UNIVERSITY OF CALGARY

Roosting Ecology and Landscape Genetics of Prairie Bats

by

Cori L. Lausen

A THESIS

SUBMITTED TO THE FACULTY OF GRADUATE STUDIES

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

DEPARTMENT OF BIOLOGICAL SCIENCES

CALGARY, ALBERTA

December, 2007

© Cori L. Lausen 2007

UNIVERSITY OF CALGARY

ISBN: 978-0-494-38224-0

ii

ABSTRACT

I characterized various aspects of bat ecology in a prairie landscape. I used radio- telemetry, acoustic monitoring and molecular genetics to address questions of roosting ecology and landscape genetics at fine and large scales. Additionally, I used a population genetics approach to address a question of systematics that arose due to a discrepancy between mitochondrial and nuclear DNA.

I compared the roosting ecology of female western small-footed bats, Myotis ciliolabrum, and big brown bats, Eptesicus fuscus in SE Alberta. Despite substantial differences in physical attributes of roosts, roosts were similar in microclimate. Unlike E. fuscus, M. ciliolabrum roosted more inconspicuously, did not change roost structure during reproduction, and roosted with few individuals. M. ciliolabrum was geographically clustered by relatedness on a small scale, unlike female E. fuscus who roosted in unrelated groups over a larger area.

I acoustically monitored along the Red Deer River for year-round bat activity. I determined that bats are active in all months at three locations, flying at unexpectedly cold temperatures. Using radiotelemetry, I located and described the first natural rock- crevice hibernacula for E. fuscus in the Canadian prairies. Acoustically I also determined that species composition and activity patterns along the river change seasonally, suggesting use of rivers as movement corridors.

I tested the hypothesis that bats vary in genetic population structure according to their mobility and habitat specificity. I compared three species of bats in a prairie environment where river valleys were the dominant landscape feature. Greater flight

iii

ability corresponded to less genetic structure, and roost specificity may have caused greater dependency on rivers as movement corridors.

I used population genetics to assess the systematics of little brown bats, M. lucifugus. Using nuclear microsatellites, I found that two groups differing substantially in mtDNA sequence (putative subspecies), were fully interbreeding. Sympatry occurs across western North America, making intact gene pools for each group unlikely. This, together with a lack of morphological and ecological distinction, suggests no biological basis for taxonomic distinction. Although recently proposed to be cryptic species based on mtDNA, my results suggest no taxonomic distinction is biologically warranted. I highlight the importance of investigating nuclear gene flow in widely sympatric animals suspected of being cryptic species.

iv

ACKNOWLEDGEMENTS

There are many people I would like to thank, as this research could not have been possible without them. For assistance in the field I thank Erin Baerwald, Krista

Patriquin, James Minifie, Troy Pretzlaw, Angelique Myles, Dora Alampi, Lydia Hollis,

Marc Obert, Terri Whitehead, Sandi Robertson, Kim Hughes, Michael Proctor, Susan

Sanford, Mai-Linh Huynh, and many other volunteers; in particular I thank Troy Pretzlaw for the key role he played in collecting M. ciliolabrum roost data. I also thank Krista

Patriquin and Lisa Crampton for additional forearm measurement data; Tanya Luszcz,

Donald Solick, Craig Willis and Jeff Gruver for additional genetic samples; and Gillian

Sanders for administrative support.

For logistical support and assistance in the laboratory I thank Corey Davis, Chris

Kyle, Dr. Curt Strobeck and other members of the University of Alberta Lab; in particular, I thank Jen Bonneville for additional microsatellite laboratory work. For mtDNA sequencing, I thank Dr. Isabelle Delisle (University of Alberta), Dr. Jan Zinck

(Portland University), Dr. Tanya Dewey (University of Michigan) and Dr. Maarten

Vonhof (University of Western Michigan).

For logistical support in the field, I thank S. and E. Jensen, A. Newman, E.

Courtnage, G. and J. Mattheis, and the staffs at Onefour Agricultural Substation, Pinhorn

Grazing Reserve (C. Stryker), Bow Island Grazing Reserve (B. and S. Schmidt), Havre

Agricultural Research Station, McClelland Ferry Crossing, Big Knife Provincial Park

(especially V. Hulbert), Dry Island Buffalo Jump Provincial Park, Dinosaur Provincial

Park (especially P. Hofer, R. Hugill, and M. Macdonald), Writing on Stone Provinical

v

Park (especially B. Moffet), Midland Provincial Park (G. Martin), Flowserve (M.

Hildebrand), City of Drumheller, Cypress Hills County, Village of Donalda, Village of

Empress, Bindloss/Empress Agricultural Society (especially D. Martin), and Special

Areas Municipal Offices. For access to bats I thank many landowners, including B.

Herman, L. Fowlie, L. Fowlie, R. Howe, D. Clements, L. and D. Bell, D. Wallwork, A. and K. Reece, D. Sorenson, S. Knox, E. and C. Alderdice, R. and L. Gillespie,

Hargrave’s Ranching, L. Shagnon, L. Adams, L. and T. Fryberger, G. Eide, B.

Vanderloh, A. and S. Pollom, Content Bridge Campground, Pinter’s Campground, T.

Van Dellen, D. Harvey, and Forks Prairie Orchard (C. Cocks).

My research was funded in part by scholarships/research allowances from

NSERC (Postgraduate Scholarhip), Alberta Ingenuity, and Izaak Walton Killam

Memorial awards; by grants from Alberta Sport, Parks, Recreation, and Wildlife

Foundation (Community Development Initiative), Alberta Conservation Association

(Biodiversity Challenge grant), University of Calgary (Thesis Research Grant) and

Mountain Equipment Co-op (Studentship); and a Discovery grant from the Natural

Sciences and Engineering Research Council (NSERC) of Canada to Dr. Robert Barclay.

I’d like to especially thank several very special people whose support I valued tremendously; I thank my supervisor, Robert Barclay, for guidance, support, patience, editing, and more editing, and the surprising willness to let me hang around the lab for yet another degree! I thank my family for their support throughout my university career, and their endless understanding of the missed family events and the paucity of visits.

And to my husband, Michael Proctor, I can’t possibly list all that I’d like to thank him for, but his support has been keystone in the completion of this thesis: our late night

vi

discussions, endless walks together by the lake, and working side-by-side provided me with invaluable motivation and inspiration.

For constructive comments on, and editing of, this thesis, I thank Dr. Curt

Strobeck, Dr. Isabelle Delisle, Dr. Robert M.R. Barclay, and Dr. Michael Proctor. I would also like to thank my committee members, Dr. Curt Strobeck (Univeristy of

Alberta), Dr. Robin Owen (Mount Royal College), Dr. Dave Coltman (Univeristy of

Alberta), Dr. Shelley Alexander (Department of Geography), and Dr. John Post

(Department of Biological Sciences) for their reviews and constructive comments.

vii

TABLE OF CONTENTS

Abstract...... iii Acknowledgements...... v Table of Contents...... viii List of Tables ...... xi List of Figures...... xii CHAPTER 1: General Introduction...... 13 CHAPTER 2: Roosting Ecology of Western Small-Footed (Myotis ciliolabrum) and Big Brown (Eptesicus fuscus) Bats in the Canadian Prairies ...... 28 Introduction...... 28 Materials and Methods...... 32 Study Site...... 32 Study Species...... 33 Capture...... 34 Radio-telemetry and Roost Characterization...... 35 Molecular Genetics ...... 37 mtDNA Sequences...... 37 Microsatellite Genotyping ...... 37 Statistical Analyses ...... 38 Results...... 39 Captures ...... 39 Radio-tracking...... 40 Myotis ciliolabrum...... 40 Eptesicus fuscus ...... 41 Roost Structure...... 42 Myotis ciliolabrum...... 42 Comparison with Female E. fuscus ...... 43 Roost Microclimate...... 43 Myotis ciliolabrum...... 43 Comparison with E. fuscus ...... 45 Genetics...... 46 Discussion...... 48 Roosts and Roosting Patterns...... 48 Relatedness Patterns...... 53 Synthesis ...... 56 CHAPTER 3: Winter Bat Activity in the Canadian Prairies ...... 69 Introduction...... 69 Materials and Methods...... 72 Results...... 76 Acoustic Detection...... 76 Bat Captures...... 78 Insects and Weather ...... 79 Discussion...... 80 CHAPTER 4: The Effect of Landscape on the Population Structure of Prairie Bats...... 90

viii

Introduction...... 90 Methods...... 94 Study Species...... 94 Study Area and Samples ...... 95 Genetic Analysis ...... 97 mtDNA Sequencing ...... 98 mtDNA Analyses ...... 99 Microsatellite Genotyping ...... 101 Microsatellite Analyses...... 102 Results...... 104 Samples...... 104 Sequencing...... 104 Genotyping...... 105 Genetic Analyses ...... 105 Discussion...... 113 CHAPTER 5: Beyond mtDNA: Nuclear Gene Flow Refutes Cryptic Species in Little Brown Bats (Myotis lucifugus)...... 148 Introduction...... 148 Methods...... 151 Study Area and Samples ...... 151 mtDNA Sequences...... 152 Microsatellite Genotyping ...... 154 Forearm Analysis...... 156 Results...... 157 Sequence Haplotypes...... 157 Microsatellite Genotypes ...... 160 Discussion...... 163 CHAPTER 6: General Synthesis ...... 188 LITERATURE CITED ...... 195 APPENDIX I: Patterns in Year-Round Bat Activity in Riparian Areas of Southern Alberta...... 224 Introduction...... 224 Methods...... 226 Results and Discussion ...... 230 Seasonal Activity Patterns ...... 230 Species Detected ...... 232 Timing of the Presence of Migratory Species...... 232 Variation in Bat Activity...... 234 Placement of Anabat Detectors...... 234 Conclusions...... 235 APPENDIX II: Nightly Activity Patterns and Reproductive Rates of Four Species of Prairie Bats Derived from Capture Records...... 255 Introduction...... 255 Methods...... 258 Study Species...... 258 Captures ...... 259

ix

Analyses...... 259 Results...... 261 Discussion...... 263 Capture Patterns...... 263 Reproductive Rates ...... 267 Conclusion ...... 268

x

LIST OF TABLES

Table 1.1. Four year-round resident species of bats in southern Alberta...... 23 Table 2.1. Comparison of M. ciliolabrum (MYCI) roosts with E. fuscus (EPFU) roosts 62 Table 2.2. Analysis of variance of microclimate for roosts used by pregnant vs. lactating M. ciliolabrum ...... 64 Table 2.3. Pairwise relatedness (r; calculated from microsatellite genotypes)...... 65 Table 2.4. Pairwise relatedness (r) for M. ciliolabrum...... 66 Table 3.1. Captures of E. fuscus at Dinosaur Provincial Park from 2003 - 2005...... 86 Table 4.1. Summary of studies showing population structure and wing-loading...... 125 Table 4.2. UTM locations for the sampling sites in Fig. 5.1 ...... 128 Table 4.3. HVII mtDNA sequence fragments and microsatellite alleles per locus ...... 130 Table 4.4. Pairwise Fst for all three species...... 131 Table 4.5. Mean genetic distances (Nei’s minimum distance [Dm] and Slatkin’s linearized Fst) ...... 132 Table 4.6. Analysis of molecular variance (AMOVA) for E. fuscus, M. lucifugus and M. ciliolabrum...... 133 Table 4.7. Results of Mantel tests using three distance matrices...... 135 Table 4.8. A comparison of within-site and among-site haplotype matches...... 137 Table 4.9. Nuclear relatedness (r) among individuals, and number of mtDNA nucleotide differences within colonies...... 138 Table 5.1. UTM locations for sampling sites in Fig. 5.2 ...... 172 Table 5.2. Subspecies designations, M. l. carissima (MYLU.CA) and M. l. lucifugus (MYLU.LU) ...... 174 Table 5.3. Forearm comparisons for M. lucifugus (A), and M. ciliolabrum and E. fuscus (B)...... 175 Table I.1. Summary of bat activity in 2004-2006 recorded at five locations along the Red Deer River...... 237 Table I.2. Earliest and latest detections of the migratory species L. cinereus and L. borealis at 5 locations...... 240 Table II.1 Morphological measurements of adult non-pregnant females...... 269 Table II.2. Captures of bats in Alberta and Montana for which capture times were known ...... 270 Table II.3. Odds ratios for captures of reproductive females in the mid-night period. . 271 Table II.4. Number of reproductive and nonreproductive females of each species ...... 271

xi

LIST OF FIGURES

Figure 1.1. North American distributions of E. fuscus, M. ciliolabrum, M. lucifugus, and M. evotis...... 24 Figure 1.2. Map of Alberta...... 27 Figure 2.1. Daily roost maximum and minimum temperatures...... 67 Figure 3.1. Winter bat passes per night and associated temperatures...... 87 Figure 3.2. Winter bat passes per night versus temperature at emergence light levels. .. 89 Figure 4.1. Map of sample locations...... 140 Figure 4.2. TCS-generated haplotype networks for E. fuscus (A), M. lucifugus (B), and M. ciliolabrum (C)...... 142 Figure 4.3. Scatterplots of Fst vs. geographic distance...... 146 Figure 5.1. Map of M. lucifugus subspecies boundaries as originally defined using morphology...... 178 Figure 5.2. Map of sample locations...... 179 Figure 5.3. HVII haplotypes (26) for M. lucifugus...... 181 Figure 5.4. Neighbor-joining trees showing population structure...... 183 Figure 5.5. Factorial correspondence analysis plots of M. lucifugus...... 185 Figure 5.6. Regression of M. lucifugus male and female forearm lengths...... 187 Figure I.1. Bat activity (passes per night) at East Coulee (A) and Dinosaur Provincial Park (B) (October through May)...... 242 Figure I.2. Total bat activity (passes per night) compared between sites along the Red Deer River during spring 2005 (A) and fall 2005 (B)...... 243 Figure I.3. Year-to-year variation in bat activity at DPP (A) and East Coulee (B), Alberta...... 246 Figure I.4. Patterns of activity of late summer and fall migratory-bat species, L. cinereus and L. borealis...... 250 Figure I.5. Bat activity in summer through to fall at East Coulee (A) and DPP (B) in 2005...... 251 Figure I.6. Bat activity in spring 2005 at DPP using two detectors. Percentage of bat passes detected by each detector...... 254

xii 13

CHAPTER 1: General Introduction

Animals are intimately tied to their environment, with species evolving physiologically, morphologically and behaviourally in response to their surroundings

(Ricklefs 1990). The study of each aspect of a species’ ecology requires a different approach and a unique set of tools. Because of this, calling oneself an “ecologist” today is about as ambiguous as the label “engineer.” Modern ecology encompasses many disciplines, from population dynamics to evolutionary ecology, and most notable to my research, behavioural ecology, landscape ecology, molecular ecology (ecological genetics), and the combination of the latter two, landscape genetics. Each field contributes to the overall understanding of how an organism interacts with its environment, and they are thus invariably linked.

The study of how landscape structure determines abundance and distribution of organisms has in the past focused on the macro-scale, but more recently the importance of fine-scale patterns and processes has been recognized (Ludwig 2005). The recognition that individuals are the foundation for population processes has led to a bottom-up approach to understanding ecology at a macro-scale (Goss-Custard and Sutherland 1997).

Using a holistic approach, this thesis draws from several disciplines of ecology and examines both fine and macro-scale patterns and processes, to provide a multi-scaled and multi-species analysis of bat ecology in the prairies. I begin with fine-scale habitat selection, examine seasonal movements of species in a natural landscape, and use these foundational elements to synthesize a broad-scale landscape perspective of population structure.

14

Because landscapes are continuously changing due to human activities, there is

increasing pressure to predict ecosystem response to anthropogenic disturbance. In

contrast with most forested areas, the prairie landscape is naturally fragmented from the

perspective of many species, for example, tree-dwelling birds or rock-roosting bats. A

multi-scaled investigation of prairie species’ ecology is likely to frame habitat selection

and population structure in an evolutionary context involving natural fragmentation;

application of this understanding to rapidly fragmenting landscapes may be instructive in

predicting animal response and adaptation. Bats, due to their ability for flight and

attraction to riparian habitats in prairie regions (Holloway and Barclay 2000), may be

good models for understanding landscape structuring in prairie areas, with applications to

riparian birds and mammals.

My research, and this thesis, begins with fine-scale habitat selection, the

foundation for landscape-level investigations. Bats spend the majority of the day in a

roost, and because their physiology is intimately linked with their environment, roost

conditions influence fitness (Thomas and Speakman 2006). As refugia, roosts should be

selected to minimize the risk of predation (Lewis 1996), and provide a micro-

environment suitable to physiological needs (McNab 1982). With tight energy budgets

(Kurta et al. 1989a), bats may select thermal environments that optimize

thermoregulatory behaviour (Willis 2006). Based on the importance of roosts to bats,

roost selection is likely to play a major role in determining their distribution. In fact, where organisms occur can be thought of as the result of an interplay between fitness and spatial environmental variation. Much research effort has gone into the study of this

“spatial fitness” for bats, with fine-scale landscape use by bats having been described for

15

many species (e.g. Myotis septentrionalis, Sasse and Pekins 1996; Eptesicus fuscus,

Brigham et al. 1997, Lausen and Barclay 2002; Plecotus auritus, Entwistle et al. 1997;

M. evotis, Chruszcz and Barclay 2002). However, most often these roost-selection and

behaviour studies encompass fine-scale summer roosting habitat only, or winter roosts

only (but see Clark et al. 1996), and rarely address habitat use on larger landscape and

temporal scales. Bats select roosts from what is available, yet because resource selection

occurs in a hierarchical manner, this availability has already been “pre-selected” by the

presence of the individuals in that geographic location (Manly et al. 2002). In an effort to

understand bat distribution on the landscape with a bottom-up approach, roosting ecology

provides the starting point for a landscape-level investigation.

As small, nocturnal, and vagile animals, bats are not easily studied at a landscape

level. Although during the summer, female, and sometimes male (e.g. Burland et al.

2001) philopatry allows for traditional radio-telemetry investigation of movement and

roosting behaviours, studies of bats during spring and fall migrations and dispersal

events, or winter roosting are substantially more difficult and are therefore under-

represented in the literature. Generally, in temperate species, bats move between summer

and winter roosts each spring and fall, with mating occurring in the fall (Nagorsen and

Brigham 1993). To study seasonal migrations of individual bats using traditional radio- telemetry techniques is difficult, and would greatly benefit from larger scale techniques such as GPS radio-telemetry, which is, unfortunately, not yet developed for animals as small as bats. Instead, more indirect tools such as acoustic monitoring and molecular genetics can be used to study seasonal bat movements, dispersal and landscape use. My

PhD research is, therefore, a combination of a fine scale study of roosting ecology, using

16

radio-telemetry, acoustic, and molecular genetics techniques, and a large scale study of

landscape ecology, using both acoustics and population genetics.

When I first began my PhD research, the literature contained many more studies

of the ecology of forest-roosting and building-roosting bats than of rock-roosting species.

Early descriptions of rock-roost selection were few (e.g. Vaughan and O’Shea 1976), likely due to lower accessibility together with the perceived lower conservation risk in more permanent roosting structures. But in the late 1990’s in southeast Alberta, a concerted effort to describe rock-roosting behaviour in bats began; Holloway (1998) provided cursory descriptions of rock roosts of western long-eared bats (Myotis evotis),

little brown bats (M. lucifugus), big brown bats (Eptesicus fuscus) and western small-

footed bats (M. ciliolabrum); Chruszcz (1999) described roost selection of female M.

evotis in the context of their physiology, and roosting and foraging behaviours; and

Lausen (2001) described roost selection of female E. fuscus, in the context of their

physiology and roosting behaviour. I wanted to continue this exploration of rock- roosting ecology, describing the fine-scale roost selection and roosting behaviours of prairie bat species, and then exploring the link between roost selection and population structure at the landscape level to ultimately improve our understanding of the distribution and abundance of prairie bat species.

The four most abundant resident bat species of the Alberta prairies are M.

lucifugus, M. ciliolabrum, M. evotis, and E. fuscus (Table 1.1; Fig. 1.1). Female M. lucifugus tend to roost mainly in buildings (Fenton and Barclay 1980), although males are often found roosting in rock crevices (pers. obs.). Both males and females of the other species are commonly found in rock-crevice roosts in the prairies. Selection of rock

17

roosts by M. evotis, and E. fuscus has been described (Chruszcz and Barclay 2002,

Lausen and Barclay 2002), but a formal description of roost use by M. ciliolabrum is

lacking. I therefore, start this thesis with an examination of the roosting ecology of M.

ciliolabrum compared to that of the other rock-roosting species, with particular emphasis

on E. fuscus (study location FC in southeastern AB, Fig. 1.2; Chapter 2). I predicted that

the small, solitary-roosting M. ciliolabrum would choose roosts with different size and

thermal properties than the larger, colonial-roosting E. fuscus, and that M. ciliolabrum,

with its slower flight (Holloway and Barclay 2001), would have smaller roosting home-

ranges. These data could then be used to develop predictions and understand results at

the landscape level (Chapter 4).

With an understanding of summer roosts for all four prairie species, I then turned

my attention to locating winter roosts. If bats breed and hibernate in summer roosting

areas, questions of land-use and mating patterns may be relatively straightforward, and

not require consideration of spatial and temporal scales. However, due to seasonal

migrations to mate and hibernate, bat distribution changes between summer and winter,

forcing the land-use scale of investigation to increase. Because mixing of bats from different areas is likely to occur, relatedness patterns as observed during summer are likely to reflect spatial patterns in the landscape. To synthesize a large-scale picture of

year-round land-use requires knowing fine-scale roost selection during both summer and

winter, and an indication of possible movement corridors between suitable habitats.

Knowing that bats in the prairies tend to concentrate along rivers in the summer

(Holloway and Barclay 2000), I acoustically monitored river valleys in an attempt to

locate fall/winter habitat for prairie bats.

18

When I first began my research, where bats from the Canadian prairies hibernated

was unknown. In the Rocky Mountains (>200 km away; Fig. 1.2), a few natural caves

were known to house hibernating bats (Alberta Fish and Wildlife 1994), but given that E.

fuscus, for example, tends to move distances <100 km between summer and winter roosts

(Mills et al. 1975), such a long distance migration against prevailing winds seemed

unlikely. I hypothesized that bats were hibernating either in small caves or in rock

crevices in river valley walls. My goal was to locate areas with late fall activity, then

using acoustics and radio-telemetry determine whether bats were hibernating in the area

and describe winter roosts and roosting behaviour (Chapter 3; published as Lausen and

Barclay 2006a). Results from this work stimulated many additional questions related to

hibernal physiology. The physiology of hibernation and hibernal arousal is not well

understood, but is likely to influence individual behaviour, roost selection and overall

habitat selection at a landscape level.

If one increases the scale of focus from that of the valley-wall crevices, to that of the entire river and river system, then the prairie landscape takes on an even more

complex spatial pattern. The river valleys are narrow corridors of water containing

patches of trees and rocks, winding through a matrix of relatively flat grasslands and

agricultural fields. To a bat looking for rock crevices to roost in, water to drink, and

clusters of insects to feed upon, the prairie landscape is a naturally fragmented mosaic of

suitable habitat patches embedded in an inhospitable matrix. Given that prairie bats tend

to move seasonally for hibernation and mating, have specific roost requirements

stemming from physiological, morphological and behavioural forces, and are limited in

movement by night length and flight ability, landscape is likely to structure bat

19

populations, and structure different species to varying degrees. The pattern and degree of

population structuring among prairie bat species is the focus of Chapter 4.

Several landscape ecology and metapopulation biology (Hanski and Simberloff

1997) questions arise when considering bat movement patterns on a larger, prairie-

landscape level: How do bats move between suitable habitat patches? Do bats follow

rivers as corridors from patch to patch during seasonal movements and dispersal, or do

they cross over the grasslands matrix? How great can the interpatch distance be before

movement is limited? An increasingly prevailing theme in landscape ecology is that patch context matters (Wiens 1997); for an animal that makes seasonal migrations for fall mating and winter hibernation, in addition to dispersal events, population structure is likely to reflect the interconnectedness of suitable summer and winter habitat patches and the availability of suitable roosts en route. I hypothesized that flight ability would influence population structure. Because habitat specialists may avoid unsuitable matrix, seeing interpatch areas as barriers to gene flow (Wolff 1999), I also hypothesized that roost specificity would influence population structure. And finally, I hypothesized that roost selection and social structure would together influence dispersal behaviour and result in differing gene flow patterns among the species and sexes. In Chapter 4 I test these hypotheses for three prairie bat species.

While the focus of my molecular genetics analysis was at the landscape level, I

also investigated individual relatedness at a finer scale, to reveal genetic patterns

associated with roost selection. In particular, I asked whether patterns of relatedness

among colonial female E. fuscus differed from relatedness patterns among the more

solitary female M. ciliolabrum. I overlay this relatedness layer on the roost selection data

20

(Chapter 2) to determine whether, on a fine scale, there is a relationship between roosting

ecology and kinship patterns. This fine scale pattern of relatedness could also be applied

to understanding abundance and distribution at the larger landscape level.

During the landscape genetics analysis, it became evident that two distinct genetic

groups of M. lucifugus were present, based on mitochondrial DNA. This was not

unexpected due to the fact that my study area was at one point proposed to be a potential

overlap zone for two subspecies (Hall 1981, but see Smith and Schowalter 1979), and

because another recent genetic study of M. lucifugus also found extreme mtDNA

divergence among M. lucifugus subspecies (Dewey 2006). A lack of obvious

morphological and ecological differences between the two groups of M. lucifugus led me

to question the biological validity of this distinction. Because of its mode of inheritance

(i.e. inherited as one unit, maternal transmission), mtDNA does not reflect contemporary

gene flow, unlike nuclear genes which are mixed each generation and are bi-parentally

inherited (Avise 1994). My goal was to test whether the two distinct mtDNA groups of

M. lucifugus reflected true, ecologically-separate species, as proposed by Dewey (2006),

or whether the groups were interbreeding, with mtDNA sequence divergence merely a

remnant of historical separation. I predicted that using nuclear DNA I would find that these putative species were interbreeding. In Chapter 5, I test this using a population genetics approach; I integrate evidence from microsatellite analyses with a cursory analysis of roosting ecology and morphology. I also discuss whether genetic variation, in the absence of ecological or morphological variation, warrants taxonomic recognition.

While bats were known to concentrate in riparian areas during the summer

(Holloway and Barclay 2000), it was not known whether bats are present in the river

21

valleys during seasonal migration. In the process of finding winter bat habitat along

prairie rivers, I acoustically monitored a number of riparian sites (Fig. 1.2). To test my

hypothesis that rivers were used as seasonal migration corridors in addition to places to

roost, I expanded my acoustic monitoring to include areas that did not contain suitable

rock-roosting habitat. I predicted that a paucity of roosting habitat would restrict bat

activity in grassy-sloped areas of river valleys to spring and late summer/fall, if bats were

seasonally migrating along rivers to some exent. I also predicted that the species

assemblages detected would change seasonally, and that species that are long-distance migrators would be detected mainly during the spring and late summer/fall during migration. This is what I found (Appendix I).

In the course of my research, I captured bats at various prairie sites from 2001 –

2006. From this large dataset, I was able to extract a number of behavioural and

morphological patterns noteworthy both on a local and species level. I present forearm

size data that demonstrate clinal patterns as one moves north in the prairies, and capture

times that reveal foraging patterns among sexes and species (Appendix II).

This thesis has at its core, the ecology of prairie bats. However, its application

extends beyond bats, to the many other species that concentrate in prairie river valleys

(e.g. Savoy 1991, Rhodes 1991) and potentially use riparian areas as movement corridors.

My goals were to advance our understanding of both fine and large scale patterns in the

ecology of habitat selection and population structure, in the context of the prairie

landscape. The prairie topography with its mosaic of suitable bat habitat interwoven

among unsuitable matrix, is likely to present unique challenges even to highly vagile

species, such as bats, dispersing and moving seasonally across the landscape.

22

Understanding how species interact with this landscape requires a multi-disciplinary approach. The following chapters are a compilation of these various ecological investigations. Starting with summer roost selection, then the discovery of winter habitat, and possible use of rivers as movement corridors, this thesis culminates with a large-scale landscape investigation of population structure.

23

Table 1.1 Description of the four abundant year-round resident bat species in southern Alberta.

Mean Mean Common Forearm ± Mass ± Species Name Foraging Roosting1 SD (mm)2 SD (g)2 buildings, rock 17.9 ± Eptesicus big brown aerial hawker*,(Kurta and crevices, tree 47.4 ± SD 3.2 (n = fuscus bat Baker 1990) crevices 2.5 (n = 37) 28) aerial hawker* but capable of gleaning** buildings, rock Myotis little (Ratcliffe and Dawson crevices, tree 37.0 ± 1.4 7.9 ± 1.6 lucifugus brown bat 2003) crevices (n = 90) (n = 35) rock crevices (occasionally western buildings; M. small- aerial hawker* (Holloway Holloway and 32.2 ± 0.68 4.9 ± 1.1 ciliolabrum footed bat and Barclay 2001) Barclay 2001) (n = 50) (n = 50)

western aerial hawker* and rock crevices, long- gleaner** (Nagorsen and tree crevices, 38.6 ± 1.1 6.7 ± 1.5 M. evotis eared bat Brigham 1993) buildings (n = 38) (n = 52) *bat captures insects that are in the air while flying **bat captures insects that are resting on vegetation 1Nagorsen and Brigham 1993 2van zyll de Jong 1985

24

Figure 1.1. North American distributions of E. fuscus (A; Kurta and Baker 1990; distance bar = 1000 km, numbers are 11 subspecies), M. ciliolabrum (B; Holloway and

Barclay 2001; numbers are two subspecies), M. lucifugus (C; see Chapter 5 for six subspecies), and M. evotis (D; Manning and Jones 1989).

25

A.

B.

Figure 1.1

26

C.

D.

Figure 1.1 (cont’d)

27

N

o 49 N o 154 W

Figure 1.2. Map of Alberta. The main study took place in the Prairie Ecozone, east of the Rocky Mountains (Mountain Ecozone). Main Alberta rivers included the Milk River (Milk R), South Saskatchewan River (SSR) and Red Deer River (RDR). Main acoustic monitoring sites (stars; Appendix I) were along the RDR: Dry Island Buffalo Jump Provincial Park (DIBJ), East Coulee (EC), Finnegan Ferry (FF), Dinosaur Provincial Park (DPP) and Bindloss Campground (BL). Some acoustic monitoring also took place at Big Knife Provincial Park (BKPP) which is located on the Battle River (not shown). The roosting ecology study (Chapter 2) took place at Bindloss Ferry Crossing (FC) on the SSR (black dot). The Prairie Ecozone continues south and east into Montana (Chapter 4).

28

CHAPTER 2: Roosting Ecology of Western Small-Footed (Myotis ciliolabrum) and Big

Brown (Eptesicus fuscus) Bats in the Canadian Prairies

Introduction

While investigation of habitat selection in controlled experimental conditions has

elucidated several important contributing factors to habitat selection, such as interspecific

competition (e.g. Larson 1980), and predation risk (e.g. Abramsky et al. 1998, Schofield

2003), in field environments complex interactions of a wide range of factors make habitat

selection studies challenging (Rosenzweig 1981). While many studies attempt to

incorporate descriptions of competition and predation forces (e.g. reviewed by Jones

2001), single-species habitat-use studies dominate the literature. Factors such as

physiology and social behaviour, also influence habitat selection (Caughley and Sinclair

1994), and should therefore be considered. Meta-analyses and studies building upon

previous work are challenged by insufficient contextual information such as conspecific

habitat use, observed social behaviour and structure, and possible predation risks, making

roosting ecology patterns difficult to discern. Describing a species’ habitat requirements

and the underlying evolutionary forces driving habitat selection is unlikely to be accomplished in a single study or studies of single species (Caughley and Sinclair 1994,

Jones 2001). Instead, a compilation of field studies from different habitats, scales, and

competition/predation contexts is required, with the complexity within each study

described as fully as possible.

The purpose of my study was to describe the roosting ecology of two species of

prairie bats in a comprehensive manner to allow for future comparisons and to further our

29

understanding of the evolution of roosting behaviour and habitat selection. As small

mammals with large surface-area-to-volume ratios and high energetic demands due to

flight (reviewed in Altringham 1996), bats have tight energy and water budgets (Kurta et

al. 1989a and b) and show a close physiological relationship with their environment

(McNab 1982). As such, roosts are likely to be important determinants of fitness, with

roost selection having evolved to reflect physiological, behavioural, and morphological

adaptations (Kunz 1982). For example, small temperate species of bats use torpor as an

energy and water-saving strategy, but incur certain costs in doing so. Costs include

increased risk of predation due to the inability to move quickly (Schmidt-Nielsen 1990),

and, for reproductive female bats, slower fetal development (Racey 1973) and milk

production (Wilde et al. 1999). For bats with a defined reproductive season, lengthened

gestation or growth periods can result in decreased fitness (Lewis 1993). As such, use of

torpor by reproductive bats may be limited. To balance their energy budgets, bats may

use a combination of behavioural (e.g. clustering) and physiological (e.g. torpor)

strategies, and this combination may change as the cost:benefit ratio of torpor changes.

In summary, bats may roost in clusters to minimize heat and water loss, and reproductive

females may select roosts with different properties at different stages of reproduction to

adjust for changing thermoregulatory demands, group size, and environmental conditions

(McNab 1982).

It is clear, that roost selection by bats is by no means random and that many

factors influence roost selection. While physiological considerations have dominated the

roost-selection literature, only recently has the role of social interactions taken on greater consideration (McCracken et al. 2006). In fact, it is difficult to discuss roosting ecology

30 without reference to social organization given that choice of roost site, movements between roosts, and the association of individual bats using multiple roosts are intimately linked and not random (Lumsden and Bennett 2006). Discoveries of fission-fusion patterns among highly related and unrelated females of some species are becoming increasingly common (e.g. Myotis bechsteinii Kerth 2006, Chalinobus tuberculatus

O’Donnell and Sedgeley 2006, Eptesicus fuscus Willis and Brigham 2004), but the role that interdependence among individuals plays in roosting ecology is not yet understood.

Molecular genetic tools are making the study of social structure more informative, as kin- based behaviours and relationships are determined (e.g. Kerth et al. 2002). It is unlikely that roost selection has evolved independent of social behaviour and its inclusion in investigations of roosting ecology is therefore important for a comprehensive description.

I examined the roosting ecology of a population of western small-footed bat,

Myotis ciliolabrum, in relation to the sympatric rock-roosting population of big brown bat, Eptesicus fuscus. The former species roosts alone or in small groups in clumped, disjunct populations along prairie rivers, most often in rock crevices (Holloway and

Barclay 2001), while the latter species is widespread and colonial, roosting in rock crevices, trees, or buildings (Kurta and Baker 1990). I compared the roost selection and roosting behaviour of M. ciliolabrum and E. fuscus in a rocky, riparian area in southern

Alberta (Lausen and Barclay 2002), building upon studies of E. fuscus (Lausen and

Barclay 2002, 2003, 2006b) and western long-eared bats, M. evotis (Chruszcz and

Barclay 2002, Holloway 1998) previously carried out in this study area. My comparison combined measurement of the physical and microclimate attributes of roosts with

31

behaviourial considerations represented by roost switching and genetic relatedness

among females and males.

Because the rock-roosts of E. fuscus are known to differ from those of sympatric

rock-roosting M. evotis (Chruszcz and Barclay 2002, Lausen and Barclay 2002), I

hypothesized that due to competition and resource partitioning, or differing

morphological, physiological and behavioural characteristics, E. fuscus and M.

ciliolabrum roosts would also differ in both physical attributes and microclimate. I also

hypothesized that roosting behaviour and genetic relatedness patterns in E. fuscus and M.

ciliolabrum would differ given their colonial versus solitary roosting differences. I included both males and females in my investigation, because male roosting ecology is a missing link in understanding many bat species’ ecologies, and because males often mediate gene flow in mammals (Greenwood 1980), providing an important component to understanding relatedness patterns. Because male-biased dispersal has been documented in a number of bat species (e.g. M. bechsteinii Kerth et al. 2002, Nyctalus noctula Petit and Mayer 1999, Macroderma gigas Worthington Wilmer et al. 1999), I predicted that

females in the study area would be related, while males would not be related to each

other or to females from the area, and would not show fidelity to roosts or the study area.

The comparison of roosting ecology and social behaviour in two sympatric

species of bats provides a unique opportunity to test several hypotheses about the

evolution of roosting behaviours. To date, studies of relatedness within maternity

colonies of bats have focused on colonies roosting in buildings or bat boxes (e.g. Plecotus

auritus Burland et al. 2001, M. bechsteinii Kerth et al. 2002, Rhinolophus ferrumequinum

Rossiter et al. 2002) and all have reported colonies comprised of mixed matrilines, where

32

closely related groups of females roost with unrelated females. However, building roosts

tend to be spacious and accommodate large numbers of bats, and there may be fitness

advantages for bats roosting in buildings rather than natural roosts (Esberard et al. 2005,

Lausen and Barclay 2006b, Law and Chidel 2007). As such, an investigation of

relatedness in natural roosts is necessary to determine whether mixed matrilines are an

artifact of spacious building roosts, or whether this social organization evolved naturally

in the species. To the best of my knowledge, relatedness patterns in solitary species have not been described or considered in their roost selection. Armed with a full complement of roosting ecology and relatedness information, I explore possible relationships among roost selection, roosting behaviour and relatedness patterns.

Materials and Methods

Study Site

I captured bats and characterized roosts along a short (2.5 km) NE – SW stretch of

the South Saskatchewan River valley, near Bindloss, Alberta (Bindloss Ferry Crossing N

50o38’, W 110o11’; Fig.5.1 of Chapter 4) in May - August 2000 and 2001 (for E. fuscus)

and 2002 (for M. ciliolabrum). The region is characterized by steep coulees (drainages

running perpendicular to the river) with short-grass prairie above, and a few cottonwood

trees (Populus spp.) in the valley bottom, at either end of the study area. Erosion of

bentonitic sandstones (hard cemented sandstone boulder and non-cemented mudstone) forms badlands features of hoodoos and cliffs. There were no human-made structures within 5 km of the study site and few trees, so roost choice by bats in the study area was limited to rock crevices or mudstone erosion holes. The 2.5 km river section had

33

numerous rock crevices, and was bordered by rolling grassy slopes upriver and

downriver. The main predators on bats at this location are thought to be snakes and owls,

as roost switching in association with owl presence, and direct predation of two radio-

tagged bats by a bull snake was documented (Lausen 2001). The climate of the study

area is arid (average precipitation June – August 13.7 cm), with mean daily minimum and

maximum temperature normals of 9.8oC/24.1 oC, 11.9oC /27.1 oC, and 10.9 oC /26.7 oC for

June, July and August respectively (Environment Canada 2000).

Captures in the study area took place at various locations, including among trees

(foraging areas) and near rock (roosting areas). Captures therefore took place along the

river and away from the river in coulees. One coulee (“Big Coulee”) in particular was

long (~1 km from river to prairie), wide (~250 m) and easily netted. Because this coulee was well delineated and I was able to catch many bats there, I present some results specific to this coulee in relation to the rest of the study area.

Study Species

Myotis ciliolabrum occupies arid regions of western North America. In Alberta, it

is found in disjunct patches of eroded rock and mudstone along river valleys (ASRD

2006, Holloway and Barclay 2001). M. ciliolabrum is one of the smallest bats in North

America, with non-pregnant adult females having a mean mass of 5.4 ± 0.1 g (mean ±

SE; n = 65 bats from southeastern Alberta), versus 21 ± 0.3 g (n = 59) for E. fuscus. Both

species have one young per litter in the study area, with parturition from late June to late

July depending on weather (C.L.L. pers. obs., Holloway and Barclay 2001, Kurta and

Baker 1990). Reproductive female M. ciliolabrum typically roost alone or in small

34 groups (e.g. 1 – 6 in southern AB, Holloway 1998). E. fuscus females typically roost in groups of 25 – 75 (reviewed in Kurta and Baker 1990), and colony size at the study site was up to 37 adults (or 52 adults and juveniles; Lausen 2001, Lausen and Barclay 2002).

Testes in males of both species are visibly enlarged by July, spermatogenesis occurs throughout the summer, then testes regress by September, sperm are stored in the caudae epididymides, and mating occurs in autumn (van Zyll de Jong 1985).

Capture

I captured individuals using mist-nets, and determined sex, reproductive status

(Racey 1988) and relative age (Anthony 1988). I took a 2 mm diameter wing-tissue biopsy, and in some cases placed a plastic split-ring band on the forearm (Barclay and

Bell 1988). Individuals were banded at the study site in 1998 (17 female M. ciliolabrum,

B. Chruszcz, pers. comm.), 2000 (43 male E. fuscus), 2001 (62 female and 72 male M. ciliolabrum, 62 male E. fuscus), and 2002 (58 female and 69 male M. ciliolabrum, 29 male E. fuscus). Additionally, I banded a colony of female E. fuscus at the study site from 2000-2002 (total 34 individuals).

I radio-tracked bats to locate day roosts. Radio-transmitters were applied to female

M. ciliolabrum (see below), and the roost structure and microclimate measured. I radiotracked female E. fuscus (n = 22) and characterized roosts as described in Lausen and Barclay (2002). Male E. fuscus were radiotracked in 2002 and roosting behaviour is described here, although roosts were not measured.

For genetic analysis, all individuals sampled were adults captured between 28 May

– 10 August to try to ensure individuals were sampled in their “summer roosting” areas

35

prior to the breeding season. All animals were cared for in accordance with the principles

and guidelines of the Canadian Council of Animal Care, and appropriate animal care

permits were obtained.

Radio-telemetry and Roost Characterization

I affixed radio transmitters (Holohil Systems, Carp, ON, Canada) between the

scapula of reproductive female bats using Skinbond® surgical adhesive (Smith and

Nephew United, Inc., Largo, FL, U.S.A.). Transmitter mass ranged from 0.45 – 0.56 g

(7.6 – 10.4 % of body mass). Although for M. ciliolabrum this exceeded 5% of body mass, the recommended limit (Aldridge and Brigham 1988), radio-tagged bats were monitored closely, and recaptured 1 – 4 days after initial tagging so that transmitters could be removed. There was no directional change in mass by the time transmitters were removed. Other studies also exceeding the 5% guideline reported no negative effects (Brigham et al. 1997, Chruszcz and Barclay 2002, Kurta and Murray 2002). I located roosts during the day following release using an R-1000 digital telemetry receiver

(Communications Specialists Inc., CA). Roosts were located each day a bat retained a transmitter.

I measured distance to level ground above and below roosts as an indication of potential risk from terrestrial predators. I measured aspect of the roost, slope of the ground containing the roost, and roost opening area, depth, and orientation (vertical,

horizontal or diagonal). I also categorized roost substrate as either boulder/rock, or

solidified mud (bentonitic mudstone). In addition, I characterized 56 random crevices to

determine whether bats selected roosts based on features other than availability. I divided

36 the roost area into 25m sections along the river valley and selected one random point per section. The nearest crevice large enough for a bat to occupy was then characterized (see

Lausen and Barclay 2002 for details). I obtained azimuth and angle of the sun for the study area from http://aa.usno.navy.mil/data/docs/AltAz.html.

I recorded roost microclimate and ambient temperature (Tamb) using Thermochron iButtons ®, ModelDS1921 (±0.5oC, Dallas Semiconductor Corp., Dallas, TX, U.S.A.) and HOBO Loggers® (±0.7oC, Onset Computer Corp., Pocasset, MA, U.S.A.). After a roost was located, bats were given the opportunity to return to the roost the following day. If bats switched roosts and the roost remained empty, a data logger was placed into the roost and recorded temperature every 10 min. Tamb was recorded in a radiation shield

1m above the ground at the valley bottom. All microclimate data were separated into

‘day’ (civil twilight near sunrise until emergence), and ‘night’ (emergence to civil twilight); mean difference between roost temperature and Tamb for day and night, maximum and minimum day and night temperatures, range of roost temperature, and amount of time to reach maximum roost temperature were calculated for each roost for the first 24 hour period of monitoring (see Lausen and Barclay 2003 for more details).

I characterized roosting behaviour of M. ciliolabrum by recording the number of individuals leaving each roost at emergence and measuring the linear distances between roosts used by each individual, and between their capture location and first roost. I compared physical and microclimate attributes of M. ciliolabrum roosts to 9 pregnancy and 15 lactation roosts of E. fuscus described by Lausen and Barclay (2002, 2003).

37

Molecular Genetics

mtDNA Sequences

I extracted DNA from wing tissue using the Qiagen DNeasy Blood and Tissue

Extract Kit (Alameda, CA) using the spin-column protocol. I sequenced a 250 base-pair

region of the hypervariable region II (HVII) on the control region of the mitochondrial

genome. DNA was amplified using primers L16517 and sH651 (Fumagalli et al. 1996,

Castella et al. 2001). Fragments were sequenced using the L16517 primer only due to the

presence of the large repeat. PCR reactions were performed in a 50 µL volume using 50-

100ng template, 1X PCR buffer (50 mM KCl, 10 mM Tris-HCl pH 8.8, 0.1% Triton), 2.5

mM MgCl2, 0.16 mM dNTPs, 0.8 µM of each primer and approximately 1 – 2 U Taq

DNA polymerase (isolated as in Engelke et al. 1990) in the following cycle: initial three minutes at 94°C followed by 25 cycles of one minute at 94°C, one minute at 54°C, 1.5 minutes at 72°C. PCR product was purified using QIAquick Gel Extraction Kit (Qiagen,

Alameda, CA) and sequenced using Big Dye® Terminator v3.1 sequencing kit (Applied

Biosystems, Foster City, CA) according to the manufacturer’s directions. Sequencing products were resolved on an ABI Prism® 3100 Genetic Analyzer (Foster City, CA). I aligned sequences in Sequence Navigator (Version 1.0 Applied Biosystems, Foster City,

CA).

Microsatellite Genotyping

I genotyped extracted DNA at 10 loci. The following sets of primers were used:

for M. ciliolabrum MME24, MMG9, MMH19, MMD9, MMD15, MMF19 (Castella and

Ruedi 2000), MYBE15, MYBE22 (Kerth et al. 2002), EF5 (Vonhof et al. 2002) and

38

Paur05 (Burland et al. 1998); for E. fuscus MMG9, MMG25 (Castella and Ruedi 2000),

MYBE22 (Kerth et al. 2002), EF1, EF6, EF14, EF15, EF20 (Vonhof et al. 2002), TT20

(Vonhof et al. 2001) and NN18 (Petri et al. 1997). I used a PCR volume of 15 µL

containing 1X PCR buffer (10 mM Tris buffer, pH 8.8, 0.1% Triton X-100, 50 mM KCl

and 0.16 mg/mL bovine serum albumin), 0.8-1.5mM MgCl2, 0.12 mM dNTPs, 0.2-0.27

mM of each primer, 0.4 units of Taq DNA polymerase, and 2 µL (~ 100 ng) DNA

template. Cycling was performed in a 9600 thermal cycler (Perkin-Elmer) under the

following conditions: 1 min at 94 °C, three cycles of 30 s at 94 °C, 20 s at 47 °C, and 5 s

at 72 °C, 33 cycles of 15 s at 94 °C, 20 s at 47 °C, and 1 s at 72 °C, followed by final

extension at 72 °C for 30 min. PCR products were resolved on model 377 ABI

sequencers (Applied Biosystems, Foster City, CA), and analysed using GENESCAN

(version 3.1) and GENOTYPER (version 2.0) software (Applied Biosystems, Foster City,

CA).

Statistical Analyses

Data were transformed to meet assumptions of homoscedasticity and normality

2 (log10, square root). T-tests, ANOVAs, ANCOVAs, χ , and Fisher’s Exact tests were

carried out in Stata 9 (2005, StataCorp LP, College Station, TX). I compared roosts used

during pregnancy and lactation, and crevices used as roosts versus randomly selected

crevices. Analyses used roosts as the statistical unit, although I repeated a sub-sample of

tests using individual as the statistical unit to ensure no artifacts arose from lack of

independence between roosts used by the same individual. Tamb was used as a covariate

in all ANCOVA models. I present mean ± SE.

39

Roost aspect was compared using a nonparametric statistical procedure for circular data (Fisher 1993). The test statistic can be evaluated using the chi-square distribution. I calculated mean directions ± SE (Fisher 1993).

Using microsatellite genotypes, I calculated relatedness of individuals using the method of Lynch and Ritland (1999) as implemented in the software SPAGeDi (Hardy et

al. 2002). Only loci not containing null alleles were used in these calculations. All loci

were employed in ML-Relate (Kalinowski et al. 2006) adjusting for null alleles using

1000 randomizations, to summarize relationships as unrelated, half-sibs (HS), full-sibs

(FS), and parent-offspring (PO). I present some between- and within-species

comparisons as percentage of related individuals (i.e. percentage of pairwise relationships

that are HS, FS and PO).

Results

Captures

In 2001 and 2002, respectively, I captured 130 and 102 adult M. ciliolabrum, 62 and 47 of which were female. I found evidence of site fidelity in both female and male

M. ciliolabrum. Of 16 recaptured bats (6 males, 10 females), 9 were caught < 50 m, three were between 50 and 400 m, and one was between 400 and 1000m from their original capture locations (distance from previous capture locations could not be determined for 3 females). Fifty percent of recaptured females and 67% of recaptured males were recaptured within 50 m of their original capture location. Ten of the 16 recaptured bats

(5 males, 5 females) had been banded in previous years (6 were within year re-captures).

Of the 5 males, 4 were banded as adults and one as a volant juvenile.

40

I also found site fidelity among male and female E. fuscus, with evidence of male site fidelity from year to year. In 2001 and 2002, out of 56 and 31 adult male captures, respectively, 4 and 7 were recaptures, with one within season recapture each year. Of the

9 between-year recaptures, 8 had associated capture histories; 50% (4 of 8 male recaptures) were banded as pups the previous year, and the other 50% were banded as adults; three had enlarged testes when banded, confirming philopatry of reproductive males. One male recaptured within 2002 was actually recaptured 3 times: 14 July, 27

August, and 12 September, suggesting that males may remain at a site for the duration of the summer. This male was reproductive, as confirmed by enlarged testes in the first two recaptures. Males born the previous year were found roosting with the maternity colony as late as 10 July.

Radio-tracking

Myotis ciliolabrum

I affixed radio-transmitters to 5 pregnant and 10 lactating M. ciliolabrum between

25 May and 13 August 2002. I also affixed transmitters to 2 non-reproductive and 1 post-lactating female, and discovered one male roost opportunistically; these roosts were used only in the analysis of roost substrate.

Reproductive females typically roosted alone or in pairs, with a mean group size of

1.4 ± 0.2 bats (pregnancy roost group size 1.4 ± 0.2, n = 8 roost groups, range 1 – 5; lactation roost group size 1.3 ± 0.2, n = 9 roost groups, range 1 – 5). Individuals switched roosts frequently; individuals used the same roost on consecutive days on only 3 of 21 occasions. Mean distance between consecutive roosts for an individual was 45 ± 6

41

m (n = 16 individuals; 2-4 roosts/individual; range 6.4 – 106 m). In this latter

comparison, a banded roost-mate not carrying a transmitter, but banded in one roost and

recaptured in another the following day, allowed for an additional roost distance to be

used in the mean calculation. Mean distance between capture location and the first roost

for an individual was 146 ± 23m (n = 15 individuals; range 4-580 m). To test that the

proximity of subsequent roost locations was not an artifact of transmitter bias, I released

two bats with transmitters 1 km from their previous roosts. They had been captured at

their roosts where their roost-mates previously wore the transmitter. Bats returned to

their “roosting area” in both cases, selecting roosts < 100 m from their original capture crevices.

Eptesicus fuscus

I affixed transmitters to 2 male E. fuscus on 8 and 12 July 2001. These bats were monitored opportunistically until 2 August. Because roosts were not verified every day, it

is possible that some roost switches between close crevices were not noted. The male

tagged on 12 July roosted in the same mudstone crevice (~1 km from his original capture

location) until 28 July, at which time the transmitter either stopped working or the bat left

the area. The bat was last confirmed foraging on 22 July, meaning that he was in the

same roost for at least 11 days. The other male initially roosted alone ~27 m from his

capture location but switched roosts at least twice, moving 190 m and ~400 m from his

initial roosting location. He returned to this initial roost by 2 August at which time he

was captured at his roost, confirmed to again be roosting alone, and the transmitter was

removed. This male, therefore, switched roosts 4 times using 3 roosts over the course of

42

25 days. Female E. fuscus were radio-tracked and their roosts were described by Lausen

and Barclay (2002, 2003, 2006b).

Roost Structure

Myotis ciliolabrum

Radio-tracking yielded 30 roosts used by M. ciliolabrum: 11 roosts used during

pregnancy and 19 during lactation. Roosts of pregnant and lactating females were

compared to random crevices (Table 2.1, MYCI vs. RCrev). Opening size differed

significantly (F2,65 = 26.02, p < 0.001), with the opening areas of pregnancy (20.2 ± 6.5

cm2) and lactation roosts (38.4 ± 19.6 cm2) being significantly smaller than those of

randomly selected crevices (301 ± 58 cm2; pairwise comparisons: p < 0.001). Openings

of pregnancy and lactation roosts did not differ from each other (p = 0.99). Randomly

available mudstone crevices offered significantly smaller opening sizes (147 ± 55 cm2) than crevices in solid boulders (387 ± 116 cm2; t = 2.18, d.f. = 29.2, p = 0.037). Roost

entrances faced south more often than randomly available crevices (Table 2.1), with no

difference in aspect between pregnancy (173 ± 17°) and lactation roosts (171 ± 6°; χ2

equivalent = 0.008, p > 0.50). Roost substrate was significantly more often bentonitic

mudstone than solid boulder material (Fisher’s exact test p = 0.021), but there was no

difference between the reproductive stages with regard to roost substrate.

Distance of the roost from flat ground above and below, depth, slope, and crevice

orientation were not significantly different between pregnancy and lactation roosts, or

from what was randomly available (Table 2.1). Female bats roosted on both sides of the river throughout the reproductive period. Roost characters did not differ when individual

43

was used as the sample unit and only the first roost for each individual was considered (5

pregnancy and 10 lactation roosts).

Comparison with Female E. fuscus

Myotis ciliolabrum roosts had smaller opening sizes, and were closer to flat

ground above and below, shallower, located on less steep inclines, more southerly in

orientation, more horizontal in orientation, and more often in bentonitic erosion holes, compared to E. fuscus roosts (Table 2.1). An analysis of variance for depth using reproductive stage and species as main effects, revealed that depth of E. fuscus and M. ciliolabrum roosts differed (F2,59 = 6.39, p = 0.003) during lactation (E. fuscus: 64.1 ±

7.7 cm; M. ciliolabrum: 37.4 ± 7.2 cm; p < 0.05) but not during pregnancy (40.6 ± 7.5

cm; 34.1 ± 7.5 cm; p > 0.05), when E. fuscus selects significantly shallower roosts than

lactating E. fuscus (Lausen and Barclay 2002).

Roost Microclimate

Myotis ciliolabrum

I collected microclimate data for 9 pregnancy roosts and 18 lactation roosts of M.

ciliolabrum. The temperature difference between Troost and Tamb (∆T) was recorded every

10 minutes and compared. During the day, the mean ∆T was significantly smaller during pregnancy (-0.82 ± 1.14°C, n = 9, max 11.0°C, min -11.7°C) than during lactation (1.6 ±

0.54°C, n = 18, max 14.3°C, min -7.0°C; ANOVA, reproductive stage F1, 25 = 4.7, p =

0.04). ∆T at night did not differ between the reproductive stages (pregnancy mean difference 4.65 ± 0.89°C, n = 330, max 10.6°C, min -1.69°C; lactation 5.27 ± 0.55°C,

44

max 12.9, min -0.18°C; ANOVA, reproductive stage F1,25 = 0.39, p = 0.54). Using Tamb

as a covariate, maximum night temperature (MXNT), minimum night temperature

(MNNT), maximum day temperature (MXDT), and minimum day temperature (MNDT)

did not differ between pregnancy and lactation roosts (Table 2.2). Tamb had a significant

effect in all of these models. Mean values are also presented, not adjusted for Tamb.

The mean time for roosts to warm to maximum temperature after sunrise, differed significantly between reproductive stages, with roosts warming more quickly during

2 lactation (pregnancy: 768 ± 42 min; lactation: 637 ± 44 min; F1,25 = 4.79, R = 0.16, p =

0.038). Tamb did not significantly influence the amount of time to reach maximum roost

temperature. I determined azimuth and angle of the sun from the horizon for 15 June and

30 July, 2002 to illustrate differences in solar exposure for south-facing crevices during

the pregnancy and lactation periods in the study area. On 15 June and 30 July, the angle

of solar incidence at azimuth 170° - 180° (M. ciliolabrum mean aspect 174°, slope 62°) is

58.0 ± 0.5°, and only 37.4 ± 0.9° for E. fuscus roosts when the sun is at 120 - 130° (E. fuscus mean aspect 127°, slope 92°); M. ciliolabrum roosts received nearly a 90o angle of sun incidence, but sunlight on E. fuscus roosts was more indirect, illustrating the advantage M. ciliolabrum had by roosting on south-facing gradual slopes at this latitude.

On 30 July in the first 7 hours following sunrise, the sun’s rays were 12.7 ± 0.3° (range

9.3 – 17.2°) more south than on 15 June, indicating that south-facing crevices potentially receive direct sunlight earlier in the morning during lactation than during pregnancy.

45

Comparison with E. fuscus

Comparing only roosts of pregnant and lactating individuals, E. fuscus and M. ciliolabrum roosts did not differ in MXDT or MNNT, corroborated by the similarity of

their roost temperature ranges. During the day in both species, maximum roost

temperature was sometimes above and sometimes below maximum Tamb, regardless of

reproductive stage (Fig. 2.1). To illustrate that this is not just due to the nature of the

crevices in the area, data for the other rock-roosting species in the study area (M. evotis)

are shown (Chruszcz 1999). M. evotis selects roosts with cold day temperatures. At night,

minimum roost temperatures for all three species are almost always above minimum

Tamb.

Roosts of E. fuscus and M. ciliolabrum differed in their cooling rates. Roosts

differed significantly in MXNT, the temperature at the end of the day when the bats

emerge (ANCOVA: adjusted means: E. fuscus 22.1 ± 0.5°C, M. ciliolabrum 23.2 ±

2 0.5°C, F5, 45 = 26.9, R = 0.75, p < 0.001, species p = 0.008, reproductive stage p = 0.02,

Tamb p < 0.001, species*Tamb p = 0.004, reproductive stage*Tamb p = 0.04). Significant

interactions indicated that the roost temperature at the time of emergence responded to

Tamb differently between species and reproductive stages. This may be indicative of E. fuscus selecting deeper roosts with larger opening sizes than M. ciliolabrum, especially during lactation. Species differed in the MNDT of roosts but not significantly (adjusted

2 means: E. fuscus 18.2 ± 0.6°C; M. ciliolabrum 16.8 ± 0.6°C; F2,48 = 25.3, R = 0.51, p <

0.001, species p = 0.08, Tamb p < 0.001). The fact that M. ciliolabrum roosts had cooled

more by dawn than E. fuscus roosts, corroborates findings of Lausen and Barclay (2002)

that deeper roosts retain heat. While species and reproductive stage did not explain a

46

significant amount of variation in the time to reach maximum roost temperature

2 (ANOVA F4,46 = 3.12, R = 0.21, p = 0.02, species p = 0.14, Tamb p = 0.07, reproductive stage p = 0.92, species*reproductive stage p = 0.004), the significant interaction between

reproductive stage and species occurred because pregnancy and lactation roost

temperatures have opposite patterns of temperature changes in the two species. Time to reach maximum Troost is higher in E. fuscus lactation roosts than pregnancy roosts (back- transformed adjusted means: pregnancy 595 ± 45 min; lactation 720 ± 41 min), but lower

in M. ciliolabrum (779 ± 60 min; 656 ± 36 min). This may reflect the shift of E. fuscus to

deeper crevices during lactation, while M. ciliolabrum remained in shallow, south-facing

crevices.

Genetics

I sequenced a 280 base-pair fragment of HVII in M. ciliolabrum and a 240 base-

pair fragment in E. fuscus. I found 6 HVII haplotypes (n = 35 females and 10 males;

mean nucleotide differences 2.6 ± 0.34, 0.93%, range 1 – 5) in M. ciliolabrum adults, and

14 in E. fuscus (n = 16 females and 10 males; mean nucleotide differences 3.2 ± 0.60,

1.3%, range 1 – 5) in the study area. In M. ciliolabrum, all 6 haplotypes were found in

females and 3 of the same haplotypes were found among the males. In E. fuscus, 4

haplotypes were found in the females, with 1 of these haplotypes also being found in the

males, along with 9 additional haplotypes. In E. fuscus, each male haplotype was

represented by a single male.

I microsatellite genotyped 82 M. ciliolabrum (50 females, 32 males) and 34 E.

fuscus (24 females, 10 males). I calculated the relatedness (r) among males, among

47

females, and between males and females for each species in the study area (Table 2.3).

The relatedness value for M. ciliolabrum for each of the 3 relationship groupings was

similar, whereas in E. fuscus, female relatedness was higher than among males or

between males and females.

An analysis of bats captured and genetically analyzed from “Big Coulee”, a rocky

drainage within the study area, revealed more genetic similarity among bats (especially

females) within this coulee compared to with bats from the rest of the study area (Table

2.4; comparing of confidence intervals suggests some of these differences are not

statistically significant). Matrilines showed geographic patterns; out of 15 sequenced Big

Coulee individuals (10 females, 5 males), 14 had the same mtDNA haplotype (Mci14);

the one male of a different haplotype (Mci13) was captured during the middle of the

night and may have been foraging but not roosting there. If only dawn and dusk captures

are considered, this coulee consisted of a single mtDNA haplotype. Netting in foraging

areas along the river (trees ~1 km away in either direction from Big Coulee) captured

males and females of all 6 haplotypes (Mci1, 6, 7, 8, 13, 14). Microsatellite genotypes of

individuals captured within one hour of emergence or dawn in Big Coulee produced the

relatedness (r) values in Table 2.4. Relatedness of males to females and within females

is higher in Big Coulee individuals than among the rest of the study area, suggesting

some degree of genetic clustering in this coulee, supporting the mtDNA results. Males in

Big Coulee were unrelated, but the sample sizes were too small (n = 3) to draw

conclusions.

I examined genetic similarity of M. ciliolabrum individuals roosting together to

determine whether roost-mates were relatives. Four roosts were examined and roost-

48

mates in all cases were identical in mtDNA haplotypes: Roost A females (n = 5) had

Mci7; Roost B females (n = 5) had Mci14 (roost located in Big Coulee); Roost C females

(n = 2) had Mci13; Roost D females (n = 2) had Mci13. The latter two roosts were < 15

m apart, and ~ 800 m and ~1670 m away from Roosts B and A, respectively. One female

from Roost C was highly related to the two Roost D individuals (r = 0.59 and 0.31) providing further evidence of geographic clustering of haplotypes in small subdivided areas of the study area. Nuclear relatedness for the roosts was as follows: Roost A,

0.1836 ± 0.0707; Roost B, 0.3657 ± 0.0945; Roost C, 0.2416 and Roost D, 0.3760 (mean r for roosts = 0.2778 ± 0.0556; Tables 2.3 and 5.8B).

Eptesicus fuscus sometimes roosted singly, but most often roosted in a fission- fusion pattern, ranging from one large maternity colony to several small groups varying

in group membership. Examination of mtDNA haplotypes of small groups revealed no

patterns of roostmate haplotype preference. Of 8 colony assemblages examined in 2001

where at least 2 females were mtDNA sequenced, none showed that all roost-mates had

identical haplotypes. Mean nucleotide difference among roost-mates (n = 3 roosts, 103

pairwise comparisons) was 2.22 ± 0.19 (0.92%, range 0 - 5). Mean nuclear relatedness

within roosts was r = 0.0451 ± 0.0078 (n = 7 roosts, mean r range among roosts -0.0435

– 0.0928; Table 2.3).

Discussion

Roosts and Roosting Patterns

The rock-crevice roosts of Myotis ciliolabrum rock crevice roosts differed from random crevices in having smaller opening sizes, more southerly aspects, and more often

49 bentonitic mudstone substrates. Other studies also report M. ciliolabrum roosting in mudstone (Barbour and Davis 1969, Garcia et al. 1995, Holloway 1998, Navo et al. 2002,

Tuttle and Heaney 1974), which may be selected for its insulating qualities (Lausen and

Barclay 2002). Small opening sizes may reflect this bat’s solitary nature, avoidance of predation, microclimate requirements, or some combination of all three; however, because opening sizes of randomly selected cavities are smaller in mudstone than in boulder substrate, it is not clear whether opening size or substrate is the quality being selected, or whether this combination provides optimal roost conditions.

Myotis ciliolabrum differs from some other bats in not changing roosting habits with reproductive status. In lactation, M. evotis, A. pallidus and E. fuscus shift to deeper roosts, which provide more stable temperatures (Lewis 1996, Chruszcz and Barclay 2002,

Lausen and Barclay 2002, Vaughan and O’Shea 1976). Because no structural attributes differed between pregnancy and lactation roosts of M. ciliolabrum, and because roost structure affects microclimate (Chruszcz and Barclay 2002, Lausen and Barclay 2003,

Lewis 1996, Vaughan and O’Shea 1976), it was not surprising that microclimate in pregnancy and lactation roosts was similar; interestingly all maximum and minimum roost temperatures were the same across reproductive stages, although roosts used in lactation received more direct sunlight earlier in the day and therefore warmed more quickly than roosts used in pregnancy.

While M. ciliolabrum was found most often in shallow mudstone erosion holes with small openings that were close to flat ground, and faced south, E. fuscus roosts are deeper, occur equally in mudstone and boulders, are on steep slopes far from flat ground above, have larger openings (Lausen and Barclay 2002), and are further from flat ground

50

below (unpub. data, Chapter 2). Roost substrate should affect roost microclimate most

when roosts are shallow, as deep roosts are more stable regardless of the substrate

(Chruszcz and Barclay 2002, Lausen and Barclay 2003). That E. fuscus roosts in

relatively deep crevices and does not select for substrate, while M. ciliolabrum roosts in

shallow crevices and selects mudstone is evidence that roost substrate may be a selected feature when crevices are shallow. Boulder heats up slowly, and has a high heat capacity, providing a relatively cool microenvironment during the day, and drawing heat away from the body of the bat (Lausen and Barclay 2002); M. evotis, for example, takes advantage of this by selecting boulder roosts and using torpor every day (Chruszcz and

Barclay 2002). Mudstone has a low heat capacity providing a more insulating surface

(Lausen and Barclay 2002), and would therefore allow a bat to spend less energy maintaining an active body temperature.

Despite substantial differences in physical roost attributes, microclimate differences between the roosts used by M. ciliolabrum and E. fuscus were minimal, differing only in

the temperature of roosts at the time of emergence. E. fuscus, roosting in deep roosts

with large opening sizes, experienced similar maximum and minimum roost temperatures

as M. ciliolabrum did in shallow roosts with smaller openings, likely because crevice depth and opening size influence roost microclimate in opposing ways (Lausen and

Barclay 2003). Because roost microclimate is important for thermoregulation (McNab

1982), the microclimate similarity between M. ciliolabrum and E. fuscus superficially suggests that patterns of torpor might be similar for these species; however, because M. ciliolabrum is a smaller bodied bat and smaller bodies experience greater benefits from torpor (Prothero and Jurgens 1986), they are likely to use torpor to a greater extent.

51

Smaller animals generally require shorter fetal development times and shorter juvenile

growing periods (Vaughan et al. 2000). Myotis ciliolabrum would therefore naturally

experience fewer costs from extended use of torpor which slows reproduction and growth

(McNab 1982, Racey 1973). This bat generally does not have the benefit of clustering

behaviour to minimize heat loss and remain active (Kunz 1982, McNab 1982), unlike E.

fuscus individuals who clustered as one group in the study area during lactation, roosted

in deep crevices and used only shallow torpor (Lausen and Barclay 2003). If M.

ciliolabrum uses torpor to a greater extent, then they would benefit from roosts that allow passive rewarming prior to emergence (Prothero and Jurgens 1986); this may explain why M. ciliolabrum roosts are warmer at time of emergence than E. fuscus roosts. A

comparison of thermoregulation of M. ciliolabrum and E. fuscus is needed to test this

hypothesis.

Pregnant and lactating E. fuscus in the study area roost in relatively large groups

(mean group size, 8.25 ± 0.96 bats, maximum 37 adults; Lausen 2001, Lausen and

Barclay 2002) that are often audible and easily detected by accumulations of feces

outside the roost (C.L.L., T. Pretzlaw, pers. obs.). M. ciliolabrum, roosting singly or in

small groups (≤ 5), occupies roosts that are less conspicuous, with small openings, and

few feces present at the entrance. I was never alerted to roosts by audible sounds or

odour. The more conspicuous roosts of E. fuscus may increase predation risk and explain

why they select roosts away from flat ground above and below (Lausen and Barclay

2002, unpub. data), unlike M. ciliolabrum who did not select roosts based on distance

from flat ground.

52

Myotis ciliolabrum switched roosts frequently but never moved far. Frequent roost

switching also occurs in rock-crevice dwelling M. evotis (Chruszcz and Barclay 2002), E.

fuscus (Lausen and Barclay 2002), and A. pallidus (Vaughan and O’Shea 1976). This

may be a tactic to avoid predators or ectoparasites (Lausen 2005, Lausen and Barclay

2002, Lewis 1995, 1996).

Despite low roost fidelity, male and female M. ciliolabrum and E. fuscus stayed in

a defined roost area. Although female M. ciliolabrum were not radio-tracked for more than a few days, in that time consecutive roosts were close to each other. Additionally, recapture locations and the genetic structure associated with coulees suggests that individual M. ciliolabrum have small roost ranges (<110 m), in contrast with the colony of E. fuscus at this study area which roosted over a 1.25 km stretch of the river (Lausen and Barclay 2002). “Roosting range” may be larger for E. fuscus because suitable large, deep crevices are more limited in the study area than the small mudstone roosts used by

M. ciliolabrum, and therefore involve greater distances between roosts. E. fuscus rotated through a series of roosts in the first season of research (2000), and then repeated a similar pattern of roost occupation the following two seasons (unpublished data), suggesting that although a great deal of roost switching occurs within a season (Lausen and Barclay 2002), there is year-to-year fidelity to roosts and pattern of occupation. This has also been found for C. tuberculatus in tree cavities (O’Donnell and Sedgely 2006).

Because M. ciliolabrum was only followed for a few days and within only one season, I cannot comment on roost re-use.

Although I did not follow radio-tagged M. ciliolabrum at night, given that tracked

bats always roosted ≤ 580 m from their capture locations (in a netting area stretching 2.5

53

km along the river), and bats banded in previous years were all recaptured within 1 km of their original capture locations, I suggest that the home range of individual M. ciliolabrum is relatively small. Because most foraging by bats in the prairies tends to occur around riparian trees (Holloway and Barclay 2000), this species may be restricted to roosting where rock crevices and riparian trees are in close proximity. Riparian trees for foraging were present at either end of the study site, making it difficult to determine whether distances between foraging and roosting sites were small because netting did not occur more than 2.5 km from roosts, or because M ciliolabrum truly has small foraging- roost distances and selects areas where trees and roosts are in close proximity.

Relatedness Patterns

I found evidence of site fidelity by both female and male M. ciliolabrum. While fidelity among adult females is common in many species of bats (Kunz 1982, Lewis

1995), fidelity among males is less common and few studies describe male roost behaviour in areas of natural roosts. I therefore examined male genetic relatedness to determine whether males were returning to their natal area. Male M. ciliolabrum had few haplotypes, and all of them matched female haplotypes from the area. Sequenced males and females captured in a single coulee all had identical haplotypes that differed from adjacent regions within the study area. This suggests that during the summer, at least some male M. ciliolabrum return each year to their natal area and that within a riparian roosting area, distribution of individuals is not random. M. ciliolabrum maternity roosts contained females of identical haplotypes, and roosts of matching haplotypes were found

54

adjacent to each other, while roosts of different haplotypes were hundreds of metres

apart, demonstrating geographic genetic-clustering on a small scale in this species.

In general, relatedness values were low among males and females for both species

(r near zero) across the study area, and genetic distances with neighboring areas were low

(Chapter 4) suggesting that nuclear gene flow was occurring with other bats in southern

AB, and that the clumped summer distributions did not represent breeding groups.

Relatedness values were higher when the study area was subdivided into smaller

geographic units; males from within a coulee displayed higher r values with females from

the same coulee than they did with females from the rest of the study area, providing

additional evidence that males and females are clustering in fine-scale geographic units of

relatedness.

While relatedness among M. ciliolabrum was generally low, males were more

related to each other and to the females in the study area than E. fuscus males were to each other and to E. fuscus females; this suggests more gene flow in E. fuscus and corroborates findings from a larger scale genetic study of which this area was a part

(Chapter 4). In contrast to M. ciliolabrum, E. fuscus males have extreme diversity of mtDNA haplotypes (each of 10 sequenced males had a different haplotype, only one matching a female haplotype) suggesting mixing of males with those from other areas along the river valley, also corroborating the larger-scale genetic study (Chapter 4). E.

fuscus males and females disperse more than M. ciliolabrum and gene flow in M.

ciliolabrum is likely to be mediated largely during fall migration along rivers (Appendix

I) or at hibernation sites (Chapter 4).

55

As with M. ciliolabrum, I found evidence of E. fuscus males returning to the study area, with half of the male E. fuscus recaptures being pups from the previous year.

Returning (i.e. one year old) male pups were captured roosting in the same crevices as the female colony only until 10 July, after which new pups were prevalent in the colony and no adult males were found in the maternity roosts (unpublished data). At least some males (nonreproductive and reproductive) remained in the study area for much of the summer, and as late as mid-September. Whether breeding takes place in this area is unknown, but acoustic monitoring in the fall suggests that this study area may be used for winter hibernation or is near a hibernation area (unpublished data) and males may therefore remain in the area to mate and then hibernate.

Although I expected to find little roost fidelity in males, radio-tracked E. fuscus males did not switch roosts often (>8 days), and when they did it was within a small (≤

400 m) distance. There are few studies of roost switching in natural roosts for male E. fuscus, but Brigham (1991) found males switched crevices every 2.8 days. Roost switching behaviour, often attributed to avoidance of predation and parasitization (Lewis

1995, 1996) may be minimal because of males’ tendency to roost alone. Females, as previously documented (Lausen 2001), showed fidelity to the area and to a particular set of rock-roosts. Eptesicus fuscus females roosted in a fission-fusion pattern over a 1.25 km length of the river in the study area, forming relatively small groups during pregnancy, but one large colony during lactation (Lausen and Barclay 2002). Sampling approximately half of the colony produced 4 haplotypes. Membership of roosting groups did not follow matrilineal lines, corroborating a study of E. fuscus in tree roosts (Metheny and Rüppell 2006); this is in contrast with the single matriline roosting structure of M.

56 ciliolabrum. In E. fuscus, mean nucleotide differences among colony haplotypes ranged from 1.3-6.4% in rock or tree-roosting colonies, and from 1.4 – 2.1% in building roosts

(Table 4.8B, Chapter 4), suggesting that haplotypes can be very different, and mixed matriline colonies appear to be common in this species, regardless of whether the roosts are natural or buildings.

Synthesis

Armed with a full complement of roosting ecology and relatedness information about E. fuscus and M. ciliolabrum, this study provided a unique opportunity to explore possible evolutionary mechanisms that may have led to the different patterns of observed roost selection and roosting behaviour. While the influence of physiological requirements on roost selection and roosting behaviour has been well documented (e.g. Chruszcz and

Barclay 2002, Lausen and Barclay 2003), the influence of kin selection has received less attention. Roost selection and relatedness seem correlated in M. ciliolabrum, but it is less apparent what role relatedness plays in the roosting ecology of the E. fuscus. I hypothesized that E. fuscus maternity colonies, like those of M .ciliolabrum, would comprise highly related females in natural crevice roosts due to the small roost size and small number of these roosting structures, together with the tendency for female philopatry. Such a finding would suggest that kin selection is likely to be a key determinant of roost-mate assemblages, with each relative increasing the fitness of the other by such means as thermal advantages and information transfer (Roverud and

Chappell 1991; Vaughan et al. 2000). Such a finding would also suggest that mixed

57

matrilines in buildings reflect recent merging of matrilines, an artifact of limited

availability of large suitable roosts. However, this is not what I found.

In E. fuscus, I found mixed matrilines even in small natural roost colonies, and I

found females of the same matrilines as present in the rock crevice colonies, in

neighboring colonies on the same river (Chapter 4), suggesting that at least some kin had

dispersed. This also suggests that kin may have been available to remain in the natal

area, but due to the tendency to roost with non-relatives and due to crevices that are limited in size, forced dispersal may have occurred (Kunz 1982). In other words, roosting with non-kin has been favoured. Such interdependence among mixed kin and non-kin may have evolved to guarantee a minimum number of clustering individuals for such advantages as mutual warming of reproductive adults and growing pups, and guarding of juveniles during the raising young (Kerth and Konig 1999). That mixed matrilines are present regardless of colony size or reproductive state, raises the question

“why not roost with kin when enough kin are available to meet thermal needs?”

Additionally, “Why are kin not preferentially selected for clustering within large mixed groups?” It would seem that when larger groups are not needed, for example during pregnancy when torpor is more extensive (Audet and Fenton 1988, Lausen and Barclay

2003), that kin would preferentially be selected, given the potential advantages of associating with kin (Rossiter et al. 2002, Vaughan et al. 2000). This is not the case for

E. fuscus in this study, and for other colonial-roosting bat species (e.g. Burland et al.

2001, Kerth and Konig 1999). This does, however, seem to be the case in M. ciliolabrum; small groups of females always consisted of one matriline, and this may

58

stem from its more solitary nature and consequently different evolutionary past (see

below).

A number of studies propose maintenance of social bonds as a possible reason for

roosting with non-relatives even when a large clustering group is not needed (e.g. Kerth

and Konig 1999). These “non-kin” social bonds, however, do not seem to extend beyond

the roost, with foraging associations either non-existent (e.g. Kerth et al. 2001) or existing only between relatives (e.g. Rossiter et al. 2002). Additionally, numbers of individuals of mixed matrilines present in large building roosts seem to well exceed the minimum number of individuals likely needed to realize thermal advantages. Yet, even in these larger colonies, individuals do not roost preferentially next to kin (Kerth and

Konig 1999). While the “maintenance of non-kin social bonds hypothesis” may indeed explain these observed roost associations, one cannot rule out other hypotheses. For example, the tendency to roost with non-relatives may have evolved from an evolutionary advantage of spreading predation risk out and thereby minimizing predation within a family unit. As a colonial species, E. fuscus roosts are not nearly as cryptic as those of

M. ciliolabrum. The tendency to choose non-kin roost-mates would have been selected for if predation risk was a stronger force than natal philopatry. One can imagine that predation is an important force in a natural setting, and the predation event witnessed with roosting radio-tagged individuals in this study area is evidence that predation does occur (Lausen and Barclay 2002). That roosts of E. fuscus were selected in a way that may minimize predation (selecting for distance from flat ground above and below) further corroborates this. Emergence patterns that were clumped and altered by the presence of owls or humans (pers. obs.) is further evidence that predation was a perceived

59

threat by E. fuscus in the study area. Predation risk appears to be perceived differently by

M. ciliolabrum and E. fuscus as reflected in their different choice of roosts and likely

reflects differences in their detectabilities by predators. Roosting of kin may therefore have not been selected against in M. ciliolabrum which has small group sizes, and often roosts adjacent to, rather than with, kin. Understanding the role that roosting with unrelated individuals plays in overall fitness of colonial-roosting individuals requires further study and is needed to elucidate how such a behaviour likely evolved. It is likely that such an investigation is best carried out in a natural setting where current evolutionary forces are likely similar to those of the recent past.

While much focus has been placed on understanding relatedness and social behaviour in colonial bat species, to the best of my knowledge, no study has attempted to

understand relatedness and its link to roosting ecology in solitary species. As a much

smaller bat than E. fuscus, M. ciliolabrum is likely to use more extensive torpor, and thus

not benefit from clustering; in fact to make full use of torpor, it may require roosting

alone. Microclimates in small insulated mudstone roosts that warm late in the day, could

provide conditions conducive to extensive use of torpor and facilitate additional energy

savings from passive re-warming (Prothero and Jurgens 1986). Such small roosts and

solitary roosting would also minimize predation risk making issues of siting less

important. This would open up a tremendous choice of roosts from abundant small

mudstone crevices covering the river valley walls. Such roost abundance would facilitate

seemingly unlimited philopatric behaviour in males and females, each benefiting from

remaining in an area of familiarity and retaining associations with kin (Kerth 2006,

Waser and Jones 1983). My finding of kin roosting in the immediate microgeographic

60 natal area supports this evolutionary scenario, and supports my hypothesis that relatedness influences the selection of crevices for roosts. One can see how these related

“neighborhoods” could be missed when studying solitary animals, and highlighting the importance of multi-scale spatial investigation. On an even larger scale, it is likely that this localized kin clustering could result in a low level of dispersal in both males and females of this species, as I found (Chapter 4).

It is unclear whether competition for roosts plays any role in the selection of the different types of crevices used by E. fuscus, M. ciliolabrum and M. evotis, or whether physiological or morphological factors are entirely responsible. While these three species have distinct roosting patterns in the study area, it is not clear that these patterns hold in other habitats or species assemblages. Because preferred habitat can differ with context

(e.g. Crockett and Hadow 1975, Kurta et al. 1996), and because when habitat resources are limiting, niches typically narrow (Caughley and Sinclair 1994, Rosenzweig 1981), documentation of roost selection in other habitats under another set of well-described conditions will be necessary to tease apart resource partitioning from other factors.

Interestingly, the small mudstone roosts used by M. ciliolabrum in the study area appear to have suitable microclimates for the torpor patterns used by M. evotis, as described by

Chruszcz and Barclay (2002); in the study area, M. evotis roosts exclusively in boulder cracks (Chruszcz and Barclay 2002), but in a stretch of river valley badlands ~ 200 km west of the study area, where boulders are not abundant, M. evotis roosts almost exclusively in mudstone erosion holes (J. Gruver, pers. comm.). Even further west in the

Rocky Mountains, M. evotis roosts under piles of rocks (Solick 2004). Similarly, in New

Mexico, Chung-MacCoubrey (2005) concluded that a single, generalized tree-roost

61

profile was not possible for M. evotis, M. volans (long-legged bat) or M. thysanodes

(fringed bat) as they adjust their roost selection with the type of tree roosts available.

Myotis sodalis uses different types of tree roosts in Michigan compared to Illinois and

Missouri, despite similar availability, and microclimate was suggested as a possible

reason (Kurta et al. 1996), although interspecific competition can not be ruled out. As

with other roost-selection studies, my work adds another piece to the overall puzzle.

However, the context is well described, setting the stage for a comprehensive

understanding of the roosting ecology of E. fuscus and M. ciliolabrum in the prairies.

62

Table 2.1. Comparison of M. ciliolabrum (MYCI) roosts with E. fuscus (EPFU) roosts

(Lausen and Barclay 2002) and randomly selected crevices (RCrev). All MYCI and

EPFU means are for pregnancy and lactation roosts only. All t-tests were performed for

heteroscedasty and evaluated using p = 0.025 to account for multiple comparisons. The

first test statistic represents the test of EPFU versus MYCI, the second is for MYCI

versus Random Crevices.

Attribute Mean ± SE Test statistic df p

2 Opening size EPFU: 154 ± 23 cm t = 4.6 60 <0.001* MYCI: 31.7 ± 12.6 cm2 RCrev: 301 ± 58 cm2 t = 4.5 60 <0.001*

Distance from level EPFU: 7.92 ± 2.0 m t = 3.7 49 <0.001*

ground above MYCI: 1.39 ± 0.27 m RCrev: 5.26 ± 2.0 m t = 0.27 47 0.88

Distance from level EPFU: 4.22 ± 0.63 m t = 6.2 56 <0.001*

ground below MYCI: 1.31 ± 0.21 m RCrev: 4.27 ± 1.5 m t = 0.27 53 0.79

Depth EPFU: 55.3 ± 5.9 cm t = 3.1 58 0.003* MYCI: 36.2 ± 5.2 cm RCrev: 55.4 ± 10.9 cm t = 0.92 64 0.36

Slope of ground EPFU: 92.9 ± 5.5° t = 4.5 56 <0.001* MYCI: 62.0 ± 4.0° RCrev: 67.5 ± 5.0° t = 0.85 84 0.40

Aspect EPFU: 126 ± 9° χ2 equiv. = 34.1 1 <0.001* MYCI: 174 ± 7° RCrev: 119 ± 8° χ2 equiv. = 8.2 1 <0.005*

63

Attribute Mean ± SE Test statistic df p

Orientation EPFU: n = 5 (16%) horizontal 1 0.049* MYCI: n = 12 (40%) horizontal Fisher’s Exact RCrev: n = 17 (57%) horizontal 1 0.81 Roost Substrate EPFU: n = 33 (53%) mudstone 1 0.21 MYCI: n = 25 (68%) mudstone Fisher’s Exact RCrev: n = 15 (39.5%) mudstone 1 0.021*

64

Table 2.2. Analysis of variance of microclimate for roosts used by pregnant vs. lactating

M. ciliolabrum. Mean roost attributes, adjusted for covariate where appropriate

(unadjusted values in parentheses), are presented ± S.E. for each of the reproductive

stages. Pregnancy and lactation roosts differed significantly only for the amount of time

for the roost to reach maximum temperature (MINS*). Ambient temperature was

included as a covariate in all models prior to step-wise elimination. All models reduced to

only one significant term. Other abbreviations are: MNDT, minimum day temperature,

usually occurring around dawn; MXDT, maximum day temperature; MXNT, maximum

night temperature, usually occurring at time of bat emergence after sunset; MNNT, minimum night temperature, often occurring in the later part of the night; RRNG, range in roost temperature over a 24 h period.

Roost Attribute Pregnancy Lactation F df R2 P

MNDT 16.8 ± 0.9°C 16.3 ± 0.6°C 22.7 1, 25 0.48 Tamb: 0.001 (17.1 ± 1.1oC) (16.2 ± 0.9°C)

MXDT 25.6 ± 2.0°C 27.4 ± 1.4°C 7.7 1, 25 0.23 Tamb: 0.01 (27.6 ± 2.1°C) (26.4 ± 1.5°C)

MXNT 21.6 ± 0.9°C 22.4 ± 0.6°C 68.2 1, 25 0.73 Tamb: <0.001 (22.9 ± 1.6°C) (21.8 ± 1.2°C

MNNT 17.0 ± 0.9°C 17.9 ± 0.6°C 35.0 1, 25 0.58 Tamb: <0.001 (18.1 ± 1.2°C) 17.3 ± 1.0°C)

RRNG 9.4 ± 2.2°C 11.2 ± 1.5°C 3.6 1, 25 0.13 Tamb: 0.07 MINS 768 ± 42 min 637 ± 44 min 4.8 1, 25 0.16 Reproductive stage: 0.04*

65

Table 2.3. Pairwise relatedness (r; calculated from microsatellite genotypes) among males, among females and between males and females for M. ciliolabrum and E. fuscus at the Bindloss Ferry crossing study site in southeastern Alberta. The first values for

Among Females include all females from the study area. Within Roosts is the mean relatedness among females found roosting together. In parentheses is the number of pairwise comparisons and range. Percentages of pairs that had r values indicative of half sib, full sib or parent-offspring (≥HS) are listed for each relationship and species.

Species M. ciliolabrum E. fuscus Relationship r r ≥ HS r r ≥ HS

Among 0.0551 ± 0.0035 0.0396 ± 0.0082 8.3% 10.6% Females (1485, -0.2309 – 0.8756) (276, -0.1469 – 0.6717) Within Roosts mean Within Roosts mean (n = 4 roosts) (n = 7) 0.2778 ± 0.0556 0.0451 ± 0.0078 (22 pairwise r values, 50% 8.8% (312, -0.1364 – 0.6717) -0.0656 – 0.8756) (range of mean roost r (range of mean roost r values, -0.0435 – 0.0928) values, 0.1836 – 0.3760)

Among Males 0.0438 ± 0.0069 6.0% 0.0076 ± 0.0158 2.8% (300, -0.1345 – 0.7384) (36, -0.0955 – 0.4481) Between Males and 0.0514 ± 0.0039 8.4% -0.0114 ± 0.0045 3.8% Females (1133, -0.2044 – 0.7064) (216,-0.1471 – 0.2692)

66

Table 2.4. Pairwise relatedness (r) among males, among females and between males and females for M. ciliolabrum in a roosting area away from the river ("Big Coulee") and in foraging areas closer to the river (Other FC Capture Sites). In parentheses is the number of pairwise comparisons. Relatedness calculated using microsatellite genotypes. All Big

Coulee individuals have the same mtDNA HVII sequences, while there are 6 haplotypes among the Other FC capture sites.

r n pairs range Big Coulee Among females 0.0929 ± 0.0269 55 -0.0942 – 0.8756 Among males 0.0192 ± 0.0161 3 -0.0027 – 0.0514 Between males and females 0.0811 ± 0.0286 33 -0.0701 – 0.5363 Other Capture Sites Among females 0.0565 ± 0.0045 946 -0.2309 – 0.8099 Among males 0.0471 ± 0.0080 231 -0.1345 – 0.7384 Between males and females 0.0502 ± 0.0041 968 -0.2044 – 0.7906 Comparing Big Coulee to Other Capture Sites BC males to Other females 0.0525 ± 0.0122 132 -0.1140 – 0.6149 BC males to Other males 0.0335 ± 0.0136 59 -0.0873 – 0.5139 BC females to Other females 0.0483 ± 0.0054 484 -0.2169 – 0.7156 BC females to Other males 0.0376 ± 0.0069 242 -0.1961 – 0.5973

67

Figure 2.1. Daily roost maximum (A, B, C) and minimum (D, E, F) temperatures plotted

against maximum and minimum ambient temperatures respectively. Data are all from the same study area on the South Saskatchewan River valley (BL, Fig. 4.1 of Chapter 4; near

Bindloss, AB). Graphs from left to right are E. fuscus (Lausen 2001), M. ciliolabrum (this

study) and M. evotis (Chruszcz 1999). Hollow squares are pregnancy roosts, solid

diamonds are lactation roosts, and solid triangles represent either pregnant or lactating

individuals. Line is Troost = Tamb.

68

Figure 2.1 69

CHAPTER 3: Winter Bat Activity in the Canadian Prairies1

Introduction

To understand the ecology of an animal requires knowing its year-round habitat

selection. While much effort has gone into describing summer roost selection by bats, far

less has gone into locating winter habitats. Where most bats hibernate in Alberta is not known. A few cave hibernacula in the Rocky Mountains have been found, but account for relatively small numbers of bats (<1000; Schowalter 1980).

Stimulated by late summer captures in Dinosaur Provincial Park (DPP), Alberta

(Schowalter and Allen 1981), I began autumn mistnetting in this area to determine

whether bats remained in the area for the winter. Attempts to radio-track bats in during

this fall period, however, proved fruitless. Bats were still active in the area right up until first snowfall, however, and it was not been clear how I would be able to prove conclusively that bats were hibernating in the area knowing that after the first snow fall, they would likely disappear into hibernacula for the winter. Because it has been hypothesized that bats may naturally arouse from hibernation as ambient temperatures rise (Brack and Twente 1985, Twente et al. 1985), I monitored DPP in late November

2002, during a Chinook (a warm, dry wind that occasionally blows down the east slopes of the Rocky Mountains). I detected winter bat flights. Given that temperatures earlier that month reached a low of -12oC (Environment Canada, accessed 16 April 2005, http://

www.climate.weatheroffice.ec.gc.ca), hibernation should have begun and I concluded

1 A version of this chapter has been published as: Lausen, C.L. and R.M.R. Barclay. 2006. Winter bat activity in the Canadian prairies. Canadian Journal of Zoology 84:1079-1086.

70

that bat activity at this time of year was indicative of hibernal arousal. Cold Canadian winters preclude insect activity and I assumed bat activity, also. However, when bats were detected active during warm Chinook periods, I realized that hibernation locations might be confirmed by acoustic monitoring throughout the winter. I predicted that bats would be detected during these warm Chinook periods and began to hypothesize why such arousals may occur.

All mammalian hibernators arouse periodically throughout the winter (Kayser 1965;

Willis 1982). Bouts of winter torpor are days or weeks in length (Willis 1982), and >80%

of fat depletion in hibernation can be due to arousals (Kayser 1965; Thomas et al. 1990).

The function of these energetically expensive arousals is not well understood, and

numerous physiological hypotheses have been proposed, including water intake for ionic

balance and circulatory functioning (Speakman and Racey 1989; Thomas 1995; Thomas

and Geiser 1997), to sleep (Trachsel et al. 1991), maintenance of neural circuitry (Popov

et al. 1992), lymphocyte regeneration (Burton and Reichman 1999), and more (reviewed

in Humphries et al. 2003). While most mammalian hibernators are relatively dormant

during these periods of arousal (Willis 1982), some animals, such as bats, are active,

spending additional energy to move around or leave the hibernaculum.

As small flying insectivorous hibernators, non-migratory bats face unique

thermoregulatory challenges in temperate zones. Unable to hoard food, bats are fat-

storing hibernators, with energy reserve sizes limited by small body size and the need to

fly. Bats have evolved to successfully over-winter in areas such as the Canadian prairies

where winters can be harsh; temperatures can reach < -45oC and insect abundance is low

or non-existant for >150 days each year. Winter arousal and activity of bats is well

71

documented for warmer temperate climates, such as in southern Ontario where some bats

hibernate in buildings (Brigham 1987); in Poland (Krzanowski 1959), southern England

(Avery 1985; Park et al. 1999), Minnesota (Swanson and Evans 1936), and Nevada

(O’Farrell and Bradley 1970), where bats reportedly arouse and forage during winter

months; and in Germany (Sendor et al. 2000) and Indiana (Whitaker and Rissler 1992)

where emergence has been documented without evidence of foraging. In more harsh

temperate climates, such as the Canadian prairies, winter flights of bats are not expected

due to lack of insects, although arousals within roosts may occur for physiological

reasons.

Several hypotheses have been proposed to explain why bats may be active during

hibernal arousals: to feed (Avery 1985, Willis 1982), to drink (Thomas and Geiser

1997), or to change hibernation sites (Daan 1973, Sendor et al. 2000). Feeding has been

observed during some winter bat flights (e.g. Avery 1985), however, diminishing energy

as a possible reason for arousal and winter bat flight has received less support than the

importance of water (Hays et al. 1992, Speakman and Racey 1989, Thomas 1995,

Thomas and Geiser 1997, Whitaker and Rissler 1992).

The bat activity that I detected at DPP in the winter of 2002 was unexpected. As

this was the first non-building bat hibernation area identified in the Canadian prairies, I increased my monitoring areas to include other rocky riparian regions. I hypothesized

that bats use rock crevices in the river valley as hibernacula, and arouse during warm

Chinook periods when temperatures exceed 0°C. My goals were to determine what

species were hibernating in the river valleys, at what temperatures bats were flying, what

72

the sex and body condition was of the individuals, and if possible, shed light on the

reason for winter bat flights. I also wanted to find and describe natural winter roosts.

Materials and Methods

I acoustically monitored two main locations along the Red Deer River in Alberta:

Dinosaur Provincial Park (DPP; N 50o 45’ 23’’ W 111o 31’ 02’) and East Coulee (EC; N

51o 19’ 58” W 112o 28’ 49”). Both areas consist of badlands terrain rich in sandstone rock crevices and solidified mud erosion holes, with DPP being a larger area (badlands

features extend > 5 km from the river) than EC. EC has many abandoned coal mines.

Bat species captured or acoustically detected here in summer months are: Eptesicus

fuscus, Beauvois, 1796, Myotis lucifugus, LeConte, 1831, M. ciliolabrum, Merriam,

1886, M. evotis, H. Allen, 1864, Lasionycteris noctivagans, LeConte, 1831, Lasiurus

cinereus, Beauvois, 1796, and L. borealis, Müller, 1776. This part of Alberta is arid,

averaging < 350 mm of precipitation per year, with temperature extremes ranging from

>40oC in the summer (July average daily maximum 26 oC) to <-45oC in the winter

(January average daily minimum -17 oC); mean number of days with maximum and

minimum temperatures ≤0°C are 77.8 and 194.8, respectively (Environment Canada,

1971-2000, Climate Normals, accessed 16 April 2005, http://

www.climate.weatheroffice.ec.gc.ca/climate_normals).

Preliminary data were collected from DPP in 2002-2004. Additionally, I

opportunistically monitored Writing on Stone Provincial Park campground on the Milk

River (N 49o 03’ 25”, W 111o 37’ 28”) 2002-2003. This arid rocky riparian area features

73 hoodoos, eroded pillars of sandstone. My main data collection took place 2004-2005 at

DPP and EC.

I used remote acoustic zero-crossing analysis to detect and identify bats (Corben

2002). These detection systems consisted of an AnaBat detector (Titley Electronics,

Australia) with a tape recorder in 2002 - 2003, and with a Compact Flash ZCA Interface

Module (CF ZCAIM; Titley Electronics, Australia) for all subsequent work. I powered the system with an external battery charged by a solar panel. I monitored DPP and EC continuously during winter 2004 - 2005, beginning 16 October 2004. Data were collected until 17 June and 4 May 2005 for EC and DPP, respectively. Previous to this, in 2002 - 2003, limited by the tape recording system, I monitored opportunistically in late fall and early spring, and in the winter only during warm Chinook winds at DPP and at

Writing on Stone Provincial Park.

I digitized call sequences from tapes using AnaBat 6.0, and visualized all calls

(from tape and CF ZCAIM) using Analook 4.9j (C. Corben, 2004, http://www.hoarybat.com). Big brown bat (Eptesicus fuscus) calls (minimum frequency

(Fmin) < 30 kHz) were identified visually, and Myotis calls (Fmin > 30 kHz) were analyzed further for species identification. While E. fuscus calls are similar to the calls of L. noctivagans, this latter species migrates south out of the province during the winter

(Smith 1993), so all ~25 kHz winter passes were assumed to be that of E. fuscus. I followed the same procedure as used by Wilson (2004), and manually selected complete search phase calls (maximum frequency [Fmax] ≥ 60 kHz), analyzed these using

Analook, and identified the species using discriminant function analyses (see below).

74

I captured bats in DPP along the Little Sandhill Creek, a small tributary draining

into the Red Deer River, using mistnets on the following winter/spring nights: 30 March,

and 23 and 24 April 2004, 3, 11, and 23 February, and 1, 4, 5, 9 March 2005. I also

captured bats periodically in the summer and fall months in DPP from 2002 - 2004. Each

bat was weighed and sexed. I recorded degree of toothwear (scale of 1 - 7, where 1 is

reserved for the sharp teeth of juveniles; Holroyd 1993), measured forearm length using

calipers, and applied a plastic split-ring band for future identification. Each bat was kept

for 1 hour to check for feces. During winter 2004-2005, in all but the first capture, I also

offered water to each bat. All animals were cared for in accordance with the principles and guidelines of the Canadian Council of Animal Care and appropriate animal care

permits were obtained.

In February 2005, I attached 0.5g radiotransmitters (Holohil, ON, Canada) to 3 E.

fuscus (2 males and 1 female) that I captured in DPP, and tracked them to their rock-

crevice hibernacula. I accessed roosts using ropes, measured outside dimensions, and

where possible, roost depth. To assess whether insect prey were available to flying bats

from 3 February to 9 March 2005, I erected a suction insect trap 1.5 m above ground each

night that I netted (n = 13). I operated the trap for the duration of netting sessions.

o I recorded ambient temperature (Ta; ± 0.7 C) using two HOBO Loggers® (Onset

Computer Corporation, Pocasset, MA, USA), encased in solar radiation shields, placed at my two main study sites. Temperature readings were taken every 15 min to 1 hour depending on the site and time of year. If a bat pass occurred between two temperature readings, and the Ta at the start of the interval was warmer than at the end, the Ta at the

75

start of the interval was used to provide a conservative estimate of flight Ta; otherwise, Ta was interpolated assuming a linear pattern.

The insect trap, AnaBat and temperature dataloggers were all located in the netting area, on a creek tributary <1 km from the Red Deer River. Nets were strung along the meandering Little Sandhill Creek in an area where the hoodoo features create a narrow

(~200 m) bottleneck in the creek valley. To the north of the netting area lies a large flat flood zone and the Red Deer River; to the south the creek, surrounded by badland features, extends >5km, where it originates as melt-water and run-off from the surrounding higher elevation grasslands.

I performed statistical analyses using S.A.S. Version 9.1 and Stata 9. I performed direct discriminant function analysis (Proc DISCRIM) on echolocation call parameters

(minimum frequency [fmin], mean frequency, characteristic slope, and [fmax- fmin]/duration) extracted using Analook. Myotis calls (i.e. those with minimum frequency > 30 kHz) that were of sufficient quality to analyze, were classified as M. ciliolabrum, M. evotis or M. lucifugus. Quadratic discriminant function analysis was performed (Tabachnick and Fidell 2001), using reference calls from locally captured M. ciliolabrum (n = 64), M. evotis (n = 26), and M. lucifugus (n = 168). I recorded reference calls using an AnaBat and audio taperecorder; all M. ciliolabrum and M. evotis were recorded after hand-release, while most (148) M. lucifugus call sequences were of free- flying individuals outside known roosts. Overall cross-validation error was 0.085 (range

0.078 - 0.100), and only passes that could be identified to species with ≥95% probability were accepted. Mass of captured bats was compared using analyses of covariance, with

76

season of capture and forearm length as the dependent variable and covariate,

respectively. All means are presented ± SE.

Results

Acoustic Detection

I carried out noncontinuous acoustic recording at DPP during 2002 and 2004, and

first detected winter bat activity on 19 November 2002, when maximum daytime and

minimum nighttime temperatures were 14oC and 2.4oC, respectively. During this

Chinook period, I detected E. fuscus and M. ciliolabrum. I detected E. fuscus on 7 March

2004, the first day that year that the detector was out. Myotis were first detected 11

March 2004 and discriminant function analyses of Myotis detected between 11 - 31

March 2004 revealed that both M. ciliolabrum and M. evotis were present. I did not

detect Myotis lucifugus. No winter bat activity was detected at Writing on Stone

Provincial Park.

Beginning 16 October and 26 September 2004, I recorded continuously at DPP

and EC, respectively, and detected E. fuscus and Myotis activity throughout November

and part of December (Fig. 3.1). In total, I recorded 535 passes at DPP and 65 passes at

EC in November, and 377 passes at DPP and 71 passes at EC in December. The last

Myotis bat pass in 2004 at both DPP and EC was 18 December when maximum day and

minimum night temperatures were 2.9/-7.3oC at DPP and -1.5/-5.8oC at EC. The last E. fuscus pass was 24 December at EC, and 25 December at DPP, when maximum day and minimum night temperatures were -1.1/-12.0oC and -1.5/-18.1oC respectively. The

longest period during which no bat was detected was 23 days (25 December - 18 January)

77

at EC and 22 days (26 December - 16 January) at DPP (Fig. 3.1). During this time

ambient temperature reached -46oC. After this cold period, E. fuscus and Myotis were

again active starting 18 January and 2 February at EC, and 17 and 30 January at DPP. At

both locations, during the winter 2004-2005 (21 December 2004 - 21 March 2005), 43 of

67 Myotis passes were analyzable, and of these, 23 could be identified to species with

≥95% probability. All passes were of M. ciliolabrum and M. evotis, with the exception of

one Myotis lucifugus pass detected at DPP on 6 March 2005. M. lucifugus was not

detected again until 3 April 2005.

I detected bats on nights when the temperature at the time of emergence (defined

as civil twilight, when the sun is 6o below the horizon) was below 0oC (Myotis: 6 nights

at EC, 12 nights at DPP; E. fuscus: 17 nights at EC and 20 nights at DPP). Myotis and E.

fuscus were not detected on nights when emergence temperatures were colder than -5.8 oC (n = 43) or -5.3oC (n = 46), respectively. I detected bats flying on nights when daytime ambient temperatures had not gone above freezing (Myotis: 1 night each at EC and DPP; E. fuscus: 8 nights at EC and 4 nights at DPP). The coldest ambient temperatures at which E. fuscus and Myotis passes occurred were -7.9oC and -6.3oC, respectively. Monitoring repeated in winter 2005-2006 revealed coldest passes at -7.9°C and -7.3°C, respectively (unpublished data). On nights when temperatures remained above these temperature limits, bat activity occurred throughout the night. During

Chinook periods, warmest temperatures did not always occur early in the evening, and peak bat activity could not be predicted; when temperatures warmed prior to sunrise, bat activity was detected at dawn, including one E. fuscus detected 19 minutes after sunrise.

Strong winds associated with Chinook periods may make bat flight difficult, and this

78

could explain why ambient temperature was not always a good predictor of bat activity

(Fig. 3.2).

Bat Captures

I captured E. fuscus opportunistically at DPP in 2003 - 2005 (Table 3.1). Most

captures (95% of 184) were male. The testes of most males (84% of 134) captured in

July were enlarged suggesting spermatogenesis. Bats captured in winter included both

young and old individuals, as indicated by the degree of tooth wear (Table 3.1).

Of the 9 E. fuscus captured in winter 2004-2005, four bats had an earthy smell,

suggesting that they may have been roosting in contact with moist soil. No feces were produced by these winter captures. The mean mass of male and female E. fuscus captured in winter 2004-2005 (n = 9, x = 16.5 ± 0.6 g) was significantly less than the mean mass of adult bats in autumn 2003 and 2004 (n = 12, x = 21.3 ± 0.8 g; F1,14 =

10.96, p = 0.005), but not significantly different than E. fuscus in spring 2004 (n = 3, x =

16.5 ± 1.0 g) and summer 2004 (n = 115, x = 17.9 ± 0.18 g; model F4,134 = 11.5, p <

0.0001; season F3,134 = 11.4, p < 0.00001; forearm length covariate F1,134 = 8.5, p = 0.004;

Tukey’s WSD test: autumn captures differed significantly from captures in the other 3

seasons which did not differ from each other).

Eight of the E. fuscus captured in winter 2004-2005 were captured on the side of

the net (north-facing) that suggested they were traveling back to hibernacula (located

south of netting area – see below) from the river. I offered water from an eyedropper to 7 of these bats before releasing them, but they did not drink. Four urinated in the mist net when handled. Only one bat was captured immediately at emergence moving towards the

79

river, suggesting it had just left its hibernaculum. This bat did not urinate in the net, and

was the only bat that drank water when offered; he consumed >3 mL of water shortly

after capture and urinated when released an hour later.

I affixed radiotransmitters to two male (3 and 23 February 2005) and one female

(11 February 2005) E. fuscus and tracked them to their winter roosts, up-creek from the

netting area. I tracked both males to the same rock-crevice roost, approximately 200 m

from their site of capture on Little Sandhill Creek. Although the female was also

captured at this same site, I radiotracked her to a roost 5 km away.

The rock crevice that the males roosted in was a crack in a boulder 10 m from flat

ground above and 22 m from flat ground below. The opening had a large deposit of bat

feces, was 15 cm high, and ranged from 1 to 2.5 cm in width. The roost faced SE (130o

aspect) on a steep (80o) slope. Due to the shape of this crevice, a depth measurement was

not possible. The crevice that the female roosted in was a tubular erosion hole in solidified mud, which was located 3 m and 3.5 m to flat ground above and below, respectively. The roost was at least 3.55 m deep, with an oval opening of 7 cm x 5 cm, faced SE (168o aspect) and was located on a steep (70o) slope. Feces deposit at entrance

was less than at male hibernaculum. Roosts were accessible only with ropes and were

likely never disturbed by other wildlife or humans. Transmitter signals did not change

location for the duration of the radiotransmitters’ battery life (12-18 days).

Insects and Weather

I caught no insect in the light-funnel trap, although I observed a total of 4 moths

flying on the nights of 5 and 9 March 2005. A male E. fuscus captured 5 March 2005

80

was kept for one hour and while he did urinate, he did not deficate, suggesting that he had

drunk but did not feed. Feces were produced by the one E. fuscus captured on 30 March

2004.

In 2004/5 the Canadian prairies experienced a winter that was on average 1.6oC warmer than normal, although it was only the 27th warmest winter since 1948

(Environment Canada, accessed April 2005, The Green Lane,

http://www.ec.gc.ca/envhome.html). I thus assume that the bat activity detected in 2004-

2005 was not unusual. DPP experienced a complete snow-melt at the end of January

2005, leaving little to no snow on the ground for most of February and March; a thin

layer of ice remained on the river, with small pockets of snow remaining in sheltered

areas of the valley. Open water (i.e., melted creek/river water) was not abundant until the

end of February.

Discussion

Little is known about hibernation of bats in the North American prairies. The

results of my study provide the first records of natural hibernacula for bats in the

Canadian prairie ecozone, of Myotis flying during mid-winter in Canada, and of

hibernating E. fuscus flying outside rock crevice hibernacula in mid-winter. I detected

bats flying at temperatures well below zero (≥ -7.9 oC). While bat activity below 0oC has been noted at high altitudes in arid regions of the southwestern United States (Nevada:

O’Farrell and Bradley 1970, Hall et al. 2005; New Mexico: L. Lewis, personal communication, 2005), this is the first time such bat activity has been recorded at low altitude and high latitude.

81

There is a universal occurrence of periodic arousals among hibernating mammals,

but most hibernators arouse for a few hours and remain relatively inactive in their

hibernaculum (Willis 1982). Periodic euthermy is thought necessary to restore

physiological balance (Willis 1982). That bats partake in energetically expensive flight

during arousal, suggests that bats may be unique in their hibernation behaviour, with

requirements beyond restoration of the euthermic state. While feeding has been

suggested as a reason for winter flight (Avery 1985), it was not the impetus for winter

flights in this study. Insects were not active at the ambient temperatures experienced in

this study (Taylor 1963), supported by the lack of insects in the light trap, and bats did

not produce feces.

Bats captured in winter weighed less than those in the fall but did not differ from

bats captured in summer. While some studies have suggested that only starving

individuals make winter flights in an attempt to attain food (Brigham 1987), my data suggest that not all bats active in winter have critically low energy reserves, which corroborates other studies (Hays et al. 1992).

Because of the large non-furred surface area of wing and tail membranes, bats

have higher evaporative water losses than other similar-sized mammals (Webb 1995). A

laboratory study of Pipistrellus pipistrellus indicated that bats must drink every 9-12 days

during hibernation (Speakman and Racey 1989). Dehydration may be a more important

arousal factor in bats than in other small fat-storing hibernators. Bats hibernating in

caves experience humid environments, may lick condensation from their fur (reviewed in

Speakman and Racey 1989), and have been observed drinking (reviewed in Davis 1970).

82

The only E. fuscus that consumed water when it was offered was the bat captured on its way out of the hibernaculum immediately at emergence, and most bats caught later in the night urinated upon capture. While it is possible that bats were urinating metabolically- produced water, voiding of this type of urine is typically observed in association with

immediate arousal from hibernation (Davis 1970, Kallen 1964). The only bat captured

on its way out of the hibernaculum did not urinate until its release one hour after

drinking. My sample size is small, but suggests that bats flying in winter during this study

may have been drinking; the source of water during part of the winter was not evident,

given the freezing temperatures and paucity of snow in the area. Dehydration has been

suggested as a factor for hibernal arousal (Thomas 1995), and I suggest it may play a role

in winter bat flight.

Relative humidity in rock-crevice roosts of arid regions is low (Lausen 2001),

compared to that of cave roosts (Davis 1970), and dehydration from evaporative water

loss may be an even greater problem for bats hibernating in arid climates (Kallen 1964,

Thomas and Cloutier 1992), such as the Canadian prairies, Nevada, and New Mexico, where subzero winter flights have been reported. Roosting in narrow crevices that would offer few microclimate options, and no water source, may force bats to fly outside the hibernaculum, and at temperatures typically considered too cold for bat activity. Because my data involve small samples sizes and anecdotal accounts, I suggest that further research is necessary to determine the possible effect that dehydration has on arousal and activity of bats hibernating in arid environments.

As indicated by relative toothwear and reproductive condition, bats active in the winter were not of a particular age class, contrary to a previous study that suggested that

83

juvenile E. fuscus are more active than adults (Brigham 1987). However, the sex ratio of my winter captures was highly skewed towards males. This may accurately reflect the number of males and females that are active in winter, if males hibernating in DPP outnumber females. However, the male bias could also be an artifact caused by sampling

< 200 m from the rock crevice where 2 males were tracked, and 5 km from the roost used by the female. This assumes that there is sexual segregation within the hibernation area, something that has been documented elsewhere for E. fuscus (Mills et al. 1975).

Winter flight activity seems to vary with bat species and location (O’Farrell and

Bradley 1970, Whitaker and Rissler 1992). E. fuscus is described as a hardy species often found roosting in cold, dry conditions (Davis 1970), and winter activity has been observed previously (Brigham 1987, Rysgaard 1942). Because the frequency of arousals increases with body mass (French 1988), the larger size of E. fuscus compared to that of

M. evotis and M. ciliolabrum may explain why I detected more winter flights of E. fuscus. However, the location of hibernacula of the different species relative to where I placed the detectors could also produce this skew. The fact that Myotis were detected at

all suggests that the need for winter flight is common among species in the study area.

I detected only one winter flight of M. lucifugus, a species which tends to cluster

in large groups (Kurta and Baker 1990), and roost in high humidity hibernacula

(reviewed in Fenton and Barclay 1980). It is possible this species does not hibernate in

large numbers in the rock crevice area of DPP, despite roosting there in the summer.

Alternatively, its acoustic absence during most of the winter might reflect less frequent

winter flights; evaporative water loss may be lower due to clustering (Studier 1970),

thereby reducing the need for winter flight (Daan 1973).

84

Where most E. fuscus hibernate is largely unknown (e.g., Mills et al. 1975). While

some hibernate in buildings (Whitaker and Gummer 2000), few natural E. fuscus

hibernacula have been identified (Kurta and Baker 1990). Indeed, it has been

hypothesized that because E. fuscus is absent or rare in most cave hibernacula, the

hibernaculum of choice for this species has historically been trees, roosts that would not

be insulative enough in cold northern areas (Whitaker and Gummer 2000). Whitaker and

Gummer (2000) thus proposed that heated buildings have provided this species with non-

tree hibernacula, allowing it to expand northward. My data indicate that hibernation by

E. fuscus in northern areas is possible without buildings, due to their use of small rock-

crevice hibernacula (this study; Neubaum et al. 2006) that are difficult to detect. Thus, I

suggest that the historic geographic range of E. fuscus was greater than hypothesized by

Whittaker and Gummer (2000), and that E. fuscus did not move northwards in response

to human-made roosts.

Because climate directly influences hibernating mammals (Humphries et al. 2001),

global climate change is likely to influence bat populations over-wintering in the

Canadian prairies. Since 1948, average yearly temperatures in Canada have been

increasingly above normal (using 1951-1980 as baseline; Environment Canada, accessed

April 2005, The Green Lane, http://www.ec.gc.ca/envhome.html). If the Canadian prairie climate continues to become drier with climate change, as predicted by Environment

Canada (1997), bats hibernating in rock crevices may be burdened with additional evaporative water loss brought about by decreased humidity and increased temperatures

(Kallen 1964). One response may be more frequent arousal and winter flights, placing

greater strain on stored fat reserves. In the Canadian prairies, where higher altitude

85

refugia are not available, distributional ranges may shift (Humphries et al. 2002).

Establishing baseline data for species composition of bats in prairie hibernation areas,

including frequency of winter flights, will allow for ongoing monitoring to determine if

changing prairie climate is altering distribution and abundance of bats (Humphries et al.

2003).

Table 3.1. Captures of E. fuscus at Dinosaur Provincial Park from 2003 - 2005. Sample size for mean mass is capture number, unless otherwise stated. Pregnant females and individuals not held for at least one hour prior to weighing are not included in mean mass.

Year Dates Captures Range of Toothclass (1 - 7) Mass (g)-- Mean and Range

Males Females Male Female

2003 4 - 22 Oct. 7 2 3 - 7 19.2±0.7 22.8±0.3

(16.3 - 20.4) (22.5 - 23.1)

2004 30 March 1 0 5 18.5 N/A

2004 23 - 24 April 1 1 2 - 7 15.9 15.2

2004 24, 27 June 20 1 2 - 7 16.3±0.6 (n = 14) 16.2

(12.6 - 20.9)

2004 14, 18 July 134 5 2 - 7 17.9±0.2 (n = 52) N/A

(14.6 - 20.8)

2004 24 Sept. 3 0 5 - 6 22.9±0.5 N/A

2005 3 Feb. - 9 March 8 1 1 - 7 16.0±0.3 21.0

(14 - 17.1) 8686

Figure 3.1. Bat passes per night and associated temperatures for Dinosaur Provincial Park, 16 October - 6 March 2005. The upper temperature line is the temperature at which light levels would have been typical of bat emergence, and the lower temperature line is minimum night temperature. Clear bars indicate E. fuscus passes, dark bars indicate Myotis passes. 8787

Myotis Activity E. fuscus Activity Temp. at Emergence Light 70 15 Night Temp. Low 10 60 5

0 Bat Passesper Night 50 C) o -5

-10 40

-15

Temperature ( 30 -20

-25 20 -30

-35 10

-40

-45 0 16-Oct 28-Oct 9-Nov 21-Nov 3-Dec 15-Dec 27-Dec 8-Jan 20-Jan 1-Feb 13-Feb 25-Feb Figure 3.1 Date (2004-2005) ` 88

89

rch C) ° saur Provincial Park, 16 October – 20 Ma ergence light levels for Dino

Temperature at Emergence Light Levels ( Light Levels at Emergence Temperature ature at em -30-20-100102030

0

70 60 50 40 30 20 10 Bat Passes per Night Night per Passes Bat 2005. Figure 3.2. Bat passes per night versus temper 90

CHAPTER 4: The Effect of Landscape on the Population Structure of Prairie Bats

Introduction

Understanding how landscape features influence animal movement and dispersal

is fundamental to ecology. This interaction of animal and environment, however, is not

always apparent, particularly when viewed from the perspective of one small study area.

In some cases it may be best understood using indirect methods, such as genetic

population structure applied to a larger, regional scale. Studies of genetic differentiation and ecology have been reciprocal ventures; how ecological and demographic processes shape genetic structure is being studied and described, which now enables biological interpretations of genetic structure to elucidate ecological and demographic processes

(e.g. Simonsen et al. 1998, Wimmer et al. 2002). In a world of stochastic events and increasing human disturbance, understanding the forces that shape populations provides us with some measure of our potential influence over future biodiversity.

Populations structured by landscape features are commonly described in the literature. For example, rattlesnake (Crotalus horridus) populations are structured by

basking habitat (Bushar et al. 1998), black-tailed prairie dogs (Cynomys ludovicianus) by

habitat loss and pest control (Roach et al. 2001), and grizzly bears (Ursus arctos) by highways and settlements (Proctor et al. 2005). Patterns of structure are becoming evident, and often predictable, depending on factors such as habitat selection, mating patterns, and degree of mobility. Population structure is expected to be greater in animals

with habitat specificity (Wolff 1999), defined social groups (van Staaden 1995), low dispersal, and low mobility (e.g. Avise 1996, Baker et al. 1995).

Given their ability to fly, bats might be expected to have reasonably little population structure, similar to non-sedentary birds (Avise 1996). While panmixia and large-scale structure has been found in some species of bats (e.g. McCracken and Gassel

1997, Webb and Tidemann 1996), structure and even complete fragmentation on relatively small scales has been found in others (e.g. Castella et al. 2000, Miller-

Butterworth et al. 2003, Rossiter et al. 2000), suggesting that population structure is variable in Chiroptera. This is to be expected given the extreme behavioural and morphological diversity in bats (Kunz and Fenton 2003). It has been hypothesized that bats that are strong fliers, with high wing-loading, will have greater gene flow and consequently less genetic structure than less capable fliers with lower wing-loading

(Entwistle et al. 2000). To the best of my knowledge, no one has directly tested this hypothesis, although a review of the literature suggests this trend exists (Table 4.1).

Therefore, I designed a study to compare species of different sizes, differing in wing- loading and consequently flight abilities. Because gene flow is also influenced by landscape features (e.g. Miller-Butterworth et al. 2003), I tested the influence of flight abilities on gene flow by examining use of landscape features. For bats, having roosting habitat available each day is critical in the movement and dispersal of animals, limiting their use of the landscape. An inability to fly long distances together with high roost specificity, for example, would likely result in a highly structured population with structure closely corresponding to landscape features providing suitable roosting habitat.

I therefore incorporate the influence of roost specificity and landscape features into my comparison of genetic population structure.

I examined the population structure of three species of prairie bats differing in their size, wing-loading and roost specificity. Big brown bats, Eptesicus fuscus, and little brown bats, Myotis lucifugus, regularly roost in both rock crevices and human-made structures during the summer, while M. ciliolabrum roosts exclusively in rock crevices.

Eptesicus fuscus is relatively large, with a higher wing-loading than the other species

(wing-loading is 6.7, 7.5 and 9.4 N/m2, respectively for M. ciliolabrum, M. lucifugus and

E. fuscus, Norberg and Rayner 1987; note name change for M. ciliolabrum from M.

leibii; van Zyll de Jong 1985). Larger bats have higher wing-loading, and must achieve faster flight speeds to maintain lift (Altringham 1996). Eptesicus fuscus is therefore

considered more capable of long distance flight than the other two. Myotis ciliolabrum

has the smallest body and wing size and the lowest wing-loading. I thus predicted, based

on differing flight abilities, that population structure would be least in E. fuscus, greater

in M. lucifugus and greater in M. ciliolabrum.

To test the effect of landscape on genetic structure, I selected a prairie landscape

because of its relatively homogeneous topography. The prairie landscape is relatively flat

and tree-less. River valleys, therefore, provide the main foraging (riparian trees) and

roosting (rock crevices) habitat for bats, although buildings scattered across the landscape

could also provide roosts for species that use man-made structures. In the prairies, during

the summer, bats concentrate in river valleys where trees, rocky terrain and water is

available (Holloway and Barclay 2000), resulting in aggregations of bats within a matrix

of grasslands. Do bats cross the grassland matrix, or remain in river valleys where

suitable roosting and foraging habitats exist? If bats are more likely to follow river corridors, do they generally remain within a river system, or do they find routes for moving between river systems? Unlike birds, bats make long-distance flights at night only, restricting the distance they can travel in a 24 hour period. Suitable roosting and perhaps foraging habitat therefore is likely to limit their movement routes. I predicted that as an obligate rock-roosting species, M. ciliolabrum would tend to remain in river valleys and show isolation-by-distance along the length of rivers. This species would therefore be less structured within a river system where rivers are naturally connected, and more structured between river systems where direct drainages between rivers do not exist. Because the other two species commonly roost in buildings, I hypothesized that they would be more likely to disperse across the prairie landscape between river corridors resulting in no or little population structure associated with rivers. I predicted that the poorer flight ability of M. lucifugus is less restricting if the use of buildings provides it with ample roosts for dispersing across the prairie; this species may therefore experience the same low genetic structure as the larger stronger flier, E. fuscus. I was uncertain as to whether the building-roosting species would show isolation-by-distance in the scale of the study area, but predicted that if they did, they would follow a straight line pattern, rather than reflecting distances along rivers.

Dispersal, or the moving of an individual away from their place of birth (Ricklefs

1990), serves the important evolutionary function of reducing the chance of inbreeding and has important demographic and genetic consequences for populations (reviewed in

Wolff 1999). What evolutionary forces cause dispersal and who should disperse, males or females? When mothers live in close proximity with their progeny, sons should

disperse (Wolff 1999). Females of the three bat species I studied appear to be philopatric

(Chapter 2, Nagorsen and Brigham 1993), returning to the same maternity areas each summer. It is unclear whether males return to their natal areas or disperse, although recaptures of male E. fuscus and M. ciliolabrum banded as both adults and juveniles

(Chapter 2) suggest some male fidelity. I predicted that male-biased dispersal, typical of mammals (Greenwood 1980), would be evident in all three species, and that females would show limited dispersal outside of their natal areas. In particular, I predicted that

M. ciliolabrum would show the least amount of dispersal by both males and females due to its small size.

Methods

Study Species

Myotis ciliolabrum occupies arid regions of western North America (Fig. 1.1) and

is one of the smallest bats in North America. In southeastern Alberta [SE AB], non-

pregnant adult females had a mean mass of 5.4 ± 0.1 g (mean ± SE; n = 65, unpublished

data), versus 8.6 ± 0.1 g (n = 86) for M. lucifugus, and 21 ± 0.3 g (n = 59) for E. fuscus.

Forearm sizes in SE AB are 32.4 ± 0.1 mm (n = 254), 37.7 ± 0.1 mm (n = 86) and 47.7 ±

0.1 (n = 220) for M. ciliolabrum, M. lucifugus and E. fuscus, respectively. Wing-loading

(N/m2) of the three species is: 6.7, 7.5 and 9.4 N/m2, respectively (Norberg and Rayner

1987). All species roost in rock-crevices in river valleys in the Alberta prairies (Table

1.1). Reproductive female M. ciliolabrum typically roost alone or in small groups (e.g. 1

– 6 in southern AB, Holloway 1998) in eroded crevices in river valley banks (Chapter 2;

Holloway and Barclay 2001). E. fuscus females typically roost in groups of 25 – 75

(reviewed in Kurta and Baker 1990), and like M. lucifugus, will use tree cavities, rock crevices, and buildings as roosts. In the study area, only the latter two roost types were typically used due to the scarcity of trees (see below). M. lucifugus, the common house bat in many parts of North America (Fenton and Barclay 1980), tends to roost in larger female groups with colonies reaching thousands of individuals (Davis and Hitchcock

1965). Reproductive patterns in all three species are similar with mating thought to occur in autumn (van Zyll de Jong 1985), followed by delayed fertilization, gestation in spring, parturition typically in late June/early July in S.E. Alberta, and lactation until mid-July to mid-August. All three species hibernate, and in Alberta, rock crevices in river valleys have been confirmed as hibernacula for E. fuscus, and are suspected for M. ciliolabrum

(Chapter 3 Lausen and Barclay 2006a). There is some evidence of M. ciliolabrum and E. fuscus males returning to their natal area after their first hibernation period (Chapter 2) and adult male fidelity has been documented in M. ciliolabrum and to a lesser extent in E. fuscus (Chapter 2). In Alberta, the longest confirmed male dispersal event is 115 km straight-line distance (350 km river distance; Chapter 3 Lausen and Barclay 2006a).

Study Area and Samples

My study area encompassed approximately 600 by 300 km of the prairies of SE

Alberta and north-central Montana (Fig. 4.1). Eight locations represent the main study sites from which both mtDNA and microsatellite genotypes were obtained (Table 4.2). I included four rivers in the study: the Red Deer (RDR) and South Saskatchewan (SSR), which converge in one river system flowing into northern Saskatchewan, and the Milk

and Missouri Rivers which converge into a separate river system which continues to flow east in Montana.

Population structure as influenced by flight ability needs to be studied at the appropriate scale. Because seasonal movement and dispersal distances are unknown for most bats, selecting an appropriate scale is challenging. I selected all of the main adjacent sampling sites to be approximately 80 - 110 km apart, based on preliminary evidence suggesting weak genetic differentiation at this scale in a small building-roosting

European bat, Plecotus auritus (Burland et al. 1999), which has a wing-loading (7.1

N/m2) mid-way between M. ciliolabrum and M. lucifugus (Norberg and Rayner 1987).

To determine whether smaller-scale genetic structure was present, I also sampled M.

ciliolabrum and E. fuscus at several auxillary sites (Fig. 4.1; Table 4.2) along the South

Saskatchewan River.

I caught bats each summer from 2001to 2006, and received additional E. fuscus

samples collected from Cypress Hills, Saskatchewan in 2002. With the exception of this

latter area, all sampling areas were in the prairie ecosystem, with grassy plains and river

valleys being the main source of bat foraging and roosting habitats; all three species were

documented roosting in rock crevices and foraging around riparian trees in the sampling

areas. Buildings near or in the river valleys provided a source of bats in some locations;

M. lucifugus was netted at building roosts at the Bow Island (BI), One Four (OF), Havre

(HV) and Coal Bank (CB) sites; E. fuscus was netted at building roosts at the Empress

(EM), Medicine Hat (MH), and HV sites (Fig. 4.1). Most males of all three species, and

all M. ciliolabrum, came from rock roosts in the river valleys.

At each of the eight main sites, ~30-50 M. lucifugus and M. ciliolabrum, and ~25

E. fuscus (males and females) were genetically sampled in a small area (≤10 km linear distance between mistnetting locations at each main study area). I collected fewer samples from the auxillary sampling sites and therefore not all sites were used in all analyses. All bats used in genetic analyses were adults captured between 25 May – 25

August to try to ensure individuals were sampled in their “summer birthing/foraging” areas prior to the breeding and hibernation seasons. Tissue from the wing near the tibia was sampled using 2 mm diameter biopsy punches, and stored in 90% ethanol for later extraction, amplification, mtDNA sequencing and nuclear microsatellite genotyping. All animals were cared for in accordance with the principles and guidelines of the Canadian

Council of Animal Care, and appropriate animal care permits were obtained.

Genetic Analysis

I used both mtDNA sequences and nuclear DNA microsatellites so that both male and female gene flow could be investigated. Because of its maternal inheritance pattern, mtDNA can be used to understand relationships among females, and male mtDNA can be used to investigate male dispersal from the natal area. However, nuclear DNA is essential to account for gene flow mediated by males (Hartl and Clark 1997). I used mtDNA specifically to understand female clustering (or lack of clustering) during the summer along rivers and to determine whether males were remaining near their natal areas, or dispersing. I used microsatellite data to understand the level and pattern of genetic mixing occurring during autumn mating. Because mating is thought to occur away from the summer pup-rearing and foraging areas, gene flow cannot be considered

evidence of dispersal, and differences between mtDNA and nuclear DNA structure cannot be equated to sex-biased dispersal, only sex-biased gene flow. Therefore, I consider the structure of male mtDNA separately to determine whether males are philopatric in summer, moving only “temporarily” in the fall to mate and hibernate.

I used both mtDNA sequences and microsatellite genotypes to examine structure as it pertains to landscape features. My two main analyses were: 1. analysis of molecular variance (AMOVA; Excoffier et al. 1992) which partitioned the molecular variation among and within groups to delineate a framework of where genetic mixing was occurring most and least; 2. isolation-by-distance using the Mantel test (Mantel 1967) which determined whether individuals/genes were moving in a pattern that reflected the river topography. I performed additional analyses to further compare species and males versus females, as described below.

mtDNA Sequencing

I extracted DNA from wing tissue using the Qiagen DNeasy Blood and Tissue

Extract Kit (Alameda, CA) using the spin-column protocol. This DNA was used for both mtDNA and nuclear DNA amplification. I sequenced a 250 base-pair region of the hypervariable region II (HVII) on the control region of the mitochondrial genome. DNA was amplified as a large fragment (~1000 bp) using primers L16517 and sH651 (Castella et al. 2001, Fumagalli et al. 1996). Fragments were then sequenced using the L16517 primer only due to the presence of the large repeats section (Fumagalli et al. 1996). PCR reactions were performed in a 50 µL volume using 50-100ng template, 1X PCR buffer

(50 mM KCl, 10 mM Tris-HCl pH 8.8, 0.1% Triton), 2.5 mM MgCl2, 0.16 mM dNTPs,

0.8 µM of each primer and approximately 1 – 2 U Taq DNA polymerase (isolated as in

Engelke et al. 1990) in the following cycle: initial three minutes at 94°C followed by 25 cycles of one minute at 94°C, one minute at 54°C, 1.5 minutes at 72°C. PCR product was purified using QIAquick Gel Extraction Kit (Qiagen, Alameda, CA) and sequenced using Big Dye® Terminator v3.1 sequencing kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s directions. Sequencing products were resolved on an

ABI Prism® 3100 Genetic Analyzer (Foster City, CA). I aligned sequences in Sequence

Navigator (Version 1.0 Applied Biosystems, Foster City, CA).

mtDNA Analyses

I constructed haplotype networks to show relationships and nucleotide differences between haplotypes for each species, using TCS 1.21 (Clement et al. 2000). I calculated pairwise nucleotide differences among haplotypes using Arlequin 3.1 (Excoffier et al.

2006), with both uncorrected comparisons and the Tamura-Nei model (Tamura and Nei

1993). Because Arlequin uses a fixed gamma shape value, allowing for unequal mutation rates among nucleotide sites, the optimal value of the shape parameter was estimated by fitting the Tamura-Nei model onto a haplotype maximum parsimony tree using PAUP*

(4.0b; Swofford 2002). I used haplotype frequency and Tamura-Nei distance data from the main sampling areas to analyze population genetic distance measured as Slatkin’s

Linearized Fst (Slatkin 1995) and Nei’s minimum distance (Dm; Nei 1972) calculated in

Arlequin and Geneclass (Piry et al. 2004), respectively.

AMOVA (Excoffier et al. 1992) and Mantel tests (Mantel 1967) were performed

in Arlequin, using 10,000 permutations for significance. Using the haploid dataset, I

tested three AMOVA models: between rivers, between river systems (Red Deer/South

Saskatchewan, and Milk/Missouri), and the Missouri River vs. the other three rivers. The latter model was tested due to the extensive stretch of river between the Missouri River sites and the other sites. In each model, variance was partitioned among and within sampling sites in addition to the highest level of structure being tested (river scale, river- systems scale, or Missouri vs. other). I used Tamura and Nei molecular distances with associated gamma values (0.13, 0.21 and 0.25 for E. fuscus, M. lucifugus and M. ciliolabrum, respectively). I tested male and female mtDNA haplotypes separately so that I could determine the influence of male dispersal using mtDNA. Because males do not pass their mtDNA on to the next generation, patterns revealed by male mtDNA are representative of first generation male dispersal. This principle was also applied to the haplotype-match distance analysis (see below), which can provide an estimate of male dispersal (M-F). Nuclear genes continue from generation to generation and therefore the microsatellite analysis allows an understanding of overall gene flow (i.e. dispersal of genes over time not just dispersal of individuals).

To determine whether there was isolation by distance in each species, I used

Mantel tests. I tested the correlation between genetic distances (Slatkin’s linearized Fst using a Tamura Nei model) and three geographic distance matrices: linear (shortest distance between sites), river (along rivers) and “best route” distances between sampling locations. The latter is a distance acquired by recognizing possible routes between rivers where suitable, albeit perhaps patchy, rock-roosting habitat existed. Thus these distances allowed travel by river part way between two sampling locations, but using a “short cut”

between rivers where suitable roosting habitat appeared to exist. These distances were shorter than river distances, but longer than direct linear ones.

As a relative measure to compare species, I examined matching haplotypes within sites (for males-males, females-females and females-males) and across sites. I present within-site data as the number of matches, and across-sites data as the geographic distance between matching haplotypes. If a haplotype did not have a match outside of its site, it was not part of the distance analysis. The M-F haplotype match distance could provide an estimate of relative male dispersal distance if females are highly philopatric

(see below).

Microsatellite Genotyping

I genotyped individuals at 10-11 loci. The following sets of primers were used: for

M. ciliolabrum MME24, MMG9, MMH19, MMD9, MMD15, MMF19 (Castella and

Ruedi 2000), MYBE15, MYBE22 (Kerth et al. 2002), EF5 (Vonhof et al. 2002) and

Paur05 (Burland et al. 1998); for E. fuscus MMG9, MMG25 (Castella and Ruedi 2000),

MYBE22 (Kerth et al. 2002), EF1, EF6, EF14, EF15, EF20 (Vonhof et al. 2002), TT20

(Vonhof et al. 2001) and NN18 (Petri et al. 1997); for M. lucifugus MME24, MMG9,

MMH19, MMD9, MMD19, MMH29, MMF19 (Castella and Ruedi 2000), MYBE22

(Kerth et al. 2002), EF21, EF5 and EF6 (Vonhof et al. 2002).

I used a PCR volume of 15 µL containing 1X PCR buffer (10 mM Tris buffer, pH

8.8, 0.1% Triton X-100, 50 mM KCl and 0.16 mg/mL bovine serum albumin), 0.8-

1.5mM MgCl2, 0.12 mM dNTPs, 0.2-0.27 mM of each primer, 0.4 units of Taq DNA

polymerase, and 2 µL (~ 100 ng) DNA template. Cycling was performed in a 9600

thermal cycler (Perkin-Elmer) under the following conditions: 1 min at 94°C, three cycles of 30 s at 94°C, 20 s at 47°C, and 5 s at 72°C, 33 cycles of 15 s at 94°C, 20 s at 47°C, and

1 s at 72°C, followed by final extension at 72 °C for 30 min. PCR products were resolved on model 377 ABI sequencers (Applied Biosystems, Foster City, CA), and analysed using GENESCAN (version 3.1) and GENOTYPER (version 2.0) software

(Applied Biosystems, Foster City, CA).

I checked all loci in all populations for deviations from Hardy-Weinberg equilibrium using GENEPOP 3.4 (Raymond and Rousset 1995) and adjusted the type I error rate for multiple comparisons using the Bonferroni method (Sokal and Rohlf 1995).

Due to deviations from Hardy-Weinberg equilibrium, loci were further examined for null alleles using Micro-checker (Van Oosterhout et al. 2004). The null frequency for each locus was estimated using the software FreeNA, and an adjusted allele frequency dataset was produced (Chapuis and Estoup 2007). I also used this software to calculate pairwise and global Fst values (INA/ENA correction; Chapuis and Estoup 2007). I performed the same analyses on corrected and uncorrected data to ensure similar results.

Where sex-specific analyses were required, the “Micro-checked dataset” was used rather than the “FreeNA dataset” as the latter is a newly generated dataset retaining no individual labels.

Microsatellite Analyses

To examine genetic population structure, I calculated pairwise population distances,

Slatkin’s Linearized Fst and Nei’s minimum distance (Dm), among the main sampling

sites using Arlequin and Geneclass, respectively. Probabilities for pairwise Fst were

based on 10,000 permutations in Arlequin. To examine isolation-by-distance, I performed Mantel tests in Arlequin as described above, using Slatkin’s linearized Fst. In

Arlequin, I examined genetic structure as it pertains to rivers and river systems using

AMOVA with the number of different alleles (Fst) as the distance method.

I calculated relatedness of individuals using the method of Lynch and Ritland

(1999). For each species, I calculated relatedness within sites, comparing male-male, female-male and female-female relatedness. Additionally, I calculated all pairwise relatedness values for each individual across the 8 main sample sites. To examine sex- biased dispersal, I followed a procedure similar to that used by Knight et al. (1999); I regressed relatedness on geographic distance for each individual, acquiring a regression coefficient (slope) for each individual. By having one datapoint per individual, I was then able to statistically compare species and sexes while avoiding pseudoreplication

(Knight et al. 1999, Prugnolle and de Meeus 2002). Because the assumption of normality was not met, I performed a nonparametric test using ANOVA of ranked regression values, and calculated a χ2 equivalent (Scheirer-Ray-Hare extension of the Kruskal-

Wallis test; Sokal and Rohlf 1995). I conducted nonparametric pairwise comparisons

using Mann-Whitney tests (two sample Wilcoxon rank-sum). Because it is not clear that

the appropriate degrees of freedom are used with the procedure as outlined in Knight et

al. 1999 (C. Strobeck, pers. comm.), I also tested the significance of these regression

values against a probability distribution generated using 200 datasets of randomly

assigned categories (male vs female, or M. lucifugus vs. E. fuscus vs. M. ciliolabrum)

keeping ratios constant. Regressions, ANOVA Mann-Whitney U tests, and generation of

probability distributions were performed in Stata 9.0 (2005, StataCorp LP, College

Station, TX). I calculated inbreeding coefficients (Fis) for all three species using FSTAT

2.9.3.2 (Goudet 2001); standard errors were calculated using jackknifing over loci.

Results

Samples

I captured and genetically analyzed 486 M. ciliolabrum, 401 M. lucifugus and 315

E. fuscus, of which 427, 401 and 271 were microsatellite genotyped, and 142, 91 and 140

were mtDNA sequenced, for each species respectively (Table 4.2). Sample sizes in the 8

main study sites ranged from 31 – 74 per site for the former two species and 22 – 53 for

the latter species, which tends to be found in smaller numbers in the study area and is often more difficult to catch as individuals fly higher than the other two species.

Sequencing

The number of polymorphic sites, haplotypes, and sequence divergence for the

HVII fragments of E. fuscus, M. lucifugus and M. ciliolabrum can be found in Table 4.3.

When all haplotypes were compared, mean uncorrected pairwise sequence difference (±

SE) was greatest in M. lucifugus (6.1%) and least in M. ciliolabrum (1.2%). Each species

had two to three obviously divergent haplotypes or haplogroups (Fig. 4.2). The mean

sequence divergence (uncorrected) between these haplogroups was 22.2 ± 0.16 base pairs

(9.2% pairwise difference; range 20-25), 35.6 ± 0.2 base pairs (14.3%, range 30-39

bases), and 7.56 ± 0.18 (1.2%; range 5-12) for E. fuscus, M. lucifugus and M.

ciliolabrum, respectively.

Genotyping

I found deviations from Hardy-Weinberg equilibrium in MMH29 for M. lucifugus following null allele correction in FreeNA, and removed it from all subsequent analyses.

I therefore had 10 microsatellite loci for each species (Table 4.3) in the FreeNA corrected datasets. In the sex-specific analyses where Micro-checked loci were used, M. ciliolabrum, M. lucifugus and E. fuscus datasets consisted of 6, 6 and 8 loci respectively due to possible null alleles in the other loci. I removed the loci MYBE15, MME24,

MMH19 and MMD9 from the M. ciliolabrum Micro-checked dataset, MYBE22,

MMH29, MMD9, EF5 and EF6 from the M. lucifugus dataset, and EF6 and G9 from the

E. fuscus dataset.

Genetic Analyses

Pairwise Fst values between all sampling locations (110 permutations, Bonferroni adjusted alpha values) for each species are found in Table 4.4. No pairwise Fst values were significant for E. fuscus; three Fst values (between river systems) were significant for M. lucifugus; all between-river-system pairwise comparisons were significant for M. ciliolabrum, as were all but one of the between-river pairwise comparisons.

Mean genetic distances for each species (Table 4.5) show similar patterns regardless of the measure of genetic distance or type of DNA: M. ciliolabrum has the largest genetic distances, and E. fuscus and M. lucifugus have similarly lower genetic distances between sites. Global Fst (Weir 1996) was also calculated for microsatellite data using FreeNA. INA and ENA correction (Estoup and Chapuis 2007) was compared to determine the effect of estimated null allele frequencies on distance and Fst estimates.

Global Fst for M. ciliolabrum, M. lucifugus and E. fuscus (INA/ENA correction) were

0.0104/0.0101, 0.0023/0.0021 and 0.0029/0.0027, respectively. These differences were small and affected each species similarly; because estimated null allele frequencies were less than 0.20 in all but 7 of 280 loci·locations (maximum 0.277), Fst and distances calculated using the null allele corrected data were assumed to be good estimates

(Estoup and Chapuis 2007).

The mtDNA-based distances are much higher than those for nuclear DNA (Table

4.5). One expects differences in the order of two to four fold due to the inherent differences between these two types of markers (Birky et al. 1983, 1989, Castella et al.

2001), but distance ratios of mtDNA:microsatellite ranged from 16 – 44, 10 – 18, and 14

– 22 times for M. ciliolabrum, M. lucifugus and E. fuscus, respectively. The percentage of pairwise population Fst measurements that were significantly different from zero after

Bonferroni correction for M. ciliolabrum, M. lucifugus and E. fuscus were 93% (26/28 pairwise comparisons were significant), 21% (6/28) and 0% (0/28), respectively, indicating that M. ciliolabrum is highly structured relative to the other two species. Non- significant pairwise comparisons for M. ciliolabrum were within rivers only (HV and MR within the Milk River, and CB and MC within the Missouri River; Fig. 4.1), suggesting less structure within rivers, whereas, in M. lucifugus all within-river sites were nonsignficant in addition to 18 between-river comparisons indicating minimal structure.

The results of the three models of AMOVA for each species using both microsatellite and mtDNA sequence datasets are presented in Table 4.6. Eptesicus fuscus showed some genetic structure in relation to rivers (p = 0.055) when nuclear alleles were examined, but structure due to rivers was most apparent in female mtDNA data (p =

0.011). When males were included, this mtDNA structure shifted to the level of the river systems, indicating that first generation males generally dispersed beyond the rivers but not beyond the river systems. M. lucifugus was also structured by rivers and river- systems. Females preferentially remain within rivers (p = 0.046), and females on the

Missouri River in particular were separated somewhat from the other three rivers (p =

0.048). When males were added to the mtDNA analysis, this structure became stronger

(p = 0.038, 0.035), suggesting that first generation males generally remain within their natal river. Mating occurs in such a way that nuclear genes move into adjoining rivers, as evidenced by the shift in significant structure from rivers in mtDNA to river systems in nuclear DNA (p = 0.029). Nuclear gene flow is limited between the Missouri River and the other three rivers (p = 0.035). Female M. ciliolabrum distribution is structured mainly by the Missouri River (p = 0.050). When the males are added into this mtDNA analysis, this same structure exists (p = 0.047). That significant structure is not seen at the level of the rivers for this small bodied bat at first seems surprising, but close inspection of the mtDNA data revealed extremely low mtDNA haplotype overlap among sampling sites. In fact, of the 25 haplotypes found in the main sampling locations (7 sites, 78 individuals sequenced), 22 are unique to the site in which they were sampled

(12% of the haplotypes are found at more than one site). This is in contrast to E. fuscus and M. lucifugus in which 40% (12 of 30 haplotypes; 113 sequenced individuals) and

33% (8 of 24 haplotypes; 91 sequenced individuals) of the haplotypes, respectively, were found at more than one site. When the auxillary sampling locations (Table 4.2) for M. ciliolabrum were examined on the SSR (SP, LD, PL, KR, along with FC; Fig. 4.1), I found that 50% (6 of 12) of the haplotypes overlapped within this 37 km stretch of river,

suggesting that overlap occurred on a more regional scale and that a river scale may have been too large to examine mtDNA structure in M. ciliolabrum. That structure in M. ciliolabrum mtDNA occurs at the level of the Missouri vs. other rivers is not surprising given that there is no overlap of M. ciliolabrum haplotypes between the Missouri River and the other rivers, whereas all major sampling sites on the RDR, SSR and Milk River share one haplotype, Mci13.

Using Mantel tests, I tested three distance matrices against genetic distance to determine whether there was isolation by distance in each species, and what distance matrix best correlated with the genetic distances (Table 4.7). Scatterplots of the pairwise genetic distances for each species are presented for the “best route” distances (Fig. 4.3).

High correlation is interpreted as isolation-by-distance over the particular distances of each matrix. There was no evidence of isolation by distance in E. fuscus, regardless of the distance matrix or type of DNA.

Using either mtDNA or microsatellites, all three distance matrices correlated significantly in M. lucifugus, although when females were analyzed separately, linear distance did not correlate with genetic distance, suggesting that females more likely disperse along rivers rather than across the prairie. The AMOVA results (above) suggested that male M. lucifugus have limited dispersal between rivers, and these Mantel results support this conclusion: linear distance is not significantly correlated with mtDNA genetic distance unless males are included and the best-fit Mantel distance matrix for the mtDNA data was river distance. However, the largest determination of genetic distance by geographic distance, the largest correlation coefficients, and the smallest p values, were all associated with best route distances for nuclear genotypes. In

other words, both the Mantel and AMOVA models suggest that M. lucifugus mtDNA tends to move along rivers and remain largely in rivers, but nuclear DNA moves out of the rivers, preferentially remaining in a river system. This nuclear gene flow between rivers follows a complex non-linear fashion (“best route” distance pattern).

Isolation-by-distance was not found in M. ciliolabrum, stemming from the high degree of structure in the mtDNA of this species (see above). Nuclear gene flow, however, occurs, with the most significant correlation of microsatellite genetic distance with the best-route distance matrix, just as in M. lucifugus. The isolation-by-distance slopes of M. ciliolabrum Mantel regressions were 3 – 12.5 times those of the other two species, with M. lucifugus and E. fuscus having similar slopes (ratio ranges from 0.5 – 2).

This adds further support for the idea of less gene flow in M. ciliolabrum compared to the other two species.

To better compare genetic structure among the species and gene flow among the sexes, I examined relatedness of individuals over distance. I calculated regression slopes for each individual using “best route” distances, and then compared these slopes for each species. The mean (± SD) relatedness-by-distance slopes were -4.70x10-6 ± 1.39x10-6 (n

= 270), -4.72x10-6 ± 1.04x10-6 (n = 401) and -16.3x10-6 ± 1.67x10-6 (n = 414) for E.

fuscus, M. lucifugus and M. ciliolabrum, respectively. I also calculated sex-specific

mean relatedness-by-distance regression slopes for each species: females were -8.94x10-

6 ± 1.95x10-6, -7.14 x10-6 ± 1.12 x10-6, and -18.4x10-6 ± 2.06x10-6, respectively; males

were -1.74 x10-6 ± 1.90 x 10-6, -3.71x10-6 ± 0.986 x10-6, and -13.9x10-6 ± 2.711x10-6, respectively. Using ANOVA of ranks, I found that slopes differed significantly among the species (χ2 equivalent = 338, df = 2, p < 0.001) and sexes (χ2 equivalent = 57.3, df =

1, p < 0.001). M. ciliolabrum differed significantly from E. fuscus (z = 5.48, p < 0.001) and M. lucifugus (z = 6.58, p < 0.001), but the latter two species did not differ from each other (z = 3.48, p = 0.73). Regression slopes differed significantly between males and females in M. lucifugus (z = 1.98, p = 0.047), and E. fuscus (z = 2.44, p = 0.015) but not in M. ciliolabrum (z = 1.06, p = 0.291) , indicating significant sex-biased gene flow within the study area for the former two species (Knight et al. 1999). I confirmed the significance of these results using my own probability distributions created through random assignment of species and sex in the same ratios as the original datasets. Despite lower gene flow and lack of sex-biased dispersal in M. ciliolabrum, this species does not show a higher inbreeding coefficient; Fis for M. ciliolabrum, M. lucifugus and E. fuscus are 0.021±0.015, 0.027±0.011, and 0.037±0.017, respectively.

I calculated the number of haplotype matches found within sites and the mean distances between haplotype matches among sites for males-males, females-females and males-females (Table 4.8). Because of the ubiquitous and therefore uninformative nature of Mci13 in M. ciliolabrum, it is not clear that this haplotype should be included in the haplotype distance calculations for this species. I present results with and without the inclusion of Mci13. The processes which have led to this haplotype being regularly scattered across the landscape are not known, and its ubiquitous nature is in stark contrast to the isolation of the other haplotypes which are unique over a small area. Because the inclusion of this single haplotype greatly inflates the mean haplotype match distances calculated using the other 28 haplotypes, I base conclusions from the haplotype-match- distances analysis on the calculations that exclude Mci13. The number of within-site haplotype matches in all cases was largest for M. ciliolabrum, and the distances between

haplotype matches among sites was smallest for M. ciliolabrum in all cases. Because

AMOVA and Mantel results suggest little movement of female M. ciliolabrum, the M-F haplotype match distance can be used as an approximation of male dispersal distance (70 km) in this species. Interestingly, and in contrast to the other two species, this value is slightly lower than the F-F haplo-match distance (77 km). Male-male haplotype match distance was extremely low (18 km), and haplotype matches occurred only among the stretch of SSR where adjacent auxillary sites were sequenced, and not among the main sampling sites, suggesting limited male dispersal from natal areas, but also suggesting the need for larger sampling of male haplotypes at each site.

Because haplotype distances suggested low dispersal in both male and female M. ciliolabrum, corroborating banding recapture records (Chapter 2), I investigated haplotype matches on a finer scale (SSR). I compared M. ciliolabrum and E. fuscus, using the Bindloss Ferry Crossing area (FC, Table 4.2; Fig. 4.1) as a focal site given the large sample sizes from this area. At FC, a rock-roosting colony of E. fuscus with ~37 adult females (Lausen and Barclay 2001) consisted of 4 haplotypes (out of 14 sequenced individuals). A small (<10 adult females) adjacent rock-roosting E. fuscus colony

(Pipeline Crossing Rocks, PL, Table 4.9B) 7 km downriver of FC had 3 haplotypes (out of 4 sequenced females), none of which overlapped with the 4 haplotypes of the FC colony. The PL colony had one matching haplotype with a building roost 30 km further downriver (EH, ~200 adult females, 7 haplotypes out of 12 sequenced females; Table

4.9B), and the FC colony had 2 matching haplotypes with this same building roost.

Similarly, a building roost 75 km upriver (ESS) contained ~150 adult females, and of 2 haplotypes found there (out of 10 sequenced females), 1 matched a haplotype at the FC

site. In other words, while the two adjacent rock crevice colonies did not have overlapping haplotypes, they did have haplotype matches with building roosts up- and down-river. When the same up- and downriver stretch of river was examined for M. ciliolabrum female haplotype matches, I found 6 haplotypes (out of 35 sequenced females) at FC and 5 of these 6 haplotypes were also found at the adjacent PL area (5 haplotypes from 11 sequenced females). In an area 13 km downriver of FC (Sandy Point

Campground; Table 4.2, Fig. 4.1) there were 6 haplotypes (out of 16 sequenced females) of which only 2 overlapped with the FC area. As mentioned above, with the exception of

Mci13, a ubiquitous haplotype in the non-Missouri sites, there was extremely limited haplotype overlap in M. ciliolabrum among the main study sites. In other words, high haplotype overlap occurred in the immediate area of the FC site, but matches diminished rapidly with distance up and down river. These data suggest that female dispersal distances in M.ciliolabrum are shorter than in E. fuscus and support the relatedness-by- distance slopes. Additionally, these data suggest that female dispersal in M. ciliolabrum is to adjacent areas, while E. fuscus females may not disperse to adjacent rocky areas, but rather to larger building roosts further away. This makes the assumption that bats existed in the area prior to the buildings.

Mean relatedness among males and females, among males, and among females was low for all species, ranging from -0.0010 – 0.0155, although higher r values were found within maternity colonies (Table 4.9). M. ciliolabrum had higher r values than M. lucifugus and E. fuscus. Roost-mates of M. ciliolabrum were of identical haplotype, unlike E. fuscus colonies which had a mean and maximum sequence divergence among roost-mates of 1.9% and 9.6% respectively, and M. lucifugus with 6.0% and 14.4% mean

and maximum divergence (Table 4.9B). Female M. lucifugus roost-mates were the least related of any of the maternity colonies investigated. In fact, relatedness among female

M. lucifugus at all sites (mean r = 0.0028; Table 4.9A) was even lower than the mean relatedness among male M. ciliolabrum (r = 0.0069). This corroborates the lower haplotype matches and match distances for M. lucifugus (Table 4.8).

Discussion

As predicted, I found limited dispersal in M. ciliolabrum. I also found no

dispersal between the Missouri River and the other three rivers. Females had high

relatedness within roosts and all roost-mates had identical haplotypes. Due to this high

level of structure and low dispersal, there was no isolation-by-distance. The complete

fracture of male and female haplotypes between the Missouri and other rivers, suggests

that neither females nor males move between the Missouri River and the more northern

rivers during the summer. Because this fracture also exists in the nuclear gene flow, I suggest that a barrier exists between the Missouri and other rivers, preventing dispersal and movement of males and females between these areas seasonally when mating occurs.

The fact that most male-male haplotype matches occurred within sites or among adjacent sites, together with a low male-female haplotype match distance (70 - 134 km; Table

4.8), suggests low male dispersal and high fidelity to natal areas. This corroborates banding recapture data (Chapter 2), and the relatedness-by-distance slopes analysis that found no sex-biased dispersal in M. ciliolabrum. This is somewhat unusual given that male-biased dispersal is common among mammals (Greenwood 1980), and in bats (e.g.

Myotis myotis, Petri et al. 1997; Nyctalus noctula, Petit et al. 2001; Myotis bechsteinii,

Kerth et al. 2002). However, male philopatry has been documented in some species of bats (e.g. Miniopterus schreibersii natalensis, Miller-Butterworth et al. 2003). While males return to their natal area, overall pair-wise genetic distance is relatively low

(Slatkin’s Fst 0.012) suggesting that male-mediated gene flow is occurring through mating with individuals outside the maternity area. Although not commonly the case, gene flow without dispersal has been documented in other animals (green sea turtles,

Chelonia mydas, Fitzsimmons et al. 1997) and in other bat species (Plecotus auritus,

Burland et al. 1999). In M. ciliolabrum, it appears that gene flow is likely being mediated by mating during fall migration or at hibernation areas. An isolation-by-distance pattern in the microsatellite data suggests that genes flow among rivers in a complex way, not following river corridors completely, but likely short-cutting between rivers along suitable habitat corridors. Mixing of M. ciliolabrum during mating preferentially occurs within rivers, but between rivers to some degree, creating weak structure at the level of the river systems. That there is some gene flow between the Missouri and other rivers is best explained by mating in the fall when individuals of these river groups must temporarily mix. This would explain the flow of nuclear genes within and between rivers despite the highly structured summer groupings of both males and females.

My data suggest that when female M. ciliolabrum disperse, they do so to adjacent rocky areas, while E. fuscus females may not disperse to adjacent rocky areas, but rather to larger building roosts further away. Roost selection and availability may explain this pattern of dispersal. Eptesicus fuscus females roost in groups and select rock roosts which are deep and have large openings (Lausen and Barclay 2001), whereas, M. ciliolabrum females roost alone and select mudstone crevices that are much smaller, are more

shallow, and appear to be abundant in rocky riparian areas (pers. obs.). The resident rock-roosting E. fuscus colony in the FC area switches roosts throughout the season over a 1.25 km area, and reuses roosts from year-to-year. Year-to-year roost fidelity of M. ciliolabrum at this site has not been investigated, but roost switching behaviour within season suggests that individuals and their close relatives use adjacent crevices and show a microgeographic structure, remaining in small (<106 m) roosting areas along the river

(Chapter 2). At FC, male E. fuscus roost alone and do not switch roosts often (Chapter

2). Male M. ciliolabrum also roost alone and tend to be found in mudstone (Chapter 2).

With this roost selection as background, one can hypothesize how different dispersal patterns may have evolved in these two species in the prairies. Deep, large suitable rock crevices for E. fuscus colonies are not likely to be abundant, and limited crevice size and the tendency to roost with non-kin (Chapter 2) is likely to limit the degree of female philopatry in an area (Kunz 1982). Adjacent rock colonies, experiencing the same limitations, may require that some females disperse to areas where roosts are more abundant, or space within suitable roosts is greater, such as in buildings.

Males, roosting alone and unlikely to experience limited roost resources may disperse only to the extent to avoid inbreeding and increase mating opportunities (reviewed in

Wolff 1999). Because mating occurs during the autumn, away from natal areas, however, the need for dispersal from the natal area may be relatively small for males, and familiarity with roosts and foraging areas may favour some males staying in their natal area. Limited availability of suitable maternity roosts along the rivers is likely to increase female dispersal and therefore decrease overall genetic structure, given that both males and females move. Year-to-year and within-season female philopatry in building-

roosting E. fuscus is common (pers. obs., reviewed in Kurta and Baker 1990). Building roosts tend to be much larger than natural crevice roosts, and may accommodate a greater proportion of returning females, decreasing the tendency for female dispersal and thereby increasing the dispersal bias between the sexes in areas where building roosts are common. Further comparison of male and female dispersal in areas of natural versus human-made roosts would be necessary to directly test this hypothesis.

All three species showed a large discrepancy between mtDNA and nuclear DNA genetic distance, indicative of sex-biased dispersal (Prugnolle and de Meeus 2002).

However, all three species are likely to undergo gene flow mainly in the autumn period meaning that this comparison of nuclear markers is indicative of sex-bias in gene flow, but not necessarily a sex-bias in dispersal behaviour. When one considers the relatedness-by-distance slopes, females of M. lucifugus and E. fuscus have significantly steeper slopes than males, suggesting that males are moving greater distances than females to mate in the fall; low male mtDNA structure in the summer suggests this sex- bias also reflects dispersal behaviour. This is in contrast to M. ciliolabrum, whose male and female relatedness-by-distance slopes do not differ; the structure evident in male mtDNA during the summer confirms male philopatry and low dispersal relative to the other two species. Lower gene flow within M. ciliolabrum relative to the other three species also supports this finding. This species roosts solitarily in abundant small mudstone roosts. Habitat saturation is a balance between resource availability and survivorship (Waser and Jones 1983), and in this species a combination of abundant roosts and perhaps low survivorship or high predation, keep crevice occupancy low enough for maximum natal philopatry (of both males and females). Survivorship in M.

ciliolabrum, however, is unknown. If rock-roost resources are not limiting philopatry, and inbreeding is avoided through seasonal migration for mating, then advantages of dispersal are few and benefits of philopatry such as familiarity with habitat, may be greater. Unlike E. fuscus and M. lucifugus, M. ciliolabrum roost primarily in rock crevices (Holloway and Barclay 2001), and as such, male and female dispersal patterns in this study are likely to reflect those of the species, or at least those in the northern part of their range where arid and semi-arid habitat is used (Holloway and Barclay 2001).

Unlike M. ciliolabrum, M. lucifugus females showed isolation-by-distance, with female dispersal occurring along river corridors. There was significant structure between rivers, suggesting that dispersal is generally within the natal river. Relatedness within colonies was low, and the presence of very divergent haplotypes (Chapter 5) within the same colony suggests that these large female clusters are not kin-based. This corroborates findings from other studies that suggest congregations of large numbers of

M. lucifugus do not reflect underlying social structure (Fenton and Barclay 1980).

Colony composition may therefore be based on advantages of group associations rather than kin selection (Burland et al. 2001). The AMOVA results suggest that male M. lucifugus tend to remain in their natal rivers during the summer, although there was weak evidence to suggest some movement out of the rivers but within river systems. As with

M. ciliolabrum, nuclear gene flow occurs to a larger extent than movement of mtDNA.

Gene flow is more extensive in M. lucifugus than M. ciliolabrum, with M. lucifugus structuring strongly at the level of the river system, rather than the rivers. Nuclear gene flow in M. lucifugus, as with M. ciliolabrum, generally follows a “best route” distance pattern and is limited between the Missouri and other rivers. In both of these species,

mating occurring part-way between sampling sites within river systems may explain the significance of the best route correlation with genetic distance. That gene flow between the Missouri River and the other rivers is limited may be explained by long river distances and long stretches of unsuitable roosting habitat (see below).

I found that female E. fuscus were structured according to rivers, with females preferentially remaining in their natal rivers, although there was weak evidence to suggest some female dispersal within river systems. Males tended to disperse from their natal river, but remain within their associated river system. Nuclear gene flow in this species was high, with only weak evidence to suggest some preferential mating among individuals from the same river. The lack of isolation-by-distance in E. fuscus suggests that the scale of the study area may have been too small to properly characterize genetic structure in this stronger flier, and that nuclear genes were moving across the entire area.

I predicted that M. ciliolabrum, due to its small size and roost specificity, would be more easily structured by the natural landscape than the more generalist roosting species (M. lucifugus and E. fuscus). This is what I found. I also predicted that as a larger, stronger flier, E. fuscus would have little structure associated with rivers. Overall genetic structure was indeed low, although both males and females did show some structuring by rivers. That E. fuscus showed no isolation-by-distance pattern, suggests that gene flow at the scale of the study area was not limited by distance and likely reflects this species’ ability for longer distance flights to mating and hibernation sites.

Houses on the prairie landscape likely opened up new potential for long distance dispersal by some species of bats, allowing even relatively small bats (e. g. M. lucifugus) to disperse further than they could have historically. Wing-loading, body and forearm

sizes of M. lucifugus are more similar to those of M. ciliolabrum than to those of E. fuscus, yet M. lucifugus shows substantially less genetic structure than M. ciliolabrum, and more similarity to E. fuscus in many measures of genetic structure., such as haplotype-match distances, lower pairwise genetic distances, and more shallow relatedness-by-distance slopes than M. ciliolabrum. I therefore propose that the ability to roost in buildings may be decreasing genetic structure in this species, producing a greater difference in structure between M. lucifugus and M. ciliolabrum than would otherwise be present based on flight abilities alone. However, this requires further investigation to tease apart the genetic structuring effects of roost specificity from wing- loading differences.

Wing-loading values in this study were negatively correlated with the amount of population genetic structure, supporting the pattern apparent in the literature (Table 4.1).

For example, Pteropus scapulatus (32.8 N/m2; Norberg and Rayner 1987), a high wing-

loaded species, is panmictic across Australia (Sinclair et al. 1996), while Plecotus

auritus, with a much smaller wing-loading value in (7.1; Norberg and Rayner 1987) is associated with structure on a small scale (~50 km; Burland et al. 1999). My results show genetic structure intermediate to these studies for M. lucifugus and E. fuscus, congruent with their intermediate wing-loading values (7.5 and 9.4 N/m2, respectively),

and for M. ciliolabrum, structure on a smaller scale than that of Plecotus auritus, as

would be predicted by its lower wing-loading value (6.7 N/m2).

Navigation using the Earth’s magnetic field has recently been established for bats

(E. fuscus; Holland et al. 2006), meaning that bats may follow river corridors less for

navigation, and more to locate roosting and perhaps foraging habitat. This would be most

important for the movement and dispersal of smaller bats limited by short nightly flight distances, and may explain the asymmetrical river structure associated with the Missouri

River. Because of a diminished degree of gene flow between the Missouri and other rivers, M. ciliolabrum and M. lucifugus on the Missouri River must have preferentially mated within the river, or with individuals not sampled in this study (e. g. Marias River, an upriver tributary on the Missouri River, or Musselshell River, a downriver tributary), rather than with others in the study area. Why this structure was not present in E. fuscus may in part be explained by an examination of river topography. Unlike on the RDR,

SSR, and Milk Rivers, where stretches of unsuitable rock-roosting habitat are relatively short (<100 km), there is nearly 1000 km of river distance separating the Missouri and

Milk River sampling sites, with much of this composed of grassy river valley banks. On the Milk River, the river valley between Havre and its confluence with the Missouri

River is approximately 500 km of grassy slopes with only scattered towns, and little to no rocky riparian habitat. Additionally, rocky riparian areas have been flooded along the

Missouri (Ft. Peck Reservoir) from its confluence with the Milk to approximately 200 km upriver (Fig. 4.1). This flooded area is the Charlie M. Russell National Wildlife Refuge, and is largely devoid of buildings. However, M. lucifugus and M. ciliolabrum have been captured along this flooded stretch of river valley (J. Stewart, pers. comm.), suggesting that some roosting and foraging habitats are available in this part of the Missouri River.

Given the paucity of rock crevices along the Milk River from Havre downriver, flight along the river corridor between the Missouri and Milk sampling locations may be challenging for a bat roosting exclusively in rocks, and even more so for a small bat. This may in part explain why M. ciliolabrum on the Missouri are isolated from individuals in

the rest of the study area. For a bat capable of roosting in buildings, the flight between the Missouri and Milk River sites would be somewhat easier; buildings are quite prevalent on the Milk River downstream of Havre, although, on the Missouri River, abandoned homesteads within the Charlie Russell Wildlife Refuge are unsuitable as roosts (J. Stewart, pers. comm.). For a bat such as E. fuscus that can fly fast, and is capable of making a long distance flight in one night, this becomes even easier. In fact, for a bat that can fly long distances, roost in buildings, and navigate using the Earth’s magnetic field (Holland et al. 2006), there may be no reason to follow the river corridor at all; the Missouri and Milk River sampling sites are a flight over prairie landscape of only ~60 km. In Alberta, E. fuscus has been reported to forage >13 km away in a single direction from its roost in a night (Wilkinson and Barclay 1997); in Colorado a radiotagged E. fuscus moved 56 km between maternity roost and mountain rock crevice roost in a single night just prior to hibernation (Neubaum et al. 2006), suggesting that a

60 km dispersal event may be possible in one night.

Male M. lucifugus and M. ciliolabrum had significantly less-steep relatedness-by- distance regression slopes than females did, suggesting male-biased dispersal (Knight et al. 1999). However, that M. lucifugus shows less difference between mtDNA-based and microsatellite-based genetic distances (mtDNA:msat, 10 – 18 times) than M. ciliolabrum

(16 – 44 times; Table 4.5), confirms that female M. lucifugus are also moving relatively far for fall mating, enhancing gene flow and making the male bias in gene flow less in this species.

Establishing an appropriate scale for comparison among species differing in vagility is difficult. For example, while the haplotype-match distance results provided a

useful way of illustrating the difference in mtDNA patterns between M. ciliolabrum and the other two species, the mean distances for E. fuscus and M. lucifugus were not different from the mean sample site distances, suggesting the scale may have been too small to properly characterize these species’ relatedness-by-distance and dispersal patterns. Overall, the main study sites represented an adequate scale for investigation of

M. lucifugus, but as evidenced by the paucity of haplotype overlap among M. ciliolabrum and the need to include closer auxillary sites, the scale was too large for M. cilolabrum.

The lack of isolation by distance, together with the absence of any significant pairwise

Fst values in E. fuscus, suggested that the scale was too small to properly characterize genetic structure in that species. While I tried to minimize the area from which I drew samples at each site, I selected a 10 km stretch of river valley as a sampling area for logistical reasons. The low within-site female relatedness values compared to within- colony values (Table 4.9A), especially for M. ciliolabrum, suggests that my within-scale sampling distance was too large for this species. I compensated by sampling adjacent areas in the SSR to properly characterize the small-scale structure of this species. Female

M. ciliolabrum have small home ranges in the study area (Chapter 2); in consecutive days, females tend to roost in rock crevices that are within a few hundred metres of each other, and forage within a few kilometers of their roosts (Chapter 2). In my study area, reproductive female E. fuscus generally forage in a 2.7 km2 area around their roost

(Wilkinson and Barclay1997), and in one study, colony members roosted in crevices

along a 1.25 km length of river (Lausen and Barclay 2002). The 10 km scale was

therefore most appropriate for this species, but may have been too large for M. ciliolabrum, grouping many unrelated females together. That landscape features on a

smaller scale than rivers may be influencing M. ciliolabrum may explain their disjunct distribution across the prairies (ASRD 2006, Holloway and Barclay 2001). For example, their tendency to roost in south-facing crevices on shallow slopes and adjacent to relatives may limit the number of riparian areas suitable for this species (Chapter 2).

Two maternity colonies of M. lucifugus included a mixture of very divergent mtDNA lineages (14.3% HVII sequence difference between MYLU.CA and MYLU.LU haplogroups; Chapter 5), and mean nuclear relatedness within roosts was low. This corroborates findings from other studies that suggest that congregations of large numbers of M. lucifugus do not reflect underlying social structure (Fenton and Barclay 1980).

This species “quickly locat[es] and exploits new roosts” (Fenton and Barclay 1980), and therefore mixed-lineage colony composition may be a result of limited roost availability, or advantages accrued from group associations rather than kin selection (Burland et al.

2001). That r values of M. lucifugus are lower than those of the other two species may suggest that kinship and roost fidelity is less important in M. lucifugus, but may alternatively stem from the nature of larger colonies ultimately having more unrelated pairwise comparisons among small groups of related roost-mates. M. ciliolabrum and E. fuscus roosts had smaller or no HVII nucleotide differences among roost-mates, and more haplotype matches among females within sites. However, overall mean relatedness among females at each site was low for all three species in general. Colony relatedness values near zero are often reported for bat colonies (e.g. Plecotus auritus Burland et al.

2001, M. bechsteinii Kerth et al. 2002, Rhinolophus ferrumequinum Rossiter et al. 2002), and mean colony relatedness has been suggested a poor predictor of colony structure

given that closely related females often exist within the colony proper (this study, Table

4.7B; Kerth et al. 2002).

In summary, bats, like terrestrial mammals and birds, vary in genetic structure according to their relative mobility. As predicted, I found that E. fuscus, with its relatively large size and high wing-loading and consequently greater capacity for long distance flight, was less structured by rivers than the other two species, although females showed fidelity to their natal rivers. M. ciliolabrum was the most structured, with the females’ distribution completely fractured by a long stretch of river valley (> 500 km) on the Milk and Missouri river system devoid of suitable rock-crevice roosting habitat, despite the fact that this unsuitable stretch of river could be by-passed with a 60 km over- prairie flight. Like females, males remained close to their natal area in the summer, with short dispersal distances (~70 km) within rivers. Gene flow during autumn mating was extensive within rivers and occurred to some degree between rivers within river systems, although gene flow between the Missouri and Milk River was limited. M. lucifugus followed a complex isolation-by-distance pattern of gene flow with river distance. Its genetic structure corresponded to rivers (mitochondrial DNA) and river systems (nuclear

DNA), and its gene flow may in part be limited by the long grassy stretch of the

Milk/Missouri river valley. I conclude that building-roosting may have minimized the genetic structure in this species, but the influence of landscape on genetic structure was evident. Overall, smaller body size and wing-loading of a species corresponds to a greater degree of genetic structure, and roost specificity appears to contribute to a greater dependency on rivers as movement corridors.

Table 4.1. Summary of studies showing presence or absence of population structure, and the wing-loading values (N/m2) of the focal species.

Only species for which there are known wing-loading values (Norberg and Rayner 1987) were included. Studies of species known to be experiencing low population numbers or range contraction due habitat destruction were not included. Genetic structure across substantial geographic barriers (eg. mountains, ocean) was not included.

Pairwise Sampling Range Species N/m2 Population Structure Notes Roost/ Behaviour (km)* Study Pteropus 32.8 None. Low pairwise Fst similar to birds trees/ migratory 38-2686 Sinclair et al. scapulatus Panmictic across species (Webb and Tidemann 1996); no 1996 range in Australia isolation by distance (Burland and Worthington Wilmer 2001). Allozymes. Nyctalus 16.1 Very weak. Weak structure across Europe; trees and other/ 18-4027 Petit and noctula Nearly panmictic population high male dispersal but Alp migratory Mayer 1999 pattern (Burland and Mountains may act as barrier; Worthington Wilmer 2001) no isolation by distance (Burland and Worthington Wilmer 2001). MtDNA and microsatellites

125

Pairwise Sampling Range Species N/m2 Population Structure Notes Roost/ Behaviour (km)* Study Desmodus 14 Weak. Some population structure caves, hollow 29-168 Wilkinson rotundus Extensive male dispersal. evident at larger scales (248- trees/non-migratory 1985 2075 km; Honeycutt et al. 1981; Burland and Worthington Wilmer 2001 Allozymes. Cynopteru s 13.1 Weak. No isolation by distance across trees/non-migratory 98-509 Peterson and brachyotis High gene flow between Phillipines. Heaney 1993 islands.

Tadarida 11.5 Weak. No isolation by distance caves and 99-1861 McCracken brasiliensis No structure across southern (Burland and Worthington buildings/migratory et al. 1994 U.S.A. in summer; weak Wilmer 2001); non-migratory structure in winter. populations also unstructured (McCracken and Gassel 1997). Allozymes Myotis 11.2 Weak. Populations on either side of buildings/ 130-770 Castella et myotis No significant structure within Gibraltar Strait are genetically migratory al. 2000 Spain or N. Africa. distinct from each other (Spain vs. N. Africa). MtDNA and microsatellites

126

Pairwise Sampling Range Species N/m2 Population Structure Notes Roost/ Behaviour (km)* Study Hipposideros 8.9 Some structure Relatively high gene flow across caves and 3-200 Rajan and speoris range in southern India; Fst buildings/ non- Marimuthu suggests some structuring. migratory 2000 RAPDs Plecotus 7.1 Structured Isolation within 50 km stretch of buildings/non- 0.1-100 Burland et auritus Microgeographical genetic river. migratory al. 1999 isolation by distance. Microsatellites Saccopteryx 5.9 Structured. Harem society results in high trees/non-migratory 8-43 McMcracken bilineata Genetic differentiation among male structuring. Structure 1984 colonies on microgeographic present in females also. scale. Allozymes *Distances obtained from Burland and Worthington Wilmer 2001 if not in original paper.

127

Table 4.2. UTM locations for the sampling sites in Fig. 4.1 in descending order of latitude. SSR is the South Saskatchewan River; SK is

Saskatchewan. Type of data collected: microsatellite genotypes (msat), mitochondrial DNA sequences (mtDNA). Sites 1 – 8 are the main sampling locations where all 3 species were captured and in most cases both microsatellite and mtDNA were analyzed; A1 – A6 are auxillary sampling sites where not all species were captured and/or both microsatellite and mtDNA analyses were not completed. Sample numbers are listed as Females, Males.

UTM (Zone 12) Species Sampled Location Map River Site# Easting Northing E. fuscus M. lucifugus M. ciliolabrum Red Drumheller, AB DH 1 387979 5697124 6, 19 msat; 4, 5 mtDNA 39, 15 msat; 7, 4 mtDNA 32, 18 msat Deer Empress, AB EM SSR 3 571627 5640100 12, 0 mtDNA 19, 12 msat; 4, 4 mtDNA Dinosaur Red Provincial Park, DP 2 464336 5624406 17, 32 msat; 9, 13 mtDNA 1, 48 msat; 1, 11 mtDNA 46, 12 msat; 6, 0 mtDNA Deer AB Sandy Point SP SSR A1 565398 5620269 16, 8 mtDNA Campground Pipeline Crossing PL SSR A2 561329 5614789 4, 0 mtDNA 11, 6 mtDNA Bindloss Ferry FC SSR 3 557276 5610380 24, 10 msat; 17, 10 mtDNA 50, 25 msat; 35, 9 mtDNA Crossing , AB Near Suffield LD SSR A3 550976 5606493 9, 4 mtDNA Army Base Near Suffield KR SSR A4 543100 5601848 5, 5 mtDNA Army Base Medicine Hat MH SSR A5 524333 5542498 10, 0 mtDNA Bow Island, AB BI SSR 4 478815 5542421 1, 28 msat 47, 6 msat; 7, 4 mtDNA 11, 26 msat; 5, 4 mtDNA

128

UTM (Zone 12) Species Sampled Location Map River Site#Easting Northing E. fuscus M. lucifugus M. ciliolabrum Cypress Hills, SK CH n/a A6 579165 5497015 4, 0 mtDNA Pinhorn Grazing MR Milk 5 507336 5441473 8, 18 msat; 8, 7 mtDNA 20, 33 msat; 4, 0 mtDNA Reserve, AB Onefour, AB OF Milk 5 538581 5440986 43, 7 msat; 6, 5 mtDNA Havre, MT HV Milk 6 588599 5373321 36, 14 msat; 8, 5 mtDNA 36, 19 msat; 6, 5 mtDNA 2, 50 msat; 1, 4 mtDNA Missour Coal Bank, MT CB 7 558746 5320250 16, 6 msat; 9, 5 mtDNA 26, 33 msat; 5, 8 mtDNA 44, 6 msat; 5, 4 mtDNA i McClelland Ferry, Missour MC 8 620042 5284197 3, 33 msat; 3, 6 mtDNA 23, 27 msat; 8, 6 mtDNA 30, 22 msat; 5, 0 mtDNA MT i 111, 160 msat; 235, 192 msat; TOTAL 234, 167 msat; 44, 47 mtDNA 88, 51 mtDNA 102, 40 mtDNA

129

Table 4.3. Length of HVII mtDNA sequence fragments (nucleotide base pairs), number of polymorphic sites, number of haplotypes (total and per female and male), and mean number of pairwise nucleotide (nt) differences between haplotypes (± SE) and range of nt differences for

E. fuscus, M. lucifugus and M. ciliolabrum. Mean number of microsatellite alleles per locus (± SD), range of alleles, and mean heterozygosity

(± SD) per locus are also provided for the 10 loci typed for each species.

Source DNA Description E. fuscus M. lucifugus M. ciliolabrum mtDNA HVII fragment length 240 250 280 Polymorphic sites 38 51 28 Total no. haplotypes 33 24 29 No. of female samples 93 44 107 No. female haplotypes (per female) 24 (0.26) 12 (0.27) 24 (0.22) No. of male samples 67 47 46 No. male haplotypes (per male) 25 (0.37) 22 (0.47) 14 (0.30) Mean pairwise difference 5.8 ± 0.26 (2.4%) 15.4 ± 0.84 (6.1%) 3.5 ± 0.10 (1.2%) Range of no. of nt differences 1 - 25 1 - 39 1 - 12 nuclear Mean no. of alleles per locus 26.4 ± 12.5 35.0 ± 11.4 29.6 ± 11.1 microsatellite Range of no. of alleles 8 - 42 22 - 50 15 - 52 Mean heterozygosity per locus 0.82 ± 0.11 0.92 ± 0.02 0.89 ± 0.03 Total samples 271 404 427

130

Table 4.4. Matrix of pairwise Fst for E. fuscus, M. lucifugus and M. ciliolabrum at the 8 main sampling locations using Microchecked dataset. Fst values significant after Bonferroni correction are indicated with an asterisk (*). Refer to Fig. 4.1 and Table 4.2 for sampling names/locations.

Sampling Species locations DH DP FC/EM BI OF/MR HV MC E. fuscus 0.002 M. lucifugus DP 0.003 M. ciliolabrum 0.005 E. fuscus 0.007 0.007 M. lucifugus FC/EM 0.000 0.004 M. ciliolabrum 0.014* 0.008* E. fuscus 0.004 0.000 0.002 M. lucifugus BI 0.002 0.001 0.006 M. ciliolabrum 0.020* 0.004 0.013* E. fuscus -0.006 0.001 0.013 0.004 M. lucifugus OF/MR 0.003 0.003 0.009* 0.004 M. ciliolabrum 0.014* 0.013* 0.009* 0.015* E. fuscus 0.003 0.004 0.008 0.001 0.000 M. lucifugus HV -0.001 0.002 0.004 0.004 0.002 M. ciliolabrum 0.020* 0.015* 0.010* 0.008* 0.002 E. fuscus 0.002 0.004 0.002 0.001 0.004 0.001 M. lucifugus MC 0.002 0.002 0.006 0.005 0.004 0.000 M. ciliolabrum 0.025* 0.012* 0.014* 0.007* 0.012* 0.009* E. fuscus 0.006 0.004 0.007 0.008 0.007 0.008 0.001 M. lucifugus CB 0.005* 0.001 0.008* 0.003 0.003 0.002 0.002 M. ciliolabrum 0.041* 0.027* 0.008* 0.018* 0.021* 0.021* 0.004 131

Table 4.5. Mean measures of genetic differentiation (Nei’s minimum distance [Dm] and Slatkin’s linearized Fst) between sampling locations for E. fuscus, M. lucifugus and M. ciliolabrum using both mtDNA sequences and nuclear microsatellite (msat) genotypes. Locations within each comparison were the same between species, and the number of pairwise comparisons are given (n).

Distance DNA Mean Genetic Method Source Species Distance SE Range n

Dm mtDNA E. fuscus 0.225 0.024 0.061 - 0.432 21 M. lucifugus 0.223 0.024 0.035 - 0.400 21 M. ciliolabrum 0.346 0.041 0.108 - 0.720 21 Slatkin's mtDNA E. fuscus 0.199 0.051 0 - 0.70 21 M. lucifugus 0.320 0.090 0 - 1.19 21 M. ciliolabrum 0.507 0.203 0 - 3.94 21

Dm msats E. fuscus 0.0162 0.00078 0.011 - 0.025 28 M. lucifugus 0.0123 0.00060 0.007 - 0.018 28 M. ciliolabrum 0.0215 0.00140 0.011 - 0.045 28 Slatkin's msats E. fuscus 0.00392 0.00056 0 - 0.110 28 M. lucifugus 0.00330 0.00027 0 - 0.007 28 M. ciliolabrum 0.01160 0.00102 0.003 - 0.027 28

132

Table 4.6. Analysis of molecular variance (AMOVA) for E. fuscus, M. lucifugus and M. ciliolabrum using both microsatellite (msat) and mtDNA sequence data. Distance method was number of different alleles (Alleles) for msat analyses and Tamura Nei (TamNei, with species- specific gamma value) for sequence data. Variance (V) is partitioned into a, b, c: among groups, among populations within groups, and within groups, respectively. Raw variance (percentage and degrees of freedom in parentheses) and associated probabilities are listed for each model. Groups were defined by the model being applied: the level of the rivers, river systems, or Missouri vs. other rivers. There were four rivers (Red Deer [RDR], South Saskatchewan [SSR], Milk and Missouri [Miss]), and two river systems (RDR/SSR and Milk/Miss). In general, two populations (sampling sites) were used from each river.

Variance Components Species Source Model Va P* Vb1 Va1 E. fuscus msat Rivers 0.008 (0.2, 3) 0.055 0.007 (0.2, 4) 4.1 (99.6, 528) River systems 0.002 (0.05, 1) 0.226 0.01 (0.3, 6) 4.1 (99.6, 528) Miss. vs. other 0.007 (0.2, 1) 0.072 0.01 (0.3, 6) 4.1 (99.6, 522) mtDNA Rivers 0.03 (1.3, 3) 0.140 0.2 (8.7, 5) 2.0 (90.0, 143) River systems 0.06 (2.6, 1) 0.023* 0.2 (8.3, 7) 2.0 (89.0, 143) Miss. vs. other 0.02 (0.9, 1) 0.310 0.21 (7.5, 7) 2.0 (89.6, 143) Female only mtDNA Rivers 0.23 (12.3, 3) 0.011* 0.23 (12.1, 4) 1.4 (75.7, 61) River systems 0.24 (11.9, 1) 0.057 0.29 (14.9, 6) 1.4 (73.2, 61) Miss. vs. other 0.16 (8.2, 1) 0.143 0.37 (18.8, 6) 1.4 (72.9, 61)

133

M. lucifugus msat Rivers 0.003 (0.07, 3) 0.190 0.01 (0.3, 4) 4.6 (99.7, 798)

River systems 0.006 (0.1, 1) 0.029* 0.01 (0.3, 6) 4.6 (99.6, 798)

Miss. vs. other 0.006 (0.1, 1) 0.035* 0.1 (0.3, 6) 4.6 (99.6, 798) mtDNA Rivers 3.8 (26.6, 3) 0.038* 0.7 (5.2, 4) 9.8 (68.2, 83) River systems 2.4 (16.5, 1) 0.084 2.6 (17.6, 6) 9.8 (65.9, 83) Miss. vs. other 7.3 (41.3, 1) 0.035* 0.5 (3.1, 6) 9.8 (55.7, 83)

Female only mtDNA Rivers 8.8 (49.2, 3) 0.046* -0.6 (-3.5, 3)1 9.7 (54.4, 39) River systems 3.2 (17.6, 1) 0.200 5.2 (28.5, 5) 9.7 (53.9, 39)

Miss. vs. other 15 (60.9, 1) 0.048* -0.3 (-1.3, 5)1 9.7 (40.4, 39)

M. ciliolabrum msat Rivers 0.03 (0.6, 3) 0.010* 0.03 (0.6, 4) 4.5 (98.8, 820) River systems 0.02 (0.4, 1) 0.055 0.04 (0.9, 6) 4.4 (98.7, 820) Miss. vs. other 0.03 (0.6, 1) 0.037* 0.04 (0.9, 6) 4.4 (98.5, 820) mtDNA Rivers -0.01 (-0.7, 3) 0.450 0.3 (19.6, 3) 1.1 (81.1, 83)

River systems 0.04 (2.9, 1) 0.087 0.2 (17.2, 5) 1.1 (79.9, 83) Miss. vs. other 0.10 (8.4, 1) 0.047* 0.2 (14.8, 5) 1.1 (76.8, 83)

Female only mtDNA Rivers 0.03 (2.5, 3) 0.410 0.2 (13.8, 5) 1.2 (83.7, 54) 1 River systems 0.07 (4.8, 1) 0.115 0.2 (12.8, 5) 1.2 (82.4, 54) Miss. vs. other 0.15 (9.9, 1) 0.050* 0.2 (10.8, 5) 1.2 (79.3, 54)

1all variance b and c components were significant in each model with the exception of Vb for female M. lucifugus in tests of rivers and river systems, and M. ciliolabrum in river systems.

* significant models, with noteworthy although not statistically significant models in italics.

134

Table 4.7. Results of Mantel tests using three distance matrices (Linear, River and "Best Route" distances) with genetic distance (Slatkin's linearized Fst, using Tamura Nei model of mutation and species-specific gamma values for mtDNA). Both mtDNA sequence and microsatellite (msat) data were used, and females were analyzed separately for the mtDNA tests. Correlation (Corr.) and regression coefficients (Slope) are presented with associated probability and percent determination (Det.) of genetic distance by geographic distance.

Ratios of regression coefficients of same sign (slopes both positive) are presented for each species pair: E. fuscus (EPFU), M. ciliolabrum

(MYCI) and M. lucifugus (MYLU).

Geog. Ratio of Slopes Source Distance Species Corr. Coeff. Slope (x10-4) Det. (%) p (both sexes) mtDNA River E. fuscus -0.002 -0.01 0.0003 0.420 MYLU:EPFU n/a Females only -0.124 -0.54 1.5 0.610 MYCI:MYLU 1.08 M. lucifugus 0.623 4.87 39.0 0.018* MYCI:EPFU n/a Females only 0.688 8.70 47.4 0.023* M. ciliolabrum 0.289 5.28 8.3 0.169 Females only 0.398 4.99 15.8 0.064 mtDNA Linear E. fuscus -0.140 -2.67 2.1 0.750 MYLU:EPFU n/a Females only -0.319 -4.56 10.2 0.942 MYCI:MYLU n/a M. lucifugus 0.400 14.50 16.3 0.035* MYCI:EPFU n/a Females only 0.145 8.44 2.1 0.233 M. ciliolabrum -0.236 -22.80 5.6 0.880 Females only -0.103 -6.84 1.1 0.632

135

Geog. Ratio of Slopes Source Distance Species Corr. Coeff. Slope (x10-4) Det. (%) p (both sexes) mtDNA B. route E. fuscus -0.050 0.74 0.3 0.555 MYLU:EPFU n/a Females only -0.133 -0.99 1.8 0.697 MYCI:MYLU 0.75 M. lucifugus 0.454 6.21 20.6 0.025* MYCI:EPFU n/a Females only 0.476 10.5 22.6 0.034* M. ciliolabrum 0.135 4.64 1.8 0.276 Females only 0.302 7.13 9.1 0.134 msats River E. fuscus 0.289 0.002 8.4 0.110 MYLU:EPFU 0.50 M. lucifugus 0.393 0.001 15.5 0.023* MYCI:MYLU 6.00 M. ciliolabrum 0.632 0.006 39.9 0.0003* MYCI:EPFU 3.00 msats Linear E. fuscus 0.070 0.002 0.48 0.360 MYLU:EPFU 2.00 M. lucifugus 0.341 0.004 11.7 0.050* MYCI:MYLU 6.25 M. ciliolabrum 0.540 0.025 29.1 0.010* MYCI:EPFU 12.50 msats B. route E. fuscus 0.186 0.002 3.5 0.182 MYLU:EPFU 1.00 M. lucifugus 0.485 0.002 23.6 0.010* MYCI:MYLU 6.50 M. ciliolabrum 0.740 0.013 54.9 0.000* MYCI:EPFU 6.50

136

Table 4.8. A comparison of within-site and among-site haplotype matches within males (M-M), within females (F-F), and between males and females (M-F) for each species. Within-site values are mean number of haplotype matches; among-site values are mean distances (best route, km) between the haplotype matches. For M. ciliolabrum, this latter value is presented both without and with (in parentheses) the inclusion of the ubiquitous Mci13 haplotype. Number of sites varied from 8 - 11, using auxillary sampling sites in addition to the main ones, such that number of sequenced individuals for each species was similar and haplotype distances for M. ciliolabrum could be appropriately measured given their clumped nature.

E. fuscus M. lucifugus M. ciliolabrum

M-M 0.50 ± 0.15 0.69 ± 0.19 1.59 ± 0.33 Mean no. of haplotype F-F 1.67 ± 0.36 1.33 ± 0.34 2.00 ± 0.51 matches within sites M-F 0.77 ± 0.24 1.21 ± 0.30 2.79 ± 0.79

Mean distance between 18.3 ± 3.2 542 ± 69 639 ± 70 M-M (98.4 ± 42.1) haplotype matches among 77.6 ± 25.4 396 ± 50 459 ± 57 F-F (96.5 ± 23.7) sites (km) 70.1 ± 16.3 463 ± 41 597 ± 52 M-F (133.8 ± 19.8) Number of sampling sites 10 8 11

Mean distance among sampling sites (km) 476 ± 46 562 ± 62 401 ± 40

Mean individuals sequenced per site 13.6 ± 2.16 11.4 ± 0.56 13.4 ± 3.63

137

Table 4.9. Nuclear relatedness (r; ML-Relate) among individuals at study sites and within colonies, and number of mtDNA nucleotide differences within colonies. A. Mean relatedness (r) within males, females and between males and females at each site (≤ 10 km length of river). B. Mean relatedness and nucleotide differences within maternity colonies (roosts) of each species (MYLU, M. lucifugus; EPFU, E. fuscus; MYCI, M. ciliolabrum). Number of sequenced individuals, number of haplotypes, mean and range of nucleotide differences (nt), overall mean number of nucleotide differences, number of individuals from the roost genotyped, mean and range of relatedness values and overall mean relatedness are presented. Means per colony and per species are presented ± SE and percentages are calculated based on 240, 250 and 280 sequenced base pairs for E. fuscus, M. lucifugus and M. ciliolabrum, respectively. Not all colonies were both sequenced and microsatellite genotyped.

A. Relationship Species Mean r SE Range No. pairwise comp.

Male-Male E. fuscus -0.0010 0.0020 -0.2066 - 0.5792 1868 M. lucifugus 0.0014 0.0018 -0.1520 - 0.6838 2385 M. ciliolabrum 0.0069 0.0020 -0.2416 - 0.7787 2796 Female-Female E. fuscus 0.0058 0.0030 -0.1900 - 0.8980 1103 M. lucifugus 0.0028 0.0052 -0.2055 – 0.7791 4104

M. ciliolabrum 0.0155 0.0018 -0.2389 – 0.9290 3899 Male-Female E. fuscus 0.0019 0.0020 -0.1778 – 0.5793 1700 M. lucifugus 0.0025 0.0015 -01763 – 0.6936 3613

M. ciliolabrum 0.0102 0.0018 -0.2581 – 0.7800 3995

138 139

± ± 0.0072 0.0072 0.0019 0.0321 0.0556 mean r Overall ± 0.0036 0.2778

------0.8756 0.7785 0.5161 0.7352 0.6717 0.5964 0.6617 0.3131 -0.0145 – -0.0145 -0.1820 – -0.1820 – -0.1226 -0.2058 – -0.2058 -0.1364 – -0.1364 -0.0656 – -0.0656 -0.1591 – -0.1591 -0.1540 – -0.1540 Range Range 1 r

-- -- 0.0945 0.0034 0.0103 0.0032 0.0707 0.0037 0.0064 0.2416 Mean 0.0090 0.0408 ± 0.3657 ± 0.0063 ± 0.0066 ± 0.0037 ± 0.1836 ± 0.0009 ± 0.0009 ± d

2 0.3760 5 2 ------13 35 47 24 20 28 No. ind. geno’

) ) ) ± 2.9 0.81 4.50 15.0 ± (1.9% (6.0% 0 (0%) diff (% Overall mean nt . ------5 ------1 - 7 3 - 5 1 - 5 2 - 15 2 - 38 2 - 23 nt diffs Range of Range ) ) ) ) )

0.60 . (%) 0 0 0 0 -- (1.3% (1.4% (1.7% (6.4% (9.0% (3.5%) 5 (2.1%) 5 4.0 ± 0.6 8.8 ± 1.7 diffs 11 (4.4%) 15.3 ± 6.4 22.5 ± 6.4 3.17 ± 36 (14.4%) (14.4%) 36 3.38 ± 0.37 Mean no. nt s 2 2 4 5 7 1 -- No. haplo 7 6 4 2 4 3 4 3 5 1 6 4 5 1 2 2 1 -- 10 14 12 No. ind. seq. rees FC ssing ssing Virgelle Bank, CB Rocks Rocks Onefour Onefour Hall Ag. Havre Building Wallwork Barn, BI Pipeline Cro Empress Hall FC Rocks Cypress Hills T Ag. Havre Building MHat School slab, Rock FC Big Coul. rock, 2K rock 1, FC 2K rock 2, FC Colony Colony Species MYLU EPFU MYCI B. 140

Figure 4.1. Map of sample locations. A. Inset shows general North American location of sample area in the prairies of Alberta (AB) and Montana (MT). B. Black dots are main sampling locations and grey dots are auxillary sampling locations (refer to Table 4.2 for location names). The four main rivers are Red Deer, South Saskatchewan (Sask.), Milk and Missouri. The former two make one river system and the latter two make another.

141

A.

N

B. Red Deer 110oW

River

South Sask.

River

o 49 N Milk River

100 km

Missouri River

Figure 4.1

142

Figure 4.2. TCS-generated haplotype networks for E. fuscus (A), M. lucifugus (B), and

M. ciliolabrum (C). Haplogroup roots are in rectangles. Each circular node signifies one base pair change. Lengths of the connection lines are not drawn proportional to divergence distance. A. Two distinct haplogroups are present, which do not correspond to known subspecies or geographic separations. Nodes were added manually between the

2 divergent groups. B. Three distinct haplogroups are present. The group radiating from

Mlu17 represents M. l. carissima (Chapter 5), and these haplotypes were found exclusively on the Milk and Missouri Rivers. Nodes were added manually between the 3 divergent haplogroups to show relationships. C. All haplotypes originated from Mci13 and are <10 base pair changes from this root. Haplotypes with asterisks(*) are those

found on the Missouri River only; all other haplotypes are found only on the other 3

rivers (Red Deer, South Saskatchewan and Milk).

143

A

.

Figure 4.2A

144

B.

Figure 4.2B

145

C.

Figure 4.2C

146

Figure 4.3 Scatterplots of Slatkin’s Linearized Fst versus geographic distance (“Best

Route”; see text) for M. ciliolabrum (A), M. lucifugus (B), and E. fuscus (C). 147 A. 0.028 0.024

0.020

0.016

0.012

0.008

0.004

0.000 0 200 400 600 800 1000 1200 1400

0.014 B. 0.012

0.010

0.008

0.006 Pairwise Fst/(1-Fst) Pairwise Fst/(1-Fst)

0.004

0.002

0.000 0 200 400 600 800 1000 1200 1400 C. 0.014 0.012

0.010

0.008

0.006

0.004

0.002

0.000 0 200 400 600 800 1000 1200 1400 Geographic Distance (km) Figure 4.3

148

CHAPTER 5: Beyond mtDNA: Nuclear Gene Flow Refutes Cryptic Species in Little

Brown Bats (Myotis lucifugus).

Introduction

The increasing availability of genetic characters is bringing about revisions of the classification of many taxonomic groups. The current ease with which mitochondrial

DNA (mtDNA) sequences are obtained, and the small sample sizes needed for phylogenetic studies, has resulted in an accumulation of mtDNA sequence data for a large number of species. Most recently, the “DNA Barcoding Project” has generated interest in acquiring cytochrome oxidase I (COI) mtDNA sequences for much of the world’s biodiversity (Hebert et al. 2003). MtDNA gene sequences such as cytochrome b

(cytb) or COI/II can be informative for the reconstruction of phylogenetic histories (c.f.

Shaw 2002), but contemporary gene flow is unlikely to be understood using this marker

(Avise 1994). Whether a mitochondrial gene sequence can provide enough information to delineate new biological species is uncertain and debated (DNA Barcoding, reviewed in Rubinoff et al. 2006). Conclusions based solely on mtDNA sequence data can be erroneous or misleading due to the nature of its inheritance pattern (e.g. Paetkau et al.

1998 vs. Talbot and Shields 1996 and Cronin et al. 1991, Worthington Wilmer et al. 1994 vs. 1999), and mtDNA studies often employ small sample sizes and limited geographic areas (e.g. Piaggio et al. 2002 vs. Valdez et al. 1999), furthering the potential for erroneous conclusions.

In most animals, mitochondrial DNA acts as a single gene, and is passed on as one unit to the next generation only by females. Nuclear genes, on chromosomes, are

149

shuffled each generation (Avise 1994). It is now well understood that examination of bi-

parentally inherited nuclear markers, in addition to mtDNA, is necessary to get a full

account of the history and contemporary breeding patterns of closely related taxonomic groups. Taxonomic classifications must therefore be based on multiple lines of evidence,

from molecular, morphological, ecological and behavioral data (Mayer and von

Helversen 2001).

In animals whose morphology may be highly constrained evolutionarily, convergent morphology has made taxonomy difficult (Lowe et al. 2004). In such cases,

having all components of the taxonomic picture is critical. Microchiropteran bats are a

classic example of shared ecological constraints; flight and nocturnal behavior are

limiting forces on morphological variation (Kawai et al. 2003, Norberg 1994, Ruedi and

Mayer 2001). Convergent evolutionary traits or phenotypic plasticity may influence fur

color, ear length, wing and body dimensions, and echolocation (e.g. Owen 1988, Siemers

et al. 2001) – traits that are often used to differentiate species (Miller and Allen 1928, van

Zyll de Jong 1985). As such, genetic evidence is being sought as a way of

differentiating morphologically and ecologically similar species of bats that are difficult

to identify in the field. For many species of bats, such as the little brown bat, M.

lucifugus, this is a work in progress.

As a widespread bat in North America (Wilson and Ruff 1999), M. lucifugus has been the focus of much physiological, behavioral, and ecological research (Fenton and

Barclay 1980, Humphries et al. 2002, Kunz et al. 1998, Reynolds and Kunz 2000). Six

North American subspecies are generally recognized: M. l. lucifugus, M. l. alascensis, M.

l. relictus, M. l. carissima, M. l. occultus, and M. l. pernox Fenton and Barclay 1980).

150

Original definitions were based mainly on pelage color and forearm length (Hall and

Kelson 1959, Miller and Allen 1928). Interestingly, these early subspecies definitions,

with the exception of M. l. occultus (Barbour and Davis 1970, Findley and Jones 1967),

have not been reviewed. There has been a long-standing recognition that “geographic

variation in this species needs to be analyzed and the subspecies re-evaluated” (van Zyll

de Jong 1985), however, even today, field biologists assume subspecies designation based on the geographic location of capture in relation to the historic subspecies map

((Fig. 5.1; Hall 1981). Recently, it has been determined that subspecies boundaries in M.

lucifugus do not hold, with M. l. lucifugus found throughout the other subspecies range

(Dewey 2006). Dewey (2006) has also suggested paraphylly within M. lucifugus, with

the M. l. lucifugus lineage sister to a clade made of M. l. carissima, M. l. alascensis, M. l.

relictus, and three long-eared Myotis species. This apparent paraphylly, obtained in a

strongly supported phylogenetic analysis of cytb sequences, led Dewey (2006) to suggest

these four subspecies of M. lucifugus may be distinct species. This recent suggestion questions the validity and comparability of previous research carried out on M. lucifugus

(Dewey 2006).

My goal was to evaluate this taxonomic recommendation for M. l. carissima and

M. l. lucifugus, using contemporary population genetics methods. I wanted to assess

nuclear gene flow in a region where these two putative subspecies (or species) are

sympatric. I also wanted to assess whether forearm length and pelage color could be used

to differentiate between the two mitochondrially distinct groups of M. lucifugus. I

hypothesized that the two groups interbreed as one population, detectable through the investigation of bi-parentally inherited markers. Additionally, I hypothesized that the

151 two groups could not be reliably differentiated morphologically, and that only mtDNA differences would exist, reflecting historically-separated maternal lineages. Because the

M. l. lucifugus subspecies is now known to occur sympatrically across the full range of

M. l. carissima (Fig. 5.1; Dewey 2006), complete interbreeding in my study area would suggest widespread interbreeding between these two M. lucifugus groups, homogenizing the nuclear gene pools and making taxonomic distinction unnecessary.

Methods

Study Area and Samples

I sampled M. lucifugus from Alberta (AB) and north-central Montana (MT; Fig.

5.2), obtaining forearm measurements and tissue for genetic analysis (Table 5.1). The north-central MT study sites, together with all southern AB sites are part of the grasslands or Prairie Ecozone (Gauthier et al. 2003). Sites north of and including Dry

Island Buffalo Jump are moister and partially treed, falling within the Aspen Parkland

Zone of the northern-most prairies. Sites north of Edmonton are well treed and belong to the Boreal Ecozone (Environment Canada 2003).

I caught bats each summer from 2001- 2006. Sex, species, and relative age (adult versus juvenile; Anthony 1988) were recorded. Pelage color was noted in individuals that were particularly dark or light. I measured forearm length of adult males and females using calipers. Tissue from the wing near the tibia was sampled using 2 mm diameter biopsy punches and stored in 90% ethanol for later extraction, amplification, mtDNA sequencing and nuclear microsatellite genotyping. All bats used in genetic

152

analyses were adults captured with mistnets between 2 June – 4 August each year, with

the exception of one volant juvenile male from Onefour, AB.

Eight locations make up the main study sites (Fig. 5.2) from which both mtDNA

and microsatellite genotypes were obtained. At each of these main sites, ~30-50 male and female M. lucifugus were sampled in a small area (<10 km linear distance between mistnetting locations at each main study area). I collected fewer samples from the other sampling sites and therefore not all sites were used in all analyses. For comparison

purposes, I obtained forearm data from additional areas in AB (Table 5.1, see footnotes

1-3), and additional genetic samples from AB, British Columbia (B.C.) and Washington

(WA; Table 5.1, see footnotes 4-5).

To demonstrate that any patterns in morphology were clinal, related to

environment rather than taxonomy, I analyzed forearm data from two other species:

Eptesicus fuscus and Myotis ciliolabrum. All analyses were done for males and females

separately given that sexual dimorphism is common. All work conformed to ASM

guidelines, and the legal requirements of Canada regarding conservation and animal

welfare.

mtDNA Sequences

I extracted DNA from wing tissue using the Qiagen DNeasy Blood and Tissue

Extract Kit (Alameda, CA) using the spin-column protocol. I sequenced a 250 base-pair

region of the hypervariable region II (HVII) on the control region of the mitochondrial

genome. DNA was amplified as a large fragment of variable length (~1000 bp) using

primers L16517 and sH651 (Castella et al. 2001, Fumagalli et al. 1996). Fragments were

153

then sequenced unidirectionally using the L16517 primer only due to the presence of the large repeats section (Fumagalli et al. 1996). PCR reactions were performed in a 50 µL volume using 50-100ng template, 1X PCR buffer (50 mM KCl, 10 mM Tris-HCl pH 8.8,

0.1% Triton), 2.5 mM MgCl2, 0.16 mM dNTPs, 0.8 µM each primer and approximately 1

– 2 U Taq DNA polymerase (isolated as in Engelke et al. 1990) in the following cycle:

initial three minutes at 94°C followed by 25 cycles of one minute at 94°C, one minute at

54°C, 1.5 minutes at 72°C. PCR product was purified using QIAquick Gel Extraction Kit

(Qiagen, Alameda, CA) and sequenced using Big Dye® Terminator v3.1 sequencing kit

(Applied Biosystems, Foster City, CA) according to manufacturer’s directions.

Sequencing products were resolved on ABI Prism® 3100 Genetic Analyzer (Foster City,

CA). I aligned sequences in Sequence Navigator (Version 1.0 Applied Biosystems,

Foster City, CA).

To ensure the HVII haplotypes corresponded to the subspecies delineations as determined by other mtDNA sequences, I calibrated the HVII markers to the putative

subspecies definitions by having a subset of samples independently ascribed using

cytochrome b (Dewey 2006; T. Dewey, pers. comm.),16S ribosomal subunit (Zinck et al.

2004; J. Zinck, pers. comm.), or cytochrome oxidase I (M. Vonhof and A. Borisenko, pers. comm.) sequences. All three mtDNA loci are able to resolve the putative subspecies of M. lucifugus to varying degrees, and generally produce congruent results

(Dewey 2006).

I examined relationships among HVII haplotypes using parsimony and neighbor- joining (NJ, using Tamura-Nei distances, see below) analyses in PAUP* (4.0b; Swofford

2002), and evaluated correspondence of HVII haplotypes with subspecies designations as

154

defined by T. Dewey and J. Zinck (unpublished data). Pairwise nucleotide differences among haplotypes were calculated in Arlequin 3.1 (Excoffier et al. 2006), using both the

Tamura-Nei model (Tamura and Nei 1993) and uncorrected comparisons. Because

Arlequin uses a fixed gamma shape value, allowing for unequal mutation rates among nucleotide sites, the optimal value of the shape parameter was estimated by fitting the

Tamura-Nei model onto the maximum parsimony tree. I used haplotype frequency data from the main sampling areas to analyze population distance measured as Slatkin’s

Linearized Fst (Slatkin 1995) calculated in Arlequin. These Fst distances were used to generate a neighbor-joining tree of population structure (using PAUP*).

Microsatellite Genotyping

I genotyped selected individuals at 11 microsatellite loci. The following sets of

primers were used: MME24, MMG9, MMH19, MMD9, MMD19, MMH29, MMF19

(Castella and Ruedi 2000), MYBE22 (Kerth et al. 2002), EF21, EF5 and EF6 (Vonhof et

al. 2002). I used a PCR volume of 15 µL containing 1X PCR buffer, 0.8-1.5mM MgCl2,

0.12 mM dNTPs, 0.2-0.27 mM of each primer, 0.4 units of Taq DNA polymerase, and 2

µL (~ 100 ng) DNA template. Cycling was performed under the following conditions: 1 min at 94 °C, three cycles of 30 s at 94 °C, 20 s at 47 °C, and 5 s at 72 °C, 33 cycles of 15 s at 94 °C, 20 s at 47 °C, and 1 s at 72 °C, followed by final extension at 72 °C for 30 min. PCR products were resolved on model 377 ABI sequencers (Applied Biosystems,

Foster City, CA), and analysed using GENESCAN (version 3.1) and GENOTYPER

(version 2.0) software (Applied Biosystems, Foster City, CA).

155

I checked all loci in all populations for deviations from Hardy-Weinberg equilibrium using GENEPOP 3.4 (Raymond and Rousset 1995) and adjusted the type I

error rate for multiple comparisons using the Bonferroni method (Sokal and Rolf 1995).

Due to deviations from Hardy-Weinberg equilibrium, loci were further examined for null

alleles using Micro-checker (Van Oosterhout et al. 2004). The null frequency for each locus was estimated using the software FreeNA, and an adjusted allele frequency dataset was produced (Chapuis and Estoup 2007). I performed the same analyses on corrected and uncorrected data to ensure similar results (see Results). I calculated pairwise population distances (Slatkin’s Linearized Fst) among main sampling sites using

ARLEQUIN, and visualized these relationships by constructing a neighbor joining tree using PAUP.

To determine whether individuals from different subspecies interbreed, I tested for

the presence of genetic structure within M. lucifugus along the Missouri River, where M.

l. lucifugus and M. l. carissima haplotypes occur at an approximately 1:1 ratio (see

below). I predicted that if these groups were independent breeding units, they should

possess unique microsatellite alleles, and have different allelic frequencies. This would

be detected as genetic structure, and would support the hypothesis that M. lucifugus

lucifugus and M. l. carissima are different species, and thus separate breeding populations along the Missouri River. I estimated population structure using the program

STRUCTURE (version 2.1; Pritchard et al. 2000) which is best suited to this type of

analysis (Latch et al. 2006) because it makes no a priori assumptions of group

membership, and uses a model-based clustering method to infer population structure and

assign individuals to breeding groups. A collection of the highest individual-assignment

156

probabilities occurs when the appropriate number of groups (k) has been identified. I ran

STRUCTURE for 5 competing models, k = 1 through 5, with 10 iterations of each. I

used a burn-in period of 250,000 with the same number of Monte Carlo Markov Chain

repetitions after burn-in. All combinations of Admixture and No Admixture models of

ancestry with correlated and independent allele frequencies were used as parameter sets.

I calculated the best model from posterior probabilities using Bayes Rule (Sokal and

Rohlf 1995).

To visualize the relationships between M. lucifugus on the Missouri River, I entered

the microsatellite genotypes into the program Genetix (Belkhir 1999), which performs a

factorial correspondence analysis of the alleles for each individual and plots the first three

factors in a three dimensional space. For comparison purposes and to illustrate what true

undisputed species’ separation should look like, I also plotted the relationship between M.

lucifugus and M. ciliolabrum, sympatric on the Missouri River using 7 shared

microsatellite loci (EF5, MMD9, MMH19, MYBE22, MMF19, MMD15, MMG9) that

were simultaneously corrected for null alleles using FreeNA. This same two-species 7

loci dataset was analyzed using STRUCTURE for additional support that differences in

allele frequencies at these microsatellite loci could be detected when present.

Forearm Analysis

I analyzed forearm lengths of adult males and females using analysis of variance

(ANOVA) and Scheffe’s pairwise comparisons in Stata 9.0 (2005, StataCorp LP, College

Station, TX). When assumptions of normality were not met, I performed a nonparametric test using ANOVA of ranked forearm measurements, and calculated a χ2 equivalent

157

(Sokal and Rohlf 1995). I conducted nonparametric pairwise comparisons using Mann-

Whitney tests (two sample Wilcoxon rank-sum; Stata 9.0), and adjusted the alpha value for multiple comparisons using the Bonferroni method. Where sample sizes allowed, areas were tested separately, and individuals from neighboring areas that did not differ significantly were pooled. All mean values are presented ± SE.

Results

From 2001-2006, I captured 1039 M. lucifugus individuals, and used subsets of

these for various analyses. I microsatellite genotyped 399 individuals from the main

study area, and produced HVII sequence for 91 of these plus an additional three samples

from B.C., WA and Sheep River, AB. Forty-four of the HVII sequenced M. lucifugus

and an additional 25 were sequenced at one or both of the cytb and 16S loci of the

mtDNA to determine correspondence with subspecies designations as defined by mtDNA

and geography (Dewey 2006). Thirty-three (46%) were sequenced at both the cytb and

16S loci because of an inability of cytb sequence to differentiate M. lucifugus from M.

volans in some (11, 31%; field identification verified using 16S instead) cases, and an

inability of 16S sequence to fully resolve M. l. carissima. While I did not take voucher

specimens for this study, species identifications were verified by calibration with

sequences from voucher specimens collected by Dewey (2006).

Sequence Haplotypes

In the 250 bp fragment sequenced from the HVII region, I found 61 polymorphic

sites and 26 haplotypes (Genbank accession numbers: EF471399-EF471445), 23 of

158

which were from the main study area, with the other 3 representing the B.C., WA and

Sheep River samples. When all sequences were compared, they had a mean uncorrected

pairwise difference of 17.1 ± 0.77 base pairs (6.8% pairwise difference; range 1 – 39

bases), and nucleotide and gene diversities were 0.061 ± 0.031 and 0.93 ± 0.01,

respectively.

To estimate relationships among haplotypes, an unrooted maximum parsimony

tree was constructed with 1000 bootstrap replicates (using PAUP*). Bootstrap support

for the main nodes appear on a neighbor joining tree (Fig. 5.3), allowing the relative

genetic distance between haplotypes and haploclusters to be visualized. Three distinct

haplogroups were apparent and by comparing the cytb/16S sequences with the HVII

haplotypes, it was determined that all individuals defined as M. l. carissima clustered

together (MYLU.CA), while the M. l. lucifugus samples clustered together and into two

sub-haplogroups (MYLU.LU-A and MYLU.LU-B). In other words, each HVII

haplotype could be assigned a “subspecies” designation, consistent with Dewey (2006),

allowing all HVII sequenced individuals to be assigned to one of the two subspecies

groups. The MYLU.CA haplotypes cluster (mean sequence divergence within

MYLU.CA is 10.0 ± 3.1 base pairs, 4.0%, range 1 – 22 bases) is distant from the 2

MYLU.LU haplogroups which are more closely related to each other (mean sequence

divergence within MYLU.LU is 8.2 ± 0.4, 3.3%, range 1 – 22 bases). The MYLU.CA

and MYLU.LU haplogroups differ by a mean sequence divergence of 35.6 ± 0.2 base

pairs (14.3%, range 30 – 39 bases), which is ~4x the divergence level found within each

group. The B.C. and WA samples (haplotypes 20 and 22, “outliers”) fall between the

haplogroups but appear to cluster with MYLU.CA and MYLU.LU, respectively; that they

159

do not cluster tightly within each group likely reflects their distance (~800 km) from the

rest of the samples (Fig. 5.2). These outlier samples were sequenced at the cytb/16S to

verify their subspecies designations. The sample from Sheep River, AB came from ~300

km away from the main sampling sites, yet grouped within the MYLU.LU-A haplogroup.

Haplotype Mlu26 is a female from within the main study area who differed from other

MYLU.LU-B haplotypes (1.8% sequence divergence) and therefore did not cluster as

tightly in that haplogroup. This individual was assigned to M. l. lucifugus based on the

COI locus sequence (M. Vonhof, pers. comm.). Because MYLU.LU-A and MYLU.LU-

B subgroups contained both males and females, and were widespread across all sampling

areas, this distinction is dropped from further analyses, using MYLU.LU as one

haplogroup only.

Myotis l. lucifugus haplotypes were found at all locations where genetic samples

were obtained (Table 5.2). In AB, M. l. carissima haplotypes were found only along the

Milk River. There, both putative subspecies were found in the same maternity colony at

Onefour (1 male [juvenile] and 1 female MYLU.CA and 6 female MYLU.LU).

Similarly, in MT, both subspecies were found along the Missouri River, with mixed subspecies in a maternity roost at Coal Bank (3 female MYLU.CA and 1 female

MYLU.LU). At a maternity colony in Havre, only M. l. lucifugus was found (6 females

sequenced).

I calculated pairwise population distances between the 8 main sampling sites

using the haplotype data and constructed a neighbor-joining tree (Fig. 5.4A). The three

study sites where MYLU.CA haplotypes were present are separated from the other sites

160

with a distance relative to the frequency of these haplotypes in the population. This tree

is in stark contrast to its equivalent using microsatellite data (see below).

Microsatellite Genotypes

A subsample of captured individuals (n = 399) were genotyped at 11

microsatellite loci. Number of alleles per locus ranged from 19 – 49 (mean 32.2 ± 3.7).

When microsatellite alleles of individuals for which subspecies designation was known

(i.e. those sequenced) were compared, 1.80% (7) were found only in MYLU.CA

individuals. However, when the comparison was expanded to include individuals from

the northern part of the study area where no MYLU.CA haplotypes were found, all

MYLU.CA alleles were also found in MYLU.LU individuals, suggesting no unique

microsatellite alleles exist within M. l. carissima.

Using GENEPOP, I found that five (EF5, EF6, MMH29, MMD9, MYBE22) of

the 11 loci were not in Hardy-Weinberg equilibrium. Because patterns did not suggest

Wahlund effect (Hartl and Clark 1997), I suspected null alleles. Using Micro-checker I

determined that all 5 loci contained null alleles. I estimated null allele frequencies using

FreeNA, producing a corrected dataset of allele frequencies for each locus. Using

GENEPOP on this corrected dataset, I found all loci to be in Hardy-Weinberg equilibrium, except MMH29, which was subsequently dropped from the dataset. Based on these results, I created three datasets: a small dataset containing only the 6 loci not

found to contain null alleles, a large uncorrected dataset which contained all 11 loci with

no allele frequency adjustments, and a large corrected dataset of 10 loci with the null

allele frequencies estimated and all allele frequencies corrected. There is a tendency for

161

genetic structure to be over-estimated when null alleles are present in a dataset (Chapuis

and Estoup 2007). As such I performed all analyses with all datasets; conclusions were

the same in all cases.

I calculated population distances (Slatkin’s linearized Fst) between sampling

locations and created a neighbor-joining (NJ) tree (Fig. 5.4B). Unlike the NJ tree using

mtDNA, the microsatellite distance tree shows an isolation-by-distance pattern. As such,

locations with MYLU.CA haplotypes are cluster together and separately on the mtDNA

tree, but in the microsatellite tree all locations spatially arranged similarly to how they are

geographically distributed.

Using the program STRUCTURE, I found no population structure along the

Missouri River, meaning that the genotype distribution is best explained by all samples coming from one population. All parameter sets and datasets produced the same results, with k = 1 being the most probable model. Using the large corrected dataset, and 10 replicates of each of k = 1 through 5, the average ln Pr (X|K) and percent probability for k = 1 was -6740 and 99.99%, respectively. I visualized this absence of structure using

Genetix, and compared it to the structure seen between M. ciliolabrum and M. lucifugus, two morphologically distinct species on the Missouri River (Fig. 5.5). I obtained this same clearly differentiated genetic structure between these two species using shared microsatellite loci in STRUCTURE; I ran 3 replicates of each of k = 1 through 5, and found that k = 1 (average ln Pr (X|K) = -9627) had the lowest probability (<<8.48x10-49), and k = 2 (-8040) had the highest probability >99.99%.

Forearm lengths of M. lucifugus from the southern part of the study area (South;

Table 5.3), the central part, and the northern part, differed significantly (n = 617; Table

162

5.3A), with forearms getting longer from south to north (furthest straight-line distance

730 km). From South to Central, males and females increased in mean forearm length

0.6 and 0.7 mm, respectively. From Central to North, the increase was 1.2 and 0.8 mm,

respectively. Males had significantly shorter forearms than females in the South and

Central regions (p < 0.001), but not in the North region (p = 0.91) where males and

females had similar means; males compared across the three areas differed significantly,

as did females. The South study region contained all Milk and Missouri River samples

and was therefore the only area known to contain M. l. carissima samples. When forearm

lengths were regressed against UTM coordinate northings, the regressions for males and

2 females were significant (Fig. 5.6; p < 0.001; males F1,307 = 107.3, p < 0.001, R = 0.26, y

-6 -7 2 = 22.8 ± 1.4 + [2.64x10 ± 2.6X10 ]x; females F1,309 = 46.6, p < 0.001, R = 0.13, y =

26.6 ± 1.7 + [2.01x10-6 ± 3.0X10-7]x ).

Forearm lengths of M. ciliolabrum and E. fuscus (Table 5.3B) were only available for South and Central regions. In both species, forearm lengths were significantly larger

in the more northerly samples, showing the same pattern as in M. lucifugus. All

differences were significant with the exception of male E. fuscus, which did not differ significantly between the South and Central study regions (p = 0.12).

Myotis lucifugus varied greatly in pelage color. I captured individuals that ranged from dark brown to light blonde at the same sampling sites on the same nights. This

variation in pelage color existed mainly in the southern part of the study area (arid prairie

region), with coloration being dark brown in the northern area (moister more treed region). I genetically sampled two bats from each of the Milk and Missouri Rivers; I

chose one sample from each river that was particularly blonde and one that was

163

particularly dark brown. Because M. l. carissima is described as the palest of all of the

subspecies (Miller and Allen 1928, van Zyll de Jong 1985), I hypothesized that the

blonde individuals from each river would be M. l. carissima and the dark individuals

would be M. l. lucifugus. However, the mtDNA sequences of both samples from the Milk

River were that of M. l. lucifugus, and both Missouri River samples were M. l. carissima.

Along the Milk River, I occasionally observed individuals with white outlines to their

wing margins, a trait thought to characterize M. l. carissima (Miller and Allen 1928);

however, an M. l. lucifugus (based on genotype) with this white wing margin was also

found north of Edmonton, AB (collected by B. McClymont, AB Fish and Wildlife,

Edmonton; deposited at Royal ON Museum by C.L.L., ROM Collection #F54056; COI

sequence provided by A. Borisenko, Bar Code Project; subspecies ID provided by M.

Vonhof, unpublished data), suggesting that this trait also does not distinguish these two

subspecies.

Discussion

Myotis lucifugus in southern Alberta and north central Montana consisted of two different mtDNA groups (14.3% uncorrected mean sequence divergence), showing a full

range of light and dark brown pelage coloration, and having smaller forearm lengths than

more northern individuals. Dewey (2006) found that cytb sequence divergence

corresponds with most of the original subspecies designations for M. lucifugus, including

M. l. carissima and M. l. lucifugus. MtDNA delineation of M. l. lucifugus and M. l.

carissima was the same regardless of what section of the mtDNA sequence was used

(HVII, cytb, 16S, COI), allowing for a clear delineation of the two putative subspecies for

164

use in my “interbreeding test.” The study area is in a suspected region of overlap for

these two putative subspecies according to the original subspecies definitions (Hall

1981). The substantial mtDNA-sequence nucleotide differences between M. l. lucifugus

and M. l. carissima haplotypes suggest that they were indeed separate races in the past.

However, I found contemporary nuclear gene flow between them on the Missouri River,

where M. l. lucifugus and M. l. carissima haplotypes are sympatric, each making up approximately 50% of the population, and I found that they are breeding as one population. Therefore, mtDNA divergence appears to be the result of historic separation of populations, rather than evidence for cryptic species. Because extensive interbreeding is occurring in the study area, it is likely that interbreeding is occurring between these two groups wherever they are sympatric. It is now known that these subspecies are fully sympatric across the range previously thought to be that of M. l. carissima only (Fig. 5.1), making it highly likely that these two groups of M. lucifugus comprise one nuclear gene pool only, and are therefore one, not two, taxonomic units.

With the exception of a morphological review of M. l. occultus (Barbour and

Davis 1970, Findley and Jones 1967), the original subspecies definitions (Harris 1974,

Hollister 1909, 1911, Miller and Allen 1928, Thomas 1904) and geography for M.

lucifugus have not been reviewed (van Zyll de Jong 1985). Subspecies identification in the field has been based on historical geography according to the map by Hall (1981).

Myotis l. carissima supposedly has the lightest pelage color of any subspecies, and shorter forearms than M. l. lucifugus (Miller and Allen 1928). However, I found that forearm length was clinal, increasing from south to north across the study area and further into northern AB. Similar patterns also exist for E. fuscus and M. ciliolabrum and

165

do not correspond to any subspecies boundaries (van Zyll de Jong 1985). This

corroborates findings of clinal bat morphology from other studies (Bogdanowicz 1990,

Burnett 1983, Findley and Jones 1967, Patriquin 2001). A snapshot of the smaller

forearms in the southern portion of the province in relation to the larger ones in the

northern part, together with darker pelage of more northern individuals, prompted Smith

and Schowalter (1979) to propose that the range of M. l. carissima includes all M.

lucifugus of the AB prairies, rather than just the small southeastern corner of the province

as was proposed by Hall and Kelson (1959).

Clinal variation in forearm length makes it an unreliable taxonomic character for

differentiating M. l. lucifugus from M. l. carissima in the study area. A thorough review of other morphological features (e.g. skull characters) in M. l. lucifugus versus M. l.

carissima across their range would be needed to confirm that other morphological

differences truly do not exist between these groups.

Pelage color varies greatly among and within populations of M. lucifugus (this

study; Nagorsen and Brigham 1993); in southern AB, colonies tended to have mixed light and dark forms, while more northerly individuals were dark ( unpublished data; Smith and Schowalter 1979). Although M. l. carissima is supposed to be paler than M. l.

lucifugus (Miller and Allen 1928), I found that color was not an indicator of subspecies,

with light blonde and dark brown individuals being found in each haplogroup. In

southern Alberta, M. ciliolabrum pelage also varies greatly in color (pers. obs.) with

blonde and dark-or red- brown individuals being found at the same locations. There are

no subspecies boundaries for M. ciliolabrum known in the study area, suggesting these

color morphs represent individual variation within the populations. While I have an

166

admittedly small sample size for testing the usefulness of pelage color in differentiating

subspecies, my findings corroborate results about color morphs in other genetic studies of

bats (e.g. Jacobs et al. 2004), suggesting that this character may not be a reliable

taxonomic criterion for some species of bats. In the study area, even the unusual trait of

having a white wing border was found in both M. l. lucifugus and M. l. carissima

haplotypes, a morphological feature that was previously thought to exist only in the latter

subspecies (Miller and Allen 1928, Smith and Schowalter 1979).

Just as M. lucifugus is the most widespread bat in North America (Wilson and

Ruff 1999), M. daubentonii is the most widespread in Europe. In Europe, an extensive

re-evaluation of this species’ morphology found an increase in body size from south to

north and phenetic overlap of morphological measures between subspecies; this resulted

in the proposal to drop two of the four recognized subspecies (Bogdanowicz 1990,

Kruskop 2004). Similar morphological re-examinations in relation to climate and genetics have resulted in radical taxonomic consolidation in other taxa as well (e.g. Ursus

arctos Paetkau et al. 1998, Ammodramus caudacutus Rising and Avise 1993, Schwartz et

al 2003). The recommendation for such a review has been made for other species (e.g.

Cooper et al. 1998, 2001), recognizing that a complete morphological and genetic profile

is necessary to warrant systematic re-classifications.

In animals such as bats, whose morphology is highly constrained evolutionarily

(Norberg 1994), convergent morphological traits may falsely group unlike individuals

(Dewey 2006, Ruedi and Mayer 2001, Stadelmann et al. 2007). As such, sequence

divergence may be indicative of morphologically cryptic species (Baker and Bradley

2006a). Upon examination of genetic and ecological characters, new bat species are

167

being discovered among morphologically similar animals; for example, in sympatric

morphologically identical European pipistrelle bats, two distinct mitochondrial DNA

lineages were found (Barratt et al. 1997). Combined with the fact that the groups do not

roost together, that each is a clearly distinct echolocation phonic group (Weid and von

Helversen 1987), and that previously unnoticed ecological and behavioral differences

have subsequently been found (Davidson-Watts and Jones 2005), they have been elevated

to species status. Morphologically cryptic genetic species may be prevalent within

Chiroptera (e.g. Jacobs et al. 2006). Baker and Bradley (2006b) predicted that based on

mtDNA alone, there are 885 unidentified genetic species of bats. But, as they acknowledge, mtDNA on its own can be misleading. Its discordance with morphological and geographic divisions among well defined species (e.g. Sorex spp. Demboski and

Cook 2001, Eptesicus spp. Mayer and von Helversen 2001), together with its inability to represent contemporary gene flow, serves caution to those interpreting mtDNA sequence divergence in a taxonomic context. Taxonomic over-splitting could occur if mtDNA evidence alone is used to define species.

Mitochondrial DNA sequence divergence can provide invaluable insight into past separation and evolutionary events (Avise 1994, c.f. Shaw 2002), but interpretation of such divergence in a modern taxonomic timeframe requires the context of biological,

geographic and additional genetic (nDNA) data. When sequence divergence is large,

taxonomic suggestions or management recommendations based solely on mtDNA are

tempting (e.g. Lloyd 2003, Piaggio et al. 2002), but as I show here, should be treated with

caution. While most mtDNA phylogenetic studies that allude to cryptic genetic species

acknowledge the need for nuclear data (e.g. Dewey 2006, Mayer and von Helversen

168

2001), caveats are sometimes ignored and new species nomenclature prematurely adopted (e.g. Dawson 2006).

Taxonomically designating groups of individuals with large mtDNA sequence divergences requires a thorough examination of haplogroup distribution and nuclear gene flow. While many systematics studies compare mtDNA and morphology (e.g. Cooper et al. 2001, Rising and Avise 1993), few include a study of nuclear gene flow to determine whether seemingly distinct genetic clusters of sympatric animals interbreed (c.f. Castella et al. 2000). Hybridization or introgression between species or genetically differentiated populations is common in zones of overlap. The extent of interbreeding between differentiated forms follows a continuum making it difficult to clearly delineate groups when zones of overlap are large (Avise 1994). When two “differentiated” groups come together, nuclear genes are shuffled and allele frequencies homogenize, although pre- existing mtDNA haplotype differences will remain. When sympatry and hence interbreeding is extensive enough, a metapopulation (Hanski and Simberloff 1997) or eventually panmixia occurs, and intact nuclear gene pools of the “differentiated forms” no longer exist, with only remnant mtDNA sequence divergence remaining.

The two sympatric divergent M. lucifugus haplogroups fail to meet the biological species concept (Mayr 1963), because they interbreed. By the genetic species concept

(GSC; reviewed and refined by Baker and Bradley 2006a), which allows some introgression at overlap zones, M. l. carissima and M. l. lucifugus fail as species due to the extent of interbreeding and the extensive overlap of the subspecies’ ranges; because

M. l. lucifugus haplotypes have recently been found sympatric with all other subspecies

(Dewey 2006), the GSC requirement of “protection of the integrity of [each group’s]

169

gene pool” (Baker and Bradley 2006a, p. 654) is unlikely to be met. As there is as yet no

“barcode-based species concept” to define a species based exclusively on mtDNA

divergence (Rubinoff et al. 2006), these two groups of bats appear to be simply one

species.

Thorough sampling of putative groups is critical in determining allopatry versus

sympatry, especially where one may be less common. Initially, southern AB was thought

to have only M. l. lucifugus haplotypes (Dewey 2006; M. Vohnof, pers. com.), but my

more thorough sampling found M. l. carissima in small numbers, and in roosts where M.

l. lucifugus had already been found. Most mtDNA studies employ a small number of

samples (<5) from each geographic area (e.g. Bilgin et al. 2006, Cooper et al. 2001,

Mayer and von Helversen 2001), and/or include limited geographic sampling areas (e.g.

Mayer and von Helversen 2001, Piaggio et al. 2002), making allopatry a weak

assumption, and conclusions about systematics difficult or questionable.

Based on my results, I do not support the suggestion that M. l. carissima and M. l.

lucifugus represent distinct species (Dewey 2006). Should they retain subspecies status?

If the goal is to “recognize only truly distinctive allopatric populations as ‘subspecies’ and reduce the vague, somewhat sympatric, interbreeding populations to a lower rank”

(Edwards 1954, p. 2), then the designation M. l. carissima should cease to be used, and individuals with such haplotypes would be referred to as M. l. lucifugus. Recognizing that an observed shared nuclear gene pool among sympatric individuals defines a population (Lowe et al. 2004), I follow the lead of others (e.g. Gill et al. 1999) and do not provide a Latin name for a group differing only in mtDNA sequence from another such

170

group that it interbreeds with. I therefore suggest that the carissima designation be dropped.

Substantial genetic, protein, and morphological data have been collected for another M. lucifugus subspecies, M. [lucifugus] occultus. Once considered a separate

species (Hollister 1909), there is much disagreement over the specific status of this little

brown bat (reviewed in Piaggio et al. 2002). An extensive review of morphology of little

brown bat body size in the southeastern U.S. where M. [l.] occultus is found, revealed

that this trait is correlated with climate (Findley and Jones 1967); this, together with the

discovery of individuals “almost exactly intermediate in all characteristics examined”

(Barbour and Davis 1970, p. 150) between M. l. carissima and M. [l.] occultus, led to the subsuming of M. occultus to M. l. occultus. Allozyme analysis shows extensive gene flow among M. l. carissima and M. l. occultus (Valdez et al. 1999). This is in contrast to mtDNA sequence (cytb and COII; 5-6% sequence divergence) data that show substantial sequence divergence between these two races, prompting Piaggio et al. (2002) to suggest re-elevation to species status. Most recently, Dewey (2006) found that there is little mtDNA sequence divergence (cytb) between M. [l.] occultus and M. l. lucifugus, and that individuals with M. l. occultus, M. l. lucifugus and M. l. carissima mtDNA haplotypes do

not have separate defined geographic ranges (Dewey 2006). Stadelmann et al. (2007),

using nuclear DNA sequence (Rag 2) found that M. [l.] occultus and M. l. carissima

possess identical sequences. In summary, biparentally inherited markers (allozymes and

nDNA) suggest that “occultus” forms do not differ from other M. lucifugus, and

morphology used to differentiate “occultus” is clinal and intergraded with M. l.

carissima. It is therefore likely that the occultus designation may be an unnecessary

171

taxonomic distinction as well; however, an examination of nuclear gene flow using

microsatellites would be useful to describe the extent of gene flow.

Much morphological and ecological data exist across North America for M. lucifugus due to its commonness, and mtDNA data are available (Dewey 2006; Barcode

Project, unpublished data; M. Vonhof, unpublished data; J. Zinck, unpublished data). As nuclear gene flow data are compiled for M. lucifugus beyond my study area, the relationships among the other subspecies can be fully understood and the taxonomy for the species resolved.

My study highlights the importance of investigating contemporary gene flow in

widely sympatric animals suspected of being cryptic species, and has important

implications for application of the DNA Barcoding Project (Hebert et al. 2003) to

mammalian taxonomy. An increasing number of “new species” are being described

based on mtDNA (COI) sequence only (reviewed in Prendini 2005, Rubinoff et al. 2006).

To date, most barcoding applications have been with invertebrates, birds, and fish, but a

recent application to mammals revealed >6 new bat species in Guyana (Clare et al. 2007).

High intraspecific mtDNA sequence divergence (>2.5%) in 6 of 87 species of bats

examined prompted the conclusion of cryptic species complexes (Clare et al. 2007). My

study shows that such claims may be too hasty. An investigation of nuclear gene flow

and extent of sympatry would be able to determine whether these potential new species

are indeed real or merely a signature of remnant mtDNA variation.

Table 5.1. UTM locations for sampling sites in Fig. 5.2 in descending order of latitude. River abbreviations are: South Saskatchewan River

(SSR) and North Saskatchewan River (NSR). For M. ciliolabrum and E. fuscus, only forearm data are presented in this study. Type of data collected for M. lucifugus: microsatellite genotypes (msat), mitochondrial DNA sequences (mtDNA), and forearm lengths (FA). The shaded sites are the 8 main sampling areas referred to in the text. The last three sites under the dotted line were outside of study area and used for mtDNA comparison only.

UTM Species Sampled M. Location River Zone Easting Northing M. lucifugus ciliolabrum E. fuscus near Peace River (EMEND), AB 1 n/a 11 415986 6292662 FA Lac La Biche, AB 2 n/a 12 0372072 - 0500000 5983522 - 6096621 FA Drayton Valley, AB 3 near NSR 11 634213 5874142 FA Donalda, AB (on Meeting Creek) Battle 12 395945 5826600 mtDNA, FA FA Big Knife Provincial Park, AB Battle 12 417028 5816359 mtDNA, FA Content Bridge, AB Red Deer 12 358373 5797285 FA McKenzie Crossing , AB Red Deer 12 366277 5765649 FA FA Dry Island Buffalo Jump Provincial Park, AB Red Deer 12 364705 5755366 mtDNA, FA FA FA Tolman Bridge, AB Red Deer 12 361048 5744424 FA Drumheller, AB Red Deer 12 387979 5697124 msat, mtDNA, FA FA FA Empress, AB SSR 12 571627 5640100 msat, mtDNA, FA FA FA Dinosaur Provincial Park, AB Red Deer 12 464336 5624406 msat, mtDNA, FA FA FA Bindloss Ferry Crossing , AB SSR 12 557276 5610380 FA FA Near Suffield Army Base*, AB SSR 12 0543100 - 0565398 5601849 - 5620269 FA FA

172

UTM Species Sampled M. Location River Zone Easting Northing M. lucifugus ciliolabrum E. fuscus Medicine Hat, AB SSR 12 524333 5542498 FA Bow Island, AB SSR 12 478815 5542421 msat, mtDNA, FA FA FA Pinhorn Grazing Reserve, AB Milk 12 507336 5441473 FA FA FA Onefour, AB Milk 12 538581 5440986 msat, mtDNA, FA FA Writing on Stone Provincial Park, AB Milk 12 454864 5436798 FA FA FA Havre, MT Milk 12 588599 5373321 msat, mtDNA, FA FA FA Coal Bank, MT Missouri 12 558746 5320250 msat, mtDNA, FA FA FA Judith Landing, MT Missouri 12 602471 5287566 FA McClelland Ferry, MT Missouri 12 620042 5284197 msat, mtDNA, FA FA FA Kananaskis, Rocky Mountains, AB 4 near Sheep 12 666127 5613540 mtDNA Skagit Valley, B.C.5 n/a 10 614611 5451864 mtDNA North Cascades National Park, WA 5 n/a 10 639898 5452167 mtDNA

1 data from K. Patriquin, 2 data from L. Crampton , 3 Lippert 2001, 4 sample from D. Solick, 5 samples from T. Luszcz

*various locations; not on map in Fig. 5.2

173

Table 5.2. Subspecies designations, M. l. carissima (MYLU.CA) and M. l. lucifugus (MYLU.LU) of sequenced males and females. Sites sampled by authors are in gray shaded rows and are listed in order from north to south. Sites: Big Knife Provincial Park (BKPP), northmost Red Deer River (Content Bridge, McKenzie Crossing, Dry Island Buffalo Jump Provincial Park), Drumheller (DH), Dinosaur Provincial Park (DPP), Empress (EM), Bow Island (BI), Writing on Stone Provincial Park (WOS), Onefour (OF), Havre (HV), Coal Bank (CB), McClelland Ferry Crossing (MC). The latter three sites are in MT, the former are in AB. Samples were obtained from others in the outlying regions (British Columbia, Washington, and Sheep River of the Rocky Mountains, AB).

MYLU.CA MYLU.LU RIVER SITE Male Female Male Female Total Sequenced Percent MYLU.CA Battle Donalda 0 0 1 1 2 0% Battle BKPP 0 0 2 2 4 0% 0% Red Deer northmost 0 0 3 0 3 0% Red Deer DH 0 0 5 7 12 0% 0% Red Deer DPP 0 0 11 1 12 0% S.Sask EM 0 0 4 4 8 0% 0% S.Sask BI 0 0 4 7 11 0% Milk WOS 0 0 1 1 2 0% Milk OF 1* 1 4 6 12 17% 7% Milk HV 0 0 8 7 15 0% Missouri CB 2 3 7 5 17 29% 43% Missouri MC 4 6 4 4 18 56% outside study area B.C. 0 1 0 0 1 n/a n/a outside study area WA 0 0 0 1 1 n/a n/a outside study area AB Rocky Mnts. 0 0 0 1 1 n/a n/a TOTAL: 7 11 54 47 119 *juvenile

174

Table 5.3. Forearm comparisons for M. lucifugus (A), and M. ciliolabrum and E. fuscus (B). Adult males (M) and females (F) from three

main areas (North, Central and South study regions) were compared. For M. lucifugus, the North region includes samples from the Battle

River, Drayton Valley, near Peace River and Lac LaBiche. The Central region for all species includes the Red Deer River sampling sites.

For M. ciliolabrum this Central region also includes the South Saskatchewan River sites. The South region for all species includes the Milk

and Missouri River sampling sites. For M. lucifugus this South region also includes Bow Island of the South Saskatchewan River; for E. fuscus this South region includes Bow Island and Medicine Hat, South Saskatchewan River. Analysis of variance was parametric for E. fuscus and nonparametric for M. lucifugus and M. ciliolabrum. Means ±SE, ranges, and sample sizes (in parentheses) are in shaded areas.

All results were significant (*) with the exception of male E. fuscus in Central versus South regions and males versus female M. lucifugus in the North region.

175

Table 5.3 A.

Region/ M. lucifugus Comparison Males Females Both Sexes M vs. F North 38.8 ± 0.2 38.9 ± 0.1 38.8 ±0.1 z = 0.119 34.5 - 41.6 37.0 - 41.2 34.5 - 41.2 p = 0.91 (n = 69) (n = 58) (n = 127)

Central 37.6 ± 0.1 38.1 ± 0.1 37.9 ± 0.08 z = 3.494 37.0 - 40.0 32.0 - 40.4 37.0 - 40.4 p < 0.001* (n = 119) (n = 112) (n = 231) South 37.0 ± 0.1 37.4 ± 0.1 37.3 ±0.07 z = 3.603 32.1 - 39.9 32.3 - 40.3 32.1 - 40.3 p < 0.001* (n = 120) (n = 140) (n = 260) North-Central z = 6.181 z = 3.649 p < 0.0001* p < 0.001* ANOVA: Sex χ2 = 20.9, df = 1, Central-South z = 4.366 z = 4.347 p < 0.001 p < 0.0001* p < 0.0001* Region χ2 = 139.8, North-South z = 8.912 z = 6.903 df = 2, p < 0.001 p <0.0001* p < 0.0001*

176 177 n/a

146.4 122.0 M vs. F M vs. = 10.0 Scheffe = Scheffe = p < 0.001* p < 0.001* = 112.8, 1,482 F 1,482 ANOVA: p < 0.001* p = 0.002* F 0.2 0.09 ±

± Sex n/a Region (n = 321) (n = 321) (n = 164) 46.2 47.0 42.0 - 51.4 41.1 - 51.1 Both Sexes Sexes Both E. fuscus 0.1 0.2 ± ±

n/a 146.4 (n = 90) (n = 90) Females Females (n = 163) (n = 163) Scheffe = 47.8 47.2 p < 0.001* 42.4 - 51.4 41.1 - 51.1 0.1 0.2 ± ±

n/a n/a n/a 30.4 Males (n = 74) (n = 74) p = 0.12 (n = 158) (n = 158) Scheffe = 46.2 45.9 42.0 - 50.3 41.2 - 49.1

n/a

M vs. F M vs. z = 8.028 z = 8.028 z = 7.376 p < 0.001* p < 0.001* = 14.5, df 1, 2 = 11.6, df 1, χ ANOVA: 2 p < 0.001* p < 0.001* χ 0.06 0.05

± ± n/a Sex (n = 417) (n = 417) (n = 275) Region 29.0 - 38.6 32.5 32.0 29.0 - 34.8 Both Sexes Sexes Both 0.08 0.08 M. ciliolabrum M. ciliolabrum

± ± n/a Females Females (n = 203) (n = 203) (n = 134) z = 2.360 z = 2.360 p = 0.018* 30.3 - 38.6 29.3 - 34.8 32.5 32.1 0.08 0.08

± ± n/a n/a n/a Males (n = 214) (n = 214) (n = 141) z = 3.530 z = 3.530 p < 0.001* 29.0 - 38.9 29.0 - 33.2 31.7 31.2

North South Central Region/ Region/ Comparison Comparison North-South North-South North-Central North-Central Central-South Central-South Table 5.3B. 178

Figure 5.1. Map of M. lucifugus subspecies boundaries as originally defined using

morphology, adapted from Hall (1981). 1. M. l. alascensis, 2. M. l. carissima, 3. M. l.

lucifugus, 4. M. l. occultus, 5. M l. pernox, 6. M. l. relictus. My study area is the dark

rectangle, spanning the M. l. lucifugus and M. l. carissima ranges. Triangles are locations

where at least one M. l. lucifugus haplotype has been located indicating that many “out of

range” samples have been documented; white triangles are from this study and dark triangles are from Dewey (2006).

179

Figure 5.2. Map of sample locations. A. Samples/data collected by others (black dots):

1. Skagit Valley, B.C., 2. North Cascades National Park, WA, 3. Sheep River, AB, 4.

Drayton Valley, AB, 4. near Peace River, AB, 5. Lac La Biche. The main study area is

the rectangle and this is expanded in B. B. Grey dots are areas where sequence data

and/or forearm measurements were collected; at black dot locations microsatellite data

were collected in addition to the latter data. Site abbreviations: BKPP, Big Knife

Provincial Park; Don, Donalda; CNTB, Content Bridge; McKC, McKenzie Crossing;

DIBJ, Dry Island Buffalo Jump; TLM, Tolman Bridge; DH, Drumheller; DPP, Dinosaur

Provincial Park; EM, Empress; FC, Bindloss Ferry Crossing; MHat, Medicine Hat; BI,

Bow Island; WOS, Writing on Stone Provincial Park; OF, Onefour; HV, Havre; CB, Coal

Bank Landing; MC, McClelland Ferry Crossing.

180

Figure 5.2

181

Figure 5.3. HVII haplotypes (26) for M. lucifugus throughout the main study area, and

including 3 samples from outside the study area, two of which are unique (Mlu22 and

Mlu20, WA and B.C. samples, respectively, marked with *). Bootstrap support (1000

replicates from maximum parsimony tree) appears on major nodes of the neighbor-

joining phylogram. Branch lengths reflect genetic distance in changes per nucleotide

(Tamura-Nei model used; Tamura and Nei 1993). Two main clusters of haplotypes:

MYLU.CA haplotypes are from specimens identified as M. l. carissima through

cytb/16S; MYLU.LU are M. l. lucifugus as identified by cytb/16S/COI (J. Zinck, T.

Dewey, M. Vonhof, unpublished data). The MYLU.LU haplogroup can be further

divided into two HVII clusters: MYLU.LU-A and MYLU.LU-B, for which no

geographic, morphological or other feature can be found that correlates with these

clusters.

182

93

73

99

Figure 5.3

183

Figure 5.4. Neighbor-joining trees showing population structure within the main study

area based on HVII haplotypes (A) and microsatellite genotypes (B). These phylograms reflect relative genetic distances, calculated using Tamura-Nei model (A) and Slatkin’s

Linearized Fst (B). The arrow in A shows the displacement of the Havre study site from its geographic position in the tree. Three sites (MC, CB, OF; respectively, in order of south to north and decreasing MYLU.CA percentage in samples) were the only sites to have MYLU.CA haplotypes and they clustered away from the other sites where only

MYLU.LU haplotypes were present. The population structure illustrated in B resembles the actual geographic orientation of the sites (see Fig. 5.2).

A. B.

Figure 5.4

184 185

Figure 5.5. Factorial correspondence analysis plots of M. lucifugus from the Missouri

River (A) and for comparison, M. lucifugus and M. cililolabrum (B) from these same sampling sites. In this highly polymorphic system of microsatellite loci, the first three factorial axes captured 7.62% (using 10 loci) and 7.28% (using 7 loci) of the variation in

(A) and (B) respectively. A) Discrimination between the two sampling sites, Coal Bank

(dark squares) and McClelland (light squares) is not possible. One cluster of individuals illustrates that there is no discrimination between supposed M. lucifugus subspecies, despite the finding that the MYLU.CA and MYLU.LU haplotypes occur in about equal proportions within these two sites. B) The two species M. lucifugus (light squares) and

M. ciliolabrum (dark squares) are easily differentiated genetically indicating that the highly variable microsatellite markers used in this study can resolve species differences when present.

186

A. A.

B. B.

Figure 5.5

187

42

41 40

39 males 38

37 females 36 35 Linear (males) Forearm length (mm) Forearm length 34

33 Linear (females)

32 5250000 5500000 5750000 6000000 6250000 6500000

UTM Northing

Figure 5.6. Regression of M. lucifugus male and female forearm lengths on UTM northings. Refer to Table 5.3A for sample sizes.

188

CHAPTER 6: General Synthesis

Understanding landscape-level processes is analogous to assembling a jig-saw

puzzle; the more fine-scale pieces that are in place, the better the picture. Using a

bottom-up approach, this thesis attempted to assemble an overall picture of prairie bat ecology in the context of riverine landscapes. While some missing pieces remain, an

overall picture did emerge.

During both summer and winter, M. ciliolabrum, M. evotis, E. fuscus and to a

lesser extent M. lucifugus, roost in rocky riparian areas. Selection of summer roosts

selection differs among species. While winter roosts for only one species were located

(acoustic data verified M. ciliolabrum and M. evotis presence in rocky riparian habitat during winter), these roosts shed light on winter bat ecology in the prairies. Unlike typical hibernacula in the Rocky Mountains or in central Canada, winter roosts in the prairies are humid caves. By remaining in the prairies, dehydration during winter may be a greater risk than in other environments, but bats may be compensating with winter flights that allow drinking during the winter. Substantial movement of bats through unsuitable roosting areas in spring and fall suggests that bats are commuting between

their summer and winter roosts using rivers as movement corridors to some extent.

While knowledge of this fine-scale habitat use was sufficient to set the stage for a

landscape-level investigation, there are a number of fundamental pieces in the puzzle of

prairie-bat landscape ecology that are still missing. Reproductive M. lucifugus females

are rare in rock crevices, but have been documented (pers. obs.; Holloway 1998). Roosts

used by colonies of M. lucifugus in my study area were mainly buildings, although two

189 roosts in the ceiling of a cave have been documented (Holloway 1998). A roost selection study directed at rock-roosting female M. lucifugus would increase our understanding of how this species uses the natural prairie landscape, and may shed light on why it is more commonly found in buildings. Likewise, at three locations on the Red Deer River, I documented M. cilioabrum, M. evotis and E. fuscus, but not M. lucifugus, residing in rocky riparian areas during winter. This apparent absence stimulates two questions: 1.

Are M. lucifugus hibernating in natural rocky riparian areas? 2. If so, why are they not acoustically detected during winter? It is possible that this species has different physiological requirements or behaviours that cause it to select different winter habitat from the other prairie bat species; describing these unique habitat needs would be necessary to fully understand the ecology of this species on the prairies.

As I discuss in Chapter 2, an animal’s environment shapes its behaviour, necessitating the study of unique habitats and community assemblages to fully characterise a species’ ecology. Therefore, having only characterized roost selection in one area is a limitation of this research. I describe the habitat selection of M. ciliolabrum and E. fuscus in as great a detail as possible to allow for comparison with other studies which could eventually lead to an overall understanding of true habitat preferences, however, my research falls short of including the range of study conditions necessary to do this in one study. Similarly, I have described landscape level processes, such as dispersal, in only one habitat, the prairies. If forces such as availability of suitable roosts vary greatly between males and females or between habitat types, dispersal behaviour and in particular, the degree of sex-biased dispersal, is likely to vary also. Because climate and topographic complexity impact population structure (e.g. Clegg et al. 1998),

190 replication of landscape-level investigations in other habitats for M. lucifugus, M. evotis and E. fuscus, the species occurring outside of xeric habitats, would be warranted.

From a landscape perspective, consideration of both spatial and temporal scales is requisite for understanding habitat use by species (Crow 2005). That bats in the prairies shift their distribution seasonally, use rock-crevice roosts during summer, seem to congregate in patches of substantial rock-crevice habitat during fall and winter, and fly along rivers to some extent, outlined the parameters for a landscape-level investigation of the role that rivers play in prairie bat ecology. Using molecular genetics, I determined that the riverine landscape was structuring bat populations, and this structure varied among species according to flight capabilities and roost specificity.

Understanding gene flow requires background knowledge of how an animal interacts with its environment. Consideration of landscape characteristics in the context of an animal’s ecology, allows for identification and appropriate interpretation of structure in populations (Manel et al. 2003). As I demonstrated in both the landscape genetics chapter (Chapter 4) and in the chapter on the analysis of M. lucifugus subspecies

(Chapter 5), mtDNA and nuclear DNA can provide very different pictures of gene flow patterns. Interpretation of gene flow with prior knowledge of movement patterns was necessary to realize the difference between sex-biased dispersal and seasonal movements for mating. Similarly, differences in results between mtDNA and nuclear DNA analyses in M. lucifugus lacked biological relevance when examined in the context of morphological and ecological variables, and resulted in a recognition of taxonomic incongruence. Using the background knowledge of roosting and behavioural ecology for each of the species, I specifically tested the congruence between riverine landscape and

191

gene flow patterns to elucidate the structuring influence of rivers on prairie bats.

Knowledge of roost selection further allowed me to speculate on the cause of observed

breaks in gene flow. I also demonstrated that the consideration of roost selection,

physiology, relatedness, and social behaviour in the context of the evolution of dispersal

could explain differences in gene flow between the sexes and the species.

While a number of studies of bat population structure have correlated genetic

structure with large topographic barriers to gene flow (e.g. Gibraltar Straight, Castella et

al. 2000; Alp Mountains, Petit and Mayer 1999), few have tested for genetic structure

from more subtle land features (e.g. Miller-Butterworth et al. 2003), and none have

compared species to determine whether landscape features similarly influence bats with

differing morphology and behaviour. While it seems intuitive that landscape features should structure movement of individuals and therefore distribution of genes, to

document genetic structure on the relatively small scale used in this study, is somewhat unexpected for volant animals. Of course, whether population structure is noteworthy is

a matter of scale; a 4 m strip of barren ground is a barrier to gene flow in gray-tailed

voles (Microtus canicaudus; Wolff et al. 1997), while panmixia can exist over thousands

of kilometers in most non-sedentary birds (Avise 1996) and cetaceans (Hoelzel 1998).

Little structure has been found in large bat species (e.g. Pteropus scapulatus, Webb and

Tidemann 1996), while small scale structure has been reported in some small species

(e.g. Plecotus auritus, Burland et al. 1999), further evidence that although all bats are

volant, relative flight abilities can dramatically affect the degree to which the landscape

structures them.

192

Given the selection of rock-crevice roosts, use of rivers as potential movement

corridors, and the previous work of Holloway (1998) that demonstrated bat concentration

in the river valleys of the prairies, riverine topography embedded in a relatively

inhospitable grasslands matrix seems to be the likely structuring force for bat distribution

in the prairies. However, the distribution of buildings on the landscape is also likely to

influence E. fuscus and M. lucifugus population structures. My analyses did not directly address buildings; I made the assumption that use of the grasslands matrix by bats would be facilitated by the presence of buildings for building-roosting species, and that buildings close to rivers could facilitate movement along rivers where natural roosts were uncommon. However, because buildings are patchily distributed just as natural rocky riparian areas are, determining the direct influence of buildings on seasonal movement for mating and on dispersal, may enhance our understanding of population structure in building-roosting species at the landscape level. Additional study design limitations

include the number of sampling locations per river. While more than two sites per river

was desirable and would have undoubtedly added strength to my conclusions, the logistics of capturing enough individuals of each species at each site to adequately determine allele frequencies limited the number of sampling locations. Similarly, inclusion of M. evotis, a species similar in size to M. lucifugus, but tending to roost exclusively in rocks in the prairies, would have been ideal to have included in the landscape level investigation, but limited sample sizes prevented its inclusion.

At the fine scale, it is not always apparent how one piece of the jigsaw puzzle adds to the overall picture; but once assembled, it is clear that the landscape level

perspective is invaluable not just for understanding a species’ ecology, but for

193

conservation and management of groups of species. If one does not understand how

animals move across and interact with the landscape, then effects of landscape changes cannot be predicted and long-term persistence of species cannot be fostered. In an ever- changing landscape, protection of critical habitat requires an understanding of land-use by wildlife on large and small spatial and temporal scales (Rolstad 2005). In the prairies of southeastern Alberta and north-central Montana where this research took place, buildings and dams were present, but less so here than in other areas of North America due to relatively sparse human population. As such, landscape genetics patterns deduced from my research likely reflect those from a somewhat naturally fragmented landscape.

That volant animals such as bats could be structured and fragmented by the prairie landscape on a relatively small scale suggests that other highly mobile animals could be similarly structured. How the prairie landscape will continue to be altered in the future is uncertain, but change is happening rapidly (Stelfox 2004). As riparian cottonwoods continue to disappear (Bradley and Smith 1986), wind-turbine developments (Arnett

2005) increase on the prairies, and drought puts additional pressure on governments to dam and divert rivers (Schindler and Donahue 2006), the prairie landscape is almost certain to change in a way that will influence many species, including bats. Because riparian areas are centres for biodiversity in the prairies (e.g. Savoy 1991), multi-species approaches to identify and conserve critical habitat in river basins are needed, and have already begun (e.g. MULTISAR, Quinlan et al. 2004, Landry 2004). This landscape- level habitat modeling requires that basic spatial and temporal habitat selection and behaviour be known for each species. I therefore hope that with the multidimensional understanding of prairie bat ecology that is slowly culminating from my work and

194 others’, it will be possible for bats to be considered in decisions of prairie river alterations, and overall prairie riparian conservation strategies.

195

LITERATURE CITED

Abramsky, Z., M.L. Rosenweig, and A. Subach. 1998. Do gerbils care more about competition or predation? Oikos 83:75-84.

Alberta Fish and Wildlife. 1994. Bats of Alberta: the real story. Alberta Environmental Protection and Alberta Agriculture, Food and Rural Development. Edmonton, AB. 16 pp.

Aldridge, H.D.J.N and R.M. Brigham. 1988. Load carrying and maneuverability in an insectivorous bat: a test of the 5% “rule” of radiotelemetry. Journal of Mammalogy 69:379-382.

Altringham, J.D. 1996. Bats - biology and behaviour. Oxford University Press, Boston, MA. 262 pp.

Anthony, E.L.P. 1988. Age determination in bats. In: Ecological and behavioral methods for the study of bats (T.H. Kunz, eds.). Smithsonian Institute Press, Washington, D.C., pp. 31–46.

Anthony, E.L.T., M.H. Stack, and T.H. Kunz. 1981. Night roosting and the nocturnal time budget of the little brown bat, Myotis lucifugus: effects of reproductive status, prey density, and environmental conditions. Oecologia 51:151-156.

Arnett, E.B. (Technical editor). 2005. Relationship between bats and wind turbines in Pennsylvania and West Virginia: an assessment of fatality search protocols, patterns of fatality, and behavioral interactions with wind turbines. A final report submitted to the Bats and Wind Energy Cooperative. Bat Conservation International. Austin, Texas. 187 pp.

ASRD. 2006. The general status of Alberta wild species 2005. Fish and Wildlife Service, Alberta Sustainable Resource Development, Edmonton, AB. Accessible at: http://srd.alberta.ca/fishwildlife/wildspecies/Species%20At%20Risk%202005.pdf

Audet, D. and M.B. Fenton. 1988. Heterothermy and the use of torpor by the bat Eptesicus fuscus (Chiroptera: Vespertilionidae): A field study. Physiological Zoology 61:197-204.

196

Avery, M.I. 1985. Winter activity of Pipistrelle bats. Journal of Animal Ecology. 54:721- 738.

Avise, J.C. 1994. Molecular markers, natural history and evolution. Chapman and Hall, New York. 511 pp.

Avise, J.C. 1996. Three fundamental contributions of molecular genetics to avian ecology and evolution. Ibis 138:16-25.

Baker, A.J., C.H. Daugherty, R. Colbourne, and J.L. McLennan. 1995. Flightless brown kiwis of New Zealand possess extremely subdivided population structure and cryptic species like small mammals. Proceedings of the National Academy of Sciences of the United States of America 92:8254-8258.

Baker, R.J. and R.D. Bradley. 2006a. Speciation in mammals and the genetic species concept. Journal of Mammalogy 87:643-662.

Baker, R.J. and R.D. Bradley. 2006b.A genetic definition of species: implications to bat biodiversity. Bat Research News 47:86-87.

Barbour, R.W. and W.H. Davis. 1969. Bats of America. University of Press Kentucky, Lexington. 286 pp.

Barbour, R.W. and W.H. Davis. 1970. The status of Myotis occultus. Journal of Mammalogy 51:150-151.

Barclay, R.M.R. and G. Bell. 1988. Marking and observational techniques. In: Ecological and behavioral methods for the study of bats (T.H. Kunz, ed.). Smithsonian Institution Press, Washington, D.C. pp. 59-76.

Barclay, R.M.R. and L. Harder. 2003. Life histories of bats: life in the slow lane. In: Bat Ecology (T.H. Kunz and M.B. Fenton, eds.), University of Chicago Press, Chicago, pp. 209-253.

Barclay, R.M.R., J. Ulmer, C.J.A. MacKenzie, M.S. Thompson, L. Olson, J. McCool, E. Cropley, and G. Poll. 2004. Variation in the reproductive rate of bats. Canadian Journal of Zoolgy 82:688-693.

197

Barratt, E.M., R. Deaville, T.M. Burland, M.W. Bruford, G. Jones, P.A Racey, and R.K. Wayne. 1997. DNA answers the call of pipistrelle bat species. Nature 387:128- 139.

Belkhir, K. 1999. GENETIX, v.4. 0. Laboratoire Génome, Populations, Interactions. CNRS UPR, Université Montpellier, France.

Betts, B. 1998. Effects of interindividual variation in echolocation calls on identification of big brown and silver-haired bats. Journal of Wildlife Management 62:1003- 1010.

Bilgin, R., A. Karatas, E. Coraman, I. Pandurski, E. Papadatou, and J.C. Morales. 2006. Molecular taxonomy and phylogeography of Miniopterus schreibersii (Kuhl, 1817) (Chiroptera: Vespertilionidae), in the Eurasian transition. Biological Journal of the Linnean Society 87:577-582.

Birky, C.W., P. Fuerst, and T. Maruyama. 1989. Organelle gene diversity under migration, mutation, and drift: equilibrium expectations, approach to equilibrium, effects of heteroplasmic cells, and comparison to nuclear genes. Genetics 121:613-627.

Birky, C.W., T. Maruyama, and P. Fuerst. 1983. An approach to population and evolutionary genetic theory for genes in mitochondria and chloroplasts, and some results. Genetics 103:513-527.

Bogdanowicz, W. 1990. Geographic variation and taxonomy of Daubenton’s bat, Myotis daubentoni, in Europe. Journal Mammalogy 7:205-218.

Brack, V., Jr. and J.W. Twente. 1985. The duration of the period of hibernation of three species of vespertilionid bats. I. Field studies. Canadian Journal of Zoology 63:2952-2954.

Bradley, C.E. and D.G. Smith. 1986. Plains cottonwood recruitment and survival on a prairie meandering river floodplain, Milk River, southern Alberta and northern Montana. Canadian Journal of Botany 64:1433-1442.

Brigham, R.M. 1987. The significance of winter activity by the big brown bat (Eptesicus fuscus): the influence of energy reserves. Canadian Journal of Zoology. 65:1240- 1242.

198

Brigham, R.M. 1991. Flexibility in foraging and roosting behaviour by the big brown bat (Eptesicus fuscus). Canadian Journal of Zoology 69:117-121.

Brigham, R.M., E.K.V. Kalko, G. Jones, S. Parson, and H.J.G.A. Limpsens. 2002.Bat echolocation research: tools, techniques and analysis. Bat Conservation International, Austin, TX. 167 pp.

Brigham, R.M., M.J. Vohnof, R.M.R. Barclay, and J.C. Gwilliam. 1997. Roosting behavior and roost-site selection of forest-dwelling bats (Myotis californicus). Journal of Mammalogy 78:1231-1239.

Brooks, R.T. and W.M. Ford. 2005. Bat activity in a forest landscape of Central Massachusetts. Northeastern Naturalist 12:447-462.

Burland, T.M. and J. Worthington Wilmer. 2001. Seeing in the dark: molecular approaches to the study of bat populations. Biological Review 76:389-409.

Burland, T.M., E.M Barratt, and P.A Racey. 1998. Isolation and characterization of microsatellite loci in the brown long-eared bat, Plecotus auritus, and cross-species amplification within the family Vespertilionidae. Molecular Ecology 7:133-140.

Burland, T.M., E.M. Barratt, M.A. Beaumont, and P.A. Racey. 1999. Population genetic structure and gene flow in a gleaning bat, Plecotus auritus. Proceedings of the Royal Society, London B 266:975-980.

Burland, T.M., E.M. Barratt, R.A. Nichols, and P.A. Racey. 2001. Mating patterns, relatedness and the basis of natal philopatry in the brown long-eared bat, Plecotus auritus. Molecular Ecology 10:1309-1321.

Burnett, C.D. 1983. Geographic and climatic correlates of morphological variation in Eptesicus fuscus. Journal Mammalogy 64:437-444.

Burton, R.S. and O.J. Reichman. 1999. Does immune challenge affect torpor duration? Functional Ecology 13:232-237.

Bushar, L.M., H.I. Rener, and L. Gelbert. 1998. Genetic variation and gene flow within and between local populations of the timber rattlesnake, Crotalus horridus. Copeia 2:411-422.

199

Castella, V. and M. Ruedi. 2000. Characterization of highly variable microsatellite loci in the bat Myotis myotis (Chiroptera: Vespertilionidae). Molecular Ecology 9:1000- 1002.

Castella, V., M. Ruedi, and L. Excoffier. 2001. Contrasted patterns of mitochondrial and nuclear structure among nursery colonies of the bat Myotis myotis. Journal of Evolutionary Biology 14:708-720.

Castella, V., M. Ruedi, L. Excoffier, C. Ibanez, R. Arlettaz, and J. Hausser. 2000. Is the Gibraltar Strait a barrier to gene flow for the bat Myotis myotis (Chiroptera: Vespertilionidae)? Molecular Ecology 9:1761-1772.

Caughley, G. and A.R.E. Sinclair. 1994. Wildlife ecology and management. Blackwell Science, Cambridge, Massachusetts. 334 pp.

Chapuis, M.P. and A. Estoup. 2007. Microsatellite null alleles and estimation of population differentiation. Molecular Biology and Evolution 24:621-631.

Chruszcz, B. 1999. Foraging and thermoregulatory behaviour of the long-eared bat (Myotis evotis) roosting in a natural habitat. MSc. Thesis, University of Calgary, Calgary, Alberta. 100 pp.

Chruszcz, B. and R.M.R. Barclay. 2003. Prolonged foraging bouts of a solitary gleaning/hawking bat, Myotis evotis. Canadian Journal of Zoology 81:823-826.

Chruszcz, B.J. and R.M.R. Barclay. 2002. Thermoregulation and roost selection by a solitary bat in southeastern Alberta, Canada. Functional Ecology 16:18-26.

Chung-MacCoubrey, A.L. 2005. Differences in roost selection and behavior among M. evotis, M. volans, and M. thysanodes in pinyon-juniper woodland and ponderosa pine forests of New Mexico. Western Bat Working Group Biennial Meeting, 31 March – 2 April 2005, Portland, Oregon.

Clare, E.L., B.K. Lim, M.D. Engstrom, J.L. Eger, and P.D.N. Hebert. 2007. DNA barcoding of Neotropical bats: species identification and discovery within Guyana. Molecular Ecology Notes 7:184-190.

Clark, B.K., B.S. Clark, D.M. Leslie, and M.S. Gregory. 1996. Characteristics of caves used by the endangered Ozark big-eared bat. Wildlife Society Bulletin 24:8-14.

200

Clegg, S.M., P. Hale and C. Moritz. 1998. Molecular population genetics of the red kangaroo (Macropus rufus): mtDNA variation. Molecular Ecology 7:679-686.

Clement, M., D. Posada, and K.A. Crandall. 2000. TCS: a computer program to estimate gene genealogies. Molecular Ecology 9:1657-1660.

Cooper, S.J.B., P.R. Day, T.B. Reardon, and M. Schulz. 2001. Assessment of species boundaries in Australian Myotis (Chiroptera: Vespertilionidae) using mitochondrial DNA. Journal of Mammalogy 82:328-338.

Cooper, S.J.B., T.B. Reardon, and J. Skilins. 1998. Molecular systematics of Australian rhinolophid bats (Chiroptera: Rhinolophidae). Australian Journal of Zoology 46:203-220.

Corben, C. 2002. Zero-crossings analysis for bat identification: an overview. In: Bat Echolocation Research: tools, techniques and analysis (R.M. Brigham, E.K.V. Kalko, G. Jones, S. Parsons, and H.J.G.A. Limpens, eds.). Bat Conservation International, Austin, Texas, pp. 95-107.

Cossins, A.R. and K. Bowler. 1987. Temperature biology of animals. Chapman and Hall, London. 339 pp.

Crockett, A.B. and H.H. Hadow. 1975. Nest site selection by Williamson and Red-naped Sapsuckers. Condor 77:365 – 368.

Cronin, M.A., S.C. Amstrup, G.W. Garner, and E.R. Vyse. 1991. Interspecific and intraspecific mitochondrial DNA variation in North American bears (Ursus). Canadian Journal of Zoology 69:2985-2992.

Crow, T.R. 2005. Landscape ecology and forest management. In: Issues and Perspectives in Landscape Ecology (J. Wiens and M. Moss, eds.), Cambridge University Press, Cambridge, pp. 201-207.

Cryan, P.M. 2003. Seasonal distribution of migratory tree bats (Lasiurus and Lasionycteris) in North America. Journal of Mammalogy 84:579-593.

Cryan, P.M., M.A. Bogan, R.O. Rye, G.P. Landis, and C.L. Kester. 2004. Stable hydrogen isotope analysis of bat hair as evidence for seasonal molt and migration. Journal of Mammalogy 85:995-1001.

201

Daan, S. 1973. Activity during natural hibernation in 3 species of Vespertilionidae bats. Netherlands Journal Zoology 23:1-71.

Davidson-Watts, I. and G. Jones. 2005. Differences in foraging behaviour between Pipistrellus pipistrellus (Schreber, 1774) and Pipistrellus pygmaeus (Leach, 1825). Journal of Zoology, London 268:55-62.

Davis, W.H. 1970. Hibernation: ecology and physiological ecology. In: Biology of bats (W.A. Wimsatt, ed.). Academic Press, New York, pp. 265-300.

Davis, W.H. and H.B. Hitchcock. 1965. Biology and migration of the bat, Myotis lucifugus, in New England. Journal of Mammalogy 46:296-313.

Dawson, N. 2006. Endemic mammals: conservation strategies for an island system (Endemic mammals on the Tongass). The Wildlife Society Annual Conference, Anchorage, AK, 23-27 Sept. 2006.

Demboski, J.R. and J.A. Cook. 2001. Phylogeography of the dusky shrew, Sorex monitcolus (Insectivora, Soricidae): insight into deep and shallow history in northwestern North America. Molecular Ecology 10:1227-1240.

Dewey, T.A. 2006. Systematics and phylogeography of North American Myotis (Chiroptera: Vespertilionidae). PhD Dissertation, University of Michigan, Ann Arbor, MI. 164 pp.

Edwards, J.G. 1954. A new approach to infraspecific categories. Systematic Zoology 3:1- 20.

Engelke D.R., A. Krikos, M.E. Bruck, and D. Ginsburg. 1990. Purification of Thermus aquaticus DNA polymerase expressed in Escherischia coli. Analytical Biochemistry 191:396–400.

Entwistle, A.C., P.A. Racey, and J.R. Speakman. 1997. Roost selection by the brown long-eared bat Plecotus auritus. Journal of Applied Ecology 34:399-408.

Entwistle, A.C., P.A. Racey, and J.R. Speakman. 2000. Social and population structure of a gleaning bat, Plecotus auritus. Journal of Zoology, London 252:11-17.

202

Environment Canada. 1997. The Canada Country Study: Climate Impacts and Adaptation. Canadian Prairies Summary. Ottawa, Ontario. 10 pp.

Environment Canada. 2000. Canadian Climate Normals 1971-2000, Empress, www.climate.weatheroffice.ec.gc.ca/climate_normals/. Accessed March 20, 2006.

Environment Canada. 2003. Terrestrial Ecozones of Canada. http://www. ec. gc. ca/soer- ree/English/Framework/NarDesc/canada_e. cfm. Accessed 18 January 2007.

Esberard, C.E.L., A.C. Jesus, A.G. Motts, H.G. Bergallo, and D. Bettinger. 2005. Hesperoctenes fumarius (Hemiptera: Polyctenidae) infesting Molossus rufus (Chiroptera: Molossidae) in southeastern Brazil. Journal of Parasitology 91:465- 467.

Excoffier, L., L. Guillaume, and S. Schneider. 2006. Arlequin version 3. 1: An integrated software package for population genetics data analysis. Computational and Molecular Population Genetics Lab, Institute of Zoology, University of Berne, Berne, Switzerland. 145 pp.

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

Faure, P.A. and R.M.R. Barclay. 1994. Substrate gleaning versus aerial hawking: plasticity in the foraging and echolocation behaviour of the long-eared bat, Myotis evotis. Journal of Comparative Physiology A 174:651-660.

Fenton, M.B. and G.P. Bell. 1981. Recognition of species of insectivorous bats by their echolocation calls. Journal of Mammalogy 62:233-243.

Fenton, M.B. and R.M.R. Barclay. 1980. Myotis lucifugus. Mammalian Species 142:1-8.

Findley, J.S. and C. Jones. 1964. Seasonal distribution of the Hoary bat. Journal of Mammalogy 45:461-470.

Findley, J.S., and C. Jones. 1967. Taxonomic relationships of bats of the species Myotis fortidens, M. lucifugus, and M. occultus. Journal of Mammalogy 48:429-444.

203

Fisher, N.I. 1993. Statistical analysis of circular data. Cambridge University Press, Cambridge. 277 pp.

Fitzsimmons, N., C. Moritz, C.J. Limpus, L. Pope, and R. Prince. 1997. Geographic structure of mitochondrial and nuclear gene polymorphisms in Australian green turtle populations and male-biased gene flow. Genetics 147:1843-1854.

French, A.R. 1988. The patterns of mammalian hibernation. American Scientist 76:569- 575.

Fumagalli, L., P. Taberlet, L. Favre, and J. Hausser. 1996. Origin and evolution of homologous repeated sequences in the mitochondrial DNA control region of shrews. Molecular Biology and Evolution 13:31-46.

Garcia, P.F.J., S.A. Rasheed, and S.L. Holroyd. 1995. Status of the western small-footed Myotis in British Columbia. Wildlife Working Report No. WR-74. 14 pp. Available at: http://wlapwww.gov.bc.ca/wld/documents/statusrpts/wr74.pdf.

Gauthier, D.A., A. Lafon, T.P. Toombs, J. Hoth, and E. Wiken. 2003. Grasslands: toward a North American conservation strategy. Canadian Plains Research Center University of Regina and Commission for Environmental Cooperation, Montreal, QC. 17 pp.

Gehrt, S.D. and J.E. Chelsvig. 2004. Species-specific patterns of bat activity in an urban landscape. Ecological Applications 14:625-635.

Gill, F.B., B. Slikas, and D. Agro. 1999. Speciation in North American chickadees: II. Geography of mtDNA haplotypes in Poecile carolinensis. The Auk 116:274-277.

Gittleman, J.L. and S.D. Thompson. 1988. Energy allocation in mammalian reproduction. American Zoologist 28:863-875.

Goss-Custard, J.D. and W.J. Sutherland. 1997. Individual behaviour, populations and conservation. In: Behavioural Ecology: An Evolutionary Approach, 4th edn. (J.R. Krebs, N.B. Davies, eds.). Blackwell Science, Oxford, pp. 373–395.

Goudet, J. 2001. FSTAT, a program to estimate and test gene diversities and fixation indices (version 2.9.3). Available from http://www.unil.ch/izea/softwares/fstat.html. Updated from Goudet (1995).

204

Greenwood, P.J. 1980. Mating systems, philopatry and dispersal in birds and mammals. Animal Behavior 28:1140-1162.

Griffin, D.R. 1970. Migration and homing of bats. In: Biology of bats (W.A. Wimsatt, ed.). Academic Press, New York, pp. 233–264.

Grinevitch, L., S.L. Holroyd, and R.M.R. Barclay. 1995. Sex differences in the use of daily torpor and foraging time by big brown bats (Eptesicus fuscus) during the reproductive season. Journal of Zoology, London 235:301-309.

Hall, D., M.J. O’Farrell, and R.G. Peppard. 2005. Novel techniques to improve long-term monitoring of bats in the Nevada test site, south-central Nevada. Western Bat Working Group Biennial Conference, Portland, OR, 30 March – 2 April, 2005.

Hall, E.R. 1981. The mammals of North America. Vol. 1. John Wiley and Sons. 600 pp.

Hall, E.R., and K.R. Kelson. 1959. The mammals of North America. Vol. 1 Ronald Press Company, New York. 546 pp.

Hamilton, I. and R.M.R. Barclay. 1994. Patterns of daily torpor and day-roost selection by male and female big brown bats (Eptesicus fuscus). Canadian Journal of Zoology 72:744-749.

Hanski, I. and D. Simberloff. 1997. The metapopulation approach, its history, conceptual domain, and application to conservation. In: Metapopulation Biology: Ecology, genetics and evolution (I.A. Hanski and M.E. Gilpin, eds.). Academic Press, London, pp. 5-26.

Hardin, J.W. and J.M. Hilbe. 2007. Generalized linear models and extensions, 2nd StataCorp LP, College Station, TX. 387 pp.

Hardy, O.J. and X. Vekemans. 2002. SPAGeDi: a versatile computer program to analyze spatial genetic structure at the individual or population levels. Molecular Ecology Notes 2:618-620.

Harris, A.H. 1974. Myotis yumanensis in interior southwestern North America, with comments on Myotis lucifugus. Journal of Mammalogy 55:589-607.

205

Hartl, D. and A. Clark. 1997. Principles of population genetics, 3rd edition. Sinauer Associates, Sunderland, Massachusetts. 542 pp.

Hays, G.C., J.R. Speakman, and P.I. Webb. 1992. Why do brown long-eared bats (Plecotus auritus) fly in winter? Physiological Zoology 65:554-565.

Heavner, M., C. Talus, and A. Hall. 2006. Acoustic monitoring of southeast Alaska bats. Bat Research News 47:109.

Hebert, P.D.N, A. Cywinska, S.L. Ball, and J.R. deWaard. 2003. Biological identifications through DNA barcodes. Proceedings of the Royal Society of London. Series B, Biological Sciences 270:313-322.

Hoelzel, A.R. 1998. Genetic structure of cetacean populations in sympatry, parapatry, and mixed assemblages: Implications for conservation policy. Journal of Heredity 89:451-458.

Holland, R.A., K. Thorup, M.J. Vonhof, W.W. Cochran, and M. Wikelski. 2006. Bat orientation using Earth’s magnetic field. Nature 444:702.

Hollister, N. 1909. Two new bats from the south western U. S. Proceedings of the Biological Society of Washington 22:43-44.

Hollister, N. 1911. Two new bats from the south western U.S. Proceedings of the Biological Society of Washington 22:43-44.

Holloway, G.L. 1998. The ecology of prairie-dwelling bats in Southeastern Alberta. MSc. Thesis, University of Calgary, Alberta. 100 pp.

Holloway, G.L. and R.M.R. Barclay. 2000. Importance of prairie riparian zones to bats in southeastern Alberta. Ecoscience 7:115-122.

Holloway, G.L. and R.M.R. Barclay. 2001. Myotis ciliolabrum. Mammalian Species 670: 1-5.

Holroyd, S.L. 1993. Influences of some extrinsic and intrinsic factors on reproduction by big brown bats (Eptesicus fuscus) in southeastern Alberta. MSc. Thesis,

206

Department of Biological Sciences, University of Calgary, Calgary, Alberta. 126 pp.

Honeycutt, R. L., I.F. Greenbaum, R.J. Baker, V.M. Sarich. 1981. Molecular evolution of vampire bats. Journal of Mammalogy 62:805-811.

Hosmer, D.W. and S. Lemeshow. 2000. Applied logistic regression, 2nd edn. John Wiley and Sons, Inc., New York. 375 pp.

Humphries, M.M., D.W. Thomas, and D.L. Kramer. 2001. Torpor and digestion in food- storing hibernators. Physiological and. Biochemical Zoology 74:283-292.

Humphries, M.M., D.W. Thomas, and J.R. Speakman, 2002. Climate-mediated energetic constraints on the distribution of hibernating mammals. Nature 418:313-316.

Jacobs, D.S., G.N. Eick, E.J. Richardson, and P.J. Taylor. 2004. Genetic similarity amongst phenotypically diverse little free-tailed bats, Chaerephon pumilus. Acta Chiropterologica 6:13-21.

Jacobs, D.S., G.N. Eick, M.C. Schoeman, and C.A. Matthee. 2006. Cryptic species in an insectivorous bat, Scotophilus dinganii. Journal of Mammalogy 87:161-170.

Jones, J. 2001. Habitat selection studies in avian ecology: a critical review. Auk 118:557- 562.

Kalinowski, S.T., A.P. Wagner, and M.L. Taper. 2006. ML-Relate: a computer program for maximum likelihood estimation of relatedness and relationship. Molecular Ecology Notes 6:576-579.

Kallen, F.C. 1964. Some aspects of water balance in hibernating bats. Annales Academiæ Scientiarum Fennicæ of Science (Series A IV) 71:257-268.

Kawai, K., M. Nikaido, M. Harada, S. Matsumura, L-K. Lin, Y. Wu, M. Hasegawa, and N. Okada. 2003. The status of the Japanese and East Asian bats of the genus Myotis (Vespertilionidae) based on mitochondrial sequences. Molecular Phylogenetics and Evolution 28:297-307.

207

Kayser, C. 1965. Hibernation. In: Physiological Mammalogy. II. Mammalian reactions to stressful environments. (W. Mayer and R. van Gelder, eds.). Academic Press, New York. pp. 179-296.

Kerth, G. 2006. Relatedness, life history, and social behavior in the long-lived Bechstein’s bat, Myotis bechsteinii. In: Functional and Evolutionary Ecology of Bats (A. Zubaid, G.F. McCracken and T.H. Kunz, eds.). Oxford University Press, Oxford. pp. 199-212.

Kerth, G. and B. König. 1999. Fission, fusion and nonrandom associations in female Bechstein’s bats (Myotis bechsteinii). Behaviour 136:1187-1202.

Kerth, G., K. Safi, and B. König. 2002. Mean colony relatedness is a poor predictor of colony structure and female philopatry in the communally breeding Bechstein’s bat (Myotis bechsteinii ). Behavioral Ecology and Sociobiology. 52:203–210.

Kerth, G., M. Wagner, and B. König. 2001. Roosting together, foraging apart: information transfer about food is unlikely to explain sociality in female Bechstein’s bats (Myotis bechsteinii). Behavioral Ecology and Sociobiology 50:283-291.

Knight, M.E., M.J.H. van Oppen, H.L. Smith, C. Rico, G.M. Hewitt, and G.F. Turner. 1999. Evidence for male-biased dispersal in Lake Malawi cichlids from microsatellites. Molecular Ecology 8:1521-1527.

Kruskop, S. 2004. Subspecific structure of Myotis daubentonii (Chiroptera, Vespertilionidae) and composition of the “daubentonii” species group. Mammalia 68:299-306.

Krzanowski, A. 1959. Some major aspects of population turnover in wintering bats in the cave at Pulawy (Poland). Acta Theriologica 3:27-43.

Kunz, T.H. 1974. Feeding ecology of a temperate insectivorous bat (Myotis velifer). Ecology 55:693-711.

Kunz, T.H. 1982. Roosting ecology of bats. In: Ecology of Bats. (T.H. Kunz, ed.). Plenum Press, New York, pp. 1-55.

208

Kunz, T.H. and M.B. Fenton. 2003. Bat Ecology. University of Chicago Press, Chicago. 779 pp.

Kunz, T.H., J.A. Wrazen, and C.D. Burnett. 1998. Changes in body mass and fat reserves in pre-hibernating little brown bats (Myotis lucifugus). Ecoscience 5:8-17.

Kurta, A. and R. Baker. 1990. Eptesicus fuscus. Mammalian Species 356:1–10.

Kurta, A. and S.W. Murray. 2002. Philopatry and migration of banded Indiana bats (Myotis sodalis) and effects of radio transmitters. Journal of Mammalogy 83:585- 589.

Kurta, A., G. Bell, K. Nagy, and T.H. Kunz. 1989a. Energetics of pregnancy and lactation in free-ranging little brown bats (Myotis lucifugus). Physiological Zoology 62:804-818.

Kurta, A., G. Bell, K. Nagy, and T.H. Kunz. 1989b. Water balance of free-ranging little brown bats (Myotis lucifugus) during pregnancy and lactation. Canadian Journal of Zoology 67:2468-2472.

Kurta, A., K.J. Williams, and R. Mies. 1996. Ecological, behavioural and thermal observations of a peripheral population of Indiana bats (Myotis sodalis). In: Bats and Forests Symposium, Oct. 19-21, 1995. (R.M.R. Barclay and R.M. Brigham, eds.). Victoria, B.C., Canada. British Columbia Ministry of Forests Research Program, pp. 102-117.

Landry, J. 2004. Western small-footed Myotis (Myotis ciliolabrum ciliolabrum). In: MULTISAR: The Milk River Basin Project. Habitat suitability models for selected wildlife management species (Downey, B.A., B.L. Downey, R.W. Quinlan, O. Castelli, V.J. Remesz and P.F. Jones, eds.). Alberta Sustainable Resource Management, Fish and Wildlife Division, Species at Risk Report No. 86. Edmonton, AB, pp. 90-97.

Larson, R.J. 1980. Competition, habitat selection and the bathymetric segregation of 2 rockfish Sebastes species. Ecological Monographs 50:221-240.

Latch, E.K., G. Dharmarajan, J.C Glaubitz, and O.E. Rhodes, Jr. 2006. Relative performance of Bayesian clustering software for inferring population substructure and individual assignment at low levels of population differentiation. Conservation Genetics 7:295-302.

209

Lausen, C. L. and R.M.R. Barclay. 2002. Roosting behaviour and roost selection of female big brown bats (Eptesicus fuscus) roosting in rock crevices in southeastern Alberta. Canadian Journal of Zoology 80:1069–1076.

Lausen, C. L. and R.M.R. Barclay. 2003. Thermoregulation and roost selection by reproductive female big brown bats (Eptesicus fuscus) roosting in rock crevices. Journal of Zoology (London) 260:235-244.

Lausen, C. L. and R.M.R. Barclay. 2006a. The benefits of living in a building: comparing maternity colonies of big brown bats (Eptesicus fuscus) in natural and building roosts. Journal of Mammalogy 87:362-370.

Lausen, C. L. and R.M.R. Barclay. 2006b. Winter bat activity in the Canadian prairies. Canadian Journal of Zoology 84:1079-1086.

Lausen, C., E. Baerwald, J. Gruver, and R. Barclay. 2006. APPENDIX 5. Bats and Wind Turbines. Pre-siting and pre-construction survey protocols. In: 2005. Handbook of Inventory Methods and Standard Protocols for Surveying Bats in Alberta (M. Vonhof). Alberta Sustainable Resource Development, Fish and Wildlife Division, Edmonton, Alberta. (2005).

Lausen, C.L. 2001. Thermoregulation and roost selection by reproductive big brown bats (Eptesicus fuscus) roosting in the South Saskatchewan River Valley, Alberta: rock-roosting and building-roosting colonies. MSc. Thesis, University of Calgary, Calgary, Alberta. 119 pp.

Lausen, C.L. 2005. First record of hosts for the bat tick, Carios kelleyi (Acari: Ixodida: Argasidae) in Canada and Montana. Journal of Medical Entomology 42:497-501.

Lausen, C.L. 2006. 2005-2006 Bat Survey of the Middle Red Deer and Battle Rivers. Alberta Natural Heritage Information Centre, Parks and Protected Areas Division. Edmonton, AB. 22 pp.

Law, B.S. and M. Chidel. 2007. Bats under a hot tin roof: comparing the microclimate of eastern cave bat (Vespadelus troughtoni) roosts in a shed and cave overhang. Journal of Zoology, London 55:49-55.

Lewis, S. 1993. Effect of climatic variation on reproduction by pallid bats (Antrozous pallidus). Canadian Journal of Zoology 71:1429-1433.

210

Lewis, S. 1995. Roost fidelity of bats: A review. Journal of Mammalogy 76: 481-496.

Lewis, S. 1996. Low roost site fidelity in pallid bats: associated factors and the effect on group stability. Behavioral Ecology and Sociobiology 39:335-344.

Limpens, H.J.G.A. and K. Kapteyn. 1991. Bats, their behaviour and linear landscape elements. Myotis 29:39-47.

Lippert, H. 2001. Habitat use and roosting ecology of bats in boreal forests of Alberta in response to forest harvest. MSc. Thesis, University of Alberta, Edmonton, AB. 86 pp.

Lloyd, B.D. 2003. Intraspecific phylogeny of the New Zealand short-tailed bat Mystacina tuberculata inferred from multiple mitochondrial gene sequences. Systematic Biology 52:460–476.

Lowe, A., S. Harris, and P. Ashton. 2004. Ecological genetics: design, analysis and application. Blackwell Publishing, Oxford. 326 pp.

Ludwig, J.A. 2005. Disturbances and landscapes: the little things count. In: Issues and perspectives in landscape ecology (J. Wiens and M. Moss, eds.). Cambridge University Press, NY, pp. 42-51.

Lumsden, L.F. and A.F. Bennett. 2006. Flexibility and specificity in the roosting ecology of the lesser long-eared bat, Nyctophilus geoffroyi: a common and widespread Australian species In: Functional and Evolutionary Ecology of Bats (A. Zubaid, G.F. McCracken and T.H. Kunz, eds.). Oxford University Press, Oxford, pp. 290- 307.

Manel, S., M.K. Schwartz, G. Luikart, and P. Taberlet. 2003. Landscape genetics: combining landscape ecology and population genetics. Trends in Ecology and Evolution 18:189-197.

Manly, B.F.J., L.L. McDonald, D.L. Thomas, T.L. McDonald, and W.P. Erikson. 2002. Resource selection by animals: Statistical design and analysis for field studies, 2nd edn. Kluwer Academic Publishers, London. 221 pp.

Manning, W. and J.K. Jones, Jr. 1989. Myotis evotis. Mammalian Speces 329:1-5.

211

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

Mayer, F. and O. von Helversen. 2001. Cryptic diversity in European bats. Proceedings of the Royal Society, London 268:1825-1832.

Mayr, E. 1963. Animal species and evolution. Harvard University Press, Cambridge, MA. 811 pp.

McCracken, G. F. 1984. Social dispersion and genetic variation in two species of emballonurid bats. Zeitschrift fur Tierpsychologie 66:55-69.

McCracken, G.F. and M.F. Gassel. 1997. Genetic structure in migratory and nonmigratory populations of Brazilian free-tailed bats. Journal of Mammalogy 78:348-357.

McCracken, G.F., L.F. Lumsden, and T.H. Kunz. 2006. Roosting ecology and population biology. In: Functional and evolutionary ecology of bats (A. Zubaid, G. F. McCracken and T. H. Kunz, eds.). Oxford University Press, Oxford, pp. 179-184.

McCracken, G.F., M.K. McCracken, and A.T. Vawter. 1994. Genetic structure in migratory populations of the bat Tadarida brasiliensis mexicana. Journal of Mammalogy 75:500-514.

McNab, B.K. 1982. Evolutionary alternatives in the physiological ecology of bats. In: Ecology of bats. (T.H. Kunz, ed.). Plenum Press, New York, pp. 151-200.

Metheny, J. D. and M.C. Rüppell. 2006. A genetic analysis of the fission-fusion roosting behavior of tree-roosting maternity colonies of big brown bats (Eptesicus fuscus). Bat Research News 47:127.

Miller, G.S., and G.M. Allen. 1928. The American bats of the genera Myotis and Pizonyx. Bulletin of the U. S. National Museum 144:1-218.

Miller-Butterworth, C.M., D.S. Jacobs, and E.H. Harley. 2003. Strong population substructure is correlated with morphology and ecology in a migratory bat. Nature 424:187-191.

212

Mills, R.S., G.W. Barrett, and M.P. Farrell. 1975. Population dynamics of the big brown bat (Eptesicus fuscus) in southwestern Ohio. Journal of Mammalogy 56:591-604.

Nagorsen, D.W. and R.M. Brigham.1993. The bats of British Columbia. Volume 1: The Mammals of British Columbia. Royal B.C. Museum, Victoria, B.C. and UBC Press, Vancouver, B.C. 165 pp.

Navo, K.W., S.G. Henery, and T.E. Ingersoll. 2002. Observations of swarming by bats and band recoveries in Colorado. Western North American Naturalist 62:124-126.

Nei, M. 1972. Genetic distance between populations. American Naturalist 106:283-292.

Neubaum, D., T.J. O’Shea, and K.R. Wilson. 2006. Autumn migration and selection of rock crevices as hibernacula by big brown bats in Colorado. Journal of Mammaology 87:470-479.

Norberg, U. 1994. Wing design, flight morphology, and habitat use in bats. In: Ecological Morphology (P. C. Wainright and S. M. Reilly, eds.). Chicago University Press, Chicago, pp. 205-239.

Norberg, U.M. and J.M.V. Rayner. 1987. Ecological morphology and flight in bats (Mammalia; Chiroptera): wing adaptations, flight performance, foraging strategy and echolocation. Philosophical Transactions of the Royal Society of London B. 316:335-427.

O’Donnell, F.J. and J.A. Sedgeley. 2006. Causes and consequences of tree-cavity roosting in a temperate bat, Chalinolobus turberculatus, New Zealand. In: Functional and evolutionary ecology of bats (A. Zubaid, G.F. McCracken and T.H. Kunz, eds.). Oxford University Press, Oxford, pp. 308-328.

O’Farrell, M.J. 1998. A passive monitoring system for Anabat II using a laptop computer. Bat Research News 39:147-150.

O’Farrell, M.J., B.W. Miller, and W.L. Gannon. 1999. Qualitative identification of free- flying bats using the Anabat detector. Journal of Mammalogy 8:11-23.

O'Farrell, M.J. and W.G. Bradley. 1970. Activity patterns of bats over a desert spring. Journal of Mammalogy 51:18-26.

213

Owen, R.D. 1988. Phenetic analyses of the bat subfamily Stenodermatinae Chiroptera Phyllostomidae. Journal of Mammalogy 69:795-810.

Paetkau, D., G.F. Shields, and C. Strobeck. 1998. Gene flow between insular, coastal, and interior populations of brown bears in Alaska. Molecular Ecology 7:1283-1292.

Park, J.K., G. Jones, and R.D. Ransome. 1999. Winter activity of a population of greater horseshoe bats (Rhinolophus ferrumequinum). Journal of Zoology (London) 248: 419-427.

Patriquin, K.J. 2001. Ecology of a bat community in harvested boreal forest in northwestern Alberta. MSc. Thesis. University of Calgary, Calgary, Canada. 104 pp.

Peterson, A.T. and L.R. Heaney. 1993. Genetic differentiation in Philippine bats of the genera ynopterus and Haplonycteris. Biological Journal of the Linnean Society 49:203-218.

Petit, E. and R. Mayer. 1999. Male dispersal in the noctule bat (Nyctalus noctula): where are the limits? Proceedings of the Royal Society of London 266:1717-1722.

Petit, E., F. Balloux, and J. Goudet. 2001. Sex-biased dispersal in a migratory bat: a characterization using sex-specific demographic parameters. Evolution 55:635- 640.

Petri, B., S. Pääbo, A.V. Haeseler, and D. Tautz. 1997. Paternity assessment and population subdivision in a natural population of the large mouse-eared bat, Myotis myotis. Molecular Ecology 6:235-242.

Piaggio, J.A., E.W. Valdez, M.A. Bogan, and G.S. Spicer. 2002. Systematics of Myotis occultus (Chiroptera: Vespertilionidae) inferred from sequences of two mitochondrial genes. Journal of Mammalogy 83:386-395.

Piry, S., A. Alapetite, J.-M. Cornuet, D. Paetkau, L. Baudouin, and A. Estoup. 2004. Geneclass2: A software for genetic assignment and first-generation migrant detection. Journal of Heredity 95:536-539.

Popov, V.I., L.S. Bocharova, and A.G. Bragin. 1992. Repeated changes of dendritic morphology in the hippocampus of ground squirrels in the course of hibernation. Neuroscience 48:45-51.

214

Prendini, L. 2005. Comment on “Identifying spiders through DNA barcodes”. Canadian Journal of Zoology 83:498-504.

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

Proctor, M.F., B.N. McLellan, C. Strobeck, and R.M.R.B. Barclay. 2005. Genetic analysis reveals demographic fragmentation of grizzly bears yielding vulnerably small populations. Proceedings of the Royal Society, Biological Sciences Series B 272:2409-2416.

Prothero, J. and K.D. Jurgens. 1986. An energetic model of daily torpor in endotherms. Journal of Theoretical Biology 121:403-415.

Prugnolle, F. and T. de Meeus. 2002. Inferring sex-biased dispersal from population genetic tools: a review. Heredity 88:161-165.

Quinlan, R., P. Jones, B. Downey, B. Downey, T. Clayton, and B. Taylor. 2004. MULTISAR: a conservation strategy for species at risk in the Milk River Basin. In: Proceedings of the 7th Prairie Conservation and Endangered Species Conference (G.C. Trottier, E. Anderson, M. Steinhilber, eds.). Natural History Occasional Paper No. 26, Provincial Museum of Alberta, Edmonton, pp. 37-40.

Racey, P.A. 1973. Environmental factors affecting the length of gestation in heterothermic bats. Journal of Reproduction and Fertility, Supplemental 19:175- 189.

Racey, P.A. 1988. Reproductive assessment in bats. In: Ecological and behavioral methods for the study of bats (T.H. Kunz, ed.). Smithsonian Institute Press, Washington, D. C., pp. 31-46.

Racey, P.A. and S.M. Swift. 1985. Feeding ecology of Pipistrellus pipistrellus (Chiroptera: Vespertilionidae) during pregnancy and lactation. I. Foraging behaviour. Journal of Animal Ecology 54:205-215.

Rajan, K.E. and G. Marimuthu. 2000. Genetic diversity within and among populations of the microchiropteran bat Hipposideros speoris based on a RAPD analysis. Zeitschrift fur Saugetierkunde 65:301-306.

215

Ratcliffe, J.M. and J.W. Dawson. 2003. Behavioural flexibility: the little brown bat, Myotis lucifugus, and the northern long-eared bat, M. septentrionalis, both glean and hawk prey. Animal Behaviour 66:847–856

Raymond, M. and F. Rousset. 1995. GENEPOP (version 1. 2): population genetics software for exact tests and ecumenicism. Journal of Heredity 86:248-249.

Reynolds, D.S. and T.H. Kunz. 2000. Changes in body composition during reproduction and postnatal growth in the little brown bat Myotis lucifugus (Chiroptera: Vespertilionidae). Ecoscience 7:10-17.

Rhodes, T.R. 1991. The importance of riparian cottonwoods to wildlife. In: The Biology and Management of Southern Alberta’s Cottonwoods (S.B. Rood and J.M. Mahoney, eds.). University of Lethbridge, Alberta, pp. 31-41.

Ricklefs, R.E. 1990. Ecology, 3rd edn. W.H. Freeman and Company, New York, NY. 896 pp.

Rising, J.D. and J.C. Avise. 1993. Application of genealogical-concordance principles to the taxonomy and evolutionary history of the sharp-tailed sparrow (Ammodramus caudacutus). The Auk 110:844-856.

Roach J.L., P. Stapp, B. Van Horne, and M.F. Antolin. 2001. Genetic structure of a metapopulation of black-tailed prairie dogs. Journal of Mammalogy 82:946-959.

Rolstad, J. 2005. Landscape ecology and wildlife management. In: Issues and Perspectives in Landscape Ecology (J. Wiens and M. Moss, eds.), Cambridge University Press, Cambridge, pp. 208-216.

Rosenzweig, M.L. 1981. A theory of habitat selection. Ecology 62:327-335.

Rossiter, S.J., G. Jones, R.D. Ransome, and E.M. Barratt. 2000. Genetic variation and population structure in the endangered greater horseshoe bat Rhinolophus ferrumequinum. Molecular Ecology 9:1131-1135.

Rossiter, S.J., G. Jones, R.D. Ransome, and E.M. Barratt. 2002. Relatedness structure and kin-biased foraging in the greater horseshoe bat (Rhinolophus ferrumequinum). Behavioral Ecology and Sociobiology 51:510-518.

216

Roverud, R.C. and M.A. Chappell. 1991. Energetic and thermoregulatory aspects of clustering behavior in the neotropical bat Noctilio albiventris. Physiological Zoology 64:1527-1541.

Rubinoff, D., S.C. Cameron, and K. Will. 2006. A genomic perspective on the shortcomings of mitochondrial DNA for “Barcoding” identification. Journal of Heredity 97:581-594.

Ruedi, M. and F. Mayer. 2001. Molecular systematics of bats of the genus Myotis (Vespertilionidae) suggests deterministic ecomorphological convergences. Molecular Phylogenetics and Evolution 21:436-448.

Rydell, J. 1993. Variation in foraging activity of an aerial insectivorous bat during reproduction. Journal of Mammalogy 74:503-509.

Rysgaard, G.N. 1942. A study of the cave bats of Minnesota with especial reference to the large brown bat, Eptesicus fuscus fuscus (Beauvois). American Midland Naturalist 28:245-267.

Sasse, D.B. and P.J. Pekins. 1996. Summer roosting ecology of northern long-eared bats (Myotis septentrionalis) in the White Mountain National Forest. In: Bats and Forests Symposium, Oct. 19-21, 1995. (R.M.R. Barclay and R.M. Brigham, eds.). British Columbia Ministry of Forests Research Program, Victoria, B.C., Canada, pp. 91-101.

Savoy, E. 1991. The importance of riparian forests for prairie birds: A case study: Dinosaur Provincial Park, Alberta. In: The Biology and Management of Southern Alberta’s Cottonwoods (S.B. Rood and J.M. Mahoney, eds). University of Lethbridge, Alberta, pp. 91-97.

Schindler, D.W. and W.F. Donahue. 2006. An impending water crisis in Canada’s western prairie provinces. Proceedings of the National Academy of Science, U.S.A. 103:7210-7216.

Schmidt-Nielsen, K. 1990. Temperature Regulation. In: Animal Physiology - Adaptation and Environment, 4th ed. (K. Schmidt-Nielsen, ed.). Cambridge University Press, Cambridge, pp. 240-295.

Schofield, P.J. 2003. Habitat selection of two gobies (Microgobius gulosus, Gobiosoma robustum): influence of structural complexity, competitive interactions, and

217

presence of a predator. Journal of Experimental Marine Biology and Ecology 288: 125-137.

Schowalter, D.B. 1980. Swarming, reproduction, and early hibernation of Myotis lucifugus and M. volans in Alberta, Canada. Journal Mammalogy 61:347-350.

Schowalter, D.B. and A. Allen. 1981. Late summer activity of small-footed, long-eared and big brown bats in Dinosaur Park, Alberta. Blue Jay 39:50-53.

Schowalter, D.B., J.R. Gunson, and L.D. Harder. 1979. Life history characteristics of little brown bats (Myotis lucifugus) in Alberta. Canadian Field-Naturalist 93:243- 251.

Schwartz, C.C., S.D. Miller, and M.A. Haroldson. 2003. Grizzly bear. In: Wild Mammals of North America (G.A. Feldhammer, B.C. Thompson, and J.A. Chapman, eds.) John Hopkins University Press, Baltimore, MD, pp 556-586.

Sendor, T., K. Kugelschafter, and M. Simon. 2000. Seasonal variation of activity patterns at a pipistrelle (Pipistrellus pipistrellus) hibernaculum. Myotis 38:91-109.

Shaw, K.L. 2002. Conflict between nuclear and mitochondrial DNA phylogenies of a recent species radiation: what mtDNA reveals and conceals about modes of speciation in Hawaiian crickets. Proceedings of the National Academy of Science 99:16122-16127.

Siemers, B.M., E.K. V. Kalko, and H.U. Schnitzler. 2001. Echolocation behavior and signal plasticity in the Neotropical bat Myotis nigricans (Schinz, 1821) (Vespertilionidae): a convergent case with European species of Pipistrellus? Behavioral Ecology and Sociobiology 50:317-328.

Simonsen, B.T., H.R. Siegismund, and P. Arctander. 1998. Population structure of African buffalo inferred from mtDNA sequences and microsatellite loci: High variation but low differentiation. Molecular Ecology 7: 225-237.

Sinclair, E.A., N.J. Webb, A.D. Marchant, and C.R. Tidemann, C. R. 1996. Genetic variation in the little red flying-fox, Pteropus scapulatus (Chiroptera: Pteropodidae). Biological Conservation 76:45-50.

218

Slatkin, M. 1995. A measure of population subdivision based on microsatellite allele frequencies. Genetics 139:457–462.

Smith, H.C. 1993. Alberta mammals: An atlas and guide. The Provincial Museum of Alberta, Edmonton, Alberta. 238 pp.

Smith, H.C. and D.B. Schowalter. 1979. A new subspecies of little brown bat for Alberta. Blue Jay 37:58-62.

Sokal, R.R. and F.J. Rohlf. 1995. Biometry - The principles and practice of statistics in biological research. W.H. Freeman and Company, New York. 887 pp.

Solick, D.I. 2004. Differences in the morphology and behaviour of western long-eared bats (Myotis evotis) within and between environments. MSc. Thesis, University of Calgary, Calgary, AB. 120 pp.

Speakman, J.R. and P.A. Racey. 1987. The energetics of pregnancy and lactation in the brown long-eared bat, Plecotus auritus. In: Recent advances in the study of bats (M.B. Fenton, P.A. Racey, and J.M.V. Rayner, eds.). Cambridge University Press, Cambridge, United Kingdom, pp. 376-393.

Speakman, J.R. and P.A. Racey. 1989. Hibernal ecology of the pipistrelle bat: energy expenditure, water requirements and mass loss, implications for survival and the function of winter emergence flights. Journal Animal Ecology 58:797-813.

Stadelmann, B., L.-K. Lin, T.H. Kunz, and M. Ruedi. 2007. Molecular phylogeny of New World Myotis (Chiroptera, Vespertilionidae) inferred from mitochondrial and nuclear DNA genes. Molecular Phylogenetics and Evolution 43:32-48.

Stelfox, B. 2004. Cumulative effects of landuses in the prairie ecosystems of Alberta. In: Proceedings of the 7th Prairie Conservation and Endangered Species Conference (G.C. Trottier, E. Anderson, and M. Steinhilber, eds.). Natural History Occasional Paper No. 26, Provincial Museum of Alberta, Edmonton, p. 159.

Studier, E.H. 1970. Evaporative water loss in bats. Physiological and Biochemical Zoology 35:935-943.

Swanson, G. and C. Evans. 1936. The hibernation of certain bats in southern Minnesota. Journal of Mammalogy 17:39-43.

219

Swofford, D.L. 2002. PAUP*. Phylogenetic analysis using parsimony (*and other methods). Version 4.0 Beta. Sinauer Associates, Sunderland, MA.

Tabachnick, B.G. and L.S. Fidell. 2001. Using multivariate statistics, 4th edition. Allyn and Bacon, Boston, MA. 516 pp.

Talbot, S.L. and G.F. Shields. 1996. Phylogeography of brown bears (Ursus arctos) of Alaska and paraphyly within the Ursidae. Molecular Phylogenetics and Evolution 5:477-494.

Tamura, K. and M. Nei. 1993. Estimation of the number of nucleotide substitutions in the control region of mitochondrial DNA in humans and chimpanzees. Molecular Biology and Evolution 10:512-526.

Taylor, L.R. 1963. Analysis of the effect of temperature on insects and flight. Journal of Animal Ecology 32:99-117.

Thomas, D.W. 1995. The physiological ecology of hibernation in vespertionid bats. Symposium of the Zoological. Society, London, No. 67:233-244.

Thomas, D.W. and D. Cloutier. 1992. Evaporative water loss by hibernating little brown bats, Myotis lucifugus. Physiological Zoology 65:443-456.

Thomas, D.W. and F. Geiser. 1997. Periodic arousals in hibernating mammals: is evaporative water loss involved? Functional Ecology 11:585-591.

Thomas, D.W. and J.R. Speakman. 2006. Physiological Ecology. In Functional and Evolutionary Ecology of Bats (A. Zubaid, G.F. McCracken and T.H. Kunz, eds.). Oxford University Press, Oxford, pp. 1-4.

Thomas, D.W., M. Dorais, and J.M. Bergerson. 1990. Winter energy budgets and cost of arousals for hibernating little brown bats, Myotis lucifugus. Journal of Mammalogy 71:475-479.

Thomas, O. 1904. A new bat from the United States representing the European Myotis (Leuconoe) Daubentoni. Annals and Magazine of Natural History. Series 7, 13:382-384.

220

Trachsel, L., D.M. Edgar, and H.C. Heller. 1991. Are ground squirrels sleep deprived during hibernation? American. Journal of Physiology 260:R1123-R1129.

Trottier, G.C., E. Anderson, and M. Steinhilber. 2004. Proceedings of the 7th Prairie Conservation and Endangered Species Conference. Natural History Occasional Paper No. 26, The Provincial Museum of Alberta, Edmonton, AB, 258 pp.

Tuttle, M. and L.R. Heaney. 1974. Maternity habits of Myotis leibii in South Dakota. Southern California Academy of Science Bulletin 73:80-83.

Twente, J.W., J. Twente, and V. Brack, Jr. 1985. The duration of the period of hibernation of three species of vespertilionid bats. II. Laboratory studies. Canadian Journal of Zoology 63:2955-2961.

Valdez, E.W., J.R. Choate, M.A. Bogan, and T.L. Yates. 1999. Taxonomic status of Myotis occultus. Journal of Mammalogy 80:545-552.

Van Oosterhout, C., W.F. Hutchinson, D.P.M. Wills, and P. Shipley. 2004. Micro- checker: software for identifying and correcting genotyping errors in microsatellite data. Molecular Ecology Notes 4:535-538. van Staaden, M.J. 1995. Breeding tactics, social structure and genetic variation in mammals: problems and prospects. Acta Theriologica Supplement 3:165-182.

van Zyll de Jong, C.G. 1985. Handbook of Canadian Mammals. National Museum of Natural Sciences, Ottawa. 212 pp.

Vaughan, T.A. and T.J. O’Shea. 1976. Roosting ecology of the pallid bat, Antrozous pallidus. Journal of Mammalogy 57:19-42.

Vaughan, T.A., J.M. Ryan, and N.J. Czaplewski. 2000. Mammalogy, 4th edn. Saunders College Publishing, Fort Worth, CO. 565 pp.

Vonhof, M.J., C.S. Davis, M.B. Fenton, and C. Strobeck. 2002. Characterization of dinucleotide microsatellite loci in big brown bats (Eptesicus fuscus), and their use in other North American vespertilionid bats. Molecular Ecology Notes 2:167-169.

221

Vonhof, M.J., C.S. Davis, C. Strobeck, and M.B. Fenton. 2001. Characterization of microsatellite loci in Spix’s disk-winged bats (Thyroptera tricolor). Molecular Ecology Notes 1:73-75.

Waser, P.M. and W.T. Jones. 1983. Natal philopatry among solitary mammals. The Quarterly Review of Biology 58:355-390.

Webb, N.J. and C.R. Tidemann. 1996. Mobility of Australian flying-foxes, Pteropus spp. (Megachiroptera): evidence from genetic variation. Proceedings of the Royal Society of London B 263: 497-502.

Webb, P.I. 1995. The comparative ecophysiology of water balance in microchiropteran bats. Symposium Zoology Society, London. No. 67:203-218.

Weid, R. and O. von Helversen. 1987. Ortungsrufe europaischer Fledermause beim Jagdflug im Freiland. Myotis 25: 5-27. As cited in O. von Helverson, K.G. Heller, F. Mayer, A. Nemeth, M. Volleth, and P. Gombkoto. 2001. Cryptic mammalian species: a new species of whiskered bat (Myotis alcathoe n. sp.) in Europe. Naturwissenschaften 88:217-223.

Weir, B.S. 1996. Genetic data analysis II. Sinauer Associates, Inc. Sunderland, Mass. 445 pp.

Whitaker, J.O. and L.J. Rissler. 1992. Winter activity of bats at a mine entrance in Vermillion County, Indiana. American Midland Naturalist 127:52-59.

Whitaker, J.O. and S.L. Gummer. 2000. Population structure and dynamics of big brown bats (Eptesicus fuscus) hibernating in buildings in Indiana. American Midland Naturalist 143:389-396.

Wiens, J.A. 1997. Metapopulation dynamics and landscape ecology. In: Metapopulation biology: ecology, genetics and evolution (I. Hanski and M.E. Gilpin, eds.). Academic Press, San Diego, pp. 43-68.

Wilde, C.J., C.H. Knight, and P.A. Racey. 1999. Influence of torpor on milk protein composition and secretion in lactating bats. Journal of Experimental Zoology 284: 35-41.

222

Wilkinson, G. S. 1985. The social organisation of the common vampire bat. II. Mating system, genetic structure, and relatedness. Behavioural Ecology and Sociobiology 17:123-134.

Wilkinson, L.C. and R.M.R. Barclay. 1997. Differences in the foraging behaviour of male and female big brown bats (Eptesicus fuscus) during the reproductive period. Ecoscience 4:279-285.

Willis, C. 2006. Daily heterothermy by temperate bats using natural roosts. In: Functional and Evolutionary Ecology of Bats (A. Zubaid, G.F. McCracken and T.H. Kunz, eds.). Oxford University Press, Oxford, pp. 38-55.

Willis, C.K.R. and R.M. Brigham. 2004. Roost switching, roost sharing and social cohesion: forest-dwelling big brown bats, Eptusicus fuscus, conform to the fission-fusion model. Animal Behavior 68:495-505.

Willis, J.S. 1982. The mystery of the periodic arousal. In: Hibernation and torpor in mammals and birds. (C.P. Lyman, J.S. Willis, A. Malan, and L.C.W. Wang, eds.) Academic Press, New York, pp. 92-103.

Wilson, D.E., and S. Ruff. 1999. The Smithsonian book of North American mammals. Smithsonian Institution and UBC Press, Vancouver. 750 pp.

Wilson, J. 2004. Foraging behaviour of insectivorous bats during an outbreak of western spruce budworm. MSc.Thesis, Department of Biology, University of Calgary, Calgary, Alberta. 189 pp.

Wimmer, B., D. Tautz, and P.M. Kappeler. 2002. The genetic population structure of the gray mouse lemur (Microcebus murinus), a basal primate from Madagascar. Behavioral Ecology and Sociobiology 52: 166-175.

Wolff, J.O. 1999. Behavioral model systems. In: Landscape ecology of small mammals (G.W. Barrett and J.D. Peles, eds.). Springer-Verlag, New York, pp. 11-40.

Wolff, J.O., E.M. Schauber, and W.D. Edge. 1997. Effects of habitat fragmentation on the social dynamics of the gray-tailed vole. Conservation Biology 11:945-946.

Worthington Wilmer, J., C. Moritz, L. Hall, and J. Toop. 1994. Extreme population structuring in the threatened ghost bat, Macroderma gigas: evidence from

223

mitochondrial DNA. Proceedings of the Royal Society, London B Biological Sciences 257:193-198.

Worthington Wilmer, J., L. Hall, E. Barratt, and C. Moritz. 1999. Genetic structure and male-mediated gene flow in the ghost bat (Macroderma gigas). Evolution 53:1582-1591.

Zinck, J.M., D.A. Duffield, and P.C. Ormsbee. 2004. Primers for identification and polymorphism assessment of Vespertilionid bats in the Pacific Northwest. Molecular Ecology Notes 4:239-242.

224

APPENDIX I: Patterns in Year-Round Bat Activity in Riparian Areas of Southern

Alberta

Introduction

At the core of any landscape ecology study is the quest to describe and explain the

abundance and distribution of species. Describing a species’ year-round habitats and

movement patterns is fundamental to understanding its ecology. How great an

undertaking landscape ecological investigations are, depends on scale, the vagility of the

species, and how much baseline information is known about habitat selection. It also

depends on having the appropriate research tools to study an animal at an appropriate

scale. As volant and nocturnal small mammals, bats are difficult to study. Long distance movements are particularly difficult to track using standard radio-telemetry techniques, given the short detection range of small radio-transmitters. However, recent advancements in acoustic technologies have allowed for long-term passive recording of

the ultrasound produced by bats as they fly, making possible the investigation of activity

patterns on a landscape level.

In the prairie landscape, bats concentrate in riparian areas during the summer

(Holloway 1998). In the late summer and fall, however, bats are observed leaving their summer roosts (pers. obs.) and “aggregations” appear in some areas (Schowalter and

Allen 1981, Schowalter et al. 1979). It is unclear as to where resident prairie bats spend the winter, and how they get there from their summer roosting areas. The prairie landscape is predominantly grasslands and agricultural lands with narrow zones of riparian habitat. To a bat looking for rock crevices in which to roost, water to drink, and

225

clusters of insects to feed on, the prairie landscape is a mosaic of suitable habitat patches

embedded in an inhospitable matrix. In Europe, bats follow linear landscape features

(Limpens and Kapteyn 1991), and as such, from a bat’s perspective, rivers may appear as

winding routes through an otherwise featureless matrix. River valleys may be movement

corridors linking suitable patches of riparian habitats.

Many temperate bat species move seasonally for hibernation and mating (Barbour

and Davis 1969), have specific roost requirements stemming from physiological,

morphological and behavioural forces (Kunz 1982), and are limited in movement by

night length and flight ability. Species that hibernate for the winter (resident species in

Alberta) are likely to fly short distances (e.g. Eptesicus fuscus, typically < 80 km, Mills et

al. 1975), and roost in crevices such as caves, rocks, or buildings (Kunz 1982). Non- hibernating (migratory) species tend to roost mainly in trees (Griffin 1970), and move long distances north-south (up to several thousands of kilometers, Cryan et al. 2004), moving north into the treed areas of Canada in the spring and back south in the fall

(Cryan 2003, Findley and Jones 1964). I hypothesized that river valleys likely provide

important corridors for both hibernating and migratory bat species as they move through

the inhospitable grasslands matrix of the prairie landscape.

My goal in this study was to determine whether bats use river valleys during the

spring and fall migration events. I wanted to document seasonal changes in species

assemblages and activity patterns along rivers in areas with and without suitable roosting

habitat and areas without roosting habitat. To do this, I acoustically monitored for the echolocation calls of bats in rocky riparian areas used as summer roosting areas and suspected of being winter roosting areas (Chapter 3 Lausen and Barclay 2006a), in

226

addition to areas with gentle grassy slopes not containing suitable rock-roosting habitat. I

predicted that a paucity of roosting habitat would restrict bat activity in grassy-sloped

areas of river valleys to spring and fall, if bats were using the rivers as migration

corridors between summer and winter habitats. I also predicted that the migratory tree species (Lasiurus borealis, L. cinereus and Lasionycteris noctivagans) would be detected

only during these migratory seasons, when they are likely to be using rivers as migration

corridors to some extent. Knowing that bats periodically fly in winter (Chapter 3 Lausen

and Barclay 2006a), I also predicted that in rocky riparian areas, I would detect bat activity of resident species (Myotis ciliolabrum, M. evotis, M. lucifugus and Eptesicus fuscus) throughout the winter months if the bats use those areas to hibernate. Using acoustic data collected in the river valleys of southern Alberta over several years I document the seasonal patterns of bat activity in several species to test the above

hypotheses and predictions. I also present comparisons that illustrate important

considerations for further acoustic studies of this type.

Methods

I passively monitored (no investigator present) areas of the Red Deer River in

Alberta using Anabat II and Compact Flash ZCAIM (Titley Electronics, Australia) data-

storage units running on external batteries with solar charge. Locations that were

continuously monitored are in order from west (upriver) to east (downriver) (see Fig.

1.2): Dry Island Buffalo Jump Park (DIBJ, N 51.933o W 112.968o) 29 October 2005 – 1

May 2006; East Coulee (N 51.333o W 112.480o) 26 Sept. 2004 – 4 May 2005, 10 July

2005 – 29 April 2006; Finnegan Ferry (N 51.126o W 112.085o) 7 Mar. – 4 May 2005, 3 -

227

9 July 2005; Dinosaur Provincial Park (DPP, N 50.761o W 111.512o) 10 March – 15

April 2004, 16 Oct. 2004 – 30 April 2006; Bindloss Campground (Bindloss, N 50.900o W

110.295o) 28 March – 2 April 2004, 16 April – 3 May 2004, 24 Mar. – 25 May 2005, 15

Aug. – 25 Oct. 2005. A few nights of monitoring also took place at Big Knife Provincial

Park (BKPP, N 52.491o W 112.222o) on the Battle River 8 Aug. 2005, 22 June 2006, 12

Aug. 2006. This latter site was monitored in summer only to determine the presence of

the migratory bat species, L. borealis and L. cinereus.

Extensive badland features create rock crevices and erosion holes in mudstone

river banks at DIBJ, East Coulee and DPP. These sites were monitored as potential

hibernation sites. Bindloss and Finnegan Ferry were monitored as sites likely to be used

as movement corridors in spring and fall, but not as hibernation sites due grassy slopes

and a paucity of rock-crevice habitat. DPP and East Coulee are ~120 km river distance

from each other, and Finnegan Ferry is in between these sites, ~80 km upriver of DPP

(Fig. 1.2). Bindloss campground is 128 km downriver of DPP with patches of eroded

river valley banks in between that would likely provide suitable summer roosts, but not

necessarily winter roosts. Finnegan Ferry has a farm house with a maternity colony of M.

lucifugus and some E. fuscus. Bindloss was not monitored in the summer, but there are

no suitable buildings in the immediate area of this site, and opportunistic summer

monitoring of this site has yielded few bat passes (unpub. data). This supports findings

of Holloway and Barclay (2000) who monitored this site in the summers of 1996 and

1997, and found activity to be relatively low (mean ~6 passes/night, approximately equal

Myotis and E. fuscus passes). I positioned detectors near (<15 m) the river with the

microphones oriented towards the middle of the river valley. In typical monitoring

228

situations, detector units are placed in areas suspected of concentrated bat activity to

maximize the amount of time a bat remains within the cone of microphone reception,

thus obtaining sequences of calls of sufficient quality for species’ identification

(O’Farrell 1998). Because the goal here was not species identification, but rather the

estimation of the number of bats moving past the site along the river, sites were selected

such that concentrated bat activity was not expected. The exact identical location(s) was

used at each site throughout the study for comparability.

At DPP in 2005, two Anabat systems were used occasionally, allowing for a

comparison between placements during winter activity; one system was placed along the

bank of the Red Deer River ~500 m from the nearest roosting habitat; the other system

was ~1 km away, upstream on the Little Sandhill Creek, a tributary of the Red Deer

River, ~50 m from the nearest available rock crevices. A chi-squared test was performed to test differences in species’ detection proportions among sites.

I noted the general bat-activity patterns, along with the earliest passes of

migratory species and the presence or absence of red bats, L. borealis, a species once

thought to be accidental in the province, but now considered “rare” in Alberta (ASRD

2006).

A sequence of bat echolocation calls that is separated by at least 5 s is recorded as

one digital file (O’Farrell 1998) and I considered each file to be one bat pass. Most files

were of short sequences (<15 s long). Because the storage buffer for Anabat is only 15 s,

after the buffer is full, the call sequence is recorded and a new file begins. Therefore, if a file was 15 s in duration, it was combined with the file(s) that followed it and counted as only one bat pass, ensuring that long single passes were not mistakenly counted as more

229

than one. Some species or species groups can be differentiated based on call

characteristics, such as minimum and maximum frequencies and call duration (Brigham

et al. 2002, Fenton and Bell 1981, O’Farrell et al. 1999). Passes per night were tallied for each location and were placed into the following species categories: Myotis (Myotis species, calls with minimum frequency of ~40 kHz), EPFU/LANO (E. fuscus or L. noctivagans, calls with minimum frequency of ~25 - 30 kHz), LACI (L. cinereus, calls which are often of long duration [> 12 ms] and/or characteristic variable minimum frequency pattern, often lower than 20 kHz), and LABO (L. borealis, calls with minimum frequency ~35 – 40 kHz with shallow call body and often characteristic upturn) (Corben

2002; Analook 4.9j Library, C. Corben, 2004, http://www.hoarybat.com). Because some

L. cinereus calls can look similar to E. fuscus calls when taken out of context of the entire sequence (Analook 4.9j Library), the number of L. cinereus passes is likely underestimated, especially when call sequences were short making pattern difficult to deduce. I documented the earliest and latest recording of L. cinereus and L. borealis calls in spring and fall. Because L. noctivagans and E. fuscus have similar calls (Betts 1998), I did not attempt to differentiate these species during spring, summer and fall periods when both species could theoretically be detected.

At DIBJ a potential hibernation location, I identified winter passes to species

when possible. Recorded passes (consisting of >1 call) were analyzed by visually

identifying E. fuscus passes (Chapter 3 Lausen and Barclay 2006a), and performing

Discriminant Function Analysis (DFA) on the Myotis passes. DFA was done using

S.A.S. Version 9.1, using Proc DISCRIM on echolocation call parameters (minimum

frequency [fmin], mean frequency, characteristic slope, and [fmax-fmin]/duration)

230

extracted using Analook 4.9j. Myotis calls (i.e. those with minimum frequency > 30 kHz)

that were of sufficient quality to analyze, were classified as M. ciliolabrum, M. evotis or

M. lucifugus, because these were the species captured in this location. Quadratic

discriminant function analysis was performed (Tabachnick and Fidell 2001) using reference calls from locally captured M. ciliolabrum (n = 64), M. evotis (n = 26), and M. lucifugus (n = 168). I recorded reference calls using an Anabat and audio tape recorder; all M. ciliolabrum and M. evotis were recorded after hand-release, while most (148) M. lucifugus call sequences were of free-flying individuals outside known roosts. Overall cross-validation error was 0.085 (range 0.078 - 0.100), and only passes that could be identified to species with ≥96% probability were accepted.

Results and Discussion

Seasonal Activity Patterns

I confirmed that bats hibernate at DPP, East Coulee, and DIBJ as evidenced by

winter bat activity (see Chapter 3 Lausen and Barclay 2006a for more details on the

former two sites). As expected, overall bat activity levels at these locations decreased

late in the fall, were low during the winter months, then increased into the spring, ranging

from 1 pass/night on some winter nights to >1000 passes/night in early May (Fig. I.1).

Anabat detectors positioned in areas along the Red Deer River with little to no rock- roosting habitat started recording bat passes on 3 April (Finnegan Ferry) and 7 April

(Bindloss) suggesting that bats were leaving hibernacula (e.g. DPP and East Coulee) for summer roosting sites by this time and passing through these grassy sloped areas (Fig.

I.2A). If bats were leaving DPP to move in either direction to summer roosting locations,

231

they would pass through Finnegan Ferry before Bindloss. This may explain why bats

were detected at Finnegan Ferry a few days before they were detected at Bindloss.

However, directionality of bat movement is unknown and whether bats remain

exclusively in the river valley is also unknown. That the Bindloss Anabat did not detect

bats earlier suggests there are no hibernacula near this site. The opposite pattern from

spring occurred in the fall when activity at Bindloss tapered off while activity remained

relatively high or increased in areas known to be used for hibernation (Fig. I.2B).

Activity patterns at the three hibernation sites differed in that East Coulee had relatively low summer bat activity (~40 passes/night), DIBJ had higher activity (~100), and DPP had the highest activity (>800; Table I.1, Fig. I.3). In the fall, at DPP, activity levels decreased relative to summer levels, while at East Coulee they increased, and peaked. This suggests that many bats leave the DPP area in the fall, while additional bats move into or through the East Coulee area. In the spring, activity levels higher than summer levels were recorded at DIBJ, East Coulee and Bindloss (Table I.1). This latter site, one of the grassy-sloped areas, also shows a substantial peak in bat activity in the fall. This is to be expected due to the paucity of roosts in the area, likely making it a transportation corridor only, joining more suitable roosting areas. The other relatively grassy-sloped area, Finnegan Ferry, has a farm house with a maternity colony of M.

lucifugus and some E. fuscus, adding to the summer activity levels at this site. Even though DIBJ has substantial bat activity in the summer, acoustic monitoring in the spring here revealed higher levels of activity than summer monitoring, suggestive of bats either

moving through the area during spring migration or emerging from hibernacula.

232

Species Detected

Similar to DPP and East Coulee (identified as hibernacula in Chapter 3 Lausen

and Barclay 2006a), DIBJ had bat activity in each winter month. Through visual

inspection of calls and discriminate function analysis, I determined that 30 winter passes

were of E. fuscus, and 38 were Myotis species. Of these latter passes, 27 passes were

analyzable; because there are multiple calls per pass, one hundred and forty-eight calls

were used to determine species identification. Twenty-three of the 27 passes could be identified with >96% probability and were therefore identified as: M. evotis (17 passes) and M. ciliolabrum (6 passes). I discriminated species of Myotis for winter passes only until 10 February, after which time I grouped all Myotis passes together. I detected no M. lucifugus passes prior to 10 February. This is similar to findings for East Coulee and

DPP, where M. lucifugus passes were not detected during the winter, but were detected in spring. The apparent absence of this species during winter in these rocky riparian areas suggests that it either does not hibernate in natural rocky riparian areas, or that different physiological requirements or behaviours eliminate the need for winter flights. Further research directed at determining the winter roosting habitat of M. lucifugus is needed.

Timing of the Presence of Migratory Species

The first detection of L. cinereus and L. borealis varied among sites and years to

some degree (Table I.2), but in general, L. cinereus was first detected in late March or

early April, and was last detected at the end of September to mid-October. One

particularly late detection of L. cinereus occurred on 23 November 2005. Temperatures

had been relatively mild in southern Alberta prior to this date, with lows >-12oC, making

233

it feasible that this bat was still making its way south. I never detected L. borealis before

7 May or after the end of September. The number of migratory-bat passes per month of monitoring is presented in Table I.1. Monitoring at Bindloss, DPP, and East Coulee allowed a comparison between spring and fall activity levels, and revealed more migratory bat passes in the fall. This was most obvious at the former and latter sites; at

Bindloss there was a 30 fold higher level of migratory bat activity in the fall (147 passes mid-August through to end of September) compared with spring (5 passes, mid-March through to end of May). Similarly, at East Coulee there were no passes of migratory

species in March and April (May was not monitored), but 256 passes were detected in

August and September. Late summer and fall patterns of migratory-bat passes at East

Coulee and Bindloss are presented in Fig. I.4.

Summer monitoring at DPP and East Coulee revealed some L. cinereus activity,

but there was no L. borealis until 29 July at DPP and 10 August at East Coulee. Summer

monitoring of DIBJ also detected L. cinereus but no L. borealis (4-9 July 2005). This

suggests that some L. cinereus reside in the prairies during the summer rather than

migrating through; capture is necessary to determine whether young are being raised in

the prairies or whether these detections are of males or non-reproductive females. Both

L. cinereus and L. borealis were detected in the summer in the Parkland region, at BKPP

(late June; Table I.1), suggesting they may be raising young there (Lausen 2006). Again,

capture would be necessary to confirm this.

234

Variation in Bat Activity

There were year-to-year and extreme night-to-night differences in bat activity.

Figure I.5 illustrates activity patterns at DPP and at East Coulee for fall 2004 and 2005.

Examination of the y-axes reveals that activity patterns differed at sites from year to year.

Overall, activity levels in fall and winter in 2005 were greater at DPP and lower at East

Coulee compared to 2004 detections (Table I.1), despite consistency of placement of

Anabats between years. The peaks in activity varied in number and in timing. This suggests that one year of monitoring in an area is not sufficient to describe its bat activity.

Similarly, timing of monitoring can be important, especially in the spring and fall when bats are moving between summer and winter roosts. Figures I.1 and I.5 demonstrate how

easily peaks in activity can be missed if monitoring does not occur for a long enough

period and at the appropriate times.

Placement of Anabat Detectors

Winter activity levels differed between Anabat sites at DPP. More bat passes

were detected at the Little Sandhill Creek site than along the Red Deer River, likely

because the former site was located closer to rock crevices. Not only did the Little

Sandhill Creek Anabat detect three times more passes (23 February – 15 March 2005, n =

282 at Creek, n = 81 at River; Fig. I.6), it detected a significantly larger proportion of

Myotis passes (17.4% Myotis at Creek vs. 2.5% at River; χ 2= 11.6, d.f. = 1, p < 0.001),

suggesting that Myotis bats do not fly as far as E. fuscus during winter flights. This

highlights the importance of Anabat placement when monitoring passively in one

location for an extended period of time.

235

Conclusions

As I hypothesized, rivers are used by resident and migratory species during

seasonal movements. Riparian areas not conducive to roosting were used primarily

during spring and fall migrations. I confirmed that resident bat species are present year-

round within prairie river valleys, although whether M. lucifugus uses hibernacula in

rocky riparian areas is yet to be determined. Given that river valleys in southern Alberta

run predominantly west-east, not north-south, it is likely that migratory species do not

remain within river valleys during the extent of their migrations. Migratory-bat routes in

Alberta have yet to be determined, but given the dependency of L. cinereus and L.

borealis on tree foliage for roosting, it is likely that presence or absence of leaves on trees

influences the migratory routes in spring versus fall. Differences in spring and fall

migratory-bat activity (e.g. Table 4.1 Bindloss), lend support to the hypothesis that

northbound and southbound migratory routes differ.

While I acoustically verified that river valleys are being used for movement of

bats, and that river valleys are sites of hibernation, it is not yet clear whether resident bat

species remain in the river valleys year-round, or make use of parts of the grasslands matrix when moving between summer and winter habitats. Because I did not monitor the grasslands matrix, it is conceivable that some species capable of faster flight and/or the

ability to make use of buildings as roosts may leave the river valleys to some extent,

short-cutting the distance between patches of suitable habitat. Using molecular genetics techniques, I test this hypothesis on a more regional scale, investigating the correlation

between relatedness and prairie landscape patterns (Chapter 4).

236

My results demonstrate the utility of landscape-level investigation. While often more difficult logistically than study of fine-scale habitat use, large-scale and year-round endeavors can elucidate ecological patterns not evident at the smaller spatial and temporal scales. Specifically, I have demonstrated that strategic placement of passive- recording bat detectors can elucidate seasonal movement of bats and locate hibernation areas. Because little is known about either of these in Alberta, future work in the province should emphasize widespread passive acoustic monitoring. The recent application of this technique precludes comparison with other published studies; much acoustic monitoring to date has focused on comparing relative activity levels in different habitat types during summer (e.g. Brooks and Ford 2005, Gehrt and Chelsvig 2004,

Heavner et al. 2006), but year-round passive monitoring is becoming increasingly common (e.g. Hall et al. 2005) and is likely to add invaluable information to our understanding of bat movement and distribution on local, landscape, regional and potentially even continental scales. Passive acoustic monitoring is a tool that may be particularly beneficial as development of the wind energy sector continues to increase, requiring knowledge of migratory-bat corridors for appropriate turbine siting (Arnett

2005, ASRD 2006, Lausen et al. 2006). It may also be highly instrumental in determining critical year-round habitats for decisions concerning prairie conservation and

in particular river manipulations, such as dams and cottonwood restorations (Trottier et al. 2004).

Table I.1. Summary of bat activity in 2004-2006 recorded at five locations along the Red Deer River: East Coulee (E. Coulee),

Finnegan Ferry (F. Ferry), Dinosaur Provincial Park (DPP) and Bindloss Campground, listed from west (upriver) to east (downriver).

Big Knife (BKPP) Provincial Park on the Battle River was also monitored. Potential rock-crevice roosts are abundant at East Coulee

and DPP, but grassy slopes predominate at Finnegan Ferry and Bindloss. Shaded cells indicate when the entire month was not

monitored, and exact dates are stated when less then three quarters of the month was monitored/analyzed. Total bat activity is the total

passes over all nights that month. Activity is in four categories: Myotis spp., E. fuscus/L. noctivagans (EPFU/LANO), L. cinereus

(LACI) and L. borealis (LABO). For more details about some of the BKPP and DIBJ acoustic data, see Lausen (2006).

Total bat passes over all nights Total bat Days monitored/ Mean passes Location Dates Monitored Myotis EPFU/LANO LACI LABO activity analyzed per day BKPP 22 Jun. 2006 4 57 104 9 174 1 174 8 Aug. 2005 16 21 4 4 45 1 45 12 Aug. 2006 1 22 9 2 34 1 34 DIBJ Nov. 2005 11 15 0 0 26 30 0.9 Dec. 2005 18 5 0 0 23 31 0.7 Jan. 2006 4 8 0 0 12 31 0.4 Feb. 2006 1 5 0 0 6 28 0.2 Mar. 2006 5 1 0 0 6 31 0.2 Apr. 2006 295 781 2 0 1078 30 36.0 1-15 May 2006 ------20161------36 2 2054 18 196.2 16-18 May 2006 9082 561 8 0 1477 14-18, 22-25 June 2006 ------16591------170 0 1829 9 203.2 6-9 July 20053 242 125 37 0 404 4 101 237

Total bat passes over all nights Total bat Days monitored/ Mean passes Location Dates Monitored Myotis EPFU/LANO LACI LABO activity analyzed per day E. 26-30 Sept. 2004 146 1203 5 0 1354 5 270.8 Coulee Oct. 2004 438 2845 8 0 3291 29 113.5 Nov. 2004 7 65 0 0 72 30 2.4 Dec. 2004 8 71 0 0 79 31 2.5 Jan. 2005 0 7 0 0 7 31 0.2 Feb. 2005 7 51 0 0 58 28 2.1 Mar. 2005 3 52 0 0 55 31 1.8 Apr. 2005 251 1623 0 0 1874 30 62.5 May. 2005 not monitored Jun. 2005 Jul. 2005 542 397 18 0 957 22 43.5 Aug. 2005 963 828 225 5 2021 31 65.2 Sept. 2005 363 1460 22 4 1849 30 61.6 Oct. 2005 58 606 0 0 664 25 26.6 Nov. 2005 1 32 0 0 33 30 1.1 Dec. 2005 1 29 0 0 30 31 1.0 1-10 Jan. 2006 0 7 0 0 7 10 0.7 F. Ferry Mar. 2005 0 0 0 0 0 24 0 Apr. 2005 34 79 2 0 115 30 3.8 1-4 May 2005 40 113 0 0 143 4 35.8 Jun. 2005 not monitored 3-9 July 2005 248 67 27 0 342 7 48.9 DPP 16-31 Oct. 2004 88 98 0 0 186 16 11.6 Nov. 2004 161 374 0 0 535 30 17.8 Dec. 2004 83 293 0 0 376 31 12.1 Jan. 2005 1 80 0 0 81 31 2.6 Feb. 2005 13 131 0 0 144 28 5.1 Mar. 2005 52 216 1 0 269 31 8.7 Apr. 2005 874 925 1 0 1800 30 60.0 May. 2005 6902 1150 9 2 8063 22 366.5 1-17 Jun. 2005 4143 562 61 0 4766 17 280.4 3-13, 28-31 Jul. 2005 10092 2778 7 4 12881 15 858.7 238

Total bat passes over all nights Total bat Days monitored/ Mean passes Location Dates Monitored Myotis EPFU/LANO LACI LABO activity analyzed per day 15-30 Aug. 2005 5339 1186 3 7 6535 18 363.1 Sept. 2005 4728 875 10 7 5620 30 187.3 Oct. 2005 958 708 0 0 1666 25 66.6 Nov. 2005 139 472 1 0 612 30 20.4 Dec. 2005 74 424 0 0 498 31 16.1 Jan. 2006 72 306 0 0 378 31 12.2 Bindloss 24-31 Mar. 2005 0 0 0 0 0 8 0.0 Apr. 2005 1 17 0 0 18 30 0.6 May. 2005 151 48 1 4 204 25 8.2 Jun. 2005 not monitored ~64 Jul. 2005 15-31 Aug. 2005 702 126 67 36 931 17 54.8 Sept. 2005 551 104 22 22 699 30 23.3 Oct. 2005 13 1 0 0 14 25 0.6 1 Myotis and EPFU/LANO passes not separated in tally

2 M. evotis passes were ≥ 291 of these based on obviously steep sloped calls with minimum frequencies 30-35 kHz.

3 Temperature at midnight ranged from 15–20oC each night.

4Mean summer passes for the Bindloss Campground site as reported by Holloway and Barclay (2000) for 1996 and 1997. 239

Table I.2. Earliest and latest detections of the migratory species L. cinereus and L. borealis at 5 locations: Dinosaur Provincial Park

(DPP), East Coulee (E. Coulee), Finnegan Ferry (F. Ferry), Bindloss Campground (Bindloss) and Dry Island Buffalo Jump (DIBJ).

Not all sites were monitored each spring and fall year (n/m) and duration of monitoring varied; if spring monitoring did not extend

long enough to detect the species, or fall monitoring began too late to detect the species, the date that monitoring in the area

ceased/began is stated. For example, fall monitoring in 2004 at DPP did not begin until 16 Oct., and this was too late to detect passes

of the two migratory species. An unusually late L. cinereus pass in 2005 is placed in parentheses.

Year Location Species 2004 2005 2006

First spring DPP L. cinereus 30 Mar. 31 Mar. 4 April detection L. borealis 0 by 15 April 7 May 0 by 30 April E. Coulee L. cinereus n/m 0 by 4 May 25 April L. borealis n/m 0 by 4 May 0 by 30 April Bindloss L. cinereus 0 by 3 May 15 May n/m L. borealis 0 by 3 May 20 May n/m F. Ferry L. cinereus n/m 0 by 4 May n/m L. borealis n/m 0 by 4 May n/m DIBJ L. cinereus n/m n/m 28 April L. borealis n/m n/m 7 May Last fall DPP L. cinereus 0 after 16 Oct. 29 Sept. (23 Nov.) n/m detection L. borealis 0 after 16 Oct. 25 Sept. n/m E. Coulee L. cinereus 12 Oct. 30 Sept. n/m L. borealis 0 after 26 Sept. 6 Sept. n/m Bindloss L. cinereus n/m 9 Sept. n/m L. borealis n/m 18 Sept. n/m 240

A. 1200

1000

800

600

400

Bat activity (Passes per Night) Bat activity (Passes per

200

0 ec

2-Jan 9-Jan 3-Apr 3-Oct 7-Nov 6-Feb 6-Mar 5-Dec 1-May 16-Jan 23-Jan 30-Jan 10-Apr 17-Apr 24-Apr 10-Oct 17-Oct 24-Oct 31-Oct 14-Nov 21-Nov 28-Nov 13-Feb 20-Feb 27-Feb 13-Mar 20-Mar 27-Mar 26-D 26-Sep 12-Dec 19-Dec Date (2004-2005)

Figure I.1A 241

B. 1200

1000

800

600

400

200 Night) Bat activity (Passes per

0

-Jan -Jan -Jan 1-Jan 8-Jan 5-Feb 5-Mar 2-Apr 9-Apr 6-Nov 4-Dec 7-May 16-Oct 23-Oct 30-Oct 15 22 29 12-Feb 19-Feb 26-Feb 12-Mar 19-Mar 26-Mar 16-Apr 23-Apr 30-Apr 13-Nov 20-Nov 27-Nov 11-Dec 18-Dec 25-Dec Date (2004-2005) Figure I.1. Bat activity (passes per night) at East Coulee (A) and Dinosaur Provincial Park (B). Peaks of activity are seen at East

Coulee in both fall and spring. The peak fall activity was likely missed at DPP, but was captured in spring. Winter activity is relatively low but constant (see Chapter 3 Lausen and Barclay 2006a for more details). 242

Figure I.2. Total bat activity (passes per night) compared between sites along the Red Deer River during spring 2005 (A) and fall

2005 (B). A. DPP and East Coulee showed a relatively high level of bat activity following the winter hibernation period. Finnegan

Ferry and Bindloss lacked activity until 3 April and 7 April, respectively, then increased as the spring progressed. B. Total activity at

East Coulee, DPP and Bindloss from 15 Aug. – 25 October, 2005. The last bat activity at Bindloss was 7 October. Monitoring and activity at Dinosaur Provincial Park (DPP) and East Coulee occurred throughout the winter (see Fig. I.1). Monitoring at Finnegan

Ferry and Bindloss did not occur over the winter due to lack of rock crevices for hibernation. Maximum activity levels are indicated above bars on series of nights when activity was too high to show on graph. Note the difference in y-axis scales between the two graphs.

243

A. < 80 < 75 < 381 < 181 50 45 DPP 40 East Coulee

F. Ferry ) 35 ht Bindloss Camp g

30 er Ni

p

25 Passes (

20 y

15 Bat activit 10

5

0 ar ar pr pr

1-Apr 3-Apr 5-Apr 7-Apr 9-Apr 1-May 3-May 11-Apr 17-Apr 19-Apr 21-Apr 23-Apr 25-Apr 27-Apr 29-Apr 13-A 15-A 24-M 26-M 28-Mar 30-Mar Figure I.2A Date (2005) 244

B. < 529 < 526 < 424 < 307 < 207 < 617 < 516 < 362 200 DPP E. Coulee Bindloss 180 160 140

120 100 80 60 40 20

Bat activity (Passes per Night) Bat activity (Passes per 0

Sep -Sep - -Sep -Sep 2-Oct 4-Oct 6-Oct 8-Oct 2-Sep 4-Sep 6-Sep 8-Sep 10-Oct 12-Oct 14-Oct 16-Oct 18-Oct 20-Oct 22-Oct 24-Oct 15-Aug 17-Aug 19-Aug 21-Aug 23-Aug 25-Aug 27-Aug 29-Aug 31-Aug 10-Sep 12-Sep 14-Sep 16 18 20 22 24-Sep 26-Sep 28-Sep 30-Sep Date (2005) Figure I.2B 245

Figure I.3. Year-to-year variation in bat activity at DPP (A) and East Coulee (B), Alberta. Anabat was placed in same location along

river at each site each year. White part of bar is E. fuscus (EPFU) and dark part is Myotis spp. Top graph and bottom graphs correspond to same range of dates, top being for 2004 and bottom for 2005. Note the different scales of the y axes between top and

bottom graphs, demonstrating the year-to-year variation in bat activity.

246

600

A.

500 East Coulee Myotis East Coulee EPF U

400

300

200

Night) Bat activity (Passes per

100

0

-Sep -Sep 2-Oct 9-Oct 10-Jul 17-Jul 24-Jul 31-Jul 4-Sep 7-Aug 16-Oct 23-Oct 25-Sep 11 18 14-Aug 21-Aug 28-Aug Figure I.3A Date (2005)

247

B.1 600

1400 DPP My otis DPP EPFU

1200 Data not analyzed 1000

800

600

Night) Bat activity (Passes per 400

200

0

3-Jul 2-Oct 9-Oct 10-Jul 17-Jul 24-Jul 31-Jul 4-Sep 7-Aug 16-Oct 23-Oct 11-Sep 18-Sep 25-Sep 14-Aug 21-Aug 28-Aug Figure I.3B Date (2005) 248

A. 30 28 26

24 L. cinereus 22 L. borealis 20 18 16 14 12 10

8 Bat activity (Passes per Night) Bat activity (Passes per 6 4

2 0 Aug Sep

2-Oct 6-Oct 3-Aug 7-Aug 8-Sep 10-Jul 14-Jul 18-Jul 22-Jul 26-Jul 30-Jul 4- 10-Oct 14-Oct 18-Oct 22-Oct 11-Aug 15-Aug 19-Aug 23-Aug 27-Aug 31- 12-Sep 16-Sep 20-Sep 24-Sep 28-Sep Figure I.4A Date (2005) 249

250

28-Sep

26-Sep

24-Sep 22-Sep

in 2005 at East Coulee (A) in 2005 at East Coulee

20-Sep

18-Sep 16-Sep

and L. borealis

14-Sep 12-Sep

L. cinereus

10-Sep

8- Sep

6- S ep

4- Sep

Date (2005)

2-Sep

L. cinereus L. borealis L. 31-Aug 29-Aug

er and fall migratory-bat species, er and fall migratory-bat

27-Aug

25-Aug 23-Aug

ity of late summ

21-Aug

19-Aug

17-Aug 15-Aug

8 6 4 2 0

30 28 26 24 22 20 18 16 14 12 10 Bat activity (Passes per Night) Night) per (Passes activity Bat B. Figure I.4. Patterns of activ and Bindloss Campground (B). and Bindloss Campground 251

Figure I.5. Bat activity in summer through to fall at East Coulee (A) and DPP (B) in

2005. Two periods of activity were not analyzed, but trend can be deduced. Summer activity is high at DPP (B) compared to East Coulee (A). At DPP activity tapers off as fall approaches (A), unlike East Coulee where bat activity peaked during the fall period

(B).

252

25

DPP EPFU Fall 2004

20 DPP Myotis Fall 2004

15

10

) 5 ht g er Ni

p 0

16-Oct 17-Oct 18-Oct 19-Oct 20-Oct 21-Oct 22-Oct 23-Oct 24-Oct 25-Oct 26-Oct Passes (

y 400

DPP EPFU Fall 2005 350 Bat activit DPP Myotis Fall 2005 300

250

200

150

100

50

0

t t t t t t t t t t t

c c c c c c c c c c c

O O O O O O O O O O O

------

6 7 8 9 0 1 2 3 4 5 6 1 1 1 1 2 2 2 2 2 2 2 Date Figure I.5A

253

B. 1200 East Coulee EPFU Fall 2004

1000 East Coulee Myotis Fall 2004

800

600

400

) 200 ht g

er Ni 0 p

t t t t t t t t t t t t t t t t t t t t t t t t t t

p p p p p

c c c c c c c c c

e e e e e

O O O O O O O O O

Oc Oc Oc Oc Oc Oc Oc Oc Oc Oc Oc Oc Oc Oc Oc Oc Oc

------S S S S S ------

- - - - -

1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6

6 7 8 9 0

1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2

2 2 2 2 3 Passes (

y 600 East Coulee EPFU Fall 2005

500 East Coulee Myotis Fall 2005

Bat activit

400

300

200

100

0

1-Oct 2-Oct 3-Oct 4-Oct 5-Oct 6-Oct 7-Oct 8-Oct 9-Oct 10-Oct 11-Oct 12-Oct 13-Oct 14-Oct 15-Oct 16-Oct 17-Oct 18-Oct 19-Oct 20-Oct 21-Oct 22-Oct 23-Oct 24-Oct 25-Oct 26-Oct 26-Sep 27-Sep 28-Sep 29-Sep 30-Sep

Figure I.5B Date

254

100%

80%

60% Myotis at River Myotis at Creek

EPFU at River 40% EPFU at Creek

20% Percent Bat Passes per Night Percent Bat Passes per Night

0%

1-Mar 2-Mar 3-Mar 4-Mar 5-Mar 6-Mar 7-Mar 8-Mar 9-Mar 10-Mar 11-Mar 12-Mar 13-Mar 14-Mar 15-Mar 23-Feb 24-Feb 25-Feb 26-Feb 27-Feb 28-Feb Date (2005)

Figure I.6. Bat activity in spring 2005 at DPP using two detectors. One detector was set

at the river’s edge and another at a creek ~1 km away, nearer rock-roost habitat.

Percentage of bat passes detected by each detector, demonstrates that the two sites not

only detected different numbers of passes, but the ratio of Myotis to EPFU differed. The

bat detector closest to the rock habitat recorded >3x more bat passes and a larger ratio of

Myotis:EPFU. The river Anabat detected 81 passes (2 Myotis, 79 E. fuscus) and the

creek Anabat 282 (49 Myotis, 233 E. fuscus).

255

APPENDIX II: Nightly Activity Patterns and Reproductive Rates of Four Species of Prairie Bats Derived from Capture Records

Introduction

Patterns of nightly activity of bats reflect their energy budgets, patterns of insect activity (Anthony et al. 1981, Racey and Swift 1985), foraging strategies (Chruszcz

1999), and for females, the need to nurse pups during the night (Kunz 1974). Just as with patterns of thermoregulation (e.g. Chruszcz and Barclay 2002, Lausen and Barclay 2003), activity patterns of reproductive females are expected to differ from those of males and non-reproductive females due to energy budget differences (Gittleman and Thompson

1988, Anthony et al. 1981, Grinevitch et al. 1995, Kunz 1974). Pregnant and lactating females can have different activity patterns, reflecting the higher daily energy demand during lactation (Gittleman and Thompson 1988, Rydell 1993). Different foraging activity patterns have been reported for males, and pregnant and lactating females (e.g.

Kunz 1974).

While activity patterns of bats are often studied by radiotracking individuals (e.g.

Audet and Fenton 1988, Chruszcz 1999, Wilkinson and Barclay 1997), activity patterns can also be deduced through capture records over a large geographic and temporal scale, diluting potential influences of differing habitats, year-to-year differences in weather and prey abundance, individual differences, and radio-transmitter effects. Therefore, although indirect, this method of investigating activity patterns may help elucidate species-specific patterns.

In the course of my research, I captured a large number of individuals of several species. Here I present an analysis of capture-time data collected in the Alberta and

256

Montana prairies to illustrate patterns of nightly bat activity in river valleys. I compare

species (Eptesicus fuscus, Myotis ciliolabrum, M. evotis and M. lucifugus), sexes, and

reproductive stages. Insect prey often show a bimodal pattern of activity with peaks of

activity at dusk and dawn (Anthony et al. 1981, Racey and Swift 1985), and this is often

reflected in bat-activity patterns. Temperate insectivorous bats often night roost or feed young in the middle of the night, and thus forage mainly in the period just after dusk and just before dawn (Anthony et al. 1981, Holloway 1998, c.f. Chruszcz 1999). I thus predicted that capture rate would be lower in the middle portion of the night.

Reproductive females have additional energy demands from those of males and non- reproductive females (Gittleman and Thompson 1988), and overall food consumption has been reported to be higher in reproductive females than in males (Kunz 1974). Based on differences in energy budgets, and previous findings of foraging behaviour differences among sexes and reproductive stages (e.g. Kunz 1974, Rydell 1993), I predicted that

capture records would show differences in pattern of capture among males, reproductive

females and nonreproductive females. I predicted that based on energy requirements, overall captures of males and non-reproductive females in the middle of the night would

be less than that of reproductive females. During the mid-portion of the night, lactating

bats often feed young (Kerth 2006), and therefore, I predicted that pregnant bats would be

captured more than females nursing pups. I also predicted that species’ differences

would exist, reflecting differences in thermoregulatory, foraging, and night-roosting

strategies. Chruszcz and Barclay (2003), using radio-telemetry, found that reproductive

M. evotis foraged longer during the night than M. ciliolabrum and E. fuscus and

hypothesized that this was due to their less energy-efficient mode of foraging (gleaning).

257

I therefore predicted that M. evotis would be captured throughout the night more than the

other species.

I compared the ratios of reproductive to non-reproductive females of the species

to determine if differences exist in overall reproductive success among prairie species.

Reproductive rates among bat species, especially those at more northern latitudes, is variable (Barclay et al. 2004), and this may stem from species’ differences in costs associated with thermoregulation, differences in morphology and associated flight costs,

or different physiological and behavioural responses to shorter summer seasons and more

variable cooler climatic conditions (Schowalter et al. 1979, Barclay and Harder 2003,

Barclay et al. 2004). I therefore hypothesized that relative reproductive success would

differ among the prairie species. Because several studies have found evidence to suggest

energetic and consequently fitness advantages for bats roosting in buildings versus

natural roosts (Lausen and Barclay 2006b, Law and Chidel 2007), I predicted that the

colonial building-roosting species (E. fuscus, M. lucifugus) would have a higher

percentage of reproductive females than the solitary rock-roosting species (M.

ciliolabrum, M. evotis) in areas where potential building roosts were present. Because

smaller species of Vespertilionid bats tend to have lower reproductive rates (Barclay et al.

2004), I also predicted that M. ciliolabrum would have the lowest percentage of

reproductive females.

258

Methods

Study Species

Myotis lucifugus, M. ciliolabrum, M. evotis and E. fuscus, differ in size (Table

II.1), foraging behaviour, and roost selection (Chapter 2, Fenton and Barclay 1980,

Holloway 1998, Holloway and Barclay 2001, Kurta and Baker 1990, Manning and Knox

1989). Of the four species, M. ciliolabrum is least capable of long distance fast flight and E.

fuscus, the most (Norberg and Rayner 1987). While M. lucifugus, M. ciliolabrum and E. fuscus typically take flying prey (aerial hawking), M. evotis is also capable of gleaning insects from surfaces (Faure and Barclay 1994). In the Montana and Alberta prairie river valleys, these species roost in rock-crevices. Reproductive female M. evotis typically roost

alone (Chruszcz 1999), and M. ciliolabrum typically roost alone or in small groups (e.g. 1

– 6 in southern AB, Holloway 1998) in eroded crevices in river valley banks (Chapter 2;

Holloway and Barclay 2001). E. fuscus females typically roost in groups of 25 – 75

(reviewed in Kurta and Baker 1990), and like M. lucifugus, will use tree cavities, rock crevices, and buildings as roosts. In this study, only the latter two roost structures were typically used due to the scarcity of trees. M. lucifugus, the common house bat in many parts of North America (Fenton and Barclay 1980), tends to roost in larger female groups with colonies reaching thousands of individuals (Davis and Hitchcock 1965).

Reproductive patterns in all four species are similar, with mating thought to occur in the autumn (van Zyll de Jong 1985), gestation in the spring, parturition typically in late

June/early July in SE Alberta, and lactation until mid-July to mid-August.

259

Captures

I captured bats with mist-nets from 2001 – 2004, and included captures from late

May through to mid-September for the capture-time analyses. For the analysis of

reproductive versus nonreproductive females, I used captures only after 23 June (in AB)

when lactation had usually begun in the AB prairies, or after 10 June (in MT) when

lactation had started in MT. This ensured that pregnancy could be unequivocally established for all captures. I distinguished adults from juveniles by examining the finger joints for ossification (Anthony 1988). I classified females as non-reproductive, pregnant, lactating, or post-lactating (Racey 1988). By gently palpating the abdomen I could detect pregnancy; I considered females captured early in the season (May and the first half of June) to be of “unknown” reproductive status if pregnancy could not be confirmed. I confirmed lactation by expressing milk from enlarged teats. Post-lactating females had hair regrowth around the periphery of the teats, and I was unable to express milk. Females captured after mid-June that were not pregnant or lactating were classified as non-reproductive.

Analyses

Because bat activity in the prairie riparian areas of SE Alberta peaks twice during

the night, once at dusk and once at dawn (Holloway 1998), I compared activity levels

throughout the night by dividing the night into 2 capture periods: 1. the mid-night period,

starting 2 h after dusk (civil twilight) and ending 2 h before dawn (sunrise) and 2.

dusk/dawn period. For captures to be included in the analysis, the netting session had to

consist of general netting (i.e. no roosts or specific species targeted) and had to start at

dusk and continue for at least four hours. Mist-netting tended to occur both among rocks

260

and among trees, and both roosting and foraging habitat were therefore sampled. All

capture locations were in the AB or Montana prairies (Fig. 4.1 of Chapter 4), and were in

rocky riparian areas. With the exception of the Bindloss Ferry Crossing (FC) site,

buildings were within 5 km of each sampling area. Therefore, E. fuscus and M. lucifugus

may have been roosting in either rock crevices, or buildings in all but the FC location.

Few M. lucifugus were captured at the FC location, but a rock-roosting colony of E. fuscus was known from there (Lausen and Barclay 2002).

Total adult captures used were: 314 female and 371 male M. ciliolabrum; 415

female, 174 male M. lucifugus; 77 female, 121 male M. evotis; and 321 female and 213

male E. fuscus, in river valleys of the AB and MT prairies. A subset of the capture data

was used in the analysis of capture patterns, using only captures for which time of capture

was known. Because they are no longer providing for a fetus or pup, post-lactating

females were pooled with the non-reproductive females for the capture time analysis.

Another subset of the capture data was used in the analysis of

reproductive:nonreproductive ratios, using only female captures with known reproductive

states (after the onset of lactation).

In the capture pattern analysis, I compared capture times for all males and

females, and further compared females for which reproductive status was known. I used

generalized linear models (binomial distribution, logit regression; Hosmer and Lemeshow

2000, Hardin and Hilbe 2007) to test sex, species, and reproductive status as predictors of

captures in the mid-night period. I compared capture patterns among species using odds

ratios of captures during the mid-night period. I also tested capture pattern differences among sexes and reproductive conditions using contingency tables (Pearson χ2).

261

Using chi-squared contingency tables, I compared the proportion of reproductive

females (pregnant, lactating and post-lactating) to nonreproductive females across

species. I also tested the influence of type of roost on the proportion of reproductive to

nonreproductive females, by comparing M. lucifugus and E. fuscus captured in rocky

riparian areas to those captured from known building roosts. All statistical analyses were

performed in Stata 9.0 (2005, StataCorp LP, College Station, TX).

Results

The proportion of adult bats captured in the mid-night period versus the

dusk/dawn period differed with sex, such that males were 1.89 ± 0.223 times more likely

to be captured in the midnight period than females (z = 5.38, p < 0.001). There were also differences among the species, with proportionally fewer mid-night captures of M. ciliolabrum compared to the other three species (odds ratios ranging from 1.38 – 2.05; p values ranging from 0.042 - <0.001); additionally, the proportion of M. evotis captures occurring in the mid-night period was significantly greater than the proportion of M. lucifugus mid-night captures (1.48 ± 0.28, z = 2.06, p = 0.04).

I tested males separately for species differences, and found that capture patterns of males of all species were similar (Pearson χ2 = 4.47, d.f. = 3, p = 0.215), with the

percentage of males captured during the mid-night period consistently between 31 and

39% for each species (Table II.2). The proportion of females captured in the two night

periods depended on reproductive condition, with female captures in the mid-night period

3.01 ± 0.695 times more likely to be of nonreproductive than reproductive individuals (z

= 4.77, p < 0.001). When I tested for differences among species in the non-reproductive

262

captures, only M. ciliolabrum differed from the other species, which did not differ from

each other; a significantly lower proportion of nonreproductive M. ciliolabrum females

(22 non-reproductive females out of 82 females, 26.8% vs. all other species 31, 52.5%;

Table II.2) were captured in the mid-night period (χ2 = 9.67, d.f. = 1, p = 0.002).

I examined reproductive female captures separately and determined that there was

no difference in the capture patterns of pregnant versus lactating females (χ2 = 0.154, d.f.

= 1, p = 0.695). There was, however, substantial variation in capture pattern among

species (Table II.3); a mid-night capture of a reproductive female had the greatest

likelihood of being M. evotis, followed by M. lucifugus and then E. fuscus/M. ciliolabrum

(Table II.3). These latter two species did not have an odds ratio significantly different from one, and mid-night captures of reproductive females of these species were 5 – 8 times less likely than those of M. evotis.

Using all captures following the onset of lactation, I compared the proportion of

reproductive females among species (Table II.4) and found that when captures from general netting (“random captures”) in river valleys were compared, species did not differ

(Pearson χ2 = 1.35, d.f. = 3, p = 0.72). However, when captures from known building

roosts (2 building roosts of E. fuscus, and 4 building roosts of M. lucifugus) were

included (Table II.4), the percentage of nonreproductive females for E. fuscus and M.

lucifugus similarly decreased (E. fuscus 45.1% to 21.7%, M. lucifugus 37.5% to 25.1%;

Table II.4), and were significantly lower than for M. ciliolabrum and M. evotis (χ2 = 49.6, d.f. = 1, p < 0.001). I specifically compared female E. fuscus and M. lucifugus captured in mistnets in rocky riparian areas to those captured at known building-roosts. In both species, the randomly captured females had significantly lower reproductive rates than

263

those roosting in buildings (E. fuscus, χ2 = 21.4, d.f. = 3, p = 0.001; M. lucifugus, χ2 =

5.67, d.f. = 3, p = 0.017).

Discussion

Capture Patterns

Across all four of the species I captured in the prairies of Alberta and Montana, males were more evenly distributed throughout the night than were females. While I

caught equal numbers of males and females in the dusk and dawn period, males

outnumbered females two-to-one in the middle of the night. Across all species, captures

of reproductive females were particularly skewed towards the dusk/dawn period

compared to nonreproductive females which were more evenly caught in the two time

periods. In at attempt to understand these capture patterns, I make several assumptions;

one assumption is that captures in the mid-night period when insects are usually less

abundant (Anthony et al. 1981), are indicative of bats foraging beyond the dusk time

period, resulting in a longer time spent foraging. Similarly, I assume that captures reflect

activity levels and that lower captures equate to lower activity level, with individuals

night-roosting instead of foraging. This assumption may not hold true if bats are

changing their location of foraging during the mid-night period to areas that I did not

mistnet in. Foraging over open prairie, for example, rather than in river valleys, would result in a lower number of captures. However, conclusions would not differ as long as the sexes and species similarly make this shift. If they do not, then a skew in capture times could mistakenly reflect foraging location rather than foraging duration differences.

For example, I assume that males and females have been similarly sampled. If one sex

264

were to typically roost or forage further from the riparian sampling locations, it is

possible that skew in capture times could reflect differences in roost-foraging distances

rather than time spent foraging, as I have assumed.

In understanding why mid-night capture times, and thus longer foraging times,

occur in some species, sexes, and reproductive stages, I assume that there is a threshold

cost:benefit ratio at which foraging ceases. Different foraging times are likely to reflect

different thresholds and/or different factors which determine cost:benefit ratios. For

example, that males and nonreproductive females foraged longer than reproductive

females, suggests that mid-night foraging, despite lower insect abundance, is still

profitable for these individuals. Additional flight costs during pregnancy from the added

weight of the fetus (Rydell 1993, Speakman and Racey 1987), together with the need to

nurse pups left behind in roosts during lactation, likely mean high cost:benefit ratios for

reproductive females, and a consequential foregoing of mid-night foraging when insect

prey requires more effort to acquire. For males and non-reproductive females, with

seemingly fewer associated costs of extensive foraging, lower cost:benefit ratios may

explain longer foraging times (see below).

I predicted that reproductive females would forage more than nonreproductive

females or males to consume more food (Kunz 1974) and meet their higher energy

requirements. However, capture patterns did not support this prediction. Based on the low proportion of reproductive females captured in the mid-night period, reproductive females experienced reduced foraging times compared with males and nonreproductive females, which is incongruent with energy demands; because reproductive females have increased energy demands (Gittleman and Thompson 1988), they should benefit from

265

longer foraging times, and be captured more evenly throughout the night. That their

captures were skewed to the dusk/dawn period, suggests that threshold cost:benefit ratios

differ and/or are reached earlier in the evening than for males and nonreproductive

females, or that the males and nonreproductive females are less efficient foragers.

Wilkinson and Barclay (1997) concluded, based on radio-telemetry data, that male and

female E. fuscus forage for the same amount of time despite presumed lower energetic demands for males, because males travel further than females during foraging,

necessitating longer foraging times than females for equivalent energy gains.

Whether male M. lucifugus, M. ciliolabrum and M. evotis travel further than

females, as in E. fuscus (Wilkinson and Barclay 1997) has yet to be determined. If this were the case, then larger male foraging areas could have influenced my capture records

such that each mistnetting location was more likely to be located in a male foraging area,

than a female foraging area. Therefore, even on nights when no males or females were

sampled in the early part of the evening, longer-ranging males may have been sampled

during the mid-night period.

The capture pattern for reproductive females, with a skew towards dusk/dawn

captures, is evident regardless of reproductive stage, suggesting that pregnant and

lactating females share similar cost:benefit ratios for determination of time spent foraging. This supports the hypothesis that even though energy expenditure (in the form

of milk) is highest during lactation, lactating females may shift their metabolic use of

energy, reducing body maintenance for example, and thereby not requiring increased

food consumption (Gittleman and Thompson 1988). Pregnant and lactating females may

arrive at similar foraging cost:benefit ratios for different reasons; for pregnant females,

266

increasing body mass may steadily increase flight costs as gestation proceeds (Speakman

and Racey 1987), and for lactating females, time spent away from the roost is likely to

correlate with decreased juvenile growth (Cossins and Bowler 1987, Kerth 2006) and

ultimately a potential lowering of overall fitness. A balance between foraging time and

night-roosting with offspring, has seemingly led to foraging mainly during the most

profitable times. My data suggest that in the prairies of AB and MT, reproductive

females are able to accrue sufficient energy by foraging only during peak insect times.

However, this may not be the case in other habitats, all latitudes, or in particular years, which may explain why some studies find different foraging patterns between pregnancy and lactation (e.g. E. nilssonii, Rydell 1993).

Compared to the other species, reproductive female M. evotis were more evenly

distributed in the two periods, as I predicted. This agrees with Chruszcz and Barclay

(2003) who found that radio-tracked M. evotis forage longer throughout the night than reproductive female E. fuscus and M. ciliolabrum in the same area. This may reflect their energetically expensive gleaning mode of foraging Chruszcz and Barclay (2003).

The greatest skew towards dusk/dawn captures was seen in M. ciliolabrum, and this skew existed for all females, regardless of reproductive stage. The proportion of males

captured in the mid-night period was only slightly lower than for males of the other

species. This suggests that foraging strategy or behaviour differences exist between

males and females of this species. For example, females regardless of reproductive stage,

may night-roost as a group, a behaviour observed in other species (reviewed in Kunz

1982). This grouping may enhance social bonds among individuals and kin (Kunz 1982)

given their solitary roosting behaviour which does not allow such social grouping during

267 the day (Chapter 2). Social grouping may have advantages such as information transfer

(Kerth 2006, Waser and Jones 1983). Alternatively, males and females may have different sizes or types of foraging home ranges, and this may be reflected in skewed mid-night captures in the river valleys. Radio-tracking of both males and females may shed light on their different capture patterns.

Reproductive Rates

Variation in reproductive rates (the proportion of females reproducing in a season) of bats in temperate regions is greater than for bats at lower latitudes (Barclay et al. 2004), and I therefore predicted different reproductive rates among the four species of prairie bats. I documented relatively consistent reproductive rates among the four species when general mist-netting captures were analyzed (38 – 45% non-reproductive). Because general mistnetting samples the entire population, including non-reproductive females who sometimes roost outside maternity colonies (Hamilton and Barclay 1994), these data likely approximate population reproductive rates. In some species, females group together to raise young at maternity roosts that appear to provide thermal benefits to reproductive females (e.g. Lausen and Barclay 2002, 2006b). It has been hypothesized that measuring reproductive rates at such colonies may skew species’ values (Barclay et al. 2004). The significant increase in the proportion of reproductive female M. lucifugus and E. fuscus caused by the addition of maternity building-roost captures to the dataset supports this hypothesis and suggests that species’ reproductive rates may be over- estimated when based on maternity colony data. Additionally, building-roosts may provide energetic and consequently fitness advantages over natural roosts (Lausen and

268

Barclay 2006b, Law and Chidel 2007), and lower reproductive rates in natural rock roosts

compared with building-roosts have been reported (Lausen and Barclay 2006b).

Therefore, roost type must also be considered in the estimation of species’ reproductive

rates.

Conclusion

I have demonstrated that large capture datasets can provide ecological data

beyond their intended use. Capture times, in particular, may be useful in elucidating foraging patterns among sexes and species, and are therefore useful data to collect.

Reproductive status of individuals is also an important characteristic to record in capture databases. There exists the potential to compile datasets from many bat researchers to elucidate reproductive rates for various bat species. Barclay et al. (2004) began this process by compiling reproductive rates from the literature and testing predictions based on life history theory. Future compilations, with special reference to roosts versus random captures, are likely to further refine estimates of species’ reproductive rates and the factors that influence these rates.

269

Table II.1 Morphological measurements of adult non-pregnant females of four species of bats from southeastern Alberta. Sample sizes are in brackets after the mean ± SE.

Forearm Length Species Mass (g) (mm) M. ciliolabrum 5.4 ± 0.1 (65) 32.4 ± 0.1 (254) M. evotis 6.6 0.1 (79) 38.0 0.1 (110) M. lucifugus 8.6 ± 0.1 (86) 37.7 ± 0.1 (86) E. fuscus 21.0 ± 0.3 (59) 47.7 0.1 (220)

270

Table II.2. Captures of bats in Alberta and Montana for which capture times were known. Capture times were divided into the dusk/dawn (2 hours after dusk and 2 hours before dawn) or mid-night (middle of the night) periods for males, and both reproductive

(Repro) and nonreproductive (Non-repro) females. Reproductive status was not known for all females, especially if captures occurred early in the season when pregnancy was difficult to determine. All mistnetting sessions took place in river valleys and were away from known roosts, and each session lasted for at least the dusk and mid-night periods.

Capture Period Species Sex Group Dusk/dawn Mid-night E. fuscus M All 131 (61.5%) 82 (38.5%) F All 61 (76.2%) 19 (23.8%) Repro 36 (92.3%) 3 (7.7%) Non-repro 13 (48.1%) 14 (51.9%) M. lucifugus M All 117 (67.2%) 57 (32.8%) F All 71 (75.5%) 23 (24.5%) Repro 50 (77.8%) 14 (22.2%) Non-repro 6 (46.2%) 7 (53.8%) M. ciliolabrum M All 255 (68.7%) 116 (31.3%) F All 323 (83.9%) 62 (16.1%) Repro 154 (88.5%) 20 (11.5%) Non-repro 60 (73.2%) 22 (26.8%) M. evotis M All 74 (61.2%) 47 (38.8%) F All 64 (61.5%) 40 (38.5%) Repro 21 (58.3%) 15 (41.7%) Non-repro 9 (47.4%) 10 (52.6%) Total M All 577 (65.6%) 302 (34.4%) F All 519 (78.3%) 144 (21.7%) M+F All 1096 (71.1%) 446 (28.9%)

271

Table II.3. Odds ratios for captures of reproductive females in the mid-night period.

Denominator species Numerator Species Statistic E. fuscus M. ciliolabrum M. lucifugus M. evotis odds ratio 8.57 ± 5.91 5.5 ± 2.27 2.5 ± 1.14 z 3.12 4.13 2.02 p 0.002 0.000 0.044 M. lucifugus odds ratio 3.43 ± 2.31 2.20 ± 0.847 z 1.83 2.05 p 0.067 0.041 M. ciliolabrum odds ratio 1.56 ± 1.01 z 0.69 p 0.492

Table II.4. Number of reproductive and nonreproductive females of each species in general riparian netting (“random captures”), and when captures from known building roosts near the river valleys were included.

Species Captures Reproductive Nonreproductive % Nonreproductive E. fuscus General 50 41 45.1 + Roosts 227 63 21.7 M. lucifugus General 55 33 37.5 + Roosts 311 104 25.1 M. ciliolabrum General 197 154 43.9 M. evotis General 44 33 42.9