattered throughout the Montérégie region, what the genetic flow is between these habitable sites, and whether various landscape barriers (i.e. agricultural fields, highways and main water bodies) have an effect on their dispersal. We conducted a population genetics analysis on 11 populations using 11 microsatellite markers to identify the processes that account for the current level and distribution of their genetic variation. Tick data was collected at each site as well to provide insight on the spread of the disease. Geographic distance had little effect on the overall genetic structure of the populations, but barriers were effective. Agricultural matrices had the weakest barrier effect on dispersal, highways slowed down dispersal but were not absolute barriers, and main rivers were strong dispersal barriers. The abundance of ticks collected on mice varied within the study area and we predicted areas of greater risk for Lyme disease. Our results are pertinent to merge with ongoing Lyme disease surveillance programs in Québec to help determine the future threat of this disease in the province, and to contribute towards disease prevention and management strategies throughout fragmented landscapes in southern Canada.

RESUMÉ

Au cours des dernières décennies, la souris à pattes blanches (Peromyscus leucopus) a étendu les limites de sa distribution du nord-est des États-Unis vers le sud du Québec. Ce petit mammifère est un excellent migrateur et colonisateur de nouveaux habitats. Cette espèce présente un intérêt tout particulier en matière de santé humaine puisqu’elle est considérée comme l’un des principaux réservoirs de la bactérie causant la maladie de Lyme. À l'heure actuelle, il y a peu de données concernant les populations de souris à pattes blanches en Montérégie, à savoir : a) si elles colonisent avec succès les montagnes et forêts fragmentées réparties à travers cette région; b) s’il existe un flux génétique bas ou élevé entre les populations occupant ces différents sites en Montérégie et finalement, c) si les champs agricoles, les autoroutes et les étendues d’eau agissent en tant que barrières influençant la dispersion des individus entre les habitats. Nous avons effectué une analyse génétique de 11 populations provenant de 11 sites de la région en utilisant 11 séquences microsatellites pour identifier la distribution et le niveau de variation génétique entre ces différentes populations. Des données concernant les tiques ont aussi été recueillies à chacun de ces sites afin d’établir un portrait de la propagation de la maladie de Lyme. Nos résultats confirment que la distance géographique entre les sites a peu de répercussions sur la structure génétique de ces populations contrairement aux barrières. Les champs agricoles montrent les effets les plus faibles sur la dispersion. Les autoroutes ralentissent la dispersion des souris et ne sont donc pas des barrières absolues, contrairement aux rivières qui ont l'impact le plus fort. L’abondance de tiques trouvées sur les souris varie selon les sites étudiés et nous avons pu utiliser ces données afin de prédire les endroits qui constituent un plus grand risque de propagation de la maladie. Nos résultats sont pertinents pour le programme de surveillance de la maladie de Lyme au Québec puisqu’ils aident à déterminer les menaces futures de propagation de la maladie dans la province. Nos conclusions peuvent ainsi contribuer à une meilleure prévention et une gestion stratégique de la maladie de Lyme à travers diverses régions du sud du Canada.

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my parents and sisters for always reminding me the importance of staying in school and for showing me unwavering support during my seemingly endless years of education. This thesis resonates with the memory of my father repeating how important it is for a woman to have an education. Thank you for the invaluable advice dad.

A special thank you goes to my supervisor, Virginie Millien. Her knowledge and guidance and has helped shaped me into the research scientist that I am today. I learned how to see a project from its birth to its end, and how to have fun doing it. I look forward to applying the skills I have learned towards my future career and am grateful for being given the chance to produce these scientific contributions under her guidance.

Thank you to my co-supervisor, François-Joseph Lapointe, for reminding me of the big picture and for helping me stay positive throughout the project, and to the members of the LEMEE lab for selflessly donating their genetic knowledge to my project and for always making me laugh. The years flew by working there.

I would like to thank the members of my supervisory committee, Andrew Gonzalez and Martin Lechowicz, for helping me explore the different avenues of my project and for manuscript advice. For hard work in the field and lab, I would like to thank all field and lab assistants that helped make data collection and preparation run as smoothly as it did. A huge thank you goes to Manon Albert for helping me brainstorm ideas, overcome writer’s block, and for letting me vent all of my graduate student problems.

Finally, I would like to thank the Ouranos Consortium, the Québec Ministère des Ressources Naturelles et de la Faune, and the Québec Centre for Biodiversity Science for research funding.

LIST OF TABLES

Table 1. Location, elevation, area, perimeter and sample size for the study sites. Nearest mountains for fragments are indicated. G: Mont Rougemont, H: Mont St. Bruno, J: Mont Yamaska, K: Mont St. Hilaire

Table 2. Tick load on trapped mice. Mice from sites H and K were collected before 2011 and were not sampled for ticks.

Table 3. Sample size (n), mean number of alleles per locus (k), allelic richness

(A), number of private alleles (p), and observed (HO) and expected (HE) heterozygosity for 11 P. leucopus populations in the Montérégie.

Table 4. FST values (above the diagonal) for the 11 populations studied and the number of significant results for the Fisher Exact test (below the diagonal) ns for the eleven loci. Significant FST values are indicated by asterisks ( : p>0.05, **: p<0.01, ***: p<0.001). Probabilities were adjusted with sequential Bonferroni corrections.

Table 5. Results for the backwards elimination multiple regression analyses using

FST values as the dependant variable and dispersal barriers as the independent variable (Highways 10, 112 & 116, the Richelieu and the Yamaska River) when either including or excluding sites D and H. Analyses on whether the total number of barriers between populations (i.e. an all- barrier matrix) affected genetic distance are shown as well, ns: p>0.05, *: p<0.05, **: p<0.01.

Table 6. AMOVA results show significant among group variation when dividing the mouse populations into the three-group hypothesis determined by the MRM analysis. Θ indices are analogous to F statistics; **: p<0.01.

Table 7. Effective number of migrants (Nm) for all mountain populations and their closest fragment (source-sink pairs identified by matching symbols). Total Nm is summed as total immigrationa and total emmigrationb rates for each population.

Table 8. Gene flow symmetry (GFS) indices between source-sink populations. If gene flow moved asymmetrically from site j to site i, then we would expect private alleles to accumulate faster at site j thus creating a GFS index of greater than 1. Asymmetrical gene flow is observed between two mountain populations (site i) and their nearest forest fragment populations (site j): sites D & H and sites C & J.

LIST OF FIGURES

Figure 1. Geographic locations of the 11 study sites and the main geographic barriers among them. Four Monteregian hills were considered: Mont St. Bruno (H), Mont St. Hilaire (K), Mont Rougemont (G), and Mont Yamaska (J). Populations were grouped according to the significance of the barriers determined through MRM analyses, and the dotted square includes the five sites whose only habitat between them is an agricultural matrix. A greater forest density is observed in the southern portion of the Montérégie (mapped with SIEF data, MRNQ 2000).

Figure 2. STRUCTURE’s population clustering analysis suggesting two population clusters (K=2). Each line represents an individual, and Q is the proportion of the individual's genome which is part of each K subpopulation.

Figure 3. Results of the A) hierarchical clustering tree (UPGMA), B) the isolation

by distance analysis using FST values, and C) the gene flow network map (mapped with SIEF data, MRNQ 2000) showing the connection between populations.

CONTRIBUTION OF AUTHORS

This thesis is comprised of two chapters and represents my own intellectual work. This research stemmed from an ongoing research program focusing on the evolution of small mammals in Québec, conducted in V. Millien’s laboratory.

I conducted the literature review and am responsible for the writing of the first chapter.

The second chapter was prepared and formatted for submission to the journal Molecular Ecology. It is co-authored by N. Tessier, F.-J. Lapointe, and V. Millien.

I performed a majority of the field sampling and most of the laboratory work. I also conducted a large amount of the data analyses and interpretations, as well as the writing of the manuscript. N. Tessier contributed towards laboratory work, data analysis and presentation, and manuscript writing. V. Millien and F.-J. Lapointe contributed to the formation of research questions and hypotheses, sampling design, data analysis and presentation, and manuscript writing. V. Millien also contributed to field sampling and all research funding.

GENERAL INTRODUCTION

Population genetic studies play an important role in understanding the spread of zoonotic diseases (Wilcox and Gubler, 2005). By examining the movements of the organisms involved in its propagation and transmission, effective control strategies may be developed in areas where the disease is in emergence (Tabachnick and Black, 1995). Climate warming has recently expanded the northernmost limit of tick and Lyme disease emergence into southern Québec (Leighton et al., 2012), and has allowed the white-footed mouse, considered the primary reservoir for the bacteria responsible for Lyme disease, to expand its range into Canada from the north-eastern United States (Myers et al., 2009). These combined factors are creating an increasingly greater risk for the human population to acquire the disease within Québec, yet not much is known about its local spread. In this thesis, I evaluated the migration patterns and degree of connectivity between mouse populations in a fragmented landscape of southern Québec. Information on the local movements of the mice, in the context of range expansion due to climate change, will help shed insight on the patterns of the spread of the disease within the province.

LITERATURE REVIEW

Climate change has always been an area of interest in the field of biology, and scientific records on its effect on plant and animal range shifts date back to the mid-18th century (Parmesan, 2006). Over the last century, the Earth’s climate has increased by an average of 0.3 to 0.6 C, but the greatest climatic changes have been occurring at higher latitudes.⁰ Annual⁰ temperatures in southern Canada have increased between 0.5 to 1.5 C over the last 100 years, as has annual precipitation (5% to 35% increase)⁰ (Almaraz⁰ et al., 2008). The Montérégie region in southern Québec in particular has had a 0.8 C increase in its mean growing season temperature since 1973, yet autumn months⁰ in particular (especially September) have seen increases of up to 2.8 C (Almaraz et al., 2008).

⁰ 10

Climate change affects all levels of ecological organization including population densities, shifts in the geographic range of species, changes in phenology (e.g. migration, egg-laying, flowering), shifts in genetic frequencies, and both morphological and behavioural changes may occur as well (Root et al., 2003). Studies show that over the last few decades, climate warming in the Montérégie has already affected various plant and animal species, agricultural activities and biodiversity (Reid and Strain, 1994; Hudon, 2004; Ouellet et al., 2005; Almaraz et al., 2008; Duchesne et al., 2009; Auzel et al., 2012). It has also contributed towards a rapid range shift of the white-footed mouse, Peromyscus leucopus, in Michigan, USA and Ontario, Canada over the last few decades (Myers et al., 2005, 2009).

The Montérégie is very important agriculturally. It contains 23% of Québec’s farmlands and produces 60% of the province’s total corn grains (Almaraz et al., 2008). Agricultural practices arose in the mid 1800’s, but truly soared in the 1940’s which led to a 70% decrease of the local forests (Wampach, 1988). Today, the region’s landscape is largely composed of mountains, formed by magma intrusions dated at approximately 130 Mya, and scattered forest fragments surrounded by agricultural fields and urban habitat. The Monteregian hills include Oka, Mont Royal, Mont Saint-Bruno, Mont Saint-Hilaire, Mont Saint-Grégoire, Mont Rougemont, Mont Yamaska, Mont Shefford, Mont Brome, and Mont Mégantic.

Forest fragmentation due to human land use has shown to have negative effects on the densities and dispersal of many plant and animal species, as well as on the biodiversity level within each resulting patch (Andren, 1994; Fahrig, 2003; Ghosh, 2004; Jha et al., 2005). These forest patches of decreasing area and increasing levels of isolation lead to the disappearance of habitat specialists and the proliferation of generalists (Devictor et al., 2008). Generalist species thrive under a wide range of habitats and are less prone to extinction due to their greater dispersal success into, and establishment within, “less desirable” habitats (Marvier et al., 2004). Some of these generalists are problematic towards the local

11 ecosystem (e.g. exotic invasive species), or towards human health (e.g. reservoirs for zoonotic diseases) (Childs et al., 1998; Marvier et al., 2004). Peromyscus leucopus, the white-footed mouse, is considered one of these zoonotic reservoirs and is currently increasing in abundance in southern Québec.

Peromyscus in Québec

The genus Peromyscus is a rodent that belongs to the family Cricetidae (Wilson and Reeder, 2005). Peromyscines are extremely abundant throughout North America and have been used as research organisms in a wide range of disciplines for over a century (e.g. physiology and endocrinology, evolution and systematics, toxicology, parasitology and epidemiology, behaviour, and ecology). Their overall importance has designated them as “the Drosophila of North American mammalogy” (Dewey and Dawson, 2001).

Two species are found in Québec: the white-footed mouse (Peromyscus leucopus Rafinesque, 1818) and the deer mouse (Peromyscus maniculatus Wagner, 1845). Both species are very similar behaviourally, morphologically and ecologically. They share the same foods, use the same microhabitats and similar nest sites, are territorial (both intra- and inter-specifically), and are affected by similar parasitic infections (Wolff, 1985). Of the 56 species of Peromyscus, P. leucopus and P. maniculatus have been the most studied due to their ease of observation in the laboratory, their ability to populate a wide range of habitats, and they are ideal research organisms for a variety of studies including human diseases, infectious disease vectors, and ecological niche adaptations (Joyner et al., 1998). Various morphological traits are used to help distinguish the two species from one another including tail size and colour pattern, pelage colouration, and craniometric variables (Rich et al., 1996), but species identification using molecular markers allows for more accurate results (Aquadro and Patton, 1980; Tessier et al., 2004). These markers also enable juveniles,

12 museum specimens and degraded specimens to be identified which may be very difficult when using morphological traits (Tessier et al., 2004).

Over the last 15 years, the white-footed mouse has expanded its northern limit into southern Québec (i.e. the Montérégie region), whilst the deer mouse is on the decline. This northern range expansion of P. leucopus has been linked to changes in climate, and it has been expanding at an astonishing rate considering the mouse’s size: approximately 15 km/year (Myers et al., 2005, 2009). Before the warming of the Montérégie region, southern Québec was considered the northernmost limit of P. leucopus’ range. A long-term field study in the 1960’s and 1970’s by Grant (1976) showed that most of the rodents caught on Mont St. Hilaire were P. maniculatus species, but in the last few decades this statistic has significantly changed. The number of trapped P. leucopus mice has steadily increased and today, they are the highest captured rodent in the field (Rogic et al., Chapter 2 of this thesis, Millien et al. unpublished).

Peromyscus leucopus success

P. maniculatus is better adapted to cold weather than P. leucopus. They build larger and better insulated nests, store more food for the winter season, and undergo torpor more efficiently that P. leucopus (Pierce and Vogt, 1993). Warmer winters and earlier springs have a negative effect on P. maniculatus abundance but a positive one on P. leucopus (Rowland, 2003; Myers et al., 2005). With the recent warming of the Montérégie region in Québec, it is likely that P. maniculatus has now pushed its range further north into the province, and that the warming created ideal climatic conditions for P. leucopus, allowing them to push their range north and increase in abundance in southern Québec.

The white-footed mouse’s success may also be attributed to their foraging and habitat preferences. They are considered extreme dietary and habitat generalists and show more flexibility in terms of food and habitat choices than P. maniculatus (Drickamer, 1972; Geier and Best, 1980; Ostfeld and Keesing, 2000).

When the two species co-exist within the same habitat, their behaviour varies. P. maniculatus spends more time in the trees than P. leucopus, linking their population densities directly to greater vertical forest stratification (i.e. canopy cover). P. leucopus, on the other hand, has been observed to concentrate most of its activity on the ground which lessens the influence of canopy cover on its abundance (Vickery, 1981; Wolff and Hurlbutt, 1982; Barry et al., 1984; Harney and Dueser, 1987). In Ohio, Cummings and Vessey (1994) observed P. leucopus to tolerate a greater range of soil and moisture conditions than any other rodent in the area, and Drickamer (1972) showed that they have greater flexibility than P. maniculatus in terms of feeding behaviour and food habits. This behavioural flexibility allows the white-footed mouse to utilize a larger range of habitats than the deer mouse. This may explain their increasing success (and the simultaneous decline of deer mice) in the mixed deciduous forest patches of the Montérégie.

Although both Peromyscus species are found in a variety of habitat types, P. maniculatus has been observed to prefer boreal forests (Myers et al., 2005). The Montérégie is part of the Great Lakes and St. Lawrence River forest system (D’Orangeville et al., 2008) and is considered a transitional zone between the deciduous forests of eastern North America and the boreal forest in northern Canada (Ontario Ministry of Natural Resources, 2012). It is composed of a mix of coniferous trees (eastern white pine, red pine, eastern hemlock, and white cedar) and deciduous broad-leaved species (yellow birch, sugar maple, red maple, basswood, and red oak). Some common boreal species are also found in this system and include white spruce, black spruce, jack pine, balsam poplar, white birch, and trembling aspen (Forest Products Association of Canada, 2012). Trees running along the Montérégie’s south-west to north-east gradient become less diversified, and more boreal tree species are found in mid to high elevations (Bouchard & Maycock, 1978). These boreal microhabitats provide a suitable habitat for P. maniculatus, but are less optimal for P. leucopus. Sites consisting of a greater amount of boreal tree species are thus important to consider for P. leucopus dispersal studies, especially in a fragmented landscape.

Fragmentation has shown to have negative effects on the densities of many mammal species, but in the case of P. leucopus the contrary has been observed (Anderson and Meikle, 2010). Numerous studies using both live-trapping and genetic data have found that P. leucopus population densities increase as forest patch size decreases, indicating that white-footed mouse numbers are enhanced by forest fragmentation (Yahner, 1992; Nupp and Swihart, 1996, 1998; Krohne and Hoch, 1999; Mossman and Waser, 2001; Wilder et al., 2005; Anderson and Meikle, 2010). The reason for this inverse relationship is not yet clearly understood, but various factors are believed to play a role. Patch edge supports greater vegetation density and complexity, resulting in greater food availability for the mice (Wilder et al., 2005). Since smaller forest patches have a greater perimeter to area ratio than larger patches, mice densities will grow with increased food availability. As a patch decreases in size, so does its ability to host a diverse number of species, therefore the mice experience less inter-specific competition and predation pressure in smaller forest patches enabling them to thrive under these conditions (Ostfeld and Keesing, 2000). Finally, emigration rates from small forest patches may be inhibited due to the patches usually being surrounded by areas of unsuitable habitat (e.g. agricultural fields and urban habitats). Mice residing within these “patch islands” have a difficult time dispersing to new environments, therefore their populations soar within the small patch (Nupp and Swihart, 1998).

Studies also show that P. leucopus does not always conform to the source- sink hypothesis (Morris and Diffendorfer, 2004; Grear and Burns, 2007; Linzey et al., 2012). The source-sink model describes how populations respond to landscapes that differ in composition (Danielson, 1992). It focuses on two habitat types: 1) an optimal habitat (i.e. the source) where individuals may achieve high fitness levels and emigration will exceed immigration, and 2) a suboptimal habitat (i.e. the sink) where a lower quality habitat results in lower reproductive rates, but the number of immigrants it receives from source populations is high (Grear and Burns, 2007). There have been conflicting reports on how difficult dispersal may be for mice moving between fragmented forest habitats (Anderson and Meikle,

2010), and this has led to the rigorous testing of various landscape features that may present as barriers.

P. leucopus is a great disperser and colonizer, and a single mating pair may establish a successful population upon entering a new habitat (Cummings and Vessey, 1994). They are known to cover a very wide range of dispersal distances throughout their lifespan and may travel between 85-867 m (mean ≈330 m) (Krohne et al., 1984). Some extreme dispersers have been found 5-14 km away from their original site of capture (Maier, 2002). But this dispersal may get hindered by various natural and human landscape barriers. In terms of dispersal within an agricultural matrix, the results are conflicting: agricultural fields have been found to act as significant dispersal barriers (Krohne and Hoch, 1999), as seasonal barriers where crop height and maturity determine whether the mice use them as dispersal routes and/or food sources (Cummings and Vessey, 1994; Anderson et al., 2006), and finally not as barriers at all (Middleton and Merriam, 1981; Mossman and Waser, 2001). Urban habitat on the other hand, consistently shows to be a strong dispersal barrier between white-footed mouse populations (Barko et al., 2003; Munshi-South and Kharchenko, 2010; Munshi-South, 2012). Various water bodies have also been found to act as significant barriers to dispersing mice including streams 3-4 m wide (Savidge, 1973), rivers (Klee et al., 2004) and between ocean island populations (Loxterman et al., 1998). Studies also report that roads are effective at slowing P. leucopus dispersal but do not act as absolute barriers (Oxley et al., 1974; Merriam et al., 1989; Clark et al., 2001; McGregor et al., 2003; McLaren et al., 2011).

Lyme disease

This interest in P. leucopus’ dispersal characteristics is largely due to their link to human health: they are considered one of the primary reservoirs for the spirochete bacterium that causes Lyme disease in humans (Ostfeld, 2011). In the 1970’s, an outbreak of what was believed to be juvenile rheumatoid arthritis in

Lyme, Connecticut (USA) led to the discovery of Borrelia burgdorferi, the bacterium responsible for Lyme disease. This bacterium is transmitted to humans through the bite of the black-legged tick (Ixodes scapularis) (Ostfeld, 2011), and shortly after infection, people experience flu-like symptoms and 60-80% may develop a red or bluish-red lesion (known as erythema migrans) at the site of the bite. If left untreated, painful joint swelling may occur eventually leading to chronic arthritis. The brain, central nervous system and heart may all be affected as well and include side effects such as lymphocytic meningitis, facial palsy, migraines, muscle weakness, an unsteady gait, chronic fatigue, and myocarditis (Stanek et al., 2002). The life cycle of I. scapularis involves many different vertebrate host species and undergoes two immature stages (larval and nymph), followed by the reproductive adult stage. The tick’s life cycle extends over two years. Once the larvae have had their first blood meal, they molt into nymphs and overwinter in the litter until the following spring. Another blood meal is required for transition into the adult stage, and a final blood meal is required for reproductive purposes in the case of the female adult tick. Although the ticks may feed on a wide range of vertebrate hosts (e.g. skunks, raccoons, white-tailed deer, opossums, foxes, etc.), there are four small mammals in particular (eastern chipmunk, short-tailed shrew, masked shrew, and white-footed mouse) that have a high reservoir competence for Borrelia burgorferi (Ostfeld, 2011). Competence is determined by how susceptible the host is to the infection once bitten, by how well the pathogen replicates inside the host, and by how efficiently the bacterium is transmitted back to vectors feeding on the host (LoGiudice et al., 2003). The white-footed mouse is considered to have one of the highest reservoir competences for Borrelia burgorferi, infecting 75-95% of the larval ticks that feed on them (Mather et al., 1989; Ostfeld, 2011).

Historically, black-legged tick populations were not established in Québec. Tick larvae may have been carried from the USA into Canada by migratory birds, but due to their sensitivity to climate, they quickly died out at those high latitudes (Ogden et al., 2005, 2010). If human population growth and greenhouse gas emissions remain much as they are today, by 2020 the ticks’

17 northernmost range will expand further north, reaching just below Québec City, and tick numbers in southern Québec (in which falls their current northernmost range limit) will double (Ogden et al., 2006). The threat of a northern range expansion of I. scapularis coupled with the current northern expansion and population increase of P. leucopus in the Montérégie region significantly increases the probability of acquiring B. burgdorferi within the bloodstream of the rodents and the ticks, thus producing a very serious threat of Lyme disease in southern Québec. Since November 2003, Québec’s health practitioners have been obligated to report any Lyme disease diagnoses to MADO (Maladies à Declaration Obligatoire, le Ministère de la Santé et des Services Sociaux du Québec). Since then, multiple indigenous-born cases of Lyme disease have been diagnosed, and approximately 10% of the ticks currently found in Québec are infected with B. burgdorferi (Trudel and Serhir, 2009).

Population genetics

Zoonotic diseases may have a very complex life cycle, and population genetic studies have become an important tool to help understand how they spread (Wilcox & Gubler, 2005). Population genetics focuses on the genetic makeup of groups of individuals, and how each group’s genetic composition changes over time (Pierce, 2008). It plays an important role within many areas of biology including plant and animal breeding, infectious diseases, DNA fingerprinting and forensics, genetic differentiation and speciation, species evolution, habitat fragmentation, and the preservation of endangered species (Halliburton, 2004). Direct and indirect measures may be employed to help estimate the genetic variation within each population, and the degree of connectivity between populations (i.e. their gene flow).

In direct estimates, the dispersal distance and breeding success of the dispersers are measured to help estimate gene flow at the time the observations were made (Slatkin, 1987). Popular techniques include direct observations

18 through mark & recapture, observations of specific life stages, and genetic marker-based parentage analyses to track and determine the area of origin of alleles over a short number of generations (Bohonak, 1999). But direct estimation calculations are limited in both space and time (i.e. limited to the size of the particular study area and to the precise time the data was collected), therefore they cannot provide historical information, cannot calculate average levels of gene flow, nor can they sense gene flow from long-distance dispersals. Indirect estimates can detect these parameters thus making them a nice complement to direct estimates of gene flow (Halliburton, 2004).

Indirect methods use allele frequencies and differences in DNA sequences to estimate historical gene flow between populations (Slatkin, 1987). The term “indirect” is used because the allele frequency patterns among subpopulations are used in a model to average all past events that led up to the current observed pattern of allele differentiation (Hamilton, 2009). Therefore, the models help predict how much gene flow must have occurred in order for the present patterns to be observed. The ideal indirect method detects patterns in allele frequencies due solely to gene flow and excludes the effects of genetic drift, natural selection and mutation (Slatkin, 1987). This is a very difficult ideal to reach and has yet to be realized, therefore there are various indirect measures that may be used to estimate the average level of gene flow between populations. Some of the factors to take into consideration when using indirect methods to estimate gene flow include the genetic model of the subpopulations, the genetic markers, and the mutation model.

The genetic model and the genetic markers used when employing indirect techniques to a population genetics study are closely linked. The genetic model determines whether the populations are haploid or diploid and whether the marker used is inherited biparentally or uniparentally. In terms of genetic markers, there are a variety of options to use including restricted fragment length polymorphisms (RFLPs), both nuclear and mitochondrial DNA sequences, allozymes, and microsatellites. RFLPs may not produce many sites (i.e. small number of

19 nucleotides), therefore direct DNA sequencing of genes may be an alternate option. With direct sequencing, many nucleotides are analyzed but fewer individuals and loci are included due to time and cost restraints. Sequencing with either RFLPs or DNA may also ignore relationships among haplotypes since they are treated as individual loci (e.g. linkage disequilibrium). Allozyme loci tend to have low mutation rates but may be affected by natural selection and usually have low resolution (i.e. few detectable alleles). Microsatellite markers help minimize these problems because they are much more variable and are located in non- coding regions, therefore are less likely to be affected by natural selection. This makes microsatellite loci powerful tools to use in gene flow studies (Halliburton, 2004).

Microsatellites are composed of short tandem repeats of 1-6 nucleotides and tend to undergo mutations at a higher rate than most other loci due to their repeat motif. The most frequent mutation to affect microsatellites is slipped strand mispairing which occurs during replication. A locus that undergoes high mutation rates results in a greater number of alleles to be formed at that locus. This high allelic variability is ideal for genetic studies on processes acting on ecological time scales, and microsatellites are one of the few markers that provide insight into fine-scaled ecological questions (Selkoe and Toonen, 2006).

Mutation models are important for determining allele frequency changes in different subpopulations that are due to mutations. The infinite alleles model, the k alleles model and the stepwise mutation model are three popular models used when generalizing how mutations affect allele frequencies. The infinite alleles model states that every mutational event creates a new allele. This indicates that any two alleles identical in state are also identical by descent. The k alleles model states that every allele (k) can mutate into every other allelic state (k-1) with equal probability. Unlike the infinite alleles model, the k alleles model has a finite number of possible allelic states that an allele may actually mutate into. Both models assume that each allele has an equal probability of mutating to any of the other allelic states, and that the new allele (produced by a mutation) is not

20 independent of the initial state of the allele (e.g. transitions are more common than transversions). The stepwise mutation model is often used for microsatellite data and assumes that alleles with larger differences between them are likely to have undergone more mutational events than alleles with fewer differences between them. This model is ideal for repeated sequences because a gain or a loss of a repeat is considered a “step”, and the fewer steps there are between two alleles, the greater the probability that they also share a recent common ancestor. Homoplasy however is a risk to this model. Multiple mutations may cause alleles from completely different lineages to be identical in state thus blurring their separate states (Hamilton, 2009). There is no single mutation model that may be used to represent the process of mutation. Rather, an array of models is generally employed to process different alleles and loci, and the overall ideal combination of genetic models, genetic markers and mutational models to use largely depends on the biological question being investigated.

Population genetic studies play an important role in understanding the spread of zoonotic diseases (Wilcox and Gubler, 2005). Knowledge of patterns of gene flow and dispersal rates are necessary to help develop effective control strategies for vector and reservoir species. Limited migration would allow for more detailed management strategies (e.g. reduction of specific populations), while wide dispersal would indicate the need for large-scale treatments (Tabachnick and Black, 1995). Although white-footed mice play an important role in the replication and spread of the bacteria responsible for Lyme disease, there is no current information on whether the mice are establishing successful populations on the mountains and forest fragments found scattered throughout the region, what the genetic flow is between these habitable sites, and how various barriers (agricultural fields, water bodies and roads) may affect their dispersal.

The objective of this thesis was to employ a population genetics study of P. leucopus mice in the Montérégie region to identify the processes that account for the current level and distribution of their genetic variation. This information can help predict the extent to which they are interacting with each other and will

21 be considered an important part of a large collaborative Québec project that was recently launched to rigorously study the effects of climate change within the province (Ouranos – Consortium on the effects of climate change). Merging this data with ongoing Lyme disease surveillance programs in Québec will help determine the future threat of the disease in the province, and will contribute towards disease prevention and management strategies within Québec, and within other Canadian provinces with similar ecological conditions, to help understand how the disease may spread.

REFERENCES

Almaraz, J. J., Mabood, F., Zhou, X., Gregorich, E. G., & Smith, D. L. (2008). Climate change, weather variability and corn yield at a higher latitude locale: Southwestern Quebec. Climatic Change, 88: 187-197.

Anderson, C., Meikle, D., Cady, A., & Schaefer, R. (2006). Annual variation in habitat use by white-footed mice, Peromyscus leucopus: the effects of forest patch size, edge and surrounding vegetation type. Canadian Field-Naturalist, 120: 192-198.

Anderson, C. S., & Meikle, D. B. (2010). Genetic estimates of immigration and emigration rates in relation to population density and forest patch area in Peromyscus leucopus. Conservation Genetics, 11: 1593-1605.

Andren, H. (1994). Effects of habitat fragmentation on birds and mammals in landscapes with different proportions of suitable habitat: A review. Oikos, 71: 355-366.

Aquadro, C. F. & Patton, J. C. (1980). Salivary amylase variation in Peromyscus: Use in species identification. Journal of Mammalogy, 61: 703-707.

Auzel, P., Gaonac’h, H., Poisson, F., Siron, R., Calmé, S., et al. (2012). Impacts of climate change on the biodiversity of Quebec: Literature review summary. CSBQ, MDDEP, Ouranos. 29 p.

Barko, V.A., Feldhammer, G.A., Nicholson, M.C., & Davie, D.K. (2003). Urban habitat: A determinant of white-footed mouse (Peromyscus leucopus) abundance in southern Illinois. Southeastern Naturalist, 2: 369-376.

Barry, R.E., Botje, M.A., & Grantham, L.B. (1984). Vertical stratification of Peromyscus leucopus and P. maniculatus in southwestern Virginia. Journal of Mammalogy, 65: 145-148.

Bohonak, A.J. (1999). Dispersal, gene flow, and population structure. The Quarterly Review of Biology, 74: 21-45.

Bouchard, A., & Maycock, P.F. (1978). Les forêts décidues et mixtes de la region appalachienne du sud Québécois. Le Naturaliste Canadien, 105: 383-415.

Childs, J., Shope, R. E., Fish, D., Meslin, F. X., Peters, C. J., Johnson, K., Debess, E., Dennis, D., & Jenkins, S. (1998). Emerging zoonoses. Emerging Infectious Diseases, 4: 453-454.

Clark, B., Johnson, L., & Haynie, M. (2001). Influence of roads on movements of small mammals. The Southwestern Naturalist, 46: 338-344.

Cummings, J., & Vessey, S. (1994). Agricultural influences on movement patterns of white-footed mice (Peromyscus leucopus). American Midland Naturalist, 132: 209-218.

D'Orangeville, L., Bouchard, A., & Cogliastro, A. (2008). Post-agricultural forests: Landscape patterns add to stand-scale factors in causing insufficient hardwood regeneration. Forest Ecology and Management, 255: 1637-1646.

Danielson, B.J. (1992). Habitat selection, interspecific interactions and landscape composition. Evolutionary Ecology, 6: 399-411.

Devictor, V., Julliard, R., & Jiguet, F. (2008). Distribution of specialist and generalist species along gradients of habitat disturbance and fragmentation. OIKOS, 117: 507-514.

Dewey, M. J., & Dawson, W. D. (2001). Deer mice: “The Drosophila of North American Mammalogy". Genesis, 29: 105-109.

Drickamer, L. (1972). Experience and selection behaviour in the food habits of Peromyscus: Use of olfaction. Behaviour, 41: 269-287.

Duchesne, L., Houle, D., Cote, M. A., & Logan, T. (2009). Modelling the effect of climate on maple syrup production in Quebec, Canada. Forest Ecology and Management, 258: 2683-2689.

Fahrig, L. (2003). Effects of habitat fragmentation on biodiversity. Annual Review of Ecology, Evolution, and Systematics, 34: 487-515.

Forest Products Association of Canada. (2012). Forest regions of Canada. Retrieved from http://www.fpac.ca/index.php/en/interactive-map/

Geier, A.R. & Best, L.B. (1980). Habitat selection by small mammals of riparian communities: Evaluating effects of habitat alterations. The Journal of Wildlife Management, 44: 16-24.

Ghosh, P. (2004). Forest fragmentation: A threat to global biodiversity. ENVIS Bulletin: Himalayan Ecology, 12: 17-26.

Grant, P. R. (1976). An 11-year study of small mammal populations at Mont St . Hilaire , Quebec. Canadian Journal of Zoology, 54: 2156-2173.

Grear, J.S., & Burns, C.E. (2007). Evaluating effects of low quality habitats on regional population growth in Peromyscus leucopus: Insights from field- parameterized spatial matrix models. Landscape Ecology, 22: 45-60.

Halliburton, R. (2004). Introduction to population genetics. Upper Saddle River, NJ, Pearson Education Inc., pp. 650.

Hamilton, M.B. (2009). Population genetics. West Sussex, UK, Wiley-Blackwell, pp. 407.

Harney, B., & Dueser, R. (1987). Vertical stratification of activity of two Peromyscus species: an experimental analysis. Ecology, 68: 1084-1091.

Hudon, C. (2004). Shift in wetland plant composition and biomass following low- level episodes in the St. Lawrence River: Looking into the future. Canadian Journal of Fisheries and Aquatic Sciences, 61: 603-617.

Jha, C. S., Goparaju, L., Tripathi, A., Gharai, B., Raghubanshi, A. S., & Singh, J. S. (2005). Forest fragmentation and its impact on species diversity: An analysis using remote sensing GIS. Biodiversity and Conservation, 14: 1681- 1698.

Joyner, C.P., Myrick, L.C., Crossland, J.P., & Dawson, W.D. (1998). Deer mice as laboratory animals. ILAR Journal, 39: 322-330.

Klee, R., Mahoney, A., Christopher, C., & Barrett, G. (2004). Riverine peninsulas: An experimental approach to homing in white-footed mice (Peromyscus leucopus). The American Midland Naturalist, 151: 408-413.

Krohne, D.T., Dubbs, B., & Baccus, R. (1984). An analysis of dispersal in an unmanipulated population of Peromyscus leucopus. American Midland Naturalist, 112: 146-156.

Krohne, D. T., & Hoch, G. A. (1999). Demography of Peromyscus leucopus populations on habitat patches: The role of dispersal. Canadian Journal of Zoology, 77: 1247-1253.

Leighton, P.A., Koffi, J.K., Pelcat, Y., Lindsay, L.R., & Ogden, N. H. (2012). Predicting the speed of tick invasion: An empirical model of range expansion for the Lyme disease vector Ixodes scapularis in Canada. Journal of Applied Ecology, 49: 457-464.

Linzey, A.V., Reed, A.W., Slade, N.A., & Kesner, M.H. (2012). Effects of habitat disturbance on a Peromyscus leucopus (Rodentia: Cricetidae) population in western Pennsylvania. Journal of Mammalogy, 93: 211-219.

LoGiudice, K., Ostfeld, R.S., & Schmidt, K.A. (2003). The ecology of infectious disease: Effects of host diversity and community composition on Lyme disease risk. PNAS, 100: 567-571.

Loxterman, J., Moncrief, N., Dueser, R., & Carlson, C. (1998). Dispersal abilities and genetic population structure of insular and mainland Oryzomys palustris and Peromyscus leucopus. Journal of Mammalogy, 79: 66-77.

Maier, T. (2002). Long-distance movements by female white-footed mice, Peromyscus lecopus, in extensive mixed-wood forest. Canadian Field Naturalist, 116: 108-111.

Marvier, M., Kareiva, P., & Neubert, M. G. (2004). Habitat destruction, fragmentation, and disturbance promote invasion by habitat generalists in a multispecies metapopulation. Risk Analysis, 24: 869-878.

Mather, T. N., Wilson, M. L., Moore, S. I., Ribiero, J. M. C., & Spielman, A. (1989). Comparing the relative potential of rodents as reservoirs of the Lyme disease spirochete (Borrelia burgdorferi). American Journal of Epidemiology, 130: 143-150.

McGregor, R., Derrane, S., Bender, D., & Fahrig, L. (2003). The effect of traffic volume on translocated small mammal movement. In C. Irwin, P. Garrett, & K. McDermott (Eds.), International Conference on Ecology & Transportation (pp. 394-396). Raleigh, North Carolina.

McLaren, A., Fahrig, L., & Waltho, N. (2011). Movement of small mammals across divided highways with vegetated medians. Canadian Journal of Zoology, 89: 1214-1222.

Merriam, G., Kozakiewicz, M., Tsuchiya, E., & Hawley, K. (1989). Barriers as boundaries for metapopulations and demes of Peromyscus leucopus in farm landscapes. Landscape Ecology, 2: 227-235.

Middleton, J., & Merriam, G. (1981). Woodland mice in a farmland mosaic. Journal of Applied Ecology, 18: 703-710.

Morris, D.W., & Diffendorfer, J.E. (2004). Reciprocating dispersal by habitat- selecting white-footed mice. OIKOS, 107: 549-558.

Mossman, C. A., & Waser, P. M. (2001). Effects of habitat fragmentation on population genetic structure in the white-footed mouse (Peromyscus leucopus). Canadian Journal of Zoology, 79: 285-295.

Munshi-South, J. (2012). Urban landscape genetics: Canopy cover predicts gene flow between white-footed mouse (Peromyscus leucopus) populations in New York City. Molecular Ecology, 21: 1360-1378.

Munshi-South, J., & Kharchenko, K. (2010). Rapid, pervasive genetic differentiation of urban white-footed mouse (Peromyscus leucopus) populations in New York city. Molecular Ecology, 19: 4242-4254.

Myers, P., Lundrigan, B., & Kopple, R.V. (2005). Climate change and the distribution of Peromyscus in Michigan: is global warming already having an impact. In E. Lacey & P. Myers (Eds.), Mammalian diversification: from chromosomes to phylogeography (a celebration of the career of James L. Patton) (Vol. 133, pp. 101-126). University of California Press.

Myers, P., Lundrigan, B. L., Hoffman, S. M. G., Haraminac, A. P., & Seto, S. H. (2009). Climate-induced changes in the small mammal communities of the Northern Great Lakes Region. Global Change Biology, 15: 1434-1454.

Nupp, T., & Swihart, R. (1996). Effect of forest patch area on population attributes of white-footed mice (Peromyscus leucopus) in fragmented landscapes. Canadian Journal of Zoology, 74: 467-472.

Nupp, T., & Swihart, R. (1998). Effects of forest fragmentation on population attributes of white-footed mice and eastern chipmunks. Journal of Mammalogy, 79: 1234-1243.

Ogden, N. H., Bigras-Poulin, M., O'Callaghan, C. J., Barker, I. K., Lindsay, L. R., et al. (2005). A dynamic population model to investigate effects of climate on geographic range and seasonality of the tick Ixodes scapularis. International Journal for Parasitology, 35: 375-389.

Ogden, N. H., Maarouf, A., Barker, I. K., Bigras-Poulin, M., Lindsay, L. R., et al. (2006). Climate change and the potential for range expansion of the Lyme disease vector Ixodes scapularis in Canada. International Journal for Parasitology, 36: 63-70.

Ogden, N. H., Bouchard, C., Kurtenbach, K., Margos, G., Lindsay, L. R., et al. (2010). Active and passive surveillance and phylogenetic analysis of Borrelia burgdorferi elucidate the process of Lyme disease risk emergence in Canada. Environmental Health Perspectives, 118: 909-914.

Ontario Ministry of Natural Resources (2012). Ontario’s forest regions: Great Lakes – St. Lawrence forest, Ontario Government. Retrieved from http://www.mnr.gov.on.ca/en/Business/Forests/2ColumnSubPage/STDPRO D_091288.html

Ostfeld, R.S. (2011). Lyme Disease: the Ecology of a Complex System. Oxford University Press Inc., New York, New York, pp. 216.

Ostfeld, R.S., & Keesing, F. (2000). Biodiversity and disease risk: The case of Lyme disease. Conservation Biology, 14: 722-728.

Ouellet, M., Mikaelian, I., Pauli, B. D., Rodrigue, J., & Green, D. M. (2005). Historical evidence of widespread chytrid infection in North American amphibian populations. Conservation Biology, 19: 1431-1440.

Oxley, D., Fenton, M., & Carmody, G. (1974). The effects of roads on populations of small mammals. Journal of Applied Ecology, 11: 51-59.

Parmesan, C. (2006). Ecological and evolutionary responses to recent climate change. Annual Review of Ecology, Evolution, and Systematics. 37: 637-669.

Pierce, B.A. (2008). Genetics: A conceptual approach (3rd ed.). New York, NY, W.H. Freeman and Company, pp. 730.

Pierce, S.S., & Vogt, F.D. (1993). Winter acclimatization in Peromyscus maniculatus gracilis, P. leucopus noveboracensis, and P. l. leucopus. Journal of Mammalogy, 74: 665-677.

Reid, C. D., & Strain, B. R. (1994). Effects of CO2 enrichment on whole-plant carbon budget of seedlings of Fagus grandifolia and Acer saccharum in low irradiance. Oecologia, 98: 31-39.

Rich, S. M., Kilpatrick, C. W., Shippee, J. L., & Crowell, K. L. (1996). Morphological differentiation and identification of Peromyscus leucopus and P . maniculatus in northeastern North America. Journal of Mammalogy, 77: 985-991.

Root, T.L., Price, J.T., Hall, K.R., Schneider, S.H., Rosenzweig, C., & Pounds, J.A. (2003). Fingerprints of global warming on wild animals. Nature, 421: 57-60.

Rowland, L. (2003). Relationship of reproductive timing and climate change to the displacement of Peromyscus maniculatus gracilis by Peromyscus leucopus noveboracensis. Unpublished masters dissertation, Miami University, Oxford, Ohio.

Savidge, I. (1973). A stream as a barrier to homing in Peromyscus leucopus. Journal of Mammalogy, 54: 982-984.

Selkoe, K.A. & Toonene, R.J. (2006). Microsatellites for ecologists: a practical guide to using and evaluating microsatellite markers. Ecology Letters, 9: 615-629.

Slatkin, M. (1987). Gene flow and the geographic structure of natural populations. Science, 236: 787-792.

Stanek, G., Strle, F., Gray, J., & Wormser, G.P. (2002). History and characteristics of Lyme Borreliosis. In J.S. Gray, O. Kahl, R.S. Lane & G. Stanek (Eds.), Lyme Borreliosis: Biology, Epidemiology and Control. Wallingford, UK: CABI Publishing, pp. 1-28.

Tabachnick, W.J., & Black, W.C. (1995). Making a case for molecular population genetic studies of arthropod vectors. Parasitology Today, 11: 27-30.

Tessier, N., Noël, S., & Lapointe, F.-J. (2004). A new method to discriminate the deer mouse (Peromyscus maniculatus) from the white-footed mouse (Peromyscus leucopus) using species-specific primers in multiplex PCR. Canadian Journal of Zoology, 82: 1832-1835.

Trudel, L., & Serhir, B. (2009). Maladie de Lyme. Institut National de Santé Publique du Québec, Gouvernement du Québec, N° de publication: 1042.

Vickery, W. (1981). Habitat use by northeastern forest rodents. American Midland Naturalist, 106: 111-118.

Wampach, J.-P. (1988). Deux siècles de croissance agricole au Québec, 1760- 1985. Recherches Sociographiques, 29: 181-199.

Wilcox, B. A. & Gubler, D. J. (2005). Disease ecology and the global emergence of zoonotic pathogens. Environmental Health and Preventative Medicine, 10: 263-272.

Wilder, S., Abtahi, A., & Meikle, D. (2005). The effects of forest fragmentation on densities of white-footed mice (Peromyscus leucopus) during the winter. The American Midland Naturalist, 153: 71-79.

Wilson, D. E. & Reeder, D. M. (2005). Mammal species of the world, 3rd edition. Smithsonian National Museum of Natural History. Retrieved from http://www.vertebrates.si.edu/msw/mswcfapp/msw/index.cfm

Wolff, J. (1985). Comparative population ecology of Peromyscus leucopus and Peromyscus maniculatus. Canadian Journal of Zoology, 63: 1548-1555.

Wolff, J.O. & Hurlbutt, B. (1982). Day refuges of Peromyscus leucopus and Peroymsucs maniculatus. Journal of Mammalogy, 63: 666-668.

Yahner, R. (1992). Dynamics of a small mammal community in a fragmented forest. American Midland Naturalist, 127: 381-391.

Genetic structure of the white-footed mouse in the context of the emergence of Lyme disease in southern Québec

1 2 2 1* Anita Rogic , Nathalie Tessier , François-Joseph Lapointe , Virginie Millien

1 Redpath Museum, McGill University, 859 Sherbrooke Street W., Montréal, Québec, H3A 0C4, Canada 2 Département de sciences biologiques, Université de Montréal, C.P. 6128, Succ. Centre-ville, Montréal, Québec, H3C 3J7, Canada

Keywords: Peromyscus leucopus, Lyme disease, gene flow, barriers, fragmentation, climate change

* Corresponding author: Virginie Millien, Address: Redpath Museum, McGill University, 859 Sherbrooke Street W., Montréal, Québec, H3A 0C4, Canada, Tel: 1- (514) 398-4849, Fax: 1-(514) 398-3185, E-mail: [email protected]

INTRODUCTION

In the 1970’s, an outbreak of what was believed to be juvenile rheumatoid arthritis in Lyme, Connecticut (USA) led to the discovery of Borrelia burgdorferi, the spirochete bacterium responsible for Lyme disease in humans. If left untreated, this disease has shown to have a number of debilitating side effects (Stanek et al., 2002).

B. burgdorferi is transmitted to humans through the bite of the black- legged tick (Ixodes scapularis), an acarian from the Ixodidae family found throughout the eastern and central United States, as well as along the eastern margin of Mexico and Canada (Almazan-Garcia, 2005). Its life cycle includes two immature stages (larval and nymph) followed by a reproductive adult stage. A single blood meal is required for transition into the next stage, or for reproductive needs for the adult female. Although the ticks may feed on a wide range of vertebrate hosts (e.g. skunks, raccoons, white-tailed deer, opossums, foxes, ground- and shrub-dwelling songbirds), there are four species of small mammals (eastern chipmunk, sort-tailed shrew, masked shrew, and white-footed mouse) that have a high reservoir competence for B. burgdorferi (Ostfeld, 2011). Reservoir competence is determined by how susceptible the host is to the infection once bitten, by how well the pathogen replicates inside the host, and by how efficiently the bacterium is transmitted back to the vectors that are feeding on the host (LoGiudice et al., 2003). The white-footed mouse (Peromyscus leucopus) is considered to have the highest reservoir competence, infecting 75-95% of the larval ticks that feed on them (Mather et al., 1989; Ostfeld, 2011). Once infected, the mice are asymptomatic and the infection will have little impact on their activities (Hofmeister et al., 1999; Schwanz et al., 2011). They also harbor the bacteria throughout the year with the number of infected mice increasing during the month of June (Anderson et al., 1987).

Environmental changes are important drivers for the emergence of vector- borne diseases, including Lyme disease (Halos et al., 2010), and both forest fragmentation and global climate change have shown to play a critical role in the spatial and temporal distribution of this disease (Ogden et al., 2008; Halos et al., 2010). Climate change allows reservoir hosts and arthropod vectors to expand their range into territories that were previously unsuitable (Brownstein et al., 2005). In addition, the fragmentation of forested landscapes due to human activities (e.g. urban development, agricultural practices) results in a significant size reduction and isolation of forest patches. Fragmentation of the habitat affects local biodiversity by favoring certain species, including habitat generalists and species that experience high population densities and small home ranges (Allan et al., 2003). A species matching this description is the white-footed mouse; both an important reservoir for B. burgdorferi and host for I. scapularis (Eisen et al., 2012).

Current global warming trends will largely influence the emergence of Lyme disease in eastern Canada (Brownstein et al., 2005; Ogden et al., 2006; 2008; Leighton et al., 2012) and the Public Health Agency of Canada has already identified certain locations in southern British Colombia, Manitoba, Ontario, New Brunswick, Nova Scotia, and Québec as endemic for Lyme disease (Government of Canada, 2012). Québec’s southern region, the Montérégie (Figure 1), is of particular interest because of its highly fragmented landscape and recent temperature increases. Over the last 40 years, the Montérégie has had a 0.8 C increase in its mean growing season temperature (Almaraz et al., 2008). Intense⁰ agricultural practices, which began in the 1940’s, has led to a 70% decrease of the local forests (Wampach, 1988) creating a landscape composed of the Monteregian hills that extend from Montréal to the United States border, and smaller forest patches surrounded by an urban or agricultural matrix. Climate warming combined with the fragmentation of the landscape may have favored the white- footed mouse to expand its range further north into southern Canada, and the

35 number of P. leucopus trapped in southern Québec has been steadily increasing over the last 15 years (Millien et al., unpublished).

White-footed mice are considered extreme generalists in terms of their dietary and habitat preferences (Ostfeld and Keesing, 2000). Although forest fragmentation generally reduces species diversity within remnant patches (Allan et al., 2003), numerous studies using both live-trapping and genetic data have found that P. leucopus population densities increased as patch size decreased, indicating that their numbers were actually enhanced by forest fragmentation (Yahner, 1992; Nupp and Swihart, 1996, 1998; Krohne and Hoch, 1999; Mossman and Waser, 2001; Wilder et al., 2005; Anderson and Meikle, 2010). The reason for this inverse relationship is not yet clearly understood. Patch edges may support greater vegetation density and complexity, resulting in greater food availability for the mice. Since smaller forest patches have a greater perimeter to area ratio than larger patches, mice densities will grow with the increased edge length and associated food availability (Wilder et al., 2005; Klein and Cameron, 2012). In addition, as a patch decreases in size, so does its ability to host a large number of species. The mice thus experience less inter-specific competition and predation pressure in smaller forest patches, enabling them to thrive (Ostfeld and Keesing, 2000). Finally, small forest patches are usually surrounded by areas of less suitable habitats (e.g. agricultural fields and urban areas), which may inhibit emigration. If mice residing within these “patch islands” have a difficult time dispersing to new environments and their populations soar within the small patch (Nupp and Swihart, 1998; Klein and Cameron, 2012).

Studies showed that P. leucopus does not always conform to the source- sink model (Morris and Diffendorfer, 2004; Grear and Burns, 2007; Linzey et al., 2012) which describes how populations may respond to landscapes that differ in composition (Danielson, 1992). It focuses on two habitat types: 1) an optimal habitat (i.e. the source) where individuals may achieve high fitness levels and emigration will exceed immigration, and 2) a suboptimal habitat (i.e. the sink)

36 where a lower quality habitat results in lower reproductive rates, but the number of immigrants it receives from source populations is high (Grear and Burns, 2007). In southern Québec, the Monteregian hills would be considered sources of emigrating mice, whereas smaller forest fragments found scattered between them would be considered the sinks. Although the model predicts the direction of dispersal, there have been conflicting reports on how difficult dispersal may be when travelling between fragmented forest habitats (Anderson and Meikle, 2010).

P. leucopus is a great disperser and colonizer, and a single mating pair may establish a successful population upon entering a new habitat (Cummings and Vessey, 1994). They have been found to expand their range at an astonishing rate considering their size: approximately 15km/year in the Great Lakes region (Myers et al., 2005, 2009) or 10km/year in southern Québec (Roy-Dufresne et al., in preparation), but their local movements may get hindered by various natural and human-made landscape barriers including agricultural fields (Krohne and Hoch, 1999; Klein and Cameron, 2012), water bodies (Savidge, 1973; Loxterman et al., 1998; Klee et al., 2004), and roads (Oxley et al., 1974; Merriam et al., 1989; Clark et al., 2001; McGregor et al., 2003; McLaren et al., 2011). Other studies, however, indicated that P. leucopus showed high levels of dispersal regardless of their surroundings (Sheppe, 1965; Middleton and Merriam, 1981; Wegner and Merriam, 1990; Mossman and Waser, 2001). If the dispersal of P. leucopus populations in the Montérégie was unhindered by environmental barriers, Lyme disease would have the potential to spread fairly quickly throughout southern Québec.

Because of the complexity of their transmission mechanisms, population genetic studies have become an important tool to help understand the spread of zoonotic diseases (Wilcox and Gubler, 2005). Knowledge of gene flow and dispersal rates are necessary to help develop effective control strategies for vector and reservoir species. Limited migration would allow for more detailed management strategies (e.g. reduction of specific populations), while wide

37 dispersal would indicate the need for large-scale treatments (Tabachnick and Black, 1995). Since November 2003, Québec’s health practitioners have been obligated to report any Lyme disease diagnoses to MADO (Maladies à Déclaration Obligatoire, le Ministère de la Santé et des Services Sociaux du Québec). Since then, multiple indigenous-born cases of Lyme disease have been diagnosed and approximately 10% of the ticks found in Québec were infected with B. burgdorferi (Trudel and Serhir, 2009). Studies also showed that of all the Canadian regions investigated, Québec had the most newly established tick populations (Leighton et al., 2012). If human population growth and greenhouse gas emissions remained much as they are today, it is predicted that I. scapularis numbers in southern Québec would double by 2020 (Ogden et al., 2006). This threat of a northern range expansion of I. scapularis coupled with the current northern expansion of P. leucopus in the Montérégie significantly increases the probability of humans becoming infected by B. burgdorferi, thus producing a very serious large-scale risk of Lyme disease in Québec. White-footed mice play an important role in the spread of the bacteria responsible for Lyme disease, yet there is no current information on (i) whether the mice are establishing successful populations on the mountains and forest fragments found scattered throughout the Montérégie, (ii) the gene flow between these habitable sites, and (iii) the relative effects of various barriers (i.e. agricultural fields, water bodies and highways) on mouse dispersal.

The objective of this study was to identify the processes that account for the current level and distribution of the genetic variation in populations of P. leucopus in southern Québec. With the advanced dispersal abilities and generalist behavior of the white-footed mouse, we hypothesized that its movements across the fragmented landscape of the Montérégie would not be structured in space and would not be affected by geographic distance or barriers. We conducted a population genetics study of 11 P. leucopus populations in the Montérégie region. We also incorporated I. scapularis data collected from every mouse to evaluate the strength of the relationships between P. leucopus gene flow, the black-legged

38 ticks and the threat of Lyme disease in the region. Our results are pertinent not only to help determine the future threat of Lyme disease in Québec, but also to contribute towards disease prevention and management strategies throughout the province and the rest of southern Canada.

MATERIALS & METHODS

Study sites

We trapped white-footed mice from 11 different sites in the Montérégie region in southern Québec (Figure 1, Table 1). Four of the sites (G, H, J, and K) are part of the Monteregian Hills. The other seven sites are forest fragments located between those hills. Overall, the study area covers 634 sq. km. Individual site areas ranged from 0.14 to 21.4 sq. km (Table 1). Four of the fragments (C, D, F and I) were selected due to their proximity to the mountains (less than 3 km). The other three forest fragments (A, B and E) formed a connection from the south to the north of the study area.

Field sampling

A total of 203 15m x 30m trapping grids of 28 traps each were set, and traps were separated from each other by a distance of 5 meters. We set 5-10 grids at each site and left the traps open for 2 nights before relocating them to another site. This helped us prevent overtrapping in an area where genetically related individuals would most likely reside within close proximity to each other. To capture a sufficient number of mice (30-35) at each site, we would set new grids at that site at a later date, selecting areas that had not yet been used for trapping. This increased our chances of analyzing the total genetic variability found within each population. Each site had traps opened between 6 to 13 nights total. Sherman live traps were baited with peanut butter and oatmeal, and all mice were euthanized via isofluorane inhalation followed by cervical dislocation. All

39 specimens were deposited as voucher specimens in the McGill’s Redpath Museum collection (Montreal, Québec, Canada). Every mouse was searched for the presence of ticks under the microscope, and all ticks were removed and assigned to larval, nymph or adult life-stage.

A total of 367 mice were collected (101 mice between 2007 and 2009, 44 mice in 2010, and 222 mice in 2011). We trapped 30-38 white-footed mice at each site except for Mont Yamaska (site J) where only 21 mice were trapped (Table 1).

DNA extraction and species identification

DNA was extracted from the liver tissue of the mice trapped in 2007 to 2009 using the invitrogen™ Purelink™ Genomic DNA Kit (Version A, February 2007). The following changes were made to the Mammalian Tissue and Mouse/Rat Tail Lysate protocol: we lysed the tissue overnight at 37 C and increased the centrifuge time to 2 minutes for the elution step. Liver tissue⁰ from the mice collected from 2010 to 2011 was extracted using a standard 3-day phenol/chloroform extraction procedure as described by Sambrook et al., (1989). Because there are two co-existing species of Peromyscus in our study area (P. leucopus and P. maniculatus), all of the mice were first identified to species using species-specific primers from the mitochondrial COIII sequence, as described in Tessier et al., (2004). Only Peromyscus leucopus were considered for further analyses.

Microsatellite amplification

Microsatellite markers were selected for their ease of amplification, their frequent use in Peromyscus population genetic studies (Mossman and Wasser, 2001; Anderson and Meikle, 2010; Munshi-South and Kharchenko, 2010; Munshi-South, 2012), and their high variability. The eleven microsatellites used were: PMl01, PMl03, PMl04, PMl05, PMl06, PMl09, PMl11, and PMl12 (Chirhart et al., 2005) as well as PLGT58, PLGT66 and PLGATA70 (Schmidt,

1999). All primers were marked with fluorescent labeling allowing the alleles to be detected during the genotyping process (Schuelke, 2000) (see Appendix).

Amplifications were carried out in 18μL volumes and included 1× reaction buffer, 2 mmol/L MgCl2, 2.5 mmol/L of each dNTP, 0.5μmol/L of each primer, 0.5 U (1 U ≈ 16.67 nkat) of Taq DNA polymerase, genomic template DNA (50-

250 ng per sample), and 13.83 μL of sterile H2O. Polymerase chain reaction (PCR) conditions were similar for all primers with annealing temperatures varying between them as follows: PMl01, PMl03, PMl09 & PMl12 = 54°C; PMl04, PMl06 & PMl11 = 58°C; PMl05, PLGT58 & PLGATA70 = 55°C; and PLGT66 = 57°C. Thermal cycling was carried out on a GeneAmp® PCR System 9700 (Applied Biosystems) under the following conditions: 95°C for 3 minutes; 35 cycles of 95°C for 30 seconds, the primer’s specific annealing temperature for 45 seconds, then 72°C for 45 seconds; and a final extension time at 72°C for 10 minutes. Amplification success was determined by staining the PCR products with SYBR® Green I nucleic acid gel stain (invitrogen™ Inc., Burlington, Canada), running them on a 2% agarose gel, and revealing the bands under UV light. Fourteen percent of all individuals were independently amplified and genotyped twice to confirm the observed genotypes.

All PCR products were sent to the McGill University Génome Québec Innovation Centre for genotyping, and all alleles were scored visually three separate times using GeneMarker v. 1.91 (SoftGenetics® LLC, Pennsylvania, USA).

Genetic variation

Within-population genetic diversity was estimated by testing for linkage disequilibrium, for departures from Hardy-Weinberg equilibrium, and by computing expected (HE) and observed heterozygosity (HO) using GENEPOP 4.0 (Raymond and Rousset, 1995; Rousset, 2008). Allelic richness (A), the number of private alleles (p), and the mean number of alleles per locus within each

41 population (k) were computed using FSTAT 2.9.3 (Goudet, 1995). MICROCHECKER 2.2.3 (Van Oosterhout et al., 2004) helped detect scoring errors and genotyping artefacts (i.e. null alleles, large allele dropouts and stutter bands) in the data. If null alleles were detected, we estimated their frequency with the method of Kalinowski and Taper (2006).

Population genetic structure

Several tests were used to examine inter-population genetic diversity and structure. We performed Fisher exact tests for pairwise comparisons of allele frequencies at each locus using GENEPOP 4.0. Genetic differentiation of populations was also evaluated by calculating pairwise FST values using Arlequin 3.5 (Excoffier et al., 2005) (significance assessed with a permutation test with 1000 iterations).

We used STRUCTURE 2.3 (Pritchard et al., 2000) to determine the most likely number of genetically distinct clusters (K) of populations (length of burn-in period at 100,000, number of MCMC replicates after burn-in at 1,100,000). Posterior probabilities were calculated for K=1 to K=n+1 (where n = number of populations sampled) using a no admixture model, and allele frequencies were assumed independent among populations (calculations were repeated 10 times for each K). The K associated with the highest ∆K value became the optimal number of populations to assign the individuals to (Evanno et al., 2005; Martien et al.,

2007). We also performed a hierarchical agglomerative clustering with our FST values using the “cluster” package in R v.2.15 (R Development Core Team, 2012) to construct a UPGMA tree and compute its agglomerative coefficient. This coefficient, ranging from 0 to 1, describes the strength of the clustering structure with a value close to 1 indicating that a very clear structure has been determined.

To assess the effect of geographical distances on the genetic structure of P. leucopus populations, we carried out an isolation-by-distance analysis using IBDWS 2.3 (10,000 permutations) (Jensen et al., 2005). We then performed a

42 multiple regression analysis on distance matrices (MRM) to determine whether the two rivers (the Richelieu and the Yamaska River) and three major highways (Highways 10, 112 and 116) in the study area acted as dispersal barriers for migrating mice. A series of dummy variables [0, 1] were used to characterize the presence or absence of ever barrier between each site. We assigned a value of 0 to pairs of populations located on the same side of a barrier and a value of 1 to pairs of populations located on opposite sides of the barrier. Distance matrices were then computed from this matrix. We considered the matrix of genetic distances

(FST values) as the dependent variable and the barrier distance matrices as the independent variables. The MRM analysis was conducted using the “ecodist” package in R v.2.15 (R Development Core Team, 2012), and both the standardized regression coefficients (b) and the coefficient of determination (R2) were tested for significance with 1000 permutations, in the same fashion as the Mantel test (Mantel, 1967). Legendre et al.’s (1994) backwards elimination procedure was used to help determine the individual matrices that contributed significantly to the model (significant p-values were determined after Bonferroni corrections). We then tested whether the total number of barriers between populations (i.e. sum of individual barrier matrices) affected genetic distance. To further evaluate how agriculture affects dispersal, we conducted an IBD analysis on the five populations not separated by highways or water bodies (Figure 1).

We grouped populations according to whether the barrier(s) surrounding them were found by the MRM analysis to be significant or not (Figure 1). The significance of these population clusters was then tested by computing an AMOVA in Arlequin (1000 permutations) (Excoffier et al., 1992; Michalakis and Excoffier, 1996). Next, we conducted assignment tests using GENECLASS2 (Piry et al., 2004) where assignments of individuals to populations were carried out using the Bayesian algorithm of Rannala and Mountain (1997). Several assignment tests were performed: in the first test, we considered the 11 populations as independent units while in a second test, we merged the populations according to the barriers deemed significant after the MRM analysis.

To visualize the genetic relatedness between pair-wise combinations of our populations, we used the Landscape Genetics ArcToolbox v. 1.2.3 (Etherington, 2011) in ArcGIS v. 10 (ESRI) to construct a gene flow network map. Edges were drawn between two populations only when the corresponding

FST values were not significant.

We calculated the effective number of migrants (Nm) between mountains and their closest forest fragment patch using MIGRATE 3.2.19 (Beerli, 2002) to test whether the different populations match the source-sink hypothesis. We used a single stepwise mutation model and assumed a migration matrix model with variable theta (Θ) and a constant mutation rate for all loci. Start values for the migration parameter were generated from the FST calculations. To ensure that the gene flow estimates were accurate, MIGRATE was run three times to verify that final chains were estimating similar values for Θ and for 4Nm. Values of 4Nm were then divided by four to compare levels of gene flow between populations (i.e. Nm), which were also summed to provide estimates for the overall immigration and emigration rates of each population.

We tested the source-sink model further using the ratio of private alleles between sites to calculate the gene flow symmetry (GFS) index, as described by Fedorka et al., (2012). GFS values were calculated using the PopGenKit (Rioux Paquette, 2011) in R v.2.15 (R Development Core Team 2012). Standard errors were estimated with 1000 bootstrap replicates (Efron and Tibshirani, 1993). Symmetrical gene flow would yield a GFS index of 1, whereas asymmetrical gene flow from site i to j would yield a GFS index < 1, or > 1 if asymmetrical from site j to i.

In all tests where multiple comparisons were computed, probability values were adjusted with sequential Bonferroni corrections (Rice, 1989) to help minimize type I errors.

RESULTS

Ticks

Approximately half of the mice trapped at sites C, D and E were carrying ticks (41%, 52% and 62% of total mice, respectively). They also hosted the largest number of ticks per mouse compared to all other sites (ranged from 1 to 8 larvae, Table 2). Site E is the southernmost population in our study and sites C and D are located at the extreme east and west ends of the map, respectively. All other sites had considerably lower numbers of mice carrying ticks (ranged from 3 - 10% of total mice trapped) as well as number of ticks per mouse (ranged from 1 to 3 larvae or 1 nymph).

Mice – Genetic variation

Results for genotypic linkage disequilibrium between microsatellite loci over our entire data set showed that 7 locus pairs were in linkage disequilibrium out of 605 pairs of loci. These significant values occurred in 5 of the 11 populations, but there was no apparent pattern in the significance of disequilibrium between locus pairs (i.e. no locus pairs were consistently in linkage disequilibrium). All loci were thus considered independent for further analyses. Hardy–Weinberg equilibrium tests detected five deviations out of 121 tests occurring in four different populations at four different loci (Site A = Pml01; Site F = Pml03 & Pml06; Site H = PMl09; Site J = PMl03). Although assortative mating or high levels of genetic drift may have induced these deviations from HWE, we also considered the possibility of the presence of null alleles in the data. MICROCHECKER 2.2.3 revealed that Pml09 may contain null alleles at site H while Pml03 was likely to contain null alleles at sites E, F, G, and J. The estimated null allele frequencies for the Pml09 locus were low (0.079-0.098), and since it was considered to possibly have null alleles at site H only, we kept this marker in the analysis. Pml03 on the other hand was considered likely to contain null alleles in four of the 11 populations, and estimated null allele frequencies

45 ranged from 0.08 to 0.22. Sites whose Pml03 locus departed from HWE (F and J) ranged on the higher end of the estimated null allele frequency spectrum (0.21 and 0.22, respectively). We conducted both intra- and inter-population analyses with and without the Pml03 locus and found no significant differences between the results of the two analyses. We therefore retained the Pml03 marker in our study.

The 11 loci were polymorphic and all populations were highly variable (Table 3). A total of 241 different alleles were obtained across all loci with the number of alleles per microsatellite ranging from 14 to 29. This high number of alleles per marker is consistent with other P. leucopus population genetic studies (Mossman and Waser, 2001; Chirhart et al., 2005; Anderson and Meikle, 2010; Munshi-South and Kharchenko, 2010).

Of the 241 alleles, 47 were private alleles found at low frequencies. The highest number of private alleles (p) was found at site C (13 alleles) followed by site D (9 alleles). These two sites are located on the western and eastern extremities of the study area, respectively. All other sites had from 0 to 4 private alleles. Allelic richness (A) was similar across all populations and ranged from 9.1 to 10.7 (Table 3).

The mean observed and expected heterozygosity values (HO and HE) were estimated at 0.873 (0.832–0.917) and 0.852 (0.830–0.869), respectively. These high allele numbers and heterozygosity values across all populations revealed high polymorphism among the 11 white-footed mouse populations.

Mice – Population genetic structure

Fisher Exact tests indicated that the two populations located to the west of the Richelieu River (sites D and H, Figure 1) exhibited the highest degree of allele frequency differentiation (Table 4). Both sites D and H each had 48 significant differences out of 110 pairwise comparison tests, yet allele frequencies were similar between them (2/11 significant differences). Besides Mont St. Bruno (site H) to the west of the Richelieu River, the other three mountains included in this

46 study had low allelic differentiation when compared with other sites. Mont Rougemont (site G) had 11 significant differences, Mont Yamaska (site J) had 13 significant differences, and Mont St. Hilaire (site K) had 15 significant differences out of 110 pairwise comparison tests performed for each population. Allele differentiation was slightly higher in the remaining forest fragments (sites A, B, C, E and F with 22, 18, 21, 23, and 18 significant differences out of 110 pairwise comparison tests, respectively), except for site I which had the lowest allele frequency of all sites (9 significant differences).

A majority of the FST values were significant (41 out of 55) (Table 4). Site D and H were the most distinct populations exhibiting differences from all other populations. Site C also exhibited differences from all other populations except for site J. Site E differed from all other populations except for site F which differed from all populations as well except sites E and G. The remaining populations varied in terms of differentiation and did not exhibit a clear pattern.

We found two distinct population clusters using STRUCTURE revealed by the highest mean posterior probability and ΔK method (Figure 2). One cluster corresponded to the sites located to the west of the Richelieu River (sites D and H), and the second cluster corresponded to all other sites (i.e. those located to the east of the Richelieu River). Within the second cluster, sites E and F also show some differentiation from all other sites but not as well defined as for sites D and H. The hierarchical clustering tree generated with the UPGMA method showed similar separation of sites D and H from all other populations as well as clustering of sites E and F. Sites C and J (located to the east of the Yamaska River) were also clustered together, and the remaining sites were all grouped together in a single cluster (agglomerative coefficient = 0.78) (Figure 3A).

We found a significant pattern of isolation by distance among the 11 populations. The pairwise genetic distances (FST) increased with the logarithm of the geographic distance (slope=0.053, r=0.32; p<0.04) (Figure 3B). A similar analysis conducted on the five populations separated by agricultural fields only

(sites A, B, G, I and K) was not significant (p=0.7). For this subsample, only a few FST values were significant (2 out of 10 values) and allelic differentiation was very low, indicating high levels of gene flow among these sites.

Barrier effect

A multiple regression analysis on distance matrices with backward elimination, using FST values as the dependant variable and geographic barriers as the independent variables, showed that the only barrier kept in the model was the Richelieu River (Table 5). We conducted a second MRM analysis without sites D and H and their associated barriers (the Richelieu River and Highway 116). This second analysis revealed that only the Yamaska River was significant in the model (Table 5). When individual barrier matrices were combined into a single matrix, both analyses (with and without sites D and H) revealed that the number of barriers between populations was significantly related to genetic distances (Table 5) indicating that the mice experience isolation by resistance within the study area (McRae, 2006). Although the multiple regression analysis did not consider the highways to be significant barriers, the inter-population differentiation of sites E and F from all other sites, and the effect of the total number of barriers on genetic variability, indicate that rather than absolute barriers, the highways may reduce dispersal between populations.

We then used the results of the MRM analyses to group the populations into three clusters separated by the Richelieu River to the west and the Yamaska River to the east (Figure 1). An AMOVA indicated that while most of the variance in the model was found within populations, the variance proportions among populations and among the three groups were also significant (Table 6).

The success rate of assigning individuals to their population of origin ranged from 59 to 97%. Sites B and I were the only populations whose assignment success rates dropped below 70%. For both sites, 19% and 20% of their individuals (respectively) were assigned to site G. When the populations

48 were grouped into three clusters (Figure 1), assignment success rates were very high for Groups 1 (96%) and 3 (93%) but significantly dropped for Group 2 (69%). Within Group 2, 43% of the individuals from site J and 21% of individuals from site C were assigned to Group 1.

The gene flow network map (Figure 3C) illustrates the connections between the pairs of populations that are not genetically distinct from one another based on FST values. The most differentiated sites occurred west of the Richelieu River, and the least differentiated sites were those found between the Richelieu and Yamaska Rivers (where only an agricultural matrix separated the sites) as well as site J (Mont Yamaska).

Source-sink

Using MIGRATE, we found low Nm values (0.18 to 0.43) between populations and low overall immigration and emigration rates (Table 7). Emigration rates from mountain populations into their closest fragments were similar to the emigration rates from fragment populations into mountains. Immigration rates also did not show any apparent patterns supporting the source-sink hypothesis. Lastly, there was no significant relation between pairwise migration rates and geographic distance between sites (p=0.2). Although MIGRATE did not detect a source-sink pattern between the mountain populations and their nearest forest fragment populations, we detected asymmetrical gene flow in that direction between two mountain-fragment pairs with a GFS index greater than one (Table 8).

DISCUSSION

Based on our knowledge of the biology of the white-footed mouse, we hypothesized that this widespread, generalist and opportunistic species would not display any strong spatial structure. Surprisingly, our population genetic analysis

49 revealed that this hypothesis did not hold and that the mice in the Montérégie were, to some degree, structured in space. Although isolation by distance was significant for the 11 populations collectively, this relation was weak. The remaining variability in our data was largely due to the number of landscape barriers between our populations (i.e. isolation by resistance) as well as the barrier effect of the largest water body found in our study area, the Richelieu River. The isolation by resistance (IBR) model considers the varying degrees of resistance that heterogeneous landscape surfaces offer to dispersing organisms, which directly affects the genetic differentiation among sampled populations (McRae, 2006). We found that as the number of barriers increased between the sampled populations, so did their genetic differentiation. This indicates that the mice experienced isolation by resistance within our study area, but the Richelieu River had the strongest barrier effect on their dispersal.

White-footed mice are able to swim but are not considered the strongest of swimmers (Klee et al., 2004), and water bodies only 3-4 m wide (ranging in depth from 10-40 cm) have proven to be significant dispersal barriers (Savidge, 1973), as have rivers (Klee et al., 2004) and ocean crossings between island populations (Loxterman et al., 1998). The Richelieu River was the strongest dispersal barrier, largely isolating the two sites to the west of it. Both of these sites showed consistent and strong genetic differentiation from all other sites.

We found that the second main water body in our study area, the Yamaska River, did not have as much influence on dispersal as the Richelieu River. Gene flow between sites located on either side of the Yamaska River was higher than expected, particularly with Mont Yamaska and multiple other sites located just west of the river. When considering the Montérégie region on a larger scale, its south-eastern landscape is composed of a greater number of forest fragments than what is observed in our study area (Figure 1) thus possibly creating an effective southern connection for dispersing mice. White-footed mouse can move great distances and have been found to cover a very wide range of dispersal distances

50 throughout their lifespan. They may travel anywhere between 85-867 m (mean ≈330 m) (Krohne et al., 1984) with some extreme dispersers found 5-14 km away from their original site of capture (Maier, 2002). Water is known to impede dispersal (Savidge, 1973; Loxterman et al., 1998; Klee et al., 2004) therefore it is likely that the rivers create strong dispersal barriers. In our study area however, mouse populations may be linked through this southern connection helping them make their way around the river. It would be important to confirm this hypothesis by trapping and analyzing mice from sites in the southern connection and from sites located in the north-eastern corner of our study area where the Yamaska River branches off and isolates scattered patches (Figure 1). The Yamaska River runs very close along the northern end of Mont Yamaska. This river is edged along its length by a riparian forest buffer, giving it great potential as a dispersal corridor for the mice. White-footed mice have been found to use a large array of corridor types when moving between fragmented forest habitats including roadside ditches (Cummings and Vessey, 1994), simple fences lined with vegetation (Merriam and Lanoue, 1990) and stone walls (Parren and Capen, 1985). Merriam and Lanoue, (1990) found that the mice used all categories of fencerows as corridors, ranging from simple fencerows less than 1 m in width with less than 10% of vegetative cover, to complex fencerows greater than 2 m in width with trees and full ground cover present in more than 10% of the fencerow’s length. Yet of all the corridors used, the mice preferred the structurally complex fencerows. Riparian zones are generally cool and wet, structurally complex and productive, and are used by multiple small mammal species, making them useful towards conservation efforts within areas of clear-cutting (Cockle and Richardson, 2003). It is likely that the riparian buffer zone along the Yamaska River provides the ideal structural complexity favoured by dispersing mice, creating an efficient corridor for western movement within our study area. A detailed population connectivity study assessing the natural and man-made corridors that link forest patches may help explain the high level of gene flow observed between sites across the Yamaska River yet not across the Richelieu River.

Mont Yamaska was also the only study site for which the lowest sample size was used (21 individuals), and where most of the land is privately owned. Thirteen of the thirty four mice we trapped on this mountain were identified as Peromyscus maniculatus, the deer mouse. The white-footed mouse and the deer mouse are both highly abundant throughout North America and have been used as important research organisms in a wide range of disciplines (Dewey and Dawson, 2001). They share the same food resources, use the same microhabitats and nest sites, are territorial, and host similar parasitic infections (Wolff, 1985). Various morphological traits are used to help distinguish the two species from one another including tail size, pelage coloration and craniometric variables (Rich et al., 1996; Millien et al., unpublished), but the two species are very similar in their morphology. Species identification using molecular markers allows for accurate results (Aquadro and Patton, 1980; Tessier et al., 2004). The white-footed mouse has expanded its northern range limit into northern USA (Myers et al., 2005) and southern Québec (Millien et al., unpublished) over the last few decades, and Mont Yamaska is the only site in our study where we caught a large number of deer mice. Although abundant in the past (Grant, 1976), the deer mouse seems to have drastically declined in abundance to the point that most Peromyscus mice we caught in the Montérégie were identified as P. leucopus.

The Montérégie is part of the Great Lakes and St. Lawrence River forest system (D’Orangeville et al., 2008). This mixed forest system is considered a transitional zone between the deciduous forests of eastern North America and the boreal forest in northern Canada (Ontario Ministry of Natural Resources, 2012). It is composed of a mix of coniferous trees (eastern white pine, red pine, eastern hemlock, and white cedar) and deciduous broad-leaved species (yellow birch, sugar maple, red maple, basswood, and red oak). Some common boreal species are also found in this system and include white spruce, black spruce, jack pine, balsam poplar, white birch, and trembling aspen (Forest Products Association of Canada, 2012). Although both species of mice are found in a variety of habitat

52 types, P. maniculatus has been observed to favor boreal forests (Myers et al., 2005) and shows better adaptation to cold weather than P. leucopus (Pierce and Vogt, 1993). Trees running along the Montérégie’s south-west to north-east gradient become less diversified, and more boreal tree species are found in mid to high elevations (Bouchard and Maycock, 1978). We observed that Mont Yamaska contained a greater number of boreal tree species than all other sites and was the tallest of the four mountains included in this study. The recent climate warming in the Montérégie region may have pushed the deer mice further north, yet a population may have remained on Mont Yamaska where cooler temperatures (due to its higher elevation), and a boreal microhabitat, provided a suitable habitat for this group. Boreal microhabitats and their associated colder winters may not provide optimal conditions for P. leucopus. Sites mainly consisting of boreal flora associated with deer mice are important to consider when assessing the dispersal rates of white-footed mice throughout southern Québec.

Agricultural fields were the weakest dispersal barriers to the white footed mice in our study area. Previously published results on the effect of agricultural fields on the movement of P. leucopus are conflicting: Agricultural fields have been found to act as significant dispersal barriers (Krohne and Hoch, 1999; Klein and Cameron, 2012), as seasonal barriers where crop height and maturity determine whether the mice use them as dispersal routes and/or food sources (Cummings and Vessey, 1994; Anderson et al., 2006), or not as barriers at all (Middleton and Merriam, 1981; Mossman and Waser, 2001). Our results contradict studies concluding that an agricultural matrix is a hard barrier for white-footed mouse dispersal. Since we trapped mice between the months of July and September when crops were high enough to provide adequate cover and food for migrating mice, it is possible that the fields were used as seasonal corridors between forest patches. Regardless, whether they are used seasonally or throughout the year, the high levels of gene flow observed in our study strongly suggest that dispersal within the agricultural matrix is effective. These results support our original hypothesis that if mouse habitats are surrounded by

53 agricultural fields, but lack major roads or water bodies between sites, their populations would not be genetically structured.

Our results indicated that highways 10 and 112 slowed dispersal but did not stop it, as seen in numerous other studies (Oxley et al., 1974; Merriam et al., 1989; Clark et al., 2001; McGregor et al., 2003; McLaren et al., 2011). We did not find any effect of highway 116 on white-footed mouse dispersal which may be due to the fact that this road runs along the Richelieu River, which was identified as the main barrier in our study system. Incorporating more populations on the western side of the Richelieu River to this dataset may help confirm the hypothesis of a negative effect of highway 116 on mouse dispersal.

Based on the effective number of migrants (Nm), the mouse populations in our study system did not conform to the source-sink model. The lack of a source- sink pattern is not uncommon for white-footed mice (Morris and Diffendorfer, 2004; Grear and Burns, 2007; Linzey et al., 2012), and sink patches have been found to regulate themselves if individuals are interactive and resident (Linzey et al., 2012). Gene flow asymmetry was nevertheless detected between certain mountain to fragment populations when considering the ratio of private alleles at every site. This may indicate subtle gene flow from source to sink populations, but Nm values were not high enough to assume recurrent extinction and recolonization of sink populations by their closest source.

Our study was designed to better understand the pattern of spread of Lyme disease within the Montérégie region and to help identify areas of greater risk for human populations. We observed that different type of landscape barriers vary in their effectiveness as disrupters to P. leucopus dispersal. Agricultural matrices may pose the greatest challenge to Lyme disease management strategies as they did not hinder gene flow between sites. Québec’s south-western landscape has largely been developed for farming purposes (for both crops and livestock) resulting in the fragmentation of a once-continuous forest (Wampach, 1988;

Raudsepp-Hearne et al., 2010). The high permeability of the agricultural matrix observed in our study suggests that P. leucopus may have the potential to spread fairly quickly. The high number of ticks found in site D also increases the chances of the Lyme disease vector spreading west along with the mice at a rapid rate. Approximately 15 km west and 27 km north of our study area is the St. Lawrence Seaway, which runs along southern Québec (just below the city of Montréal) and connects Ontario’s Great Lakes to the Atlantic Ocean. The Richelieu River had a strong barrier effect on the white-footed mouse populations in our study, therefore we can make the hypothesis that the St. Lawrence Seaway could be just as effective and impede the northern expansion of this species. Several high-traffic highways located just before the seaway may likely slow dispersal further (highways 15, 20 and 30). At first glance, south-eastern Québec may pose greater challenges for Lyme disease management strategies as there are no main water bodies intersecting it in the north-south direction. South-eastern Québec is much less fragmented than its western side (Figure 1) largely due to the ecosystem services it provides (i.e. maple syrup production, recreational hunting and general forest recreation) (Raudsepp-Hearne et al., 2010). But this increase in patch connectivity and forest patch size may provide a naturally effective control on P. leucopus population abundance.

A larger forested area and increased connectivity between patches attracts a larger variety of species that generally do poorly in small, isolated forest fragments. Larger and more connected fragments thus become suitable habitat for different species, and biodiversity within the patch will increase. Studies conducted by LoGiudice et al., (2003) show that vertebrate species richness positively correlates with forest patch size and negatively correlates with white- footed mouse abundance. Increasing the biodiversity within a patch creates a dilution effect – if infected ticks are feeding on a variety of species that have a lower reservoir competence, the bacteria will either be destroyed by the host’s immune response or will have a hard time being transmitted to future feeding ticks. This, in turn, lowers the chances of P. leucopus (whose reservoir

55 competence is quite high) from becoming infected (Van Buskirk and Ostfeld, 1995). Another positive advantage to increased biodiversity within a patch is that it will heighten resource and habitat competition as well as the number of predators that would feed on the mice, thus reducing the white-footed mouse’s home range and abundance. Finally, a phenomenon called encounter reduction may also arise – a constrained home range decreases the chances of a mouse from encountering a tick or from fighting with an infected mouse and acquiring the bacteria through bodily fluids (Brunner and Ostfeld 2008). Southern Québec has the highest biodiversity level within the province, but also the greatest number of species threatened by forest fragmentation (Auzel et al. 2012). Maintaining high biodiversity levels, or increasing biodiversity in areas dominated by agricultural fields and sparse forest habitats, could provide efficient control on P. leucopus population abundance and dispersal, thus reducing the rate of spread of Lyme disease in southern Québec.

Although the white-footed mouse plays an important role as a reservoir for the bacterium that causes Lyme disease, this bacterium also requires the black- legged tick (i.e. the disease vector) to transmit itself to vertebrate hosts for cycle completion. A survey of over 50 sites in southern Québec revealed the presence of ticks in vegetation approximately 100 km north of our study area (Bouchard et al., 2011). Tick populations may reach new regions via their vertebrate hosts, but are most likely dispersed over large geographic distances by migratory birds (Ogden et al., 2010). The success of the establishment of these ticks in their new northern environments will then depend on the suitability of their surrounding climatic conditions (Ogden et al., 2005) in addition to the presence of their hosts. Currently, the northern edge of tick and Lyme disease emergence is in southern Québec (Ogden et al., 2010), and although ticks are carried further north into the province, their populations cannot establish themselves successfully due to current climatic conditions (Ogden et al., 2006). This confirmed tick and Lyme disease emergence in southern Québec, coupled with the range expansion and population abundance of the efficient Lyme disease reservoir P. leucopus, shows

56 how important it is to merge P. leucopus population analyses like ours with ongoing Lyme disease research in Québec. Such combined efforts may then be applied towards disease management strategies within Québec and within other Canadian provinces with similar ecological conditions to help understand how the disease may proliferate.

REFERENCES

Allan, B. F., Keesing, F., & Ostfeld, R. S. (2003). Effect of forest fragmentation on Lyme disease risk. Conservation Biology, 17: 267-272.

Almazan-Garcia, C. (2005). Identification and characterization of protective antigens for control of Ixodes scapularis infestations. PhD Thesis, Oklahoma State University, Stillwater.

Anderson, C., Meikle, D., Cady, A., & Schaefer, R. (2006). Annual variation in habitat use by White-footed Mice, Peromyscus leucopus: The effects of forest patch size, edge and surrounding vegetation type. Canadian Field- Naturalist, 120: 192-198.

Anderson, J. F., Johnson, R. C., & Magnarelli, L. A. (1987). Seasonal prevalence of Borrelia burgdorferi in natural populations of white-footed mice, Peromyscus leucopus. Journal of Clinical Microbiology, 25: 1564-1566.

Aquadro, C. F. & Patton, J. C. (1980). Salivary amylase variation in Peromyscus: Use in species identification. Journal of Mammalogy, 61: 703-707.

Auzel, P., Gaonac’h, H., Poisson, F., Siron, R., Calmé, S., et al. (2012). Impacts of climate change on the biodiversity of Quebec: Literature review summary. CSBQ, MDDEP, Ouranos. 29 p.

Beerli, P. (2002). MIGRATE (version 3.2.19) – documentation and program. Available on the site http://popgen.sc.fsu.edu/Migrate/Migrate-n.html

Bouchard, A. & Maycock, P. F. (1978). Les fôrets décidues et mixtes de la région appalachienne du sud québécois. Le Naturaliste Canadien, 105: 383-415.

Bouchard, C., Beauchamp, G., Nguon, S., Trudel, L., Milord, F., et al. (2011). Associations between Ixodes scapularis ticks and small mammal hosts in a newly endemic zone in southeastern Canada: Implications for Borrelia burgdorferi transmission. Ticks and Tick-borne Diseases, 2: 183-190.

Brownstein, J. S., Holford, T. R., Fish, D. (2005). Effect of climate change on Lyme disease risk in North America. Ecohealth, 2: 38-46.

Brunner, J.L., & Ostfeld, R.S. (2008). Multiple causes of variable tick burdens on small mammal hosts. Ecology, 89: 2259-2272.

Chirhart, S. E., Honeycutt, R. L., & Greenbaum, I. F. (2005). Microsatellite variation and evolution in the Peromyscus maniculatus species group. Molecular Phylogenetics and Evolution, 34: 408-15.

Clark, B., Clark, B., Johnson, L., & Haynie, M. (2001). Influence of roads on movements of small mammals. The Southwestern Naturalist, 46: 338-344.

Cockle, K. L. & Richardson, J. S. (2003). Do riparian buffer strips mitigate the impacts of clearcutting on small mammals? Biological Conservation, 113: 133-140.

Cummings, J., & Vessey, S. (1994). Agricultural influences on movement patterns of white-footed mice (Peromyscus leucopus). American Midland Naturalist, 132: 209-218.

Danielson, B. J. (1992). Habitat selection, interspecific interactions and landscape composition. Evolutionary Ecology, 6: 399-411.

Dewey, M. J., & Dawson, W. D. (2001). Deer Mice: "The Drosophila of North American mammalogy". Genesis, 29: 105-109.

Efron, B. & Tibshirani, R. J. (1993). An introduction to the bootstrap. Chapman & Hall/CRC, New York. pp. 443.

Eisen, R., Piesman, J., Zielinski-Gutierrez, E., & Eisen, L. (2012). What do we need to know about disease ecology to prevent Lyme disease in the northeastern United States? Journal of Medical Entomology, 49: 11-22.

Etherington, T. R. (2011). Python based GIS tools for landscape genetics: Visualising genetic relatedness and measuring landscape connectivity. Methods in Ecology and Evolution, 2: 52-55.

Evanno, G., Regnaut, S., & Goudet, J. (2005). Detecting the number of clusters of individuals using the software STRUCTURE: A simulation study. Molecular Ecology, 14: 2611-2620.

Excoffier, L., Smouse, P.E., Quattro, J.M. (1992). Analysis of molecular variance inferred from metric distances among DNA haplotypes: Application to human mitochondrial DNA restriction data. Genetics, 131: 479-491.

Excoffier, L., Laval, G., & Schneider, S. (2005). Arlequin version 3.0: An integrated software package for population genetics data analysis. Evolutionary Bioinformatics Online, 1: 47-50.

Fedorka, K. M., Winterhalter, W. E., Shaw, K. L., Brogan, W. R., & Mousseau, T. A. (2012). The role of gene flow asymmetry along an environmental gradient in constraining local adaptation and range expansion. Journal of Evolutionary Biology, 25: 1676-1685.

Forest Products Association of Canada. (2012). Forest regions of Canada. Retrieved from http://www.fpac.ca/index.php/en/interactive-map/

Goudet, J. (1995). Fstat version 1.2: A computer program to calculate Fstatistics. Journal of Heredity, 86: 485-486.

Government of Canada, Public Health Agency of Canada (2012). Lyme disease and other tick-borne diseases: Information for health professionals. Retrieved from http://www.phac-aspc.gc.ca/id-mi/tickinfo-eng.php

Grant, P. R. (1976). An 11-year study of small mammal populations at Mont St . Hilaire, Quebec. Canadian Journal of Zoology, 54: 2156-2173.

Grear, J. S., & Burns, C.E. (2007). Evaluating effects of low quality habitats on regional population growth in Peromyscus leucopus: Insights from field- parameterized spatial matrix models. Landscape Ecology, 22: 45-60.

Halos, L., Bord, S., Cotté, V., Gasqui, P., Abrial, D., et al. (2010). Ecological factors characterizing the prevalence of bacterial tick-borne pathogens in Ixodes ricinus ticks in pastures and woodlands. Applied and Environmental Microbiology, 76: 4413-4420.

Hofmeister, E. K., Ellis, B. A., Glass, G. E., & Childs, J. E. (1999). Longitudinal study of infection with Borrelia burgdorferi in a population of Peromyscus leucopus at a Lyme disease-enzootic site in Maryland. The American Journal of Tropical Medicine and Hygiene, 60: 598-609.

Jensen, J. L., Bohonak, A. J., & Kelley, S. T. (2005). Isolation by distance, web service. BMC Genetics, 6: 13.

Kalinowski, S. T., & Taper, M. L. (2006). Maximum likelihood estimation of the frequency of null alleles at microsatellite loci. Conservation Genetics, 7: 991-995.

Klee, R., Mahoney, A., Christopher, C., & Barrett, G. (2004). Riverine peninsulas: An experimental approach to homing in white-footed mice (Peromyscus leucopus). American Midland Naturalist, 151: 408-413.

Klein, G. P. & Cameron, G. N. (2012). Effect of habitat gradients on space use by white-footed mice (Peromyscus leucopus). Journal of Mammalogy, 93: 706- 715.

Krohne, D. T., Dubbs, B., & Baccus, R. (1984). An analysis of dispersal in an unmanipulated population of Peromyscus leucopus. American Midland Naturalist, 112: 146-156.

Krohne, D. T., & Hoch, G. A. (1999). Demography of Peromyscus leucopus populations on habitat patches: The role of dispersal. Canadian Journal of Zoology, 77: 1247-1253.

Legendre, P., Lapointe, F.-J., & Casgrain, P. (1994). Modeling brain evolution from behavior: A permutational regression approach. Evolution, 48: 1487- 1499.

Leighton, P. A., Koffi, J. K., Pelcat, Y., Lindsay, L. R., & Ogden, N. H. (2012). Predicting the speed of tick invasion: An empirical model of range expansion for the Lyme disease vector Ixodes scapularis in Canada. Journal of Applied Ecology, 49: 457-464.

Linzey, A. V., Reed, A. W., Slade, N. A., & Kesner, M. H. (2012). Effects of habitat disturbance on a Peromyscus leucopus (Rodentia: Cricetidae) population in western Pennsylvania. Journal of Mammalogy, 93: 211-219.

LoGiudice, K., Ostfeld, R. S., Schmidt, K. A. (2003). The ecology of infectious disease: Effects of host diversity and community composition on Lyme disease risk. Proceedings of the National Academy of Sciences, 100: 567- 571.

Maier, T. (2002). Long-distance movements by female white-footed mice, Peromyscus lecopus, in extensive mixed-wood forest. Canadian Field- Naturalist, 116: 108-111.

Mantel, N. (1967). The detection of disease clustering and a generalized regression approach. Cancer Research, 27: 209-220.

Martien, K., Givens, G., & Archer, E. (2007). A note on the ability of STRUCTURE to correctly infer the number of populations for Bering- Chukchi-Beaufort Seas bowhead whales. Whaling Commission, Paper SC/5, 1-8.

McLaren, A., Fahrig, L., & Waltho, N. (2011). Movement of small mammals across divided highways with vegetated medians. Canadian Journal of Zoology, 89: 1214-1222.

McRae, B.H. (2006). Isolation by resistance. Evolution, 60: 1551-1561.

Merriam, G., Kozakiewicz, M., Tsuchiya, E., & Hawley, K. (1989). Barriers as boundaries for metapopulations and demes of Peromyscus leucopus in farm landscapes. Landscape Ecology, 2: 227-235.

Merriam, G. & Lanoue, A. (1990). Corridor use by small mammals: Field measurement for three experimental types of Peromyscus leucopus. Landscape Ecology, 4: 123-131.

Michalakis, Y., & Excoffier, L. (1996). A genetic estimation of population subdivision using distances between alleles with special reference for microsatellite loci. Genetics, 142: 1061-1064.

Middleton, J., & Merriam, G. (1981). Woodland mice in a farmland mosaic. Journal of Applied Ecology, 18: 703-710.

Morris, D. W., & Diffendorfer, J. E. (2004). Reciprocating dispersal by habitat- selecting white-footed mice. OIKOS, 107: 549-558.

Mossman, C. A., & Waser, P. M. (2001). Effects of habitat fragmentation on population genetic structure in the white-footed mouse (Peromyscus leucopus). Canadian Journal of Zoology, 79: 285-295.

MRNQ (2000). Ministère des Ressources naturelles du Québec, Système d’information écoforestière (SIEF). Forêt Québec, Direction des inventaires forestiers, Québec, Québec.

Munshi-South, J. (2012). Urban landscape genetics: Canopy cover predicts gene flow between white-footed mouse (Peromyscus leucopus) populations in New York City. Molecular Ecology, 21: 1360-1378.

Munshi-South, J., & Kharchenko, K. (2010). Rapid, pervasive genetic differentiation of urban white-footed mouse (Peromyscus leucopus) populations in New York city. Molecular Ecology, 19: 4242-4254.

Myers, P., Lundrigan, B., & Kopple, R. V. (2005). Climate change and the distribution of Peromyscus in Michigan: Is global warming already having an impact. In E. Lacey & P. Myers (Eds.), Mammalian diversification: From chromosomes to phylogeography (a celebration of the career of James L. Patton) (Vol. 133, pp. 101-126). University of California Press.

Myers, P., Lundrigan, B. L., Hoffman, S. M. G., Haraminac, A. P., & Seto, S. H. (2009). Climate-induced changes in the small mammal communities of the Northern Great Lakes region. Global Change Biology, 15: 1434-1454.

Nupp, T., & Swihart, R. (1996). Effect of forest patch area on population attributes of white-footed mice (Peromyscus leucopus) in fragmented landscapes. Canadian Journal of Zoology, 74: 467-472.

Nupp, T., & Swihart, R. (1998). Effects of forest fragmentation on population attributes of white-footed mice and eastern chipmunks. Journal of Mammalogy, 79: 1234-1243.

Ogden, N. H., Maarouf, A., Barker, I. K., Bigras-Poulin, M., Lindsay, L. R., et al. (2006). Climate change and the potential for range expansion of the Lyme

disease vector Ixodes scapularis in Canada. International Journal for Parasitology, 36: 63-70.

Ogden, N. H., St-Onge, L., Barker, I. K., Brazeau, S., Bigras-Poulin, M., et al. (2008). Risk maps for range expansion of the Lyme disease vector, Ixodes scapularis, in Canada now and with climate change. International Journal of Health Geographics, 7: 1-15.

Ostfeld, R. S. (2011). Lyme Disease: The Ecology of a Complex System. Oxford University Press Inc., New York, New York, pp. 216.

Ostfeld, R. S., & Keesing, F. (2000). Biodiversity and disease risk: The case of Lyme disease. Conservation Biology, 14: 722-728.

Oxley, D., Fenton, M., & Carmody, G. (1974). The effects of roads on populations of small mammals. Journal of Applied Ecology, 11: 51-59.

Parren, S. G. & Capen, D. E. (1985). Local distribution and coexistence of two species of Peromyscus in Vermont. Journal of Mammalogy, 66: 36-44.

Pierce, S. S., & Vogt, F. D. (1993). Winter acclimatization in Peromyscus maniculatus gracilis, P. leucopus noveboracensis, and P. l. leucopus. Journal of Mammalogy, 74: 665-677.

Piry, S., Alapetite, A., Cornuet, J.-M., Paetkau, D., Baudouin, L., & Estoup, A. (2004). GENECLASS2: A software for genetic assignment and first- generation migrant detection. The Journal of Heredity, 95: 536-539.

Pritchard, J. K., Stephens, M., & Donnelly, P. (2000). Inference of population structure using multilocus genotype data. Genetics, 155: 945-959.

R Development Core Team (2012). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.

Rannala, B., & Mountain, J. L. (1997). Detecting immigration by using multilocus genotypes. Proceedings of the National Academy of Sciences, 94: 9197- 9201.

Raudsepp-Hearne, C., Peterson, G.D., Bennett, E.M. (2010). Ecosystem service bundles for analyzing tradeoffs in diverse landscapes. Proceedings of the National Academy of Sciences, 107: 5242-5247.

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

Rice, W. (1989). Analyzing tables of statistical tests. Evolution, 43: 223-225.

Rioux Paquette, S. (2011). PopGenKit: Useful functions for (batch) file conversion and data resampling in microsatellite datasets. R package version 1.0. http://cran.r-project.org/web/packages/PopGenKit/index.html.

Rousset, F. (2008). Genepop'007: A complete reimplementation of the Genepop software for Windows and Linux. Molecular Ecology Resources, 8: 103-106.

Roy-Dufresne, E., Chmura, G., Millien, V. Northern range expansion of the white-footed mouse (Peromyscus leucopus) under climate change and its consequences on the emergence of Lyme disease in Québec. In preparation.

Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989). Molecular cloning: A laboratory manual. Chapter 6 - Preparation and Analysis of Eukaryotic Genomic DNA. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.

Savidge, I. (1973). A stream as a barrier to homing in Peromyscus leucopus. Journal of Mammalogy, 54: 982-984.

Schmidt, C.A. (1999). Variation and congruence of microsatellite markers for Peromyscus leucopus. Journal of Mammalogy, 2: 522-529.

Schuelke, M. (2000). An economic method for the fluorescent labeling of PCR fragments: A poor man’s approach to genotyping for research and high- throughput diagnostics. Nature Biotechnology, 18: 233-234.

Schwanz, L. E., Voordouw, M. J., Brisson, D., & Ostfeld, R. S. (2011). Borrelia burgdorferi has minimal impact on the Lyme disease reservoir host Peromyscus leucopus. Vector-Borne and Zoonotic Diseases, 11: 117-124.

Sheppe, W. (1965). Dispersal by swimming in Peromyscus leucopus. Journal of Mammalogy, 46: 336-337.

Stanek, G., Strle, F., Gray, J., & Wormser, G. P. (2002). History and characteristics of Lyme Borreliosis. In J.S. Gray, O. Kahl, R.S. Lane & G. Stanek (Eds.), Lyme Borreliosis: Biology, Epidemiology and Control. Wallingford, UK: CABI Publishing, pp. 1-28.

Tabachnick, W. J., & Black, W. C. (1995). Making a case for molecular population genetic studies of arthropod vectors. Parasitology Today, 11: 27- 30.

Trudel, L., & Serhir, B. (2009). Maladie de Lyme. Institut National de Santé Publique du Québec, Gouvernement du Québec, N° de publication: 1042.

Van Buskirk, J., Ostfeld, R.S. (1995). Controlling Lyme disease by modifying the density and species composition of tick hosts. Ecological Applications, 5: 1133-1140.

Van Oosterhout, C., Hutchinson, W. F., Wills, D. P. M., & Shipley, P. (2004). Micro-checker: Software for identifying and correcting genotyping errors in microsatellite data. Molecular Ecology Notes, 4: 535-538.

Wampach, J.-P. (1988). Deux siècles de croissance agricole au Québec, 1760- 1985. Recherches Sociographiques, 29: 181-199.

Wegner, J., & Merriam, G. (1990). Use of spatial elements in a farmland mosaic by a woodland rodent. Biological Conservation, 54: 263-276.

Wilcox, B. A. & Gubler, D. J. (2005). Disease ecology and the global emergence of zoonotic pathogens. Environmental Health and Preventative Medicine, 10: 263-272.

Wilder, S., Abtahi, A., & Meikle, D. (2005). The effects of forest fragmentation on densities of white-footed mice (Peromyscus leucopus) during the winter. American Midland Naturalist, 153: 71-79.

Wolff, J. (1985). Comparative population ecology of Peromyscus leucopus and Peromyscus maniculatus. Canadian Journal of Zoology, 63: 1548-1555.

Yahner, R. (1992). Dynamics of a small mammal community in a fragmented forest. American Midland Naturalist, 127: 381-391.

TABLES

Site Latitude Longitude Elevation Area (km2) Perimeter (m) Nearest mountain # P. leucopus Years collected A N45 28.91 W73 10.47 44 0.50 4787 NA 35 2010, 2011 B N45 29.01 W73 13.86 23 0.36 5624 NA 32 2010, 2011 C N45 26.97 W72 54.54 83 1.01 9806 Mont Yamaska 34 2011 D N45 29.75 W73 18.99 35 0.14 1633 Mont St. Bruno 31 2011 E N45 23.28 W73 12.09 63 4.72 23850 NA 34 2010, 2011 F N45 25.43 W73 03.92 68 0.35 3533 Mont Rougemont 35 2011 G N45 29.74 W73 04.06 89 21.4 39826 NA 35 2009, 2011 H N45 32.97 W73 18.23 90 11.6 43649 NA 37 2009 I N45 31.23 W73 12.14 59 1.50 10817 Mont St. Hilaire 35 2011 J N45 27.42 W72 51.52 295 16.3 30185 NA 21 2009, 2011 K N45 33.16 W73 09.09 244 16.8 53202 NA 38 2007-2009

72 Table 2. Tick load on trapped mice. Mice from sites H and K were collected before 2011 and were not sampled for ticks.

Site # of mice Mice with ticks Min. and max. number of ticks # % Larvae Nymphs Adults A 35 1 2.9 0 1 0 B 32 2 6.3 1 0 0 C 34 14 41.2 1 - 7 1 0 D 31 16 51.6 1 - 8 0 0 E 34 21 61.8 1 - 7 1 0 F 35 3 8.6 2 1 0 G 35 0 0 0 0 0 I 35 3 8.6 1 - 3 0 0 J 21 2 9.5 1 - 2 0 0 Total 292 62 21.2

Table 3. Sample size (n), mean number of alleles per locus (k), allelic richness (A), number of private alleles (p), and observed and expected heterozygosity (HO and HE) for 11 P. leucopus populations in the Montérégie.

Site n k A p HO HE A 35 13.7 10.7 4 0.858 0.860 B 32 11.3 9.4 4 0.846 0.844 C 34 13.3 10.7 13 0.853 0.852 D 31 11.8 9.7 2 0.902 0.830 E 34 12.0 10.0 3 0.917 0.863 F 35 13.2 10.7 9 0.874 0.869 G 35 11.9 10.2 2 0.893 0.868 H 37 10.6 9.1 3 0.861 0.853 I 35 12.1 9.9 3 0.891 0.846 J 21 9.8 9.4 0 0.872 0.845 K 38 11.8 9.7 4 0.832 0.842

Table 4. FST values (above the diagonal) for the 11 populations studied and the number of significant results for the Fisher Exact test (below the diagonal) for the ns eleven loci. Significant FST values are indicated by asterisks ( : p>0.05, **: p<0.01, ***: p<0.001). Probabilities were adjusted with sequential Bonferroni corrections.

Site A Site B Site C Site D Site E Site F Site G Site H Site I Site J Site K Site A - 0.005 ns 0.027*** 0.054*** 0.01** 0.01*** 0.004 ns 0.037*** 0.008** 0.021*** 0.016*** Site B 1/11 - 0.019*** 0.039*** 0.017*** 0.01** 0 ns 0.026*** 0.003 ns 0.01 ns 0.006 ns Site C 2/11 2/11 - 0.037*** 0.016*** 0.018*** 0.011*** 0.031*** 0.021*** 0.009 ns 0.01*** Site D 6/11 6/11 4/11 - 0.048*** 0.043*** 0.031*** 0.021*** 0.034*** 0.047*** 0.038*** Site E 2/11 2/11 2/11 6/11 - 0.006 ns 0.01*** 0.039*** 0.012*** 0.01** 0.015*** Site F 1/11 1/11 2/11 6/11 0 - 0.006 ns 0.027*** 0.012*** 0.01** 0.012*** Site G 0 0 1/11 5/11 1/11 1/11 - 0.018*** 0.002 ns 0.003 ns 0.0007 ns Site H 5/11 5/11 6/11 2/11 7/11 5/11 3/11 - 0.023*** 0.029*** 0.031*** Site I 2/11 0 1/11 3/11 1/11 1/11 0 6/11 - 0.016*** 0.006 ns Site J 1/11 1/11 0 5/11 1/11 1/11 0 4/11 0 - 0.010 ns Site K 2/11 0 1/11 5/11 1/11 0 0 5/11 1/11 0 -

Table 5. Results for the backwards elimination multiple regression analyses using FST values as the dependant variable and dispersal barriers as the independent variable (Highways 10, 112 & 116, the Richelieu and the Yamaska River) when either including or excluding sites D and H. Analyses on whether the total ns number of barriers between populations (i.e. an all-barrier matrix) affected genetic distance are shown as well, : p>0.05, *: p<0.05, **: p<0.01.

Coefficient (b) R2 F 11 populations (n = 367) All barriers separately Intercept 0.011 ns Richelieu River 0.024 * 0.71* 129.7* All barriers combined Intercept 0.006 ns Total barriers 0.013** 0.30** 23.2** 9 populations (n = 299) All barriers separately Intercept 0.005 ns Hwy 112 0.004 ns Yamaska River 0.006** 0.30 * 6.97 * All barriers combined Intercept 0.005 ns Total barriers 0.005 * 0.26 * 12.2 *

Source of variation df Variance % variance Θ

Among groups 2 0.060 1.55 ΘCT = 0.0155**

Among populations 8 0.037 0.97 ΘSC = 0.0098** within groups

Within populations 723 3.760 97.5 ΘST = 0.0251**

Site C Site D Site F Site G Site H Site I Site J Site K Total Nmb Site C 0.31 0.35 0.30 0.26 0.38 0.26 Δ 0.32 3.00 Site D 0.30 0.18 0.28 0.28~ 0.27 0.38 0.24 2.81 Site F 0.27 0.26 0.31° 0.27 0.26 0.43 0.25 2.96 Site G 0.36 0.31 0.38° 0.28 0.31 0.32 0.27 3.17 Site H 0.28 0.30~ 0.28 0.25 0.15 0.25 0.26 2.59 Site I 0.29 0.31 0.31 0.35 0.34 0.29 0.37# 3.11 Site J 0.24 Δ 0.24 0.30 0.29 0.26 0.24 0.27 2.72 Site K 0.31 0.29 0.31 0.24 0.27 0.28# 0.32 2.82 Total Nma 2.87 2.90 3.14 2.91 2.81 2.67 3.24 2.75

Site i Site j GFS index Standard error 95% confidence interval

F G 1.57 0.40 0.79 – 2.35

D H 1.49 0.23 1.04 – 1.94

C J 2.61 0.51 1.61 – 3.61

I K 1.10 0.26 0.59 – 1.61

FIGURES

Figure 3. Results of the A) hierarchical clustering tree (UPGMA), B) the isolation by distance analysis using FST values, and C) the gene flow network map (mapped with SIEF data, MRNQ 2000) showing the connection between populations.

GENERAL CONCLUSIONS

Environmental changes are important drivers for the emergence of vector-borne diseases, including Lyme disease. Both forest fragmentation and global climate change have shown to play an important role in its spatial and temporal distribution (Ogden et al., 2008; Halos et al., 2012). The Montérégie region in southern Québec has had a 0.8 C increase in its mean growing season temperature and agricultural practices have⁰ led to a 70% decrease of the local forests (Wampach, 1988) so that today, the region’s landscape is largely composed of mountains and scattered forest fragments surrounded by crops and urban habitat. This fragmented landscape and recent climate warming has allowed the white- footed mouse (Peromyscus leucopus) to expand its northern limit into the Montérégie from the northeastern United States. P. leucopus is a great disperser and colonizer and is considered one of the primary reservoirs for the bacterium that causes Lyme disease in humans (Cummings and Vessey, 1994; Ostfeld, 2011).

Although they play an important role in the replication and spread of the bacteria, there was no current information on the processes that account for the current level and distribution of their genetic variation within the Montérégie. We employed a population genetics study and found the mice, in some degree, to be structured in space. Geographic distance explained a tenth of their genetic variability, therefore we tested for the effect of barriers on their movements. Agricultural matrices did not hinder gene flow between sites, highways slowed down dispersal but were not absolute barriers, and main rivers appeared to be strong dispersal barriers.

If Lyme disease were to become endemic in southern Québec, we predict that the various water bodies dissecting it would help keep the disease fairly isolated to this end of the province. Although certain barriers would aid in slowing white-footed mouse dispersal and thus slow the spread of the disease, increasing biodiversity within the forest patches would also help directly control P. leucopus dispersal and abundance (Ostfeld, 2011). Southern Québec has the

83 highest biodiversity level within the province, but also the greatest number of species threatened by forest fragmentation (Auzel et al., 2012). Although increasing the biodiversity within an area may seem directly advantageous to its local flora and fauna, its indirect effects hold just as much importance, especially towards human well-being.

Leighton et al. (2012) have confirmed the northernmost limit of tick and Lyme disease emergence to be in southern Québec. This emergence, coupled with the range expansion and population abundance of the efficient Lyme disease reservoir P. leucopus, shows how important it is to merge P. leucopus population analyses like ours with ongoing Lyme disease research in Québec. These accumulated efforts may then be applied towards disease management strategies within Québec, and within other Canadian provinces with similar ecological conditions, to help slow the spread of the disease and efficiently treat those that are afflicted. To understand P. leucopus movements further, biodiversity studies within the Montérégie and a detailed population connectivity study assessing the natural and man-made corridors that link forest patches within the region may provide much better insight as to what factors enable or hinder their success.

REFERENCES

Auzel, P., Gaonac’h, H., Poisson, F., Siron, R., Calmé, S., et al. (2012). Impacts of climate change on the biodiversity of Quebec: Literature review summary. CSBQ, MDDEP, Ouranos. 29 p.

Cummings, J., & Vessey, S. (1994). Agricultural influences on movement patterns of white-footed mice (Peromyscus leucopus). American Midland Naturalist, 132: 209-218.

Leighton, P. A., Koffi, J. K., Pelcat, Y., Lindsay, L. R., & Ogden, N. H. (2012). Predicting the speed of tick invasion: An empirical model of range expansion for the Lyme disease vector Ixodes scapularis in Canada. Journal of Applied Ecology, 49: 457-464.

Ostfeld, R. S. (2011). Lyme Disease: The Ecology of a Complex System. Oxford University Press Inc., New York, New York, pp. 216.

Wampach, J.-P. (1988). Deux siècles de croissance agricole au Québec, 1760- 1985. Recherches Sociographiques, 29: 181-199.

APPENDIX

Primer code Sequence (5’ – 3’) Fluorescence used # of alleles Author(s) PMl01 F - CATTCAAGACCTGGCTTTTT * Carboxyfluorescein (FAM) 14 Chirhart et al. (2005) R - TGGGTTTCATCAGTGCTTCT PMl03 F - GCCATTAGTCTATGTGACAG* Hexachlorofluorescein (HEX) 12 Chirhart et al. (2005) R - GCGATGTACCCAGAAAT PMl04 F - CATAAGGTGGCTCGGAATCA * Hexachlorofluorescein (HEX) 10 Chirhart et al. (2005) R - CAGGAAGGGGAAATGACCAT PMl05 F - CTGAGCCAAAGTGGTCCTT * Hexachlorofluorescein (HEX) 12 Chirhart et al. (2005) R - TGAAGACAGCCCCTCTCTG PMl06 F - CAGGGCTGTAGAGGGAGAAC * Carboxyfluorescein (FAM) 10 Chirhart et al. (2005) R - ACTGGAGCAGAGGCATTTG PMl09 F - GAATCCATACACCCATGC * Hexachlorofluorescein (HEX) 9 Chirhart et al. (2005) R - TTGCTTTTCGTCAAGTTTT PMl11 F - ACCCCCGAGTGCTGAGATT * Carboxyfluorescein (FAM) 13 Chirhart et al. (2005) R - TTTGCTGCTTTCCCCAGAGA PMl12 F - GCAGCCTGTATTCTCTCACA * Carboxyfluorescein (FAM) 13 Chirhart et al. (2005) R - GCCAACCATTTCTTCAAGTG PLGT58 A - GATCTTGTGAACACGCTTCT Hexachlorofluorescein (HEX) 28 Schmidt (1999) B - TTGATGGCTCTGGAGAGGCT *

86 PLGT66 A - CTCTGTCTGCCACACATGCT Carboxyfluorescein (FAM) 27 Schmidt (1999) B - GTGCCATCACAGATGTGACA * PLGATA70 A - CTTGGTATGCATCGCCATCT Hexachlorofluorescein (HEX) 29 Schmidt (1999) B - TAATCTCTGTAGCTTCATGT * * fluorescently labeled