FRESHWATER PHYLOGEOGRAPHY:THE IMPACT OF LIFE HISTORY TRAITS

ON THE POST-GLACIAL DISPERSAL OF ZOOPLANKTON IN NORTH AMERICA

A Thesis

Presented to

The Faculty of Graduate Studies

of

The University of Guelph

bY

ANDREA JOAN COX

In partial hl fi Nment of the requirements

for the degree of

Master of Science

March, 200 1

(3Andrea Jorn Cox, 200 1 National Library Bibliothèque nationale If1 of Canada du Canada Acquisitions and Acquisitions et Bibliographie Seivices services bibliographiques 395 Wellington Street 395. rue Wellington Ottawa ON KIA ON4 OttawaON K1AON4 Canada Canada

The author has granted a non- L'auteur a accordé une licence non exclusive licence allowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan, distribute or sel1 reproduire, prêter, distribuer ou copies of this thesis in microform, vendre des copies de cette thèse sous paper or electronic formats. la forme de microfiche/film, de reproduction sur papier ou sur format électronique.

The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts from it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doiverit être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation. ABSTRACT

FRESHWATER PHYLOGEOGRAPHY: THE IMPACT OF LIFE HISTORY TRAITS

ON THE POST-GLACIAL DSIPERSAL OF ZOOPLANKTON IN NORTH AMERICA

.Andrea Joan Cox Advisor: University of Guelph. 200 1 Professor P.D.N. Hebert

This study used molecular markers to examine patterns of genetic variation and post-glacial dispersa1 in three species of freshwater zooplankton fiom glaciated North

America.

A survey of both mitochondrial and allozyme variation in the cladoceran Sida crysrollino revealed the presence of four allopatric assemblages derived from separate

Wisconsinan refupia. Divergence between phylogroups was deeper than that previously observed for refugial lineages of freshwater fishes subject to the same glaciation events.

Examination of mtDNA variation at wo genes in the arctic anostracans

Branchinec~apuludosa and Arterniopsis ste+soni revealed a divergent phylogeographic history. A. stefanssoni populations denve from a single glacial refuge, while B. paludosa occupied several refugia The distribution of these phylogroups supports the presence of hi& arctic refugial lineages of both species in the Canadian arctic.

Results fiom this study highlight die importance of life history characteristics including body size, diapause, and generation tirne in determinhg the ability of different tava to persist in glacial refugia during the Pleistocene. white birds cal1 across the arctic hindra voices on the wind

each bird caries our future our answer determines theirs

Dina ElCox Legacy. CAA MTB Press, 2000 ACKNOWLEDGEMENTS

This thesis represents immeasurable amounts of love and friendship that I received from many speciai people. The boundless encouragement and support that I have received from my family, and many wonderful fnends has been a constant source of strenkqh and inspiration.

Funding for this research was provided by NSERC, to Paul D.N. Hebert. Salary support was provided by Ontario Graduate Scholarships in Research and Technology,

University of Guelph Graduate Entrance Awards. and a Norman James Aquatic Ecology

Sc holarship. Addi tional support was provided by the Northem Studies Training Program

( Canada). and the Polar Continental Shelf Program (Canada).

I would like to thank my advisory cornmittee Dr. Paul Hebert. Dr. Teri Crease. and Dr. Roy Danzrnann for their guidance, input and direction. Dr. Fu and Dr. Lynn. mrmbers of my examining cornmittee. provided important suggestions which improved the final version of this thesis.

My passion for dl things arctic has been cultivated and encouraged by Paul

Hebert since 1994. His enthusiasrn for the north has certainly more than strengthened mine. I am incredibly grateful for the nurnerous opportunities that he has provided me with since 1 fint began my 'nordiem joumey? on the Arctic Ecology field course as an undergraduate. I would like to thank Paul for providing me with a graduate experience that was not only c hallenging, but highl y rewarding.

My nonhem journey also began with Chad Rowe - his constant fncndship has enriched my experience at Guelph. Jonathan Wiîi has ben both a fnend and mentor to me. and his influence is seen throughout the pages of this thesis. Chapter 2 was written in collaboration with Dr. Paul Hebert. Sara Adamowicz,

Teri Crease. Ryan Gregory and Jon Witt provided many heipful comments on early drafts of the manuscript. Rob Dooh John Havel, Sheila Potter, James Rhydderch, Chad Rowe,

Anna Sands. Young Um and Jon Witt assisted with North Amencan collections. while

Henk Ketelaars. Tom Little. and Klaus Schwenk provided samples From Europe.

Arctic collections for Chapter 3 were made possible by the superb staff of the

Polar Continental Shel f Project in Resolute and Tuktoyaktuk, and John MacDonald at the Eastern Arctic Scientific Resource Centre in Igloolik. Dr.3 Denton Belk. Richard

DutTord. and Jim Saunden provided direction and contacts regarding B. paludosa collection sites in the US.. Robert Mussleman of the U.S. Department of Agriculture

3nd Chad Rowe provided Bronchinecta from Wyoming, while Larry Weider supplied specimens from the Tatra Mountains of Slovakia. Many people helped collect fairy shrimp in the Canadian no&: Benoit Boulet. John Colboume. Alison Derry. Rob Dooh.

Tyler Hoar. Chad Rowe. Tom Little, Derek Taylor, Dave Totararn. Larry Weider. and

Jon Witt. I will never forget the 600 feet as1 conversations with Derek Taylor, John

Colboume. Tom Little. and Dave Totararn - i can't think of a better place to taik evolution vs. creationism than north of sixty in a Bell 206L. Rob Hodgson. and Debbie

'Sunshine' still give me energy.

I would like to thank many members of the Hebert Laboratory, both past and present. for providing both friendship and advice 24/7: Young Um is a tnie fiend who 1 admire and respect for her pursuit of her dreams. Sara Adamowicz Junjian Chen. Alina

Cywinska Rita De Melo. Alison Demy, Jeremy Dewaard, Rob Dooh, Teme Finston.

Ryan "the phantom" Gregory. David Hardie. Ed Remigïo, James Rhydderch. Veronika Sacherova. Liz Strazinsky. Graham Thompson. Heather Pollard, and Chris Wilson contributed much to this thesis. Tom Little and John Colbourne are first class scientists, and encouraged me from day one.

I would also like to thank many other department rnembers for their assistance and friends hip: Ron Brooks. David Barker. Mary-Anne Davis. Alison Derry, Suzanne

Gray. Diana Hamilton, Chris Henschel, Angela Hollis, Swan Mannhardt, Rob

4lacLaughlin. Seanna McTaggart. Tom Nudds. Cheryl Prokopowich. Ian Smith. Derek

Taylor. lrene Teeter. Dara Torgerson, and Kelly Wells. I thank Rob Willson for many shared Irish Crearn coffees. and philosophizing about life. Many. many thanks to Pat and

Sandy Skinner for entxusting me with "the Glen", where Chapter 3 was completed.

Jennifer Cnizet. Melinda PorteIli. Holly Spiro. Shauna Schock, Young Um. James

Rhydderch. Ken Oakes (oceans, anyone?), Kam Vlasman, Geraid Tetrault Chad

McAlpinc and Marie McGrath are like family to me and provided endless friendship. srniles and laughter - it is my pnvilege to cal1 you fnends. Vemnika Sacherova, Jeremy

Dewaard. and Rob Dooh went above and beyond the cd1 of duty as both colleayies and

Fnends - your support has meant more than any of you could possibly realize!

For always being hirnseif. 1 would especially like to thank Graeme Skinner. who reminded me to enjoy the journey dong the way.

My parents. David and Dina my sisters Pamela and Elizabeth. and my brother

David have never ceased to amaze and inspire me in their continued support (and patience!) for my pursuit of al1 things biological. It is from their unconditional love and encouragement, that 1 draw the strength and inspiration to follow my dreams. TABLE OF CONTENTS

AC KNOWLEGEMENTS ...... i

TABLE OF CONTENTS...... iv .. LIST OF TABLES...... VII

.*. LIST OF FIGURES...... WII

LIST OF APPENDICES ...... x

CHAPTER 1

General Introduction...... 1

Li terature C i ted...... 7

CHAPTER 2

Colonization, extinction, and phylogeogrnphic patterning in a freshwater

crustacean...... 1 1

Abstract...... 12

Introduction...... 13

Materials and Methods...... 15

Specimen collection...... 15

..\llozyme electrophoresis...... 17

Mitochondrial DNA analysis...... 19

Results...... 2 1 . . Species identification...... 21 . . Nuclear genetic vanation...... 2 1 . - MtDNA sequence vanation...... 23

Divergence estimates and refbgial or&@...... -27 Discussion ...... 33

Comparative phylogeography ...... 37

Literature Cited...... -40

CHAPTER 3

Phylogeograpby of arctic anostracans: diapause and the occupaney of high arctic refugia ...... -47

A bstract ...... A8

Introduction ...... 49

Materials and Methods ...... A3

Specimen collection and species distributions ...... 53

MtDNA anal ysis ...... -58

Resul ts ...... *.62 ... S pecies distributions...... 62

Gene sequence diversity ...... 62

Artemiopsis stefanssoni ...... 62

COI gene sequence diversity and phenetic analysis...... 62

16s gene sequence divenity and phenetic anaiysis...... 63

Brunchinecta paludosa ...... 67

C O1 gene sequence diversity and p henetic analysis...... 67

C ladistic analysis of sequence divergence in the COI gene ...... 67

16s sequence diversity and phenetic analysis...... 70

Divergence time estimates and refugial ongins ...... 72

Discussion...... -80 Distribution and refugial origins ...... 8 1

Polar phylogeography ...... 87

Literature Ci ted ...... -90

CHAPTER 4

Generd Conclusion ...... 100

Li terature Ci ted ...... 106

APPENDKES ...... 108 LIST OF TABLES

CHAPTER 2

2.1 Allozyme diversity data based on seven loci for 54 populations of S. crysiallina

from North Amerka...... -22

2.2 F-statistics for three-level hieruchical analysis of five polymorphic allozyme loci

among 54 North American populations of S. cry~tallina...... -. 26

CHAPTER 3

3.Mean sequence divergence and divergence estimates between B. paludosa

phylogoups for the COI and 16s genes ...... 78

vii LIST OF FIGURES

CHAPTER 2

2.1 Map of North Amenca showing sampling locations for S . crystallina ...... 16

2.2 UPGMA dendrograrn of genetic distances among one European and 54 North

Amencan populations of S. crystaIha ...... 25

7.3 Neighbour-joining tree of COI variation in S. crysfailina...... 29

2.1 Parsimony tree for COI nucleotide variation in S. crystallina ...... 30

2.5 Geographic distribution of four S. crystullina clades in North Amenca ...... 31

CMAPTER 3

1 Map illustrating the distribution of A . siefanssoni in North Amenca ...... 54

3.2 Maps illustrating the distribution of B. paludosa in North Amenca (3.2.1) and

Europe (3.2.2) ...... 56

3.3 Map showing areas sampled for A . stefamsoni and B . paludosa in the Canadian

arctic archipelago ...... 57

5.4 Neighbour-joining phenogram of COI variation in A . stefamsoni ...... 65

3.5 Xeighbour-joining phenogram of 16s rDNA variation in A . stefunssoni ...... 66

3.6 Neighbour-joining phenogram of COI variation in B. paludosa ...... 69

3.7 Parsirnony tree of COI nucleotide variation in B. paludosa ...... 71

3.8 Neighbour-joining phenogram of 16s rDNA variation in B. puludm ...... 73

3.9 Geographic distribution of A . stefanssoni populations in the Canadian arctic,

showing the location of the most and least divergent haplotypes...... 74

3.1 O Geogaphic distribution of B. paludosa arctic archipelago phylogroups, identified by

the COI and 16s genes ...... -76 3.1 1 Map illustrating proposed Wisconsinan refugial ongins for the 7 phylogroups of B.

poludosa in Norih America ...... -. . .. -77 LIST OF APPENDICES

CHAPTER 2

2.1 Locations and abbreviations for North American and European habitats with S.

crys~ullina...... IO9

2.2 Summary of allele frequencies in North Arnencan and European populations of S.

crystullina at five polymorphic allozyme loci...... 1 12

CHAPTER 3

3.1 Location details and abbreviations for populations of A. stefanîsoni (3.1.1 ) and B.

palrcdosa (3.1.2) examined in the study...... 1 17

32 Sumrnary of c haracteristics for habitats occupied by A. sfefanssoni and B. paludoso

listed in Appendix 3. 1...... -121

3.3 Pairwise sequence divergences for isolates of A. stefanrsoni for both COI (3.3.1) and

16s rDNA (3.32) mitochondnal genes...... 122

3.4 Pairwise sequence divergences for isolates of B. paludosa for both the COI (3.4.1)

and 16s rDNA (3.4.2)mitochondrial genes ...... 127 CHAPTER ONE

GENERAL INTRODUCTION "Amongst biologists it has, for obvious reawns, been customiq to regard the glaciations as periods of desolation and death...... this view should be balanced by realization of the role that ice caps have played as dispersai agents for certain animals,...... the study of the prehistory of these animals has pmvided zoogeography with one of its most fascinating c hapters"

Sven Ci. Segerstale 1976: 83:2- 15

Much of the diversity of the North Arnerican landscape reflects the -~romorphological products of repeated glacial influence (Dyke & Prest 1987: Trenhaile

1990). At the height of the Wisconsinan glaciation 18-20 000 years ago, ice sheets coverrd more than 60% of the continent. creating enormous barrien to the movements of many species. effectively dividing component populations into isolated groups residing in disj unct habitable areas (Rogers et. al 199 1 : Dawson 19%). Virtually every specits from the once ice-covered areas of North America is descended fiom ancestors who survived in these ice free areas or 'refugia' during glacial maxima (Pielou 1991). As the ice sheets receded. massive proglacial lakes presented banien to dispersal for terrestrial animals and plants. but corridors for the reintroduction of freshwater fishes and other aquatic organisms throughout Canada and no&-central USA (Black 1983; Bermingham

& Avise 1986: Pielou 199 i : Mandrak & Crossman 1992; Wilson 1995).

Previously, biogeographic studies on both arctic and temperate species relied on distnbutional data to reconstnict the colonization routes exploited by plants and animals to establish their curent distributions (Carter et. al 1980: Ho fihan 198 1; Munay 198 1;

Black 1983; Mandrak & Crossman 1992). However. the recent development of

molecular technoiogies capable of probing the genetic impacts of Mcariance events, has allowed the application ofgene genealogies to the geographic distributions of organisms. This new subdiscipline of biogeography, temed 'phylogeography', is providing novel insights as to how vicariance events have influenced the geographic distribution of genetic variation arnong modem populations (Avise et al. 1987; Avise 2000; HewÏn

2000).

Mitochondrial (mt) DNA has proven to be especiaily usefd for studies of phylogeognphy (Avise et. al 1987; Avise 1992) as the stochastic loss of mtDNA diversity during population bottlenecks. such as those occurring during post-glacial dispersa1 events. is not readily recovered in brief periods of population expansion (Avise et al. 1987: Avise 1994). In addition. the time since the Iast deglaciation has not been sufficirnt for locai dispersal and gene flow arnong populations to erode mitochondnai divergence ihat arose due to the isolation of populations in separate glacial refugia during penods of glaciîl maxima (Bermingham & Avise 1986; Hewia 2000). Depending on the size of a particular refuge and the number of populations residing within it. genetic drift and foundrr effects caused divergence of fonnerly panmictic populations that rernains rrcognizable usinç molecular marken (Hoffman 198 1 ; Hewitt 1996: 2000). As a result. mtDNA haplotype distributions cmbe used to trace species dispersal routes and distinguish between populations originating From separate refbgia

In addition to providing support for the importance of vicariance events, phylogeographic studies are revealing that life history traits. such as dispersal mechanisms, have had significant impacts on both current and historic species distributions (Carter et al. 1980: Wilson & Hebert 1998). For example. most small terrestrial mammais exhibit deep genealogicai subdivisions over broad geographic scales. attributable to their lirnited vagility and the impact of physical barriers on their disped (.Avise 7000). In contrast. the presence of genetic discontinuities in many large terrestrial and marine mamrnals are oflen better explained by behaviouml characteristics, such as philopatry and group alliances, despite their ability to disperse over vast geographic ranges (Avise 2000). Numerous studies of other species including bats and birds

( reviewed by .Avise 2000) have also revealed patterns of genetic variation strikingly discordant from that which might be predicted fiom dispersal ability alone, illustrating that the ro!e of dispersai in the production of phylogeographic patterns is complex.

Phylogeographic studies on life in freshwater habitats of North America have focused mainiy on actively dispersed freshwater fish. This work has revealed that phy logeographic patterns are generail y congruent with those expected assuming that lineagcts dispened through fluvial connections present during deglaciation (Bermingham

& Avise 1986; Bematchez & Dodson 199 1; Wilson 1995).

Although the freshwater zooplankton are substantially more speciose than fishes. little eron has ken directed towards the survey of their phylogeographic structure.

Moreover. these grooups possess several life history features that might be expected to influence phylogeographic patterns. First, due to their limited abiiity to move actively between habitats, many zooplankton species have developed desiccation and freeze- tolerant resting eggs which facilitate overland dispersal to new habitats by wind or as

'hitchhikers' in the fur. feathen and digestive tracts of marnrnalian and avian vectors

(Maguire 1 963; Proctor 1964; Proctor & Malone 1965). Consequently. zooplankton are round in many habitats from whicb fish are absent. including both high alpine and arctic ecosystems (Hebert & Ham 1986). Second, during repeated glacial cycles, these diapausing eggs would have permitted refoundhg of populations 'in-situ' fiom sediment rgg banks. bypassing the need for colonization by propagules from distant populations

(Boileau & Hebert 1991 ). Finaily, zooplankton reproduce rapidly ;species that reside in the arctic hatch. reach rnaturity and reproduce in a short four month growing season

(Jobansen 192 1 ). This attribute may lead to the greater accumulation of mutations than would be expected in vertebrate species over the same intervals of evolutionary tirne. providing the potential for detailed phylogeographic resolution in geologically 'young' reçions.

Past studies of freshwater zooplankton. which concentrated mainly on circumpolar memben of the Daphnia pulex complex, revealed little genetic divergence ocross their entire range (Colbourne et al. 1998). Although a dominant member of freshwater zooplankton assemblages, Daphnia is one of the few genera to produce tloating ephippia: the resting eggs of moa oiher fieshwater invertebrates si& or stick to substrates (Pennak 1989). As generalizations of phylogeographic histories may be limited to ecologically equivalent groups of species (Wilson 1995). it is unlikely that results for Daphnia are representative of the fieshwater zooplankton at large.

Examination of other zooplankton species with di fferent ecological attributes and distributions should provide a better understanding of the patterns of genetic diversity in freshwater organisms in general.

This thesis aims to extend understanding of phylogeography of freshwater life in glaciated North America It uses zooplankton as a mode1 system to examine how life history characteristics influence the geographic distribution of genetic variation across species' ranges. By investigating the implications of traits associated with passive dispersai following the vicariance events of glaciation, these data will help to clarify the role of dispersai strategy on the evolutionary implications of histoncal events in detemining patterns of divenity on macrogeographic scales.

Chapter 2 uses a comparative approach to examine how the life history characteristics associated with passive versus active dispersal differentidly influence the contemporary distribution of genetic variation in freshwater organisms. Both mitochondrial and nuclear markers are used to assess intraspecific phylogeography of

Sidu crys~allina.a cladoceran broadly distributed in North Arnenca Results are compared to data fiom fieshwater fish in order to advance Our understanding of how the contrasting life history strategies of these freshwater organisms have affected their rvolutionary trajectories.

Chapter 3 Merinvestigates the effects of vicanant events on the mitochondrial grnomes of passively dispened invertebrates by addrrssing genealogical concordance across two species of anostracans that occur in the Canadian arctic. Using data fiom two mitochondrial genes. this chapter examines the implications of life history traits associated with passive dispersal on freshwater organisms in a geologically 'young' region of North America and provides some of the first phylogeographic data for zooplankton which reside in the hi& arctic. LITEFUTURE CITED

Avise JC. Arnold J, Bail RM, Bermingham E, Lamb T, Neige1 JE, Reeb CA, Saunders

NC ( 1987). Intmspeci fic Phylogeography : The mitochondrial DNA bridge

between popuiation genetics and systematics. Annual Review of Ecology and

Sysirema 1ics. 18,489-522

Avise JC ( 1992) Molecular population structure and the biogeographic history of a

regional fauna: a case history with lessons for conservation biology. OiRos. 63,

62-67

.-\vise JC ( 1994) iLfolecularMarkers, lVutural His~ory,and Evolution..pp.5 1 1. Chapman

and Hall. New York.

Avise JC (2000) Phylogeography. The History and Formation oj'spcies. pp. 447.

Harvard University Press, London.

Bermingharn E. Avise JC (1986) Molecular zoogeography of freshwater fishes in the

southeastem United States. Genetics, 113.939-965.

Bematchez L, Dodson JJ ( 199 1) Phylogeographic structure in mitochondrial DNA of the

lake whitefish (Coregonus clupeafomis) as revealed by mitochondrial-DNA

restriction analysis. Evolurion. 44. 10 16- 1035.

Boileau MG. Hebert PDN ( 199 1 ) Genetic consequences of passive dispersal in pond

dwelling copepods. Evolution, J5,72 1-733.

Black GA (1983) Cysîidicola farionis (Nematoh) as an indicator of lake trout

(Saiveninus namaycush) of Be~gancestry. Canadian Journal of Fisheries and

Aquatic Sciences, 40.2034-2040.

Carter JCH. Dadswell MJ. Roff JC, Spdes WG (1980) Distribution and zoogeography of planktonic crustaceans and dipterans in glaciated eastern North

Amerka. Canadiun Joutnal of Zoologv. 58. 1 3 55- 13 87.

Colbourne K. Crease TJ, Weider Uet al. (1998) Phyiogenetics and evolution of a

circumarctic species complex (Cladocera: Daphnia pulex). Biological Journal of

the Linnean Sociem 65,347-365.

Dawson AG (1 992) Ice Age Emh: Late Quaterna- Geology and Climate. Routledge

Press. London.

Dyke AS. Prest VK ( 1987) Late Wisconsin and Holocene history of the Laurentide

ice sheet. Grogruphie Physique et Cxternaire. 41.237-263.

Heben PDN. Ham BJ ( 1986) Patterns in the composition of arctic nindra pond

m icrocmstacean communities. Canadian Journal of Fisheries and Aqwtic

Sciences, 43, 14 1 6- 1425.

Hewin GM ( 1996) Sorne genetic consequences of ice ages. and their role in divergence

and speciation. Biological Journal of the Lhnean Society, 58.247-276.

Hewitt GM (2000) The genetic legacy of the Quatemary ice ages. Nature. 405.907-913.

Hot'tinan RS (198 1 ) Dif'ferent voles for different holes: environmental restrictions on

reîùgial survival of marnmals. In: Evolurion Today: Proceedings of the Second

lnternafional Congress of Systematic and Evolutionary Biolog~(eds. Scudder GE,

Reveal JL). pp24-25. Hunt Institute for Botanicd Documentation. Pittsburgh.

Johansen F (1921) The larger fieshwater crustacea from Canada and Alaska The

Cànadian Field Narrrralist. 35,2 1 -3 0.

Maguire B Jr (1963) The passive dispersal of small aquatic organisrns and their

colonization of isolated bodies of water. Ecologicuf Monographs?33, 16 1-1 85. Mandrak NE. Crossman EJ (1 992) Postglacial dispersal of fieshwater fishes into Ontario.

Canadian Journal of Zoology, 70.2247-2259.

Murray DF ( 198 1) The role of arctic refugia in the evolution of the arctic vascular fiera

a Beringian perspective. In: Evolution Todùy: Proceedhgs of the Second

Interriational Congress of Sysrematic and Evolutiona?y Biology (eds. Scudder GE,

Reveal JL), pp 1 1-20. Hunt Institute for Botanical Documentation, Pittsburgh.

Pielou EC ( 199 1 ) AAer the Ice Age: the Remof Life to Glaciated North Arnerica. pp.

3 66. University of Chicago Press, Chicago.

Pennak RW ( 1989) Freshwater Invertebrotrs of the United States. jrd ed. John Wiley &

Sons.. New York.

Rogers RA. Rogen LA. Hofmiann RS. Martin LD (1991) Native Arnerican

biological divenity and the biogeographic influence of Ice Age refugia Journal

of.Biogeography. 18,623-630.

Trenhaile AS ( 1990) The Geomorphology of Canada. pp. 240. Oxford University Press,

Toronto

Proctor V W ( 1964) Viability of crustacean eggs recovered From ducks. Ecologv, 45.

656-658.

Proctor VW. Malone CR (1 965) Further evidence of the passive dispersal of small

aquatic organisms via the intestinai tracts of birds. Ecology. 46, 728-729.

Wilson CC ( 1995) P hylogeography and postgZacial dispersal of two North A merican

Salvelinus species. Ph.D. Thesis. University of Guelph, Guelph. Ontario.

Wilson CC. Hebert PDN (1 998) Phylogeography and postglacial dispersal of lake trout ( Salvelinus namuymh) in North Amenca. Canadian Journal of Fisheries and

Aquatic Sciences. 55. 1O 1O- 1024. CHAPTER TWO

Colonization, extinction, and phylogeographic patternhg

in a freshwater crustacean ABSTRACT

Phylogeographic analyses have revealed the importance of Pleistocene vicariance cvents in shaping the distribution of genetic diversity in Freshwater fishes. However, few studies have examined the patteming of variation in fieshwater organisms with diffenng dispersal syndromes and life histones. The present investigation addresses this gap, exarnining the phylogeography of Sida crysrallinn, a passively dispersed fieshwater crustacean common in North Amenca and Eurasia. The present analysis has revealed deep allozymr and cytochrome oxidase I mitochondrial DNA divergence between populations from North America and Europe. Moreover. North American populations are sepmted into four ailopatric assemblages, whose distribution suggests their derivation from different Pleistocene rehgia. These lineages show higher haplotype diversity and deeper sequence divergence than those of any fish fiom temperate North

America. Its distinctive life history traits have evidently sheltered lineages of S. crystallina fiom extinction, contributing to a remarkably comprehensive and high resolution phy logeographic record. INTRODUCTION

The biota of North Amerka was profoundly affected by the recurrent glacial advances and retreats during the Pleistocene. Life in formerly glaciated regions consists of immigrant taxa denved from either Benngia or ice-fiee areas in the temperate zone, while life in the south is comprised of assemblages exposed to range adjustments as species tracked shifling climate regimes (Hewitt 1996; 2000). Until recently, efforts to rrconstruct histones of habitat occupancy and rates of range extension were conjectural, based on the inspection of modem distributions and the identification of morphological discontinuities arnong populations (Dadswell 1974: Hocutt & Wiley 1986; Stemberger

1995). However. phylogeographic studies are adding new ngor to the field by enabling a cri tical evaluation of refugial origins and histones of habitat tenancy (Avise et al. 1987;

1998). Such investigations are providing novel insights conceming the interactions of ecology. life history. and abiotic factors in shaping the evolution and distribution of extant populations on both local and continental scales (Berrningharn & Moritz 1998;

Avise 2000).

The life of inland waters has been the target for a considerable arnount of phylogeographic study (Bermingham & Avise 1986; Bematchez & Wilson 1998). Most work has focused on fishes - active disperses capable of migration against currents within watershed boundaries. These phylogeographic investigations have provided some generalizations. revealing, for exarnple, that most north temperate species include a nurnber of refugial lineages (Bematchez & Dodson 1991; Wilson & Hebert 1998). These studies have also shown that the lineages of fishes which are largely restricted to glaciated regions are Young, suggesting the vigorous pruning of diversity during the Plçistocene. By contrast, species in the unglaciated regions of North Arnerica include much older lineages (Billington & Hebert 199 1; Bematchez & Wilson 1 998).

Although work on fishes has now been comprehensive enough to provide a good sense of their phylogeographic patteming, it is unlikely that these results are representative of freshwater life at large. The impact of dispersal strategy on p hy logeograp hic structure merits particular attention. The most striking axis of variation in this regard relates to the reliance of some taxa, such as fishes, on active dispersai, while othen. such as the plankton. employ passive dispersal. It has long been assumed that regional genetic variation would be much more limited in the latter groups because of their ability to cross watershed boudaries (Danuin 1859: Mayr 1963). Moreover, their production of diapausing stages which remain viable for hundreds of years (Weider et al. 1997) would seem to provide additional security against extinction, perhaps making these organisms less prone to local lineage loss. The present study had the primary goal of advancing knowledge of phylogeographic structure in passively dispersed life. It firstly tests the expectation that regional genetic divergence is Iimited in such taxa It secondarily examines whether the patteming of genetic diversity in these organisms shows any correspondence to that observed in lineages emphying active dispersai.

Sida crystallina. the target for the present analysis, is a cyclically parthenogenetic cladoceran cnistacean found in the littoral zone of lakes and ponds throughout the

Holanitic. Its broad distribution haevidently been achieved through the dispersa1 of diapausing eggs which often adhere to substrates such as macrophytes (Fryer 1996). and are both fieeze-tolerant and digestion resistanî, providing an avenue for dispersal via migratory watertowl (Korovchinsky & Boikova 1996). Eurasian and North Amencan populations of Sida have ken classified as distinct subspecies because of subtle rnorphological differences (Korovchinsky 1992), suggesting that intercontinental gene tiow is limited. There is. however. no knowledge of the paiteming or extent of gene flow arnong populations on each continent or their history of habitat tenancy. Since S. crysrallîna is eurythermal. as evidenced by its occurrence from the subtropics to low arctic of North America (Korovchinsky l992), populations may have persisted in ice-free areas. despite shifting thermal regimes during the Pleistocene. Its prevalence in regions which were both ice-covered during the Pleistocene and distant fiom glacial refkgia also indicates that many modem populations owe their origin to recent range extension.

The present study examines the extent of genetic divergence between European and North American populations to ascertain the recency of contact beiween Sida populations from these areas. As well, a comprehensive analysis of genetic population structure in North American lineages is undertaken to evaluate their ongins and the rxtent of gene flow across this continent. Finally, the patteming and extent of geographic variation in this species is cornpared with that in both freshwater fishes and life reliant on passive dispersal.

MATERIALS and METHODS

Specimen CoIIection

Populations of Sida crystallinu were collected fiom 68 habitats in North Arnenca and three sites in Europe between June 1995 and September 1997 (Figure 2.1 ;Appendix

2.1 ). Based on their geographic origin, North Amencan populations of S. crystallina were assigned to one of seven freshwater biogeographic provinces according to Burr & Figure 2.1. Map of North America showing sampling locations for S. crystal~im- Biogeogaphic provinces (fiom Burr and Mayden 1992) sampled in this snidy are abbreviated as follows: Cascadia (CD), Great Lakes (GL), Hudson Bay (HB).Mississippi (MI). North Appalachian (NA), Southeastern (SE), and Yukon-Mackenzie (YM). Numbers in parentheses indicate the number of populations obtained kom each region. The large. diagonaily striped circle in the GL region overlies 20 habitats. Mayden ( 1992): Cascadia (CD), Great Lakes (GL), Hudson Bay (HB), Mississippi (MI),

North Appalachian (NA), Southeastem (SE), and Yukon-Mackenzie (YM). Derived from surveys of fish assemblages, the boundaries of these provinces provide an ideal framework against which to examine biogeographic patteming in other freshwater life, as they reflect the historical structure of drainage systems on the continent (Hocutt & Wiley

1986). North Amencan populations were assigned a two digit code (01-68) followed by a two lrtter code indicating their biogeographic province of origin, while European collections were identified as El-E3. followed by a two letter country code. Specific details for each habitat locality including latitude and longitude are provided in Appendix

2.1.

Sarnples were obtained using a 250 pm mesh net towed through littoral zone

macrophytes. sorted in the field. and either flash-frozen in liquid nitrogen or preserved in

90% ethanol for subsequent molecular analyses. Representatives of each population were

later examined morphologically for the diagnostic features whch ordinarily differentiate

Eunsian and North Arnerican forms (Korovchinsky 1992). Due to variation in sample

size and preservation method. not ail populations were used in both DNA and allozyme

analyses (see Appendix 2.1 ).

A lloqyme EIectrop h oresLr

Levels of allozyme variation were examined in 54 populations. Whcrever

possible. 20-40 individuals were examined for variation at seven cornmonly polymorphic

and well-resolved loci using standard cellulose acetate electrophoresis (Hebert & Beaton

1 993). Enzymes screened included arginine phosphokinase ( APK, EC 2.7.3.3),

supernatant aspartate amino tramferase (sAAT, EC 2.6.1.1), fiunarate hydratase (RTM, EC 4.1.1 2).glucose-6-phosphate isomerase (GPI. EC 5.3.1.8), supernatant mdate dehydrogenase (MDH, EC 1.1.1.37). mannose-6-phosphate isomerase (MPI, EC 5.3.1 A), and phosphoglucomutase (PGM,EC 5.4.2.2). Two individuals fiom Beech Lake (49 GL) were included as reference standards in each gel ru. The dominant allele at each locus in the standard population was assigned a relative mobility (Rr ) value of 1 .O; the mobilities of al1 other alleles were scored with respect to this standard. Where sample sizes were sufficient. side-by-side cornparisons were made to confirm the presence of putative novel alleles.

Analysis of allozyme data and calculations of descriptive and hierarchical population statistics were carried out using Genetic Data Andysis (GDA) software

(Lewis & Zaykin 1999). Genotypic fiequencies were detemined for each population. and polymorphic loci (0.99 critenon) were examined for deviations (a= .OS) fiorn

Hardy- Wein berg (HW)equilibrium using Fisher's exact test, followed by sequential

Bonferonni corrections (Rice 1989). Nei's genetic distance (Nri 1978) was used to rstimate a UPGMA phenogram to examine relationships among populations.

Bootstrapping was subsequently performed over loci in order to test the significance of population clustea within the phenogmn using the program Genetic Distance and

Phylogenetic Analysis (DISPAN) (Ota 1993).

The extent of allele Bequency divergence among populations was evaluated in

GDA using a three level hierarchical andysis of Wright's FST. nie hierarchy was constructed such that divergence was measured between individuals in single populations

(FIS).between single populations within a biogeographic province (FSp),and between biogeographic provinces across the entire geographic range (Fm). Confidence intervals (95%)were estimated for each level in the hierarchy by bootstrapping across loci (5 000 replicates).

MWAAnalysis

Total DNA was isolated from a single individual From each of 32 populations, and for 2-3 individuals from another 15 populations, using modified proteinase K methods descnbed by Schwen. (1996). individuals animais were homogenized in 50 pl of H3 extraction buffer ( 10 mM Tris-HCL pH 8.3.0.05 M potassium chloride. 0.005% Tween-

20 and 0.005% NP-40; Replitherm Reaction Buffer, Biozym) and 20 pg of proteinase K. and incubated overnight in a 50°C oven. Following denaturation of proteinase K via incubation for 12 minutes in a 94'C waterbath, extracts were stored at -20°C. The polymerase chain reaction (PCR)(Saiki et al. 1988) was subsequently utilized to amplify a 680 bp fragment of the mitochondnai (mt) cytochrome c oxidase subunit I (COI)gene with the primer pair LCOI490 and HC02918 (Folmer et al. 1994). Each 50-pl PCR reaction consisted of 5 pi of genomic DNA template, 9 rnM TrisHCl (pH 8.3). 45 mM

KCI. 2.2 mM MgC12, 0.26 pM of each dNTP (C.G,A.T). 0.36 pM of each primer, and

I .O U of Taq DNA polymerase (Qiagen Inc). The thermal regime for amplification consisted of 1 cycle of 1 minute at 94°C; 40 cycles of 1 min at 94°C. 1.5 min at 45T, and 1.5 min at 72°C; followed by 1 cycle of 5 minutes at 72°C. The product was gel purified using the Qiaex II kit (Qiagen hc), subjected to dye terminator cycle-sequencing reactions (25 cycles. 55OC annealing) with primer LCOI490 and ArnpliTaqO DNA polymerase FS (Perkin Elmer). and sequenced on an AB1 377 automated sequencer

(Applied Biosystems). Electropherogams were aiigned with the Seqapp 1.9a sequence editor (Gilbert

1992). and nucleotide compositions were calculated in MEGA 1.O2 (Kumar et al. 1993).

Preliminary phylogenetic analyses utilizing the sidid Diaphanosorna brachywum as an outgroup revealed that European S.c crystallina are a divergent sister group to al1 North

Amencan populations. Accordingly, European populations of S. c. crystailina were used as an outgroup for dl subsequent analyses. As cornparisons between sequences for the multiple individuals in 15 populations revealed that, in al1 cases. these individuals showed very limited sequence divergence (mean divergence=O. 1%; maximum divergence=O.j%) relative to divergences arnong populations (4-7%), al1 subsequent phenetic and cladistic analyses were carried out using one individual per population. A distance matrix of painvise nucleotide sequence divergence was calculated using the

Kimura two-parameter mode1 (Kimura 1980). and used to estimate a Neighbour-Joining

(NJ) phenogam with confidence limits determined in MEGA (1 000 bootstrap pseudoreplicates). Maximum parsimony (W)heuristic searches were conducted in

PAUP* 4.0b2 (Swofford 1998). with steepest descent and tree bisection recomection options invoked. Transitions and transvenions were weighted equally as differential weightings did not affect the topology of the phylogeny. Confidence in the cladistic analyses was also assessed in PAüP*, both a priori, by estimation of the gl skewness statistic from 100 000 random tree length disnibutions (Hillis & Huelsenbeck IW), and a posteriori. by bootstrap analysis with 1 000 pseudoreplicates.

Estimated rates of sequence divergence for COI range from 1.4-2.6% per million years (Knowlton et al. 1993; Knowlton & Weight 1998; Schubart et al. 1998). In addition to the inherent problems in al1 rate estimates (Hillis & Mona 1990; Martin & Palumbi 1993), no currently available calibrations for COI are based on the precise frarnent- amplified in the present study. Therefore, we chose an intemediate dock rate of 2% per million years for COI divergence time estimates. which is comparable to the arthropod mtDNA dock described by Brower (1994), as well as previous work by Brown et al. ( 1979).

RESULTS

Species Idenrification

Al1 individuals from North Amerka were unambiguously assigned to the subspecies Sida crystaIlina americana. while Eutopean populations were identified as

Sida cystallinu crystallina based on differences in the number of teeth on the post- abdominai claw (Korovchins ky 1 992). Males were only detected in collections from the

Yukon-Mackenzie biogeographic province, and in one late autumn collection from the

Great Lakes province. lVuclear Geneiic Variation

Allozyme variation was detected at hve of the seven loci (Appendix 2.2). The percentage of polymorphic loci in North American populations ranged from O - 71% wirh a mean of 3 1K (Table 2.1 ). An average of 1.4 alleles per locus were detected, while individual heterozygosities ranged fiom 0-22%. with seven populations monomorphic for al1 loci. Mean heterozygosities were highest in the Great Lakes biogeographic province, moderate in Hudson Bay. Yukon-Mackenzie. and Cascade regions. and low in the North

Appalachian and Southeastern provinces (Table 2.1). The single European population examined for allozyme variation showed 3% heteroygosity and 14% polymorphic loci, Table 2.1. Allozyme divenity based on a survey of 7 loci in North American populations of S. crysrallina.. Data are grouped by biogeographic province: CD = Cascadia. GL =

Great Lakes, HB = Hudson Bay, NA = Northem Appalachian, SE = Southeastem, YM =

Yukon-Mackenzie. N = mean number of individuals analyzed per locus, P=percentage pol ymorphic loci (0.99 criterion). A, = mean number of alleles per polymorphic locus. H,, = observed heterozygosity. He = rxpected heterozygosity. * = deviation from Hardy- Weinberg expectation der sequential Bonferonni tests (k=54, P

Region N P Ap Ho He HUDSON BAY

33 HB 19.0 14 2.00 0.02 0.02 mean 25.7 26.9 2.1 O. 0.07 NORTH APPALACHIAN GREAT LAKES 60 NA 19.7 29 2.00 0.05 0.05 61NA 15.0 O - 0.00 9.00 û4 NA 54.0 O - 0.01 0.01 mean 29.6 9.7 2.0 0.02 0.02 SOUTHEASTERN 65 SE 22.0 O - 0.00 0.00 66 SE 20.0 O - 0.00 0.00 67 SE 15.0 14 2.00 0.07 0.05

YUKON-MACKENZIE 12 YM 19.9 43 2.00 13 YM* 39.3 71 2.60 14YM 28.0 14 2-00 15 YM 10.0 14 2-00 16 YM 20.0 O - 17 YM 40.0 14 2-00 18 YM 20.0 14 2.00 19 YM 30.0 43 2.00 20 YM 20.0 14 2.00 22YM 29.0 O - with fixed allelic differences at two loci (SUTand PGM) between it and ail North

American populations (Appendix 2.2). Most populations were in Hardy- Weinberg equilibrium (Table 2.1); only a single significant deviation. linked to heterozygote deficit, was revealed when the data were corrected for multiple cornparisons with sequential

Bonferonni tests (k=54.p<0.001) (Rice 1989).

A UPGMA dendrogram of Nei's genetic distance revealed marked allozyme divergence between North Amencan and European (D 2 0.70 ) populations. Most of the

North Amencan populations showed limited genetic divergence (D < 0. l), with similar allelic arrays in populations ranging fiom the Yukon-Mackenzie, Hudson Bay. North

Appalachian. and Southeastern biogeographic provinces (Figure 2.2). A few populations. such as ihose in Cascadia and some from the Hudson's Bay region. formed distinct clusters. but bootstrap support (not shown) for these groups of populations was weak (<

45%).

F-statistics indicated substantial differentiation (Table 2.2), with greater divergence among populations within single biogeographic provinces (Fsp = 0.48) than between provinces (FPT= 0.3 1 ). This high FSp value reflects the fixation of aiternate alleles at MPI in local populations. and the presence of othenvise rare GPI and PGM aileles at high frequencies in a few populations.

rntDNA Sequence Variation

The 6 14bp alignment of 47 COI sequences was unarnbiguous as no gaps were present. Fony North American and three European haplotypes (GenBank accession Figure 2.2. UPGMA dendropm of genetic distances (Nei 1978) among one European and 54 North Amencan populations of Sida crystallina. Abbreviations for population locations are in Appendix 2.1. Estimates of D are based on gene fkquencies at five polymorphic allozyme loci. No clusters were supported by bootstrap values greater than

45%. 02 CD 15 YM 04 CD 64 NA 16 ïM 22w 65 SE E6 SE 61 MA 68 SE 33 HB 36 GL aJHA N YM 14 13 VM 19 YM 67 SE 30 HB 10 CD 17 YM 7% YM 24 YM 39 GL 43 GL 06 GL 53 GL 47 GL 49 GL JB GL 54 GL 57 GL 50 GL Y GL 40 GL 5ô GL M GL 45 GL 51 GL 55 GL a4 CD 41 GL 42 GL 26 HB 29 HB 27 HB 28 HB 31 Ha 32 HB 12 YM 05 CD a5 CI) 07 CD OS CD €3 CH Table 2.2. F-statistics for three-level hierarçhical analysis of five polymorphic allozyme loci among 54 North American S. crystalha populations. FK = inbreeding coefficient for individuals within populations; FSp= coancestry coefficient among populations within a region; FPT=coancestry coefficient for different biogeographic provinces. CI = upper and lower bounds of 95% confidence intervals obtained by bootstrapping over loci (5 000 replicates)

Locus FIS FSP Fm FIT

Mpi 0.05 Pgm 0.1 3 sAat 0.iO Mdh O. 16 Gpi -0.03 ctmbers AF277849 - AF277891) were detected with 169 variable sites, 157 of which were phylogenetically informative. Neighbour-Joining analyses indicated the presence of four phy[ogeographic assemblages: Atlantic, Pacific, Mississippian, and Beringian. with strong bootstrap support for al1 four clusters (Figure 2.3).

Due to the large number of sequences, it was not possible to complete a search for parsirnonious cladograrns of dl taxa Therefore, Maximum Parsirnony (MP) analysis was carried out on three (MIS. PAC, An)to six (BER) divergent representatives from ulthin each of the major haplotype clusters identified by NJ analyses, as well as one haplotype that did not fall into a major cluster (03CD). and three haplotypes £Yom Europe. for a total of 19 taxa with 147 parsimony informative sites. MP heuristic searches yielded 36 rqually parsirnonious trees of length 243 (CI4.77. Rk0.87). Both strong phylogenetic signal (gl = -2.16; a,, = -0.09;P < 0.01) and high bootstrap values support the presence of four phylogenetically divergent clades of haplotypes in North Amerka, which correspond to the four major haplotype clusten identified in the NJ analysis. A strict consensus of the 36 trees is show in Figure 2.4. The Atlantic clade was the sister croup to the other three clades (Beringian, Mississippi, Pacific), which formed an C unresolved tnchotomy. Additional MP analyses conducted on 19 different taxa gave congruent results (data not shown), indicating that cladistic support for the major phy logeographic groups is unafTected by taxon choice.

Divergence estimaes and ref ugial origrœns

The four mitochond~alclades showed largely allopatric distributions (Figure 2-51, rach occurring in one of the regions known to have comprised former glacial refugia Figure 23. Neighbour-Joining (NI) tree based on sequence variation in mitochondrial

COI sequences (6 14 aligned sites) for a single isolate from each of 44 North Amencan populations of S. c. urnericana. Abbreviations for population locations are listed in

Appendix 2.1. The tree was rooted with three sequences obtained from European S. c. nstia Bootstrap confidence limits (1 000 replicates) greater than 70% appear above the branches. Shaded bars correspond to the four clades mapped in Figure 2.5: Beringia

(BER). Mississippian (MIS), Pacific (PAC) and Atlantic (ATL). Two phyiogeographic groups (BERI. BE=) are shown for the Beringian clade. Haplotype 03CD (*) was not assigned to a geographic cluster. The NJ tree was estimated in PAUP*4.0b2 (Swofford

1998) using a distance matrix generated with the Kimura two-parameter mode1 (Kinura

1980). BER(2)

MIS

PAC

ATL BER

MIS

PAC

An

Figure 2.1. Strict consensus of 36 most parsirnoniou trees (length 243. Ck0.77. RI=0.87) for 19 Sida crystallinu haplotypes, based on 147 phylogenetically informative characters of the COI gene. Tres were obtained using an heuristic search in PAW* -!.Ob2 (Swofford 1998). with steepest descent and tree bisection reconnection options invoked. Transitions and tramversions were weighted equally; additional analyses with diserential weightings did not affect the topology of the phylogeny. Clades are identified by lines. bootstrap values (1 000 replicates) are shown on the branches. Figure 2.5. Geographic distribution of the four major clades of

S. c-tullina within North Amerka. The Beringian clade is broken

into wophy logeogdphic groups (BERI, BER2). Shaded circles and

the single asterix correspond to haplotypes identified within the clusten

of the NJ tree in Figure 2.3. (reviewed in Pielou 1991). The Mississippian clade was dominant in the Great Lakes and

Mississippian biogeographic provinces, but exhibited little sequence divergence over this broad range. By contrast each of the other three clades was hgrnented into genetically distinct subgroups with allopatric distributions. Haplotypes in the Pacific clade were restncted to populations from the Cascadian biogeographic province, but were

subdivisable into a southem mainland group, and a group restncted to the Queen

Charlotte Islands. The Atlantic clade du, included two groups, one restncted to

populations from the Southeastem biogeographic province, and a second assemblage

which included populations From sites in the North Appalachian biogeographic province.

The Beringian clade was additionally subdivided into two phylogeographic groups

(Figure 2.5 ). although these lineages fonned an unresolved polytomy in cladistic analysis.

Populations with BER1 haplotypes occupied the Yukon-Mackenzie and Hudson Bay

biogeographic provinces. occuning predominantly in the NWT and prairies. By contrast,

haplotypes of the other Benngian group (BER2) were located rnainly in the western

Y ukon-Mackenzie and mainland Cascadian biogeographic provinces, with a few

populations scattered in the Great Lakes and Mississippian regions (see Figure 2.5).

The average percent nucleotide sequence divergence behveen the least

(Mississippian and Beringian) and most (Atlantic and Pacific) divergent clades in North

America was 4.8% and 7.4% respectively. while European populations showed 23.4-

23.8% divergence fiom the four North Amencan clades. Assumuig 2% sequence

divergence per million years. North Arnerican and European S. cr~~stalZinahave not been

in contact for more than 10 million years. while the four clades within North Amenca last

shared common ancestors between 2.4 - 3.7 Mya DISCUSSION

This study supports the prior taxonornic discrimination of S. crystallina lineages from Europe and North America; ailozyrne and mtDNA divergences were far greater between populations fiom the different continents than across North Amenca In facf this intercontinental divergence was greater than that between rnany other congeneric species of cladocerans (e.g. Taylor et al. 1998), indicating that these taxa ment recognition as distinct species. Regardless of their final taxonomic placement, the 23% sequence divergence at COI suggests that the colonization event(s) which led to the

Holarctic distribution of Sida occurred before the Pliocene. Our failure to detect the haplotypes and allozyme alleles diagnostic of Europea. lineages in No& Amenca suggests that long distance dispersai between these continents is so uncornmon that it has not impacted subsequent evolutionary trajectories. at least in North America.

Allozyme and yses suggested that North American populations now assigned to S. ctysiollina are a single species. There was. for example. no evidence of the Hardy-

Weinberg disturbances which characterize populations where sibling species occur in sympatry (Witt & Hcbert 2000). As well. allelic arrays were shared by populations across the continent. and the regionai patteming of gene fiequencies was modest.

Populations in proximity did exhibit marked gene fkequency shih. but this is typical of populations founded from a srnail nurnber of colonists capable of parthenogenesis

(Boileau er al. 1992).

The anaiysis of sequence variation at COI revealed hi& diversity with 10 haplotypes detected among the 44 North Amencan populations. Moreover, cladistic nnaiysis indicated that these haplotypes were memben of four clades showing marked (4- 7%) sequence divergence. These clades had largely allopatnc distributions, which suggested their denvation from source populations in the Mississippian, Atlantic.

Beringian and Pacific refugia. Based on a clock calibration of 2% per million years these lineages diveeified 2.4 - 3.7 million years ago, consistent with their isolation just prior to. or at the onset of the Pleistocene (Webb & Bartlein 1992). In addition to isolation imposed by advancing ice sheets during the Pleistocene, southem refugial groups were undoubtedly Merseparated by other physiographic features. For example. the clear divergence observed between Atlantic lineages of Sida and the neighbouring Mississippian refugial group io the West likely reflects isolation induced by the Appalachian Mountains. Similarly. the Western Cordillera presented a barrier to gene flow between populations on the Pacific coast and those in the prairies.

The depth of divergence between refugial goups of S. crystallina and their largely allopatnc distributions indicate that these populations exhibit a Category I phylogeographic pattern (Avise 2000). which is characterized by the presence of spatially isolated haplogroups showing marked genetic divergence. Closer inspection of taxa which exhibit such patterns often reveals Mergeographic structure within phylopoups. which may reflect isolation by distance or resaicted conternporary gene flow (Avise 2000). In fact, al1 of the phylogroups in S. crystallina excepting the

Mississippian lineage showed Mergeographic structure. The substantial COI divergence between populations of S. crystallina fiom the Queen Charlotte Islands and their mainland counterparts on the Pacific Coast suggests the fragmentation of this refùgiurn into two isolates, a conclusion reinforced by evidence of allozymic divergence in these populations. Si& crystallino popdations dong the Atlantic Coast were also separable into two subgmups, one from the Southeastem biogeographic province, and the other from the North Appdachian biogeographic province.

Despite their shallow sequence divergence, Benngian isolates were similady separated into two phylogroups, which ment particular attention from a colonization perspective. The distribution of the BER1 haplotypes suggested their dispersal fiom

Benngia through the arc of proglacial lakes which extended into modem Lake Winnipeg

(Dyke & Prest 1987). The absence of the second lineage (BER2) fiom these habitats suggests that these two haplogroups were not panmictic within Benngia during the last glaciation. but experienced restricted gene flow and subsequently utii ized separate dispersa1 comdon. The generally western distribution of the BER2 haplotypes suggests their derivation from a Nahanni refuge (Lindsey & McPhail 1986; Dyke and Prest 1987) with dispersal occurring southwards dong the Pacific Coast, in a similar fashion to that reponed in iake trout (Wilson & Hebert 1998). The presence of the BEW lineage to the southwest of the Great Lakes region likely represents its subsequent eastward dispersal, but funher sampling in the prairies is required to test this speculation.

Collectively, these haplotype distributions indicate that post-Wisconsinan corridors enabled only a unidirectional passage for S. crysfaIlina- allowing the southward movement of Beringian lineages, but not the northerly migration of

Mississippian or Pacific refugial groups. In addition, the presence of phylogeographic stnicture over smaller spatial scales within major refugial groups indicates that, despite the abi lity for overland dispersal in S. crystallina? such events are either too Uifrequent to overcorne divergence due to lineage sorting and random &fi among isolated populations, or that pre-existing populations are so large that they resist penetration by secondary coionists.

The phylogeographic groups observed in North Amencan S. crystallina are remarkable in two ways. Firstly, the various lineages show a higher degree of sequence divergence than reported in fishes (Bematchez and Wilson 1998). Secondly, the boundaries between lineages are very well defined, with little evidence of the admixture which would be expected if dispersal events were nondirectional, as might be predicted due to cheir ability to disperse overland. The general coincidence between the boundaries of S. crys~allinolineages, and biogeographical provinces in fishes, suggest that much dispersai in S. crysiullina has been accomplished through the same interconnections which facilitated gene flow in fish. However. the occurrence of rare. longdistance dispersa1 rvents is suggested by the presence of disjunct haplotypes such as the MIS rehgial lineages in the Southeastem biogeographic province. and a BER4 lineage in one habitat within the Cascade biogeographic province.

Few studies have examîned the patteming of genetic diversity in other freshwater organisms reliant on passive dispersal. Work on the genus Daphnia has largely focused on polar taxa which show subçtantial evidence of lineage admixture and long distance dispersal (Weider & Hobaek 1997; Colbourne et al. 1998; Weider et a[. 1999a.b).

Studies on a single temperate zone assemblage, the D. hevis cornplex. reveded a pattern similar to that in S. crystallina with marked genetic divergence between allopatric lineages which showed little evidence of secondary contact (Taylor et al. 1998).

Moreover. the Appalachians and Western Cordillera played the same important role in de fining phylogeographic boundaries as they did in S. crysiallina. This contrasting pattern of secondary contact between temperate and polar taxa suggests that in areas where sustained occupancy has been possible, lineages have long histories of tenure and have pre-empted die dimision of secondary colonists. By contrast, in areas remote fkom refugia the prevalence of vacant habitats has provided greater opportunities for lineage mixture.

Comparative phylogeography

There is now sufficient information to gain a preliminary sense of the response of fieshwater organisms with different dispersal strategies to the cyclic advance and retreat of ice sheets during the Pleistocene. The recurrent formation of specific refugia, as a result of either persistent physiographic barrien. such as mountain ranges. or the regeneration of habitats linked to sea level lowering, providrd a basis for the sustained differentiation of lineages (Hewitt 19%). However, the potential for the admixture of linrayes during each interglacial set the stage for the erosion of incipient divergence arnong different refbgial stocks (Bematchez & Wilson 1998). As weil, despite the repeated construction of specific refugia and the successful retreat of populations to these sites. the risk of local extinction was likely high as a joint consequence of the long duration of each glacial advance and the limited number of habitats in each refugium

(Hewitt 2000).

The mitochondrial genomes of coldwater fish species record a history of extinction events and Iineage contact throughout the Pleistocene in three ways. First, it appears that modem populations often derive from just one or two refugia (Bematchez &

Wilson 1998), indicating that many isolates penshed. Second, the number of haplotypes in each reliigial lineage is ordinarily low, indicating the loss of diveaity, as expected if the population size in each refugium was smali (Hewitt 1996). Finaily, the extent of sequence divergence among modem haplotypes from different refugia is srnall, suggesting that fiequent secondary contact achieved via active dispersal through pst- glacial meltwates, and subsequent genetic drift, led to the loss of divenity which originated earlier in the Pleistocene (Bernatchez & Wilson 1998; Wilson & Hebert 1998).

Hence. refugial lineages in coldwater fish species rarely show more than 1% sequence divergence. while those in more southerly species show more than four times this level

(Bermingham & Avise 1986; Billington & Hebert 199 1 ; Avise 2000).

The mitochondrid genome of S. crystallina records a different history of response to the repeated glacial cycles than that observed in Freshwater fishes; one which signals a much reduced sensitivity to extinction. First. there is evidence fiom this study that its

lineaagrs persisted in al1 of the major rehgia available to Freshwater life in temperate

North Amerka. As well. haplotype diversity within refugial groups of S. crystaha is

high relative to that observed within refugial groups of teleosts frorn the same regions.

Finally. even the lowest amount of sequence divergence between phylogroups is greater

than that observed in freshwater fish subject to the same glaciation events, suggesting that

phylogroups of S. crystdina successfully survived recurrent cycles of ice advance and

retreat. Considered jointly, these results indicate that refugid lineages of S. crystailina

were little impacted by either extinction events or secondas, contact throughout the

Pleistocene. Their resistance to extinction is likely a joint consequence of the production

of diapausing eggs which can suMve davourable conditions for centuries, and their

parthenogenetic mode of reproduction which allows the refounding of populations fiom a

single individuai (Hebert 1987: Korovchinsky and Boikova 1996). The potentidly large size of populations in each habitat undoubtedly provided further protection from stochastic lineage extinction. These results are important fiom a phylogeographic perspective because they suggest that Merwork on organisms sharing these life history atiributes wiil provide an unusually comprehensive phylogeographic record in cornparison with studies on fishes, where much of the historical record has been erased through lineage loss. Differential exposure to extinction may also explain an otheMlse puvling bioçeographic pattern - the absence of fish species endemic to polar freshwaters. but the presence of numerous invertebrates restricted to these settings. LITERATURE CITED

Avise JC. Arnold J. Bal1 RM et al. (1987) [ntraspecific phylogeography: the

mitochondrial DNA bridge between population genetics and systematics. Annual

Rrview of Ecology and Systematics. 18,489422.

Avise JC. Walker D. Johns GC (1 998) Speciation durations and Pleistocene effects on

vertebrate phylogeography. Proceedings of the Royal Society of London Series B.

265, 1707- 1 7 12.

Avise JC (2000) Phvlogeogrnphy. The Hisrow and Formation of Species. pp. 447.

Hmard University Press, London.

Bcrmingham E. Avise .TC (1986) Molecular zoogeography of freshwater fishes in the

southeastern United States. Genetics, 1 13,939-965.

Bermingham E. Moritz C (1998) Comparative phylogeography: concepts and

applications. Molecular Ecology. 7.367-369.

Bematchez L. Dodson JJ (1991) Phylogeographic structure in mitochondrial DNA of the

lake whitetish ( Coregonus clupeafonzis) and its relation to Pleistocene

glaciations. Evolution. 45. 101 6- 1035. C

Bematchez L. Wilson CC (1998) Comparative phylogeography of Nearctic and

Palearctic fishes. Mo[ecukar Ecologv, 7.43 1-452.

Billington N. Hebert PDN (1 99 1) Mitochondrid DNA diversity in fishes and its

implications for introductions. Cunadian Journal of Fisheries and Aquatic

Sciences. 48. ( Suppl. 1). 80-94. Boileau MG. Hebert PDN,Schwartz SS (1 992) Non-equilibrium gene fiequency

divergence: persistent founder effects in nawal populations. Journal of

Evolutionaty BioZogy. 5,2540.

Brower AVZ (1 994) Rapid morphoiogical radiation and convergence among races of the

buttertl y Heliconius erato inferred fiom patterns of mitochondnal DNA

tvolution. Proceedings of the Nationul Academy of Sciences. USA. 91.649 1 -

6495.

Brown JM. George M Jr, Wilson AC (1 979) Rapid evolution of animal mitochondnal

DNA. Proceedings of the National Academy of Sciences, CISA. 76, 1967- 197 1.

Burr BM. Mayden RL (1992) Phylogenetics and North Arnerican freshwater fishes. In:

Sysrematics. Hirrorical Ecology. and North American Freshwater Fishes (ed.

Mayden RL), pp. 1 8-75. Standford University Press. Stanford.

Colboume K. Crease TJ. Weider U et al. (1998)Phylogenetics and evolution of a

circumarctic species complex (C ladocera: Daphnia puiex). Biological Journal of

the Linnean Socie~y,66,347-365.

Dadswell MJ (1 974) Distribution, ecology and postglacial dispersion of certain

crustaceans and fishes in eastem North America. Narionai Mureum of Canada

Publications in Zoology. 11.

Darwin C ( 1 859) On the Origin of Species by Means of Nafural Selection. John Murray,

London.

Dyke AS. Prest VK (1 987) Paleogeography of northern North America, 18 000-5 000

years aga Geological Suntey of Cana&. Map 1 7034 scale 1 :12 500 000. Folmrr O. Black M. Hoeh W, Lutz R, Vrijenhoek R (1994) DNA primee for

amplification of mitochondrial cytochrome c oxidase subunit 1 From diverse

metazoan inverîebrates. Molenrlar Marine Biology and Biotechnologv, 3,294-

399.

Fryer G ( 1996) Diapause. a potent force in the evolution of Freshwater crustaceans.

Hydrohiologia. 320. 1 - 14.

Gilbert D ( 1992) SeqApp version 1.9a. A multiple sequence editorfor Macintosh

cornputers, Published electronicaily on the Internet at fly.bio.indiana.edu.

Heben PDN ( 1987) Genotypic characteristics of the Cladocem Hydrobiologia. 145, 183-

193.

He ben PDN. Beaton MJ ( 1 993 ) Merhodologies for A llocyme Anulysis using Cellulose

.4cetute E[ectrophoresis. A Practical Handbook Helena Laboratories.

Beaumont. TX.

Hewitt GM ( 1996) Some genetic consequences of ice ages. and their role in divergence

and speciation. Biological Journal ofthe Linnean Society, 58.247276.

Hewitt GM (2000) The genetic legacy of the Quatemary ice ages. Nature. 405,907-9 13.

Hillis DM. Moritz C ( 1990) An overview of application of molecular systematics. In:

Moleculor Systemutics (eds Hillis DM, Moritz C), pp. 502-5 15. Sinauer

Associates, Sunderland, MA.

Hillis DM. Huelsenbeck IP (1992) Signal. noise. and reliability in molecular

phyiogenetic analyses. 7?ze Journal of Herediy, 83. 1 89- 195.

Hocut! CH. Wiley EO ( 1986) The Zoogeography of North American Freshwater Fishes.

John Wiley and Sons, New York. Kirnura M ( 1980) A simple method for estimating evolutionary rates of base substitutions

through comparative studies of nucleotide sequences. Journal of Moiecular

Evolurion. 16. 1 l 1- 120.

Knowlton N. Wei@ L.4, Solorzano LA, Mills DK, Benningham E (1993) Divergence in

proteins. mitochondnai DNA. and reproductive cornpatibility across the Isthrnus

of Panama. Science, 260, 1639-1 632.

Knowlton N. Weight LA ( i 998) New dates and new rates for divergence across the

Isthmus of Panama Proceedings of the Royal Society of London, Series B. 256,

2257-2363.

Korovchinsky NM. Boikova OS (1996) The resting eggs of the Ctenopoda

( Cnistacea:Branchiopoda): a review. Hydrobiologia. 320. 13 1 - 140.

Korovc hi nsky NM ( 1992) Sididne and Holopediidae: (Crus~ucea:Daphniijormes). pp.

82. SPB Academic Publishing, The Hague.

Kurnar S. Tamura K. Nei M ( 1993 ) MEGA: Mole~vlarEvolufionury Generics Analysis.

Cérsion 1.02. Distributed by the authors. Pennsylvania State University,

University Park. PA.

Lewis PO. Zaykin D ( 1999) Genetic Data Analysis: a cornputer program for the anaZysis

of ullelic data. Version 1.0 (d13). Distributed by the authors on the intemet from

the GDA homepage. h~://al~evn.eeb.ucom.edul~da/.

Lindsey CC. McPhail JD (1986) Zoogeography of fishes of the Yukon and Mackenzie

bains. In: The Zoogeography of NdArnerican Freshwcler Fishes (eds. Hocutt

CH. Wiley EO), pp. 639-674. John Wiley and Sons. New York. Martin AP. Palumbi SR (1993) Body size. rnetabolic rate. generation time, and the

rnolecular dock. Proceedings of the lYotional Academy of Sciences, USA. 90,

4087409 1.

Mayr E ( 1963) Animal Species and Evolution. pp. 797. Harvard University Press,

Cambridge.

Nei M ( 1978) Estimation of average heterozygosity and genetic distance from a small

nurnber of individuals. Genetics, 89,583-590.

Ota T (1993)DISP.AN: Generic Distance and Phylogenetic Anaf'ysis. Distributed by the

Institute of Molecular Evolutionary Genetics: Pennsylvania State University,

University Park. PA.

hnp:: 1~w-w.bio.~su.eduiPeople/Facul tv/Nei/Lab/ProwI

Pielou EC ( 199 1 ) ilfer the Ice Age: the Rerurn of Life to Glaciated North America. pp.

366. University of Chicago Press. Chicago.

Rice WR ( 1989) Anaiyzing tables of statistical tests. Evolurion. 43.223-225.

Saiki RK. Gelfand DH. Stoffel S (1988) Primer-directed enzymatic amplification of

DNA with a thermostable DNA polymerase. Science, 239.487-491.

Schubart CD. Diesel R Hedges SB ( 1998) Rapid evolution to terrestrial life in Jamaican

crabs. Nature. 393,363-365.

Schwrnk K ( 1996) Evolutionary genetics of Daphnia species cornplex: hybridism in

syntopy. Publication 223 1. Netherlands Institute of Ecology, Centre for

Limnology. Nieuwersluis. The Netherlands. Stemberger RS (1 995) Pleistocene refuge areas and postglacial dispersal of copepods of

the northeastem United States. Canadian Journal of Fisheries and Aquuiic

Sciences. 52.3 197-22 10.

Swofford DL (1998) PAUP*: Phylogenetic analysis using parsimony. Version 4.0b2.

Sinauer Associates, Sunderland MA.

Taylor DJ. Finston TL, Hebert PDN (1998) Biogeography of a widespread freshwater

crustacean: pseudocongnience and cryptic endemism in the North American

Daphnia laevis complex. Evolution. 52. 1648-1670.

Webb T. Bartlein PJ (1992) Globai changes during the last 3 million years: climatic

contiols and biotic responses. Annual Review of Ecology and Systematics, 23.

141-173.

Weider LJ. Hobaek A (1997) Postglacial dispenal, glacial refugia and clonal structure in

Russian/Sibenan populations of the arctic Daphnia pulex complex. l7eredit-y. 78,

363-3 72.

Weider LJ. Lampert W. Wessels M. Colbourne K. Limburg, P ( 1997) Long-tenn

genetic shifts in a microcnistacean egg bank associated with anthroprogenic

ç hanges in the Lake Constance ecosystem. Proceedings ofthe Royal Socies, of

London, Series B, 254, t 6 1 3- 16 18.

Wcider LJ. Hobaek A. Colbourne. JK et al. (1999a) Holarctic phylogeography of an

asexual species complex 1. MtDNA variation in arctic Daphnia. Evolution, 53,

777-792. Weider LJ. Hobaek A. Hebert PDN,Crease TJ (1999b) Holarctic phylogeography of an

asexual species cornplex-II. Allozyme variation and clonal structure in arctic

Daphnia. Molecufar EcoIogy, 8. 1 - 13.

Wilson CC. Hebert PDN (1998) Phylogeography and postglacial dispersal of lake bout

(Salvelinus namaycush ) in North America. Canadian Jo urnal of Fisher ies und

Aquatic Sciences. 55, 1 0 10- 1024.

Witt JDS. Heben PDN (2000) Cryptic species diversity and evolution in the amphipod

genus Hvallela within central glaciated North America: a molecular phylogenetic

approach. Canadian Journal of Fisheries and Aquuric Sciences, 57,687-698. CHAPTER THREE

Phylogeography of arctic anostracans: diapause and the occupancy of hi& arctic rehigia ABSTRACT

Geologicai evidence for the peaistence of small ice-free areas in the high arctic during the Wisconsinan ha recently gained support from studies of passively dispeaed plants. As freshwater zooplankton share a similar dispersal strategy, populations residing in arctic habitats rnay also be compnsed of high arctic refugial lineages. 1 tested this hypothesis by investigating the phylogeographic structure of two anostracan crustaceans cornmon in arctic North Amerka. Sequence divergence at both the cytochrome oxidase I and 16s rDNA mitochondnal genes revealed that Artemiopsis sfefanssoni populations derive from a single Wisconsinan refuge, while Branchinecfapaludosa persisted in multiple refugia in North Arnenca and Europe. The distribution of these phylogroups provides evidence for the presence of high arctic refùgial lineages of freshwater zooplankton. and suggests that the life history characteristics common to passively

Jispersed organisms were key to their survival in high arctic ice-free areas. INTRODUCTION

Despite the recent deglaciation of much of its area, biotic diversity is relatively high in the Canadian arctic archipelago (Pielou 1991). Approximately 20 species of vertebrates. more than 500 species of invertebrates, and 200 species of higher plants are year-round residents of terrestrial or freshwater environments (Macpherson 1965; Hebert

& Hann 1986; Murray 1987). The provenance of most of these species is unknown.

Some taxa undoubtedly derive fiom southem rehigia, expanding their distributions northwards as the Pieistocene glaciers retreated (Pielou 199 1). Other species ciearly originate from the large ice-free regions of Alaska and the Yukon. collectively known as

"Beringia". which have long been regarded as the most important northem refbgium

(Macpherson 1965: Lindsey & McPhail 1986: Hofian 198 1 ; Hopkins et al. 1982;

W iison & Hebert 1 998). However, numerous small ice-free areas also persisted in the northemmost and eastemmost regions of the arctic archipelago during glacial maxima

(England 1976: Hodgson 1985; Dyke & Prest 1987). nie significance of these latter retùgia to polar taxa remains uncertain. but likely varied among species with diffenng bioiogical attri butes.

Previous distibutional studies of arctic mmalssuggested that several species rnay derive from high arctic Wisconsinan ice-fiee areas (Macpherson 1965). However, more recent phylogeographic investigations of vertebrates comrnon in the arctic archipdago have provided no evidence of high arctic refugiai lineages (Wilson er al.

1996: Holder et al. 1999). By contrast northern refugia were evidently important for the wvival of key elernents of the polar flora (Tremblay & Schoen 1999; Abbot et al. 1995,

3000). The conflicting importance of such refigia for different taxon assemblages may reflrct the impact of demographic factors. Because of the relatively small size and isolation of high arctic refugia, population sizes of resident taxa would have been small, especially for large organisms. setting the stage for demographic extinction (Hoffnan

198 1 : Hewitt 1996.2000). Plant populations may have been protected from extinction by both their greater abundance relative to vertebrates. and by their production of seeds, capable of diapause during unfavourable periods (Freedrnan et al. 1982; Fox 1983;

Grul ke B Bliss 1985: Murray 1987). The presence of a seed bank would have provided both protection against transient deterioration of climatic conditions, as well as the prriodic submergence of a refugium as a result of ice advance.

If resistant life history stages have been important in enabling species to persist in marginal refugia one would expect a divergence in refugid occupancy among fieshwater organisms. linked to the fact that çorne taxa possess diapausing stages. while othen do not. Pasr work on arctic fishes. al1 of which lack a diapausing stage. has provided no evidrnce for the occupancy of high arctic refugia (Wilson et al. 1996; Wilson and Hebert

1998). By contrast. many freshwater invertebrates possess a diapausing stage. but little is known of the refugial origins of these taxa The broad distribution of some species suggests they may have persisted in nurnerous refugia while the more restricted distributions of other species suggests they penisted in a single refugium (Hebert & Ham

1986). Only one group of polar invertebrates bas been studied intensively - memben of the Daphnia pulex cornplex. The distribution of genetic variation in some lineages suggests that refbgial populations likely persisted in the hi& arctic during the

Wisconsinan (VanRaay & Crease 1995: Weider et al. 1999b). In contrast, other memben of the group exhibit linle phylogeographic structure. with component species comprîsed of a number of broadly disuibuted lineages (Weider & Hobaek 1997; Colboume et al.

1998: Weider et al. 1999a). However, Daphnia species may not be typical of arctic freshwatcr invertebrates at large. Their parthenogenetic mode of reproduction and

Hoatinç ephippial eggs provide an exceptional capacity for long range dispersal and rapid range expansion. These life history characteristics likely contributed to extensive post- glacial admixture of some lineages, eroding evidence of genetic structure that might have developed during isolation in high arctic refugia.

Aside tiom cladocerans, fieshwater habitais in the high arctic are occupied by two other groups of crustaceans - copepods and anostracans. The few species of caianoid copepods now present in the high arctic lack diapausing stages. and have colonized this area using the Arctic Ocean as a dispersal comdor (Roff & Carter 1972; Dadswell 1974).

In contrast. both cyclopoid copepods and anostracans produce diapausing eggs (Tash &

Armitage 1967: Fryer 1996). enabling a test of the hypothesis that these life stages prornoted the occupancy of marginal refugia Because of their broad distributions and stable tavonomy (Hebert & Ham 1986; Belk & Brtek 1995), anostracans present the best target for analysis.

Three species of anostracans occur in the hindra regions of North America and

Greenland (Johansen 192 1. 1922; Roen 198 1; Belk & Brtek 1995). These sexually dirnorphic species share similar life histories; they are univoltine. with females producing large nurnbers of diapausing eggs (20-200), which hatch shody after ice melt and mature in approximately four weeks (Johansen 1 92 1: Daborn 1976). PoZyartemieila hazeni

(Murdoch 1885), a North American endernic, is reseicted to the low arctic, occuning in nindra ponds and frost wedge polygons in coastd regions of Alaska and the Northwest Territories (Belk & Brtek 1995). but the other two species have distributions that penetrate the hi& arctic. Artemiopsis stefanssoni (Johansen 1921) is absent f?om the western arctic. but is prevalent in ponds across the central. eastem and high arctic islands and West Greenland (Hartland-Rowe & Anderson 1968; Belk &Brtek 1995).

Brunclrinrcfupaludosa (Muller 1788) has a circumpolar distribution, occuming in tundra regions across the Holarctic. as well as southward into alpine tundra settings of both

North Amenca and Europe (Belk & Brtek 1995). Both of these species currently occur in

high arctic settings near the margins of active glaciers, where climatic regimes are similar

to ihose during Pleistocene glacial maxima.

Aside from one localized ailozyme study (Boileau et al. 1992). little is known of

yenetic diversity in anostracans from the Canadian hi& arctic. Because of their

divergent distributional patterns, it seems likely that the two anostracans found in this

area occupied differing glacial refugia during the Pleistocene. The easterly distribution

of A. stefanssoni suggests that this species did not persist in western refugia By contrast,

the circumpolar distribution of B. paludosa, coupled with its occurrence in alpine tundra

settings. suggests that this species had a broad Holarctic distribution prior to the last -elacial advance. leading to its likely isolation in several geographically isolated refugia Altematively. the broad distribution of this species may reflect its aggressive colonization

of deplaciated regions. with isolated montane populations reflecting either relict

populations or recent avian-mediated dispersal (Ferguson 1935; Dexter 1953; Dabom

1976: Saunders et al. 1993).

This study examines the patteming of genetic variation in both A. stefanssoni and

B. paludosa to determine if the curent distributions of these two anostracans in the North Amencan arctic reflect their dispend from single cr multiple Pleistocene refugia.

Specifically, this study tests the hrpothesis that the current distribution of A. stefumsoni reflects its restriction to a single refugium during the Pleistocene. while B. paludosa persisted in several refügia in North Arnerica and Europe. The extent of genetic divergence among populations of these two species was assessed by analyzing sequence variation in two mitochonàrial genes, cytochrome c oxidase 1 (COI)and 16s rDNA. If populations derive from a single refugium. the expected phylogeographic pattern includes both the widespread distribution of iineages. which exhibit limited sequence divergence among haplotypes. and the absence of geographicaily localized groups of ailied haplotypes. By contrast, if modem populations derive from two or more refugia, one would expect sevenl well supported phylogeographic lineages. whose spatial distributions are consistent with their genealogicd relationships. Finally, phylogeographic patterns in these anostracans are compared to those in other arctic organisms. and coupled with knowledge of the glacial history of the Canadian arctic to determine both the importance of high arctic refugia and the role of life history c haractenstics in the production of phylogeographic patterns in polar li fe at large.

.MATERIALS and METHODS

Specimen collection und species distributions

The geographic ranges of A. stefànssoni and B. paludosa are presented in Figures

3.1 and 3.2 respectively. Thirty-eight populations of A. stefanssoni. and 64 populations of B. paludosa were collected Frorn habitats in Alaska, the Yukon Temtory, Figure 3.1. Map illustrating the geographic distribution of A. stejonssoni in North America. nie black circle represents the sole known location for this species south of the arctic. in

Banff. Alberta Canada. (Johansen 192 1 ;Hartland-Rowe &

Anderson 1968; Dabom 1976,1978; Belk & Bretk 1995) Figure 3.2. Maps ihstrating the geographic distribution (dashed lines) of B. paludosa in

North Amenca (32.1) and Europe (32.2). Black circles indicate disjunct populations, astrrisks indicate disjunct populations sampled in this study in the Medicine Bow

Mountains of Wyoming (32.1 ) and the Tatra Mountains of Slovakia (3 2.2) (Johansen

193 1. 1922; Ferguson 1935: Dexter 1953; McLaren 1964; Andenon 1974; Dabom 1978;

Roen 1 98 1 : Belk & Brtek 1 995; Brtek & Thiery 1995; Saunders et al. 1 993; Musselman

1994).

Figure 33. Map showing areas sarnpled for A. stefanssoni and B. paludosa in the

Canadian arctic archipelago. Abbreviations for sarnpled sites are as follows: AH

(Axel Heiberg Island); BF (Baffin Island); BT (Bathurst Island); CW (Cornwallis

Island): DV (Devon Island); EL (Ellesmere Island); (FB) (Foxe Basin - includes

Igloolik. Iens Munk. Rowley and Spicer Islands); HB (Hudson Bay); JL, (Jemy Lind

Island): LC (Little Cornwallis Island); LW (Lowther Island); MI3 (Manitoba); MP

(Melville Peninsula): SO (Somerset Island); TK (Tuktoyaktuk Peninsula). the Northwest Temtories. Nunawt and Greenland berneen July 1995 and August 1998

(Figure 3.3). Additional populations of B. paludosa were obtained fiom one alpine habitat in the High Tatra Mountains of Slovakia, and fiom two alpine habitats in the

Medicine Bow Mountains of Wyoming, USA (Figure 3.2). A. siefanssoni was also collccted from the sole site where it is known south of 6OoN (2 400 m at Dolomite Pass in Banff. Alberta), but attempts to ampli@ DNA fiorn specimens of this population were unsuccessful. Specific details on collection locations are provided in Appendix 3.1.

Collections were made either fiom helicopter or on foot using either a 250 pm mesh tow. or hand-held dip nets. Samples were sorted !ive in the field, and either flash- frozen in liquid nitrogen or preserved in 90% ethanol for subsequent moleculai analyses.

When possible. the elevation above sea level. depth. and the conductivity of each habitat were recorded. in addition to latitude and longitude of the sampling location (Appendix

3.1 ). The mean and range for each of these parameten were calculated in order to summarize the habitat preferences of each species.

MIDNA analysis

In order to examine as broad a geographic range as possible for each species.

rntDNA analysis was ordinarily restricted to I individual per population, the most

efficient sampling strategy for broad scale phylogeographic surveys (Pons & Petit 1995;

Templeton et al. 1995). Cornparisons between sequences for multiple (2-3) individuals

in 14 populations (see Appendix 3.1) confirmed that this strategy was appropriate for the

current study. as individuals co-occurring in a habitat invariably showed limited sequence

divergence (mean divergence 0.2%, maximum divergence 0.4% A. stefanssoni: mem divergence 0.4%. maximum divergence 0.8% B. paludosa) relative to the divergences among populations (see Results, Table 3.1).

Total DNA was isolated from each individual using the Isoquick Nucleic Acid

Extraction Kit (ORCA Research), and resuspended in 30 pl of Tris-HCl pH 8.5. A 657 base pair (bp) fragment of the mitochondnal cytochrome c oxidase (COI)gene was mplified with the primer pair (5' GGTCAACAAATCATAAAGATATG 3') and (5'

TA~ACTTCAGGGTGACCANJWATCA3') (Folmer et al. 1994). Examination of variation at protein coding genes provides a usefûl check against the inadvertent amplification of pseudogenes (Zhang & Hewitt 1996) through the assessrnent of amino acid translations. However. although COI is appropriate for examining both genenc and species level relationships of anostracans, it may be less usehl for resolving deeper relationships among groups. due to high levels of homoplasy observed at greater levels of srquence divergence (Remigio & Hebert 2000). Therefore. individuals from each of the major COI clusten identified by preiiminary phenetic analyses were also examined for

variation in a 530 bp fragment of the mitochondnal 16s rRNA gene. using the primers

(S'CGCCTGTTTATCAAAAACAT3') and (SCCGGTCTGAACTCAGATCACGT3')

( Palum bi 1 996).

The 50 pl PCR reactions used for both COI and 16s rDNA amplification

consisted of 10-300 ng of DNA template. 9 rnM TrisHCl (pH 8.3), 50 mM KCI. 2 mM

MgCi?. 0.2 pM of each dNTP (C.G+A.T),0.2 pM of each primer, and 1.0 U of Taq DNA

polymerase (Qiagen Inc.). Two different thermal regimes were used for the amplification

of mtDNA. depending on the species and gene being amplified. Conditions for Regime

A (COI: B. paludosa; 16s rDNA B. paludosa, A. stefumsoni) were as foilows: 1.5 minutes at 94°C. 1 cycle of 30 seconds (s) at 94OC, 1 minute at 60°C. 1 minute at 72OC; 5 cycles with the annealing temperature decreased 2OC each cycle; 29 cycles of 45 seconds at 93°C. 1 minute at 50°C. 1 minute at 72OC; seven minutes at 72°C. hdividuals that

were not successfully amplified using Regime A were subsequently subjected to less

stringent annealing conditions in Regime B (COI: A. srefanrsoni): 1 cycie of 1 minute at

94OC: JO cycles of : 1 min at 94OC, 1.5 min at 45OC, and 1.5 min at 72°C; followed by 1

cycle of 5 minutes at 72OC. Al1 PCR products were ge[ purified using the Qiaex II kit

(Qiagen Inc), subjected to dye terminator cycle-sequencing reactions with AmpliTaqC3

DNA polymrrase FS (Perkin-Elmer), and sequenced on an AB1 377 automated sequencer

(Applied Biosystems).

CO 1 electropherograrns were aligned with the Seqapp 1.9a sequence editor

(Gilbert 1992). while 16s rDNA data were aligned using Sequence Navigator 1.O. 1

iMyers 1 986) with a gap penalty of 10; al1 alignments were subsequently adjusted by eye.

Translation of the COI nucleotide aiignments to arnino acids was conducted using the

Drosophila mitochondrial code in MEGA 1 .O2 (Kumar 1993). Pairwise nucleotide

sequence divergences were then caiculated in Mega 1 .O2 (Kumar et al. 1993) for both

COI and 1 6s using the Kimura 2-Parameter mode1 (Kimura 1980) of molecular

evolution. The resulting matrices were used to estimate neighbour-joining (NJ)

phenograms for both genes in each species. with divergent sister tava chosen as

outgroups according to Remigio & Hebert (2000). Confidence lirnits for the nodes in NJ

phenograms were detemined in MEGA with 1 000 bootstrap pseudoreplicates

(Felsenstein 1985). The alignments of both COI and 16s rDNA nucleotide data for A. stefamsoni, and the 16s data for B. paludosa, lacked enough phylogenetically informative sites to conduct parsimony analysis. However. Maximum Parsimony (MP) analysis was possible for the COI data set for B. paludosa. As many haplotypes were separated by only 1 or 2 nucleotide changes. divergences within terminal groups had too few informative sites to be resolved by parsimony analysis. Accordingly, MP anaiysis was carried out on divergent. representative taxa fiom dl 8 major haplotype cluters identified by NJ anal pis. MP heuristic searches were conducted in PAUP*4.02b (SwoRord 1998)

( starting tree obtained by stepwise addition. random addition sequence) employing steepest descent and tree-bisection reconnection options, with transitions and transvenions weighted equally. Each data set was examined for phyiogenetic signal using the gi skewness statistic fiom 100 000 random tree length distributions (Hillis &

Hueisenbeck 1 992). and stability of groups was assessed by bootstrap analysis ( 1 000 pseudoreplicates), with groups appearing in 2 70% of the replicates considered well supponed (Hillis and Bull 1993).

Divergence time estimates for COI were determined using a clock rate of 2% per million years (see Methods. Chapter 2 for discussion). The nucleotide substitution rate of

0.38% per million years applied to the 16s rDNA data sets corresponds to both the calibration of Cunningham et. al (1992) for the same 16s mt DNA hgment in hennit cnbs. as well as calibrations of Lynch & Jarre11 (1993). RESULTS

Species distributions

The mean and range for elevation, conductivity, depth, and geographic CO- ordinates for the habitats From which A. stefamsoni and B. paludosa were collected are summarized in Appendix 3.2. Within the Canadian arctic archipelago. the mean devations of habitats containing B. paludusa (1 92 m) was greater than those with A. srt./inssoni ( 1 O5 m). B. paludosa was present in habitats of similar depth as populations of .-1. srefirnssoni ( 1.2 vs. 1 m). and in settings with conductivities more than twice as high

(467 vs. 232 ps/cm).

'4. stefanssoni was present throughout the eastem and central islands of the

Canadian arctic archipelago. but was not present West of 95' longitude. or in collections from Greeniand and the United States. By contrat. B. paludosa was obtained frorn habitats as îàr West as 133O W longitude and ranged eastward throughout the Canadian arctic archipelago to western Greenland. Both species ranges extended into northem

Ellesmere Island (81 O N). and southward on the Melville Peninsula of continental

Canada. The collections of A. stefonrsoni from both Ellesmere and Axe1 Heiberg Islands, and B. paludosa tiom Axe1 Heiberg Island represent the first records of these species

tiom these islands.

Genc seqmce divemity

Artemiopsis stefanssoni

COI gene sequence diversity and phenetic analysis

The 639 nucleotide COI alignment of 38 A. stefanssoni isolates was unambiguou as no gaps were present; amino acid translation of the nucleotide sequences did not reveal the presence of stop codons. A total of 147 nucleotide positions were variable,

only 18 of which were parsimony informative using cladistic criteria Corrected pairwise

sequence divergence estimates ranged from 0.0 - 2.1 % (Appendix 3.3.1 ), with an overall

mean divergence of 0.7 M. 1% among ingroup isolates.

The matnx of pairwise distances calculated between isofates from 38 populations

of A. stejamsoni revealed the presence of 29 unique haplotypes (Appendix 3.3.1).

Neighbour-joining (NJ) analysis with Polyarremiella hazeni employed as an outgroup,

showed that some haplotypes from the same geographic area clustered together (Figure

3 4. However. bootstrap evaluation failed to provide support for these groupings, as

values for al1 nodes were below 70%. and only three nodes exceeded 50%.

i 6s rDNJ gene sequence diversity and phenetic analysis

The 16s rDNA alignment of 1 1 A. stefamssoni in-group isolates was

unambiguous as no gaps were present. Addition of the outgroup taxon. P. hareni.

introduced several gaps which were 1-2 nucleotides in length. The 496 nucleotide

alignment contained 90 polyrnorphic sites. only three of which were informative for

cladistic analysis. Corrected painvise divergence estimates ranged from O - 0.4%

( Appendix 3.3 2).with an overall mean of 0.2 M.1 %.

The matrix of pairwise sequence divergence (Appendix 3.3.2) revealed 7 unique

haplotypes in the 1 1 A. stefamsoni isolates. NJ analysis revealed no clustering of

haplotypes into recognizable gr~ups(Figure 3.5). Bootstrap values were Iow, with only a

single node above 50%. The topology of the 16s rDNA NJ tree differed only slightly

from the COI phenetic analysis with respect to the relative positionhg of isolates. Figure 3.1. Neighbour-joining (NJ)phenogram based on sequence variation in mitochondrial COI sequences (639 aligned sites) for a single isolate hmeach of 38 populations of '4. stefamssoni. All 38 isolates representing 29 unique haplotypes are presenied. The tree was rooted with a single sequence of P. hrueni. Bootstrap values are indicated beside the branches; values below 40% are not shown. Terminal branches marked with an asterisk (*) indicate isolates merexamined for sequence variation in the 16s rRNA gene (Figure 3.5). Population abbreviations are listed in Appendix 3.1.1.

Figure 3.5. Neighbour-joining phenogram of 16s rRNA gene variation (496 aligned sites) for a single isolate from each of 11 populations A. stefamsoni. Al1 11 isolates representing 7 unique haplotypes are shown. The tree -MErooted with a single sequence of P. hazeni. Bootstrap values appear beside the branches; values below 40% are not shown. Population abbreviations correspond to those in Figure 3.4. and are listed in

Appendix 3.1.1 . Branch inecfa paludosa

COI gene sequence diversity and phenetic analysis

The 639 nucleotide COI alignment for B. paludoso was unambiguous as no gaps were present; amino acid translation of the nucleotide sequences did not reveal the presence of stop codons. Forty-five North Amencan. three Greenlandic, and one

Slovakian haplotype(s) were detected with 205 variable sites, 1 19 of which were phyloçenrtically informative using cladistic criteria. Overall mean corrected pairwise sequence divergence among ingroup taxa was 3.3 k0.4%, with a rang of 0.0 - 13.3%

( .4ppendix 3 A. 1 ).

The NJ tree constructed fiom the matrix of sequence divergence. with

Strepîocephulus dorothae as an outgroup. revealed that the 64 isolates of B. paludma rxamined in this study were divided into five well supported groups of haplotypes (A-E;

Figure 3.6). Group A was further divided into three clusters (A 1. A2. A3). although branc hing of group A 1was not well resolved (50%). Within group B. two clusters were well supponed (97%). Group C (SL-1) hmthe Tatra Mountains of Slovakia was sister to al1 North Amencan haplotypes except those of group E. which were detected in two populations from southeastem Bfin Island (BF-1. BF-2). Groups D (fiom the Medicine

Bow Mountains in Wyoming) and E were sister to al1 North Amencan lineages. and group C.

Cludistic analysis of sequence divergence in the COIgene

Maximum Parsimony (MP)heuri stic searches of 100 cladisticaily informative characters fiom 25 representative ingroup taxa found 38 most parsimonious mes (length Figure 3.6. Neighbour-joining phenogram of COI nucleotide variation (639 aligned sites) for a single isolate from each of 64 B. paludosa populations. The tree was rooted with a single sequence of S~ptocephaZusdorothae. Al1 64 isolates representing 49 unique haplotypes are presented. Bootstmp confidence limits are indicated above each node. Lctters on branches (A-E) denote the five major phylogroups identified. Numbea indicate lineages within groups A and B. Terminal branches marked with an asterisk (*) indicate isolates further examined for sequence variation in the 16s rDNA gene (Figure

3.7). Population abbreviations are listed in Appendix 3.1 2.

= 202; CI = 0.69; Ri = 0.8 1) that differed fiom each other only in the placement of the nodes for groups A and B, and the branching order of haplotypes within cluster Al. A strict consensus is shown in Figure 3.7. Using S. dorolhae as an outgroup, both strong phylogenetic signal (gl = -0.83; ht= -0.10; P < 0.0 1) and high bootstrap values support the presence of 7 phylogenetically divergent clades of haplotypes. which correspond to groups AI. A3. B 1, B2. C. D. and E identified by phenetic analysis (Figure 3.6). As in the NJ tree. groups D and E were cleariy sister to group C fiom Slovakia and polar lineages A and B. Relationships arnong lineages fiom groups A and B were less clearly rrsolved. Although the monophyly of groups A2. A3. B( 1.2). B 1. and BZ was well supported with hi& bootstrap values, no evidence existed for the phylogenetic cohesion of group 4 1. whose members formed an unresolved polytomy with groups Aî and A3.

Additional MP heuristic searches conducted on 28 different taxa (consensus not shown) gave congruent results (iength = 205; CI = 0.69;RI = 0.8 1), indicating that cladistic support was not af5ected by taxon choice.

1 6s rDN.4 gene seqiience diversity and phenetic anafysis

The 16s rDNA alignment for 14 B. paludosa ingroup isolates was unarnbiguous as no gaps were present. although the outgroup taxon S. dorothae caused the introduction of several gaps which were 1-2 nucleotides in length. The 490 nucleotide long alignent contained 1 1 8 variable sites of which 2 1 were phylogenetically informative. Corrected painvise sequence divergence estimates ranged fiom O - 3.8% (Appendix 3.4.2) with an overall group mean of 1.4 M.3%.and identified 1 1 unique haplotypes. C

TK-2 82 TK- 1

L

BF-1 WY-2 E 100 1 WY-1 Sdarolhae

Figure 3.7. Strict consensus of 38 most parsimonious mes (length 202: CI = 0.69: RI =

0.8 1) for 24 B. paludosa haplotypes. based on 100 phylogenetically informative characters of the COI gene. Clades are identified by lines with letters corresponding to ph y lopoups identi fied by NJ anal ysis (Figure 3.6). The dotted line identifies haplotypes which formed group Al in the NJ analysis, members of which do not fom a monophyletic clade. Bootstrap values (1 000 replicates) are shown on the branches. The 16s rDNA NJ phenogram constnicted fiom the distance matnx identified 5 haplotype clusters consistent with those identified by COI phenetic andysis (Figure 3.8).

.Mihough bootstrap support was not çtrong for al1 groups, the topology of the tree was consistent with the phenogram estimated from the COI data. The unique identity of croups B-E was maintained, aithough the 16s rDNA phenogram did not differentiate C between haplotypes from the three clusters identified within group A by NJ analysis of

COI data.

Divergence theestimates and refugiiti origim

The geographically unstructured distribution of A. stefanssoni haplotypes in the

Canadian arctic archipelago. with littie divergence (4%COI; S 0.4% 16s rDNA) occumng between widely distributed haplotypes. is consistent with their derivation from a single refugial ma(Figure 3.9). Estimates of divergence times between haplotypes from both COI and 16s rDNA sequence variation suggest their recent origin, indicating that '4. stefanîsoni haplotypes divenified in the last 350 000 to 530 000 years, with a mmimum separation of approxirnately 1 .1 Mya. Their restricted distribution in the central and northem arctic archipelago suggests their origins fiom a refuge either south of the ice margins on continental North Amerka, or north of the arctic ice sheets.

In contmt, the geographic distributions of rntDNA diversity in B. puhdosu suggest its persistence in at least 8 separate refugia, 6 of which were in the arctic (Figure

3.1 0.3.1 1 ). Divergence time estimates (see Table 3.1 ) between the major groups of B. paludosa di ffered only slightly between genes - estimates for both the COI and 16s rDNA data sets were similar between two lineages identified from arctic North Amenca Figure 3.8. Neighbour-joining phenogram of 16s rRNA gene variation (490 digned sites) for a single isolate from each of 14 B. paludosa populations. The tree was rooted wit h a single sequence of Strepiocephalus dorothae. AI1 14 isolates representing 1 1 unique haplotypes are presented. Bootstrap vaiues appear above the branches; values below 65% are not shown. Population abbreviations correspond to those in Figures 3.6 and 5.7. and are liste J in Appendix 3.1 -2. Figure 3.9. Geographic distribution in the Canadian arctic archipelago of

A. stefanssoni populations examined in this study. Populations containing the most (LI) and least ( * ) divergent pairs of haplotypes identified fiom painvise divergence estimates at the COI gene (see Appendix 33.1; Figure 3.4) are indicated.. Figure 3.10. Geographic distributions of the six arctic phylogroups of B. paludosa (A3,

A 1 ;A2. E: B 1. 82) identified by NJ analysis of the COI and 16s rDNA genes.

Figure 3.11. Map illustrating proposed Wisconsinan refugid origins for the 7 phylogroups of B. paludosa in North Amenca. The white overlay represents the rxtent of the Wisconsinan glacial margins 18 000 before present. The location of

Nunatuk rehgia in the Tomgat Mountains (Hodgson 1985; Dyke & Prest 1987;

Trenhaile 1990) are indicated by x's. Original sampling locations are illustrated

in Figure 3.3. Table 3.1. Mcan sequence divergence (below diagonal) and divergence time estimates per million years (above diagonal) between groups of B. paludosa identified with the neighbour-joining method (Figures 3.6,3.8) for both the COI (Appendix 3.1.1 ) and 16s rDNA (Appendix 3.1.2) rnitochondrial genes. Mean sequence divergences within major groups apprar in bold on the diagonals. Mean sequence divergences and divergence rstimares (in brackets) for the combined groups A and B appear in labeled columns. Al1 sequence divergence estimates were corrected using the mura(1 980) two-parameter mode1 of molecular evolution. Divergence time estimates were calculated using nucleotide substitution rates of 2% (COI) and 0.38% (1 6s rDNA) sequence divergence per million years. (-4.B) and Slovakia (C). By contrast, time estimates for deeper relationships between grooupps D (Wyoming)and E (Iqaiuit) from ail other groups differed between genes by 2-3 million years.

The rstimated mean divergence time of 1.5 Mya between North Amencan pups

A and B suggests their separation during the Pleistocene glaciations. Surprisingly, these groups from the Canadian arctic exhibited lower mean sequence divergence From the sin& population of B. paludosa fiom SIovakia (group C) than to groups D and E on the same continent (Table 3.1). This indicates that lineage C last shared a common ancestor with the dominant groups in the arctic archipelago about 3 Mya, just prior to the onset of the Pleistocene. By contrast. divergence estimates for groups D and E indicate their more ancient separation from al1 other lineages 6 to 9 Mya (Coi vs. L6S rDNA). suggesting their origins in the eariy Pliocene or late Miocene respectively.

The distribution of lineage A3. from Tuktoyaktuk dong the coastal Northwest

Territories to Manitoba and Hudson Bay. and the absence of this lineage fiom the centrai and northem arctic islands. is consistent with its derivation from source populations within the Beringian refuge (Figure 3.1 1). By contrast, the restriction of A2 haplotypes to the Foxe Basin area (barring one Greenland isolate) and their absence no& of Baffin

Island. suggest that this group denves from a refbgium in the eastem high arctic. Despite some overlap with A2 haplotypes on the Melville Peninsula and Baffin Island, the dominance of Al haplotypes in the central tier of high arctic islands suggests their

survival in a western hi& arctic glacial refuge. Clade B2 lineages. which were not

observed south of Ellesmere Island, likefy penisted in a northem high arctic refuge,

while B 1 lineages. which did not occur no& of Devon Island. Iikely derive fiom a separate. more southerly refugium. The restriction of group E haplotypes to a site on eastern Baffin Island suggests that this clade may have penisted in an ice-free region in this area. while group D swived south of the Wisconsinan Ice sheets on continental

North America.

DISCUSSION

Prior genetic work on anostracans has examined either intraspecific diversity in local areas (Bohonak 1998; Boileau et of. 1992; Fugate 1992; Riddoch et al. 1994;

Davies ri al. 1997) or the phylogenetic afinities of higher tava (Hanner 1997; Remigio &

Elctbert 2000). These studies have provided linle insight into the evolutionary relationships between populations within species of anostracans, many of which have broad distributions (Bnek & Thiery 1995). The current study is the first to examine the phylogeography of anostracan crustaceans across their range in the Canadian arctic. The results. based on two mitochondrial markers. present a divergent pattern of phylogeographic structure between species. The genetic structure of A. siefunssoni suggests that the populations of this species examined in this study denve from a single refugium. By contrast. B. puludosa suMved in multiple refugia in both arctic and montane North America as well as in montane regions of Europe.

Although their precise refùgial locations are uncertain, the presence of geographicaily localized groups of ailied haplotypes of B. paludosa indicate that this species suMved the Wisconsinan in numerous refugia throughout the arctic archipelago, including sites in the eastem and hi& arctic. The generally ailopatric distribution of B. paludosa lineages from the eastem. western and northem margins of its range suggests their derivation from populations that occupied codrefùgia dong glacial margins. By contrast. regions in the central arctic (Foxe Basin, Melville Peninsula), which were the last areas to deglaciate, have been colonized by several different lineages. This co- distribution of lineages indicates contact zones between fringe refugial groups, and hrther, suggests the survival of some lineages in centrally located small ice-fiee areas or nunatuks.

As several of the most divergent clades of B. paludosa were represented by only one or nvo populations. examination of more habitats lrom their geographic origin would be usefûl for confirming their distribution. Further, although this study did not examine

reproductive isolation between lineages. the depth of sequence divergence indicates that

B. paludosa merits more intensive study. Divergence between some phylogeographic

yroups of B. poludosa at 16s rDNA (4%)is comparable to that observed between

di fferent species in the notostracan genus Lepidm at 12s rDNA (King & Hanner 1998).

a gene which evolves more rapidly than 16s rDNA. As anostracan taxonomy has

iraditionally been based on morphology, it is possible that more detailed study would

reveal the presence of species complexes. as in other fieshwater invertebrates (Heben

1998: Rowe 2000; Witt & Hebert 2000).

Distribution and refugial origins

The analysis of sequence variation in B. paludoso revealed high diversity at both

the COI and 16s rDNA genes. Although many haplotypes were separated by only one or

two nucleotide changes, NJ and cladistic analysis for COI indicated that these haplotypes

were members of eight phylogeographic groups. showing marked (1.4 - 13 %) sequence

divergence. The general congruence between phylogenies built fiom the data sets for the

wogenes indicates that the clades in B. puIzidosa represent separate phylogeographic restriction of clade B2 haplotypes to Ellesmere and Axe1 Heiberg Islands, and its failure to penetrate more southem arctic islands indicate that this group almost certainly derives from a northem high arctic refuge. The geological documentation of ice-fiee regions, which persisted throughout the Wisconsinan on both westem Ellesmere and no&- western A~elHeiberg Island (Hodgson 1985; Dyke & Prest 1987). are consistent with this assertion.

Although goup E diverged from other B. paludosa lineages pnor to the

Pleistocene. the absence of this lineage from the northem islands has severai possible sxplanations. Fossil evidence has show that now submerged regions of the costal p tains of eastern Canada once supported arctic flora (Pielou 199 1). Hence, it is possible that B. puliidoso persisted there during the Pleistocene. and followed the receding

Laurentide margins north to their current location. Altematively, lineage E may have persisted in a coastal refugium in the arctic. With the exception of alpine glaciers, the rastem coast of Bafin Island from Bylot Island, south to Frobisher Bay was ice-fiee throughout the Wisconsinan, while closer to Iqaluit. nunatuk refugia persisted in the

Torngat Mountains of northem Quebec ( Dyke & Prest 1987; Murray 1987; Pielou 1991)

Examination of Dym integrfolio chloroplast DNA haplotypes from these regions revealed that they supported ailied haplotypes (Tremblay & Schoen 1999), suggesting that these refugia may have ken connected.

The ancient separation of lineage D from polar B. paludosa phylogroups refutes previous assertions that populations on southem rnountain ranges represent lineages recently established through waterfowl dispersal (Saunders et al. 1993). The irregular presence of B. paludma in montane ponds of the Rocky Mountains may aiso be due to refounding From a glacial relict egg bank, rather than re-colonization via migratory waterfowl (Donald 1983). The cioser association of Iqaiuit (lineage E) B. puhdosa to

Wyoming (lineage D) than to other arctic archipelago lineages. provides Mersupport that while E and D rnay have pesisted south of the Laureniide Ice sheets, lineages currently occupying the arctic archipelago are the descendents of polar. not southem refugial populations.

The divergence of lineage C (Tatra Mountains Slovakia) from North Arnencan arctic archipelago lineages approximately 3 Mya is consistent with its separation just prior to the onset of the Plcistocene (Webb & Bartlein 1992). As the Tatra populations are considered to be European glacial relicts (Brtek & Thiery 1995), mitochondnal divergence between these and North American lineages may provide an upper boundary to the separation of lineages fiom these continents. The deep divergences between polar and temperate lineages of B. paludosa on a single continent. and much shailower separation between North Amencan polar lineages and European glacial relicts is a striking contrast. Col boume et al. ( 1999) suggested that due to the relative proximity of

Nearctic and Palearctic habitats, these regions may serve as a "polar bridge" between continents. Results of the current study rnirror those found for lineages of the Daphnia pukx cornplex. and support Colbourne et al.'s (1 999) conclusion that long distance dispersal between polar regions may be more common than between temperate lineages due to their closer proximity, and the absence of large oceanic barriers.

The divergences between haplotype assemblage Al and clades A2 and B 1 are suficiently deep to suggest that they sunived the Wisconsinan in separate glacial retùgia. However. overlap in the distribution of haplotypes belonging to these groups, and their distance fiom known glacial refbgia make their precise origins uncertain. The absence of these lineages Mernorth or West do however indicate that they do not have westem or northem refugial origins.

Fossil and pollen evidence for the presence of a narrow band of tundra dong the southern margins of the Laurentide ice (Dillon 1956: Argus & Davis 1962; Ritchie 1987)

has led to the general assurnption that populations of many taxa followed the receding

W isconsinan ice margins north to recolonize the arctic (Pielou 1 99 1 ). Certainly,

conditions suitable for arctic anostracans may have existed south of the ice. but the ability

of such populations to penetrate habitats on the centrai tier of islands during deglaciation

is uncertain. Although Laurentide ice margins did recede to the no& a broad glacial

lobe covered the Foxe Basin. Melville Peninsula northem Quebec. and much of Baffin

lsland until just 8 -5 000 bp (Dyke & Prest 1987). This rernnant ice sheet would have

been a significant barrier to the northward spread of southem populations (if they existed)

into habitats that became available in the central tier of arctic archipelago islands 10 - 9

000 bp. Instead. these areas may have been colonized by northern refugial groups. which

would have had earlier direct access to these regions. For exarnple, lineages A2 and B1

may derive from ice-free areas that existed on northwest Somenet lsland or on the

Borden Peninsula of Baffin Island (Dyke & Prest 1987). The proximity of these refugia

would have provided hem first oppomuiity for the colonization of Somenet Island,

westem Baffin Island, the Melville Peninsula, and the Foxe Basin Islands as they became

available. Once established, these populations wodd have pre-empted colonization by

southern lineages. The distribution of Al haplotypes north of Baffin Island and their dominance on

Bathurst Island, suggests their persistance in an ice fiee area in the north-west during the

Wisconsinan. The derivation of Al from Banks, Melville, or Prince Patrick Islands is consistent with the broad distribution of this haplotype assemblage. During range expansion as the ice rnargins receded to the southeast, descendants of Al would have pre- cmpted the southem penetration of high arctic lineages fiom Axel Heiberg and Ellesmere

Island. If so. the disjunct presence of Al haplotypes on Ellesmere Island must reflect

me long distance colonization events. Their broad distribution to the southwest is not

consistent with a derivation from northeastem Ellesmere Island. and more generally, the

presence of disjunct " tip" haplotypes is normally an indicator of ment range expansion.

nther than refugiai occupation (Schaal et al. 1998).

In contrast to the deep phylogenetic breaks between phylogroups of B. paludosa,

the populations of A. stefirnssoni exarnined in this study appear to be de~vedfrom a

single Pleistocene refuge. Both 16s rDNA and COI data showed that populations of this

species exhibit low levels of sequence divergence. with haplotypes distributed

haphazardly across the central arctic archipelago. The broad distribution of closely

related haplotypes fiom Ellesmere Island. south to the more recently deglaciated regions

of the Foxe Basin, suggest the rapid colonization of these regions From a single glacial

retùge. Further. the absence of A. stefamsoni hmwestern habitats including Beringia

suggests that this species was not present in a western refuge. The presence of the

anostracan Polyurremiella hazeni in these regions (Johansen 1922) may have

competitively excluded A. stefamsoni f?om this area (Hebert & Ham 1986). The presence of A. stefanssoni in northwestem helHeiberg and western

Ellesmere Islands is consistent with its derivation from a high arctic refuge in this area.

The nonhem limit of this species mirron that of B. paludosa lineage 82, suggesting that these two species co-occupied ice-free areas on Axe! Heiberg or Ellesmere Island during the Wisconsinan. Following deglaciation. A. stefanssoni was able to penetrate central and southem habitats, while B. paludosa were pre-empted by colonists fiom other refugia

( cg. lineage A 1 ). As these two species are known to be trophically segregated. and commonly CO-occurin the same habitat (Dabom 1979), cornpetitive displacement in a common refugium is unlikely. If B. paludosa and A. stefànssoni shared a Wisconsinan refuge in the high arctic, it follows that they might also have CO-occupiedsouthem ice- kee areas. The occurrence of A. stefanssoni in both southeastem Baffin Island and norihem Quebec rnight reflect either a second phylogeographic group of A. stefunssoni, or simply the broad southward expansion of a nonhem lineage. As this study was not able to assess the conspecific status of arctic archipelago populations with the sole population of A. stefanssoni recorded south of 60°N. its relationship to polar lineages is uncertain. However, given the deep sequence divergence between B. paludosa populations tiom similar geographic locations. examination of genetic variation in this population would be interesting.

Polar phylogeography

This study has shown that at least for some zooplankton, ice-fiee areas in the high arctic served as important refuges. Together with ment investigations of other zooplankton and arctic Bora (VanRaay & Crease 1995; Tremblay & Schoen 1999; Abbon et al. 1995: 2000). this work suggests that coastal refugia in the high arctic were occupied by both freshwater invertebrates and terrestrial plants during the Wisconsinan. This result contrats with phylogeographic studies of arctic vertebrates, which have shown no evidence of hi@ arctic refugial lineages. For example. the rock ptarmigan now present on northern Ellesmere Island derive from Benngian refuge populations (Holder et al.

1 999). Similarly. intensive investigations of arctic freshwater fishes have revealed no cvidence for hi& arctic refugial lineages (Wilson et al. 1996; Wilson & Hebert 1998).

TXe difference in refùgial occupancy between vertebrates and other groups merits

explmation. Although Macpherson ( 1965) suggested that high arctic refugia may have

supponed certain species of marnmals. phylogeopphic evidence for this assertion has

yet to be uncovered. Hohan ( 198 1) acknowledged the importance of body size of

mamrnals in dctermining refugial survival. and compared the dynamics within small

glacial refugia to those of oceanic islands. Indeed, the small size of high arctic ice-&e

areas. glacial nunatuks. and coastal rehigia during the Wisconsinan would have had

dramatic implications for the persistence of populations of large-bodied mammals, as the

carrying capacity of these refugia would be limited (Pielou 199 1).

In addition to body size, two other life history traits are common to species that

rvidently possessed high arctic refugial lineages - short generation time and a diapausing

stage. Among crustaceans, there is a general association between small body size, short

generation time. and the ability of a species to enter prolonged diapause (Haimon &

Caceres 1996). A similar relationship exists in plants, as species with the shortest adult

life span produce seeds with the longest diapause (Rees 1993: 1994). For life in remote

giacial refùgia these factors would have had important implications. First, populations

of plants and freshwater crustaceans would have had high effective population sizes, enabling the maintenance of substantial levels of polymorphisms (Avise 2000). Second, many freshwater zooplankton have a short generation time and hi& fecundity, providing for the rapid establishment of populations following colonization events (Hebert 1998).

By contrast. many large vertebrates require years to mature, providing greater chance for lineage extinction before the establishment of populations. The ability to diapause was likely the most important defense against stochastic lineage extinction for passively dispersed taxa in small refugia. Certainly. the ability to survive long periods of ice advance. and subsequently refound populations without requinng immigration from distant populations. appears to have provided Beshwater zooplankton with the ability to withstand multiple cycles of glacial advance and retreat.

Collectively. the shared features of small body size. short generation time. and the ability to diapause have contributed to the successful occupancy of high arctic refugia by both freshwater zooplankton and terrestrial plants. The absence of evidence for high arctic refugial lineages in terrestrial vertebrates and freshwater fish subject to the same glacial events (Wilson et al. 1996: Wilson & Hebert 1998; Holder et al. 1999) suggests that species lacking these characteristics were unable to persist in smdl polar refugia.

Phylogeographic investigation of other small-bodied organisms. such as arctic arthropods will further determine if these characteristics were key for the survival of other taxa in the hi& arctic. LITERATURE CITED

Abbon W. Chapman HM, Crawford RMM, Forbes DG (1995) Molecular diversity and

denvations of populations of Silene acaulis and Smrfiaga oppositifoliu From the

high Arctic and more southerly latitudes. Molecular Ecologv, 4. 199-207.

Abbott RJ. Smith LC. Milne RI. Crawford RMM, Wolff K, Balfour 1 (2000) Molecular

analysis of plant migration and refùgia in the arctic. Science, 289. 1343-1346.

Anderson RS (1974) Crustacean plankton cornmunities of 340 lakes and ponds in and

near the National Parks of the Canadian Rocky Mountains. Journal of the

Fisheries Research Board of Camda. 3 1.855-869.

Argus GW. Davis ME3 (1 962) Macrofossils from a late-glacial deposit at Cambridge.

Massachusetts. American Midland Naturalist, 67. 1 06- 1 17.

Avise JC (2000) Phylogeography. The Hisiory and Formation of Species. pp. 447.

Harvard University Press. London.

Ayala FJ (1 997) Vagaries of the rnolecular dock. Proceedings of the National Acudemy

of 'Sciences. USA. 94.7776-7783.

Belk D. Brtek J ( 1995) Checklist of the Anostraca. Hydrobiologiu.. 298,3 15-353.

Bohonak AJ ( 1998) Genetic population structure of the fairy shrimp Brunchinecta

colaradensis (Anostraca) in the Rocky Mountains of Colorado. Canadian

Journal of Zoology. 76.2049-2057.

Boileau MG. Hebert PDN. Schwartz SS (1992) Non-equilibrium gene Frequency

divergence: persistent founder e ffects in nanual populations. Jomal of

Evo2irtion-y Bîology, 5.25-39.

Bne k J. Thiery A ( 1995). The geo graphic distribution of the European Branchiopods ( Anostraca, Notostraca S pinicaudata, Laevicaudata). Hydrobiologiu,. 298,

263-280.

Colboume JK. Crease TJ. Weider U et al. (1998) Phylogenetics and evolution of a

circumarctic species complex (Cladocera: Daphnia pulex). Biological Journal of

the L inneon Society, 65.347-3 65.

Cunningham C W. Blackstone NW, Buss LW ( 1992) Evolution of king crabs fiom hennit

crab ancestors. Nature. 355,539-542.

Dabom GR ( 1976) On the distribution and biology of an arctic fairy shrimp Artemiopsis

srrfinssoni JO hansen. 192 1 (Cnistacea: Anostraca). Canadian Journal of

Zoologv. 55.280-287.

Dabom GR ( 1978) Distribution and biology of some nearctic tundra pool phyllopods.

L'ilrh. Internat. Verein. Lirnnol. 20,2442-245 1.

Dabom GR ( 1979) Limb structure and sexuai dimorphism in the Anostraca (Cnistacea).

Canadian Journal of Zoology. 57.894-900.

Dadswell MJ ( 1974) Distribution. ecology and postglacial dispersal of certain crustaceans

and fishes in aastem North America National hfweum of Canada- Publicuiions

in Zoologv. 11. 1 1Op.

Davies CP. Simovich MA, Hathaway SA (1997) Population genetic structure of a

Cal i fomia endemic branchiopod. Branchinecta sandiegonensis. Hyùro biologia?

359. 149- 158.

Dexter RW ( 1953) Studies on North Amencan fairy shrimps with the description of two

new species. nie Americon Midland Naturulist. 4,751-771.

Dillon LS ( 1 956) Wisconsin climate and life zones in North Amenca. Science, 123, 167- 176.

Donald DB (1983) Erratic occurrence of anostracans in a temporary pool: colonization

and extinction or adaptation to variations in annual weather? Canadian Journal

of~Zoology.6 1. 1492-1498.

Dyke AS. Prest VK ( 1987) Paleogeography of northem North Arnerica., 18 000-5000

years ago. Geological Survey of Canada, Map 1 7034 scale 1: 1 2 500 000.

England J ( 1976) Late Quaternary glaciation of the eastern Queen Elizabeth Islands.

Northwest Territories. Canada Quaternary Research, 6.1 85-202.

Frlsenstein J ( 1985) Confidence limits on phylogenies: an approach using the bootstrap.

Evolurion. 39,783-79 1.

Frrguson MS ( 1933) Notes on the arctic fairy shrimp Branchinecta paludosa. The

Cknadian Field Naruralist, 39-86-89.

Folmer O. Black M. Hoeh W. Lutz R. Vnjenhoek R (1 994) DNA primers for

amplification of mitochondnal cytochrome c oxidase subunit I from diverse

metazoan invertebrates. hlolecular Marine Biology and Biotechnology. 3.294-

299.

Fox JF ( 1983) Germinable seed banks of interior Alaskan tundra Arcric and Alpine

Reseurch. 15,405-4 1 1.

Freedman B. HiII N. Svoboda J. Henry G (1982)Seed banks and seedling occurrence in a

high arctic oasis at Alexander Fjord. Ellesmere Island. Canada. Canadiun Journal

of Botany. 60.2 1 12-2 1 18.

Fryer G (1996) Diapause. a potent force in the evolution of freshwater crustaceans.

Hvdrobiologia 320. 1- 14. Fugate M ( 1992) Speciat ion in the fairy shrimp genus Branchinecta (Crustacecr:

d nostraca)fiom western North Arnerica- Ph.D. Dissertation, University of

California Riverside.

Gilbert D (1992)SeqApp version 1.90. A multiple seguence ediror for Macintosh

cornputers. Published electronically on the Intemet at fly.bio.indiana.edu.

Grulke NE. Bliss LC (1985) Growth forms, carbon allocation, and reproductive pattern

of high arctic saxifrages. Arcrk and Alpine Research. 17.241-250.

Haiston NG JR. Caceres CE (1996) Distribution of crustacean diapause: micro- and

macroevolutionary pattern and process. Hydrobiologia. 320.27-4.

Hanner RH ( 1 997) Tmnornic problems wilh phylogeneric solutions derivedpom the

inf egraf ion of b iochemicaI. morphological and mo lecular data. P Ph. D. Thesis.

University of Oregon. USA.

Hartland-Rowe R. Anderson RS (1 968) An arctic fairy shnrnp (Arremiopsis

srefonssoni Iohansen 192 1 ) in southem Alberta with a note on the genus

Arremiopsis. Canodian Journal of Zooiogv. 46,423 -425.

Hebert PDN. HaBJ (1 986) Patterns in the composition of arctic hindra pond

microcrustacean communities. Cunadian Journal of Fisheries and Aquatic

Sciences, 43, 14 16- 1425,

Hebert PDN ( 1998) Variable environments and evolutionary diversification in inland

waters. In: Advances in Molecular Ecology. (ed Carvdho GR), Series A: Life

Sciences. Nato Science Senes. 306. pp. 267-290. IOS Press. Ohmsha

Hewitt GM (1 996) Some genetic consequences of ice ages, and their role in divergence

and speciation. Biological Journal of the Linnean Society, 58.247-276. Hewitt GM (2000) The genetic legacy of the Quatemary ice ages. Nature, 405,907-9 13.

Hillis DM, Huelsenbeck JP (1992) Signal, noise, and reliability in molecular

phy logenetic analyses. The Journal of Heredity, 83. 189- 195.

Hi1 1 is DM. Bu1 1 JJ ( 1 993) An empincal test of bootstrapping as a method for assessing

contidence in phylogenetic analysis. Systematic BioZogy. 42. 182-1 92.

Hodgson DA (1 985) The last glaciation of west-centrai Ellesmere Island. Arctic

Arc hi pelago. Canada. Canadian Journal of Earth Sciences. 22,347-368.

Hoffman RS ( 198 1 ) Different voles for different holes: environmental restrictions on

rehgial survivai of mammals. In: Evolurion Todqy: Proceedings of the Second

international Congress of Systematic and Evolutioncrry BioZogy (eds. Scudder GE,

Reveal JL). pp. 24-25. Hunt Institute for Botanical Documentation. Pittsburgh.

Holder K. Montgomerie R, Friesen VL (1999) A test of the glacial refugium hypothesis

using patterns of rnitochondnal and nuclear DNA sequence variation in rock

ptmigan (Lagopus mutus). Evolution, 53. 1936-1950.

Hopkins DM. Matthews .W. Schweger CE. Young SB (eds) (1982) Paieoecology of

Beringia. Academic Press. New York.

Johansen F ( 192 1) nie larger freshwater crustacea from Canada and Alaska. The

Canadian Field Naturalist, 3521 -3 0.

Johansen F ( 1 922) Euphyllopod cnistacea of the Amerîcan Arctic. Report of the

Cunadian Arctic Expedition 1 91 3- 19 18. Volume VI: Crustacea. Part G:

Eirphyllopodu. p. 1 -3 4 Ottawa Kimura M (1980) A simple method for estimating evolutionary rates of base substitutions

through comparative studies of nucleotide sequences. Journal cf Molemlir

Evolurion. 16, 1 1 1 -1 20.

King JL. Hanner R (1 998) Cryptic species in a "living fossil" lineage: taxonomie and

phylogenetic relationships within the genus Lepidunrs (Crustacea: Notosîmca) in

North Amenca Molecukar Phylogenetics und Evolution, 10,23-36.

Kumar S. Tamura K. Nei M (1993)MEGA: Molecular Evolutionary Genetics Analysis.

Version 1.02. Distributed by the authors. Pennsylvania State University,

University Park. PA.

Lindsey CC. McPhail JD (1986) Zoogeopphy of fÏshes of the Yukon and Mackenzie

bains. in: The Zoogeography of North American Freshwater Fishes (eds. Hocun

CH. Wiiey EO). pp. 639-674. John Wiley and Sons. New York.

Lynch M. Jarrell PE (1993) A method for calibrating molecular clocks and its application

to animal mitochondrial DNA. Genetics, 135. 1 197-1208.

Macpherson AH (1965) The origin of diversity in mammals of the Canadian arctic

tundn. Sysrematic Zooiogy. 14, 1 53- 173.

McLaren IA ( 1964) Zooplankton of Lake Hazen. Ellesmere Island. and a nearby pond.

with special reference to the copepod Cyclops scufifer Sm. Canadian Journal of

Zoologv. 42.6 1 3-629.

Muller OF (1788) Zoologia Danica. 2. 10.

Murdoch J (1 885) Description of seven new species of Crustacea and one worm from

arctic Alaska Proceedings, U. S. National Museum- 1984,7,5 18-522. Murray DF ( 1987) Breeding systems in the vascular flora of arctic North Arnerica. In:

Dzfjerentiation Patterns in Higher Plants, (ed. Urbanska KM), pp. 239-262.

Academic Press. New York.

Musselman RC (1 994) (technical coordinator). The Glacier Lakes Ecosystem

Experiments Site. General Technical Report. RM-249. Fort Collins, CO: US

Department of Agriculture, Forest Service, Rocky Mountain Forest and Range

Experiment Station. pp.94.

Myers E ( 1986) Sequence Nmiigator. Applied Biosystems Inc.

Palurnbi SR ( 1996) Nucleic Acids 2: The polymerase reaction. In: Molecular

Sysrematics. (eds. Hillis DM, Moritz C. Mable BK), pp. 205-248. Sinauer

ri\ssociates. Sunderland MA.

Pielou EC ( 199 1 ) Ajer the Ice Age: the Return of Lfe to GIaciated North America. pp.

366. University ofChicago Press. Chicago.

Pons O. Petit EU ( 1995) Estimation, variance and optimal sampling of gene diversity. I

Hap loid locus. Theoretical und Applied Genefics.90.462-470.

Rees M ( 1993) Trade-offs among dispersal strategies in British plants. Nature, 366,

150- 153.

Rees M (1 994) Delayed germination of seeds: A look at the effects of adult longevity.

the timing of reproduction. and population agelstage structure. American

;l'at uralist. 144.43-64.

Remigio EA. Hebert PDN (2000) anities among anostracan (Crustacex

Branc hio poda) farnilies inferred fiom phylogenetic analyses of multiple gene

sequences. Mo./olecularPhylogenetics and Evolution. 17. 11 7- 128. Riddoch BJ. Mpoloka SW?Cantrell M (1994) Genetic variation and localized gene flow

in the fairy shrimp, Branchipodopsis wolfi. in temporary rainwater pools in

southeastem Botswana. In: Genetics and Evolution of Aqualic Organism. (ed.

Beaumont AR), pp. 96-102, Chapman and Hall, New York.

Ritchie JC ( 1987) Postglacial vegetation of Canada. Cambridge University Press, New

York.

Roen U ( 1 98 1 ) Studies on the freshwater Entomostraca in Greenland V. The fauna of the

Hazen Camp snidy are%Ellesmere Island. NWT.. Canada compared to that of

the Thule area Greenland. Streenstrupia. 7.32 1-335.

Roff JC. Carter CH (1 972) Life cycle and seasonal abundance of the copepod

Lirnnocalanzis macrum Sars in a high arctic lake. Limnology and

Oçeunography. 17.363-370.

Rowe. CL (2000) Global distribution. phylogeny and taxonomy of the freshwater

zooplankton genus Holopedium. M.Sc. thesis, University of Guelph, Guelph,

Ontario.

Saunden JF III. Belk D. Dufford R (1 993) Peeistence of Brunchinecfa paludosa

(Anostraca) in southem Wyoming. with notes on zoogeography. Journal of

CrusfaceonBiology, 13. 1 84- 1 89.

Schaal BA. Hayworth DA. Olsen KM. Rauscher .lT. Smith WA (1 998) Phylogeographic

studies in plants: problems and prospects. Moleculor Ecology, 7.45344.

Stanley SM ( 1987) fitinction. Scientific Amencan Library. WH Freeman. New York.

S wofford DL (1 998) PAW*: Phylogenetic analysis ushg parsimony Version 4.0b2.

S inauer Associates. Sunderland. MA. Tash JC. Armitage KB ( 1967) Ecology of zooplankton of the Cape Thompson area,

Alaska. Ecology, 18, 129-139.

Templeton AR. Routman E. Phillips CA (1995) Separating population structure from

population history, a cladistic analysis of the geographical distribution of

mitochondnal DNA haplotypes in the tiger salamander. Ambystoma tigrinum.

Genetics, 140, 767-782.

Trem blay NO. Schoen DJ ( 1 999) Molecular phylogeography of Dryos in~egr~lia:

glacial refügia and postglacial colonization. Molecular Ecologv. 8, 1 1 87- 1 198.

Trenhaile AS ( 1990) The Geomorphology of Canada. pp. 240. Oxford University Press.

Toronto.

VanRaay TJ. Crease TJ ( 1995) Mitochondnd DNA divenity in an apomictic Daphnia

compiex from the Canadian hi& arctic. Molecular Ecolog~,4, 149- 1 6 1.

Webb T. Bartlein P.J ( 1992) Global changes during the last 3 million years: ciimatic

controls and biotic responses. Annual Review of Ecology and Systernatics, 23,

141-173.

Weider LJ. Hobaek A (1 997) Postglacial dispersal, glacial refugia, and clonal structure in

Russian/Siberian populations of the arctic Daphnia pulex complex. Heredily. 78.

363-3 72.

Weider LJ. Hobaek A. Colbourne, JK et al. ( 1999a) Holarctic phylogeography of an

asexual species complex 1. MtDNA variation in arctic Daphnia. Evolution, 53,

777-792. Weider M. Hobaek A. Hebert PDN. Crease TJ (1999b)Holarctic phylogeography of an

asexuai species cornplex-II. Allozyme variation and clonal structure in arctic

Daphnio. iClolecular Ecology, 8. 1- 13.

Wilson CC. Hebert PDN, Reist JD, Dempson JB (1996) Phylogeography and postglacial

dispenal of arctic chan Salvelinus alpinus in North Arnerica. Molecular Ecoiogy.

5. 187-197.

Wilson CC. Hebert PDN (1998) Phylogeography and postglacial dispenal of lake &out

(Salvelinus namaymh) in North America Canadian Journal of Fisheries and

.4qiratic Sciences. 55. 1 0 10- 1 024.

Witt JDS. Hebert PDN (2000) Cryptic species diversity and evolution in the arnphipod

genus Hv~llelawithin centrai glaciated North Amenca: a molecular phylogenetic

ap proach. Canadian Journal of Fisheries and Aquatic Sciences. 57.687-698.

Zhang D. Hewitt GM ( 1996) Nuclear inteptions: Challenges for mitochondrial DNA

marken. Trends in Eculogy and Evolufion, 11.347-75 1. CHAPTER FOUR

GENERAL CONCLUSION Past work has examined the relative importance of both dispersal and vicariance events in determinhg phylogeographic structure in a broad array of life (Avise 2000).

My investigations of phylogeographic structure in passively dispersed organisms have examined the importance of dispersal syndromes and life hinory characteristics in the production of broad scale phylogeographic patterns following vicariance events. This work has provided new insights conceming the processes goveming the distribution of genetic variaiion across species ranges, and has revealed that life history characteristics were a primary factor in determining the effect of Pleistocene vicariance events on the genetic structure of freshwater organisms.

Through the examination of thezooplankton species from arctic and temperate

North Arnerica and their cornparison to other inhabitants of freshwater and terrestrial systrms. my work has revealed that the life history characteristics of passively dispened oqanisms have sheltered lineages From extinction, ailowed colonization events over great distances. and provided for the retention of ancestral polymorphisms, contributing to high levels of contemporary divenity. In addition, my work has provided important suppon for the presence of hi& arctic refugial lineages of freshwater taxa during the

W isconsinan.

Dispersal syndromes and phyiogeography

Previously it was believed that the potentiai for wind and avian mediated dispersal in freshwater zooplankton would lead to their uniform distribution (Maguire 1963). My study provides suppon for more ment work which ùidicates that. while passive dispersal has contributed to occasional long distance colonization events. this dispersal syndrome has not led to the diffuse distribution of lineages in ail groups (De Meester 1996; Hebert 1998). hstead. phylogeographic results indicate that zooplankton responded to the

Pleistocene glaciations in a similar fashion as actively dispersed fieshwater organisms

(Chapter 2). Sida crysiallina lineages exhibit phylogenetic breaks in the same geographic regions as thoçe observed in fish subject to the same vicariant events. This congruent pattern illustrates the common influence of Pleistocene glaciations on the distribution of eenetic variation in freshwater species of North America. Together with similar patterns C observed in D.lnevis (Taylor et ai. 1998), these results imply that despite their ability to cross watershéd boundaries. this occurrence in Cladocera is either so rare that its importance lies mainly in the occasional colonization of distant habitats, or that pre- rmptive habitat occupancy is a significant force in preventing the establishment of new coionists (Hebert 1998).

Cornparison of divergence depths between lineages of different species highlights the implications of the life history traits associated with a passive dispersal strategy. The deeper divergences between and higher diversity within lineages of freshwater zooplankton in cornparison to fkshwater fishes From glaciated areas illustrates this point.

My work suggests that the production of a diapausing stage and the ability of parthenogenetic individuals to rapidly establish populations have provided resistance to stochastic lineage loss not available to freshwater fishes (Chapter 2). However. the fact that fish species fiom temperate regions not subject to repeated glacial cover exhibit similar divergence depths as zooplankton fiom these areas (Bematchez & Wilson 1998) suggests that the short generation tirnes of zooplankton species was not the primary factor leading to the contrasting divergence depths in giaciated regions. uistead the smali body size of freshwater zooplankton was likely more important - their ability to sustain large populations in small habitats would have allowed the retention of higher levels of variation through the maintenance of greater evolutionary effective population sizes.

Diapause and the occupancy of high orcric refugiu

Chapter 3 revealed the importance of diapausing stages in the ability of taxa to

persist in small. high arctic refùgia dunng the Wisconsinan glaciations. This work

supports the growing nurnber of studies that indicate that species whose life histones

include a stage capable of long periods of diapause. and whose small size may result in

relatively large local population sizes, were able to persist in srnall refûgia during glacial

cycles (Abbot et cil. 1995; 2000; Tremblay & Schoen 1999). Despite the short geological

time penod since deglaciation. divergences between high arctic lineages in this study are

relatively deep. indicating that these life history traits were important for the retention of

lineages in polar refugid groups. This evidence provides an explanation for the absence

of high arctic refugial goups of vertebrates (Wilson et al. 1996; Holder et al. 1999). as

large-bodied taxa with long generation times. and no diapausing stage would have been

lcss resistant to stochastic lineage loss or local population extinctions. Closer inspection

of other vertebrate species for which high arctic refugia have been suggested (e.g.

Macpherson 1965) will provide more detailed information regarding the factors

governing refûgial origins of these taxa

The presence of several refugial groups of B. paludosa in the Canadian arctic

archipelago contrasts with the single refugial group observed for A. stefumsoni, further

indicating that the impacts of Iife history on phylogeographic patterns is cornplex. The

distributions of taxa pnor to glaciations, competition within refugia, and ecological tolerances have likely also influenced the contemporary distribution of lineages (e.g.

Hebert & Hann 1986). Future examination of additional A. sfefanssni populations from the eastem and southem arctic will help to clarify the significance of this conflicting pattern.

Summary

Together. my results illustrate the importance of the life histoiy characteristics associated with passive dispersal on the distribution of genetic variation across species ranges. In general. freshwater zooplankton fiom glaciated North Arnenca exhibit high levels of divenity within. and deep divergences between refugial groups. Their apparent resistancc to extinction is likely due to the combined influences of small body size, ability to diapause. and rapid reproduction.

The presençe of hi& arctic refugial lineages of zooplankton supports the conclusion that areas of the hi& arctic were not glaciated during the Wisconsinan. In combination with recent work on plants, my work highlights the importance of rxamining evidence From numerous disciplines when developing theones about historical events. Geologists and biologists have much to lem fiom each other - even if ice-free areas did not exist in the high arctic fhroughout the Pleistocene (Hodgson 1985), evidence fiom this study confirms that in the late Wisconsinan, some regions were clearly habitable. These findings are important. as a complete understanding of the geological and climatological history of polar regions is necessary to detemine how these areas will respond to funire changes in the global environment.

Future studies should examine comparative phylogeography of both srnall and large-bodied aquaùc and terrestrial arctic organisms. Such work will provide a more detailed understanding of late Wisconsinan temperate and arctic cnvironments. and further clarify how dispersai syndromes and life history traits contribute to the impact of vicariance events on the evolutionary trajectories of different taxa. LITERATURE CITED

Abbon RJ. Chapman HM, Crawford RMM, Forbes DG (1995) Molecular diversity and

derivations of populations of Silene acaulis and Saifaga oppositifoliu fiom the

high Arctic and more southerly latitudes. Molecular Ecologv, 4. 199-207.

.4bbott RJ. Smith LC. Milne RI, Crawford RMM, Wolff K, Balfour J (2000) Molecular

analysis of plant migration and refugia in the arctic. Science, 289. 1343-1346.

.\vise JC (2000) Phylogeography. me History and Formation of Species. pp. 4-47.

Harvard University Press. London.

Bcmatc hez L. Wilson CC ( 19%) Comparative phylogeography of Nearctic and

Palearctic fishes. Molecular Ecology, 7.43 1-42.

De Meester L ( 1996) Local genetic differentiation and adaptation in freshwater

zooplankton populations: pattern and processes. Ecoscience. 3.3 85-399.

Heben PDN. Ham BJ (1986) Patterns in the composition of arctic nuidra pond

microcmstacean communities. Cunadian Journal of Fisheries and Aquatic

Sciences. 43, 14 1 6- 1425.

Hebert PDN ( 1998) Variable environments and evolutionary diversification in inland

waters. In: Advances in Molecular Ecology. (ed Carvalho GR),Senes A: Life

Sciences. Nato Science Senes, 306. pp. 267-290. IOS Press, Ohmsha.

Hodgson DA (1985) The ktglaciation of west-central Ellesmere Island, Arctic

Archipelago. Canada Camdian Jouml of Eurth Sciences, 22,347-368.

Holder K. b!ontgornene R, Friesen VL (1999) A test of the glacial refugium hypothesis

using patterns of mitochondrial and nuclear DNA sequence variation in rock

ptarmigan (Lagopus murus). Evolution, 53, 1936- 1 950. Macpherson AH ( 1965) The origin of diversity in mammals of the Canadian arctic

tundra. Systematic Zoology. 14, 1 53-173.

Maguire B Ji (1 963) The passive dispersai of srna11 aquatic organisms and their

colonization of isolated bodies of water. Ecological Monographs, 33, 161- 185.

Taylor DJ. Finston TL. Heben PDN (1998) Biogeopphy of a widespread freshwater

crustacean: pseudocongnience and cryptic endemism in the North Amencan

Duphnia taevis complex. Evolution, 52, 1648- 1670.

Tremblay NO. Schoen DJ (1999) Molecular phylogeography of Dryas integrifolia:

glacial refugia and postglacial colonization. hfolecular Ecology. 8. 1 187-1 198.

Wilson CC. Hebert PDN. Reist JD. Dempson JB (1 996) Phylogeogaphy and postglacial

di spersal of arc tic cham Saivelinus alpinus in North Arnerica. Molecular Ecology,

S. 187-197. APPENDICES Appendix 2.1: Locations and abbreviations for North Amencan and European habitats with Sida crystaflina s.1.. Populations are designated by both number and biogeographic province of origin (Burr & Mayden 1992): CD (Cascadia); CL (Great Lakes); HB

(Hudson Bay); MI (Mississippi); NA (North Appalachian); SE (Southeastem); YM

( Yukon-Mackenzie). Country abbreviations are: CA (Canada); US (United States); BE

(Belgium); NL (Netherlands); CH (Gemany). Type of data collected from each population are indicated by D = mtDNA only; A=allozymes oniy; *= mtDNA and al lozyne data. Code Sample Site Regioo C LOY) LON) Data Type Saunders Lake Oregon Fem Ridge Lake Oregon Umarned Pond Oregon Cain Lake Washington Mosquito Lake British Columbia Skidegate Lake British Columbia Pure Lake British Columbia Mayer Lake Bntish Columbia Cluculz Pond British Columbia Fraser Lake British Columbia Babine Lake British Columbia Boya Lake British Columbia Blue Lake British Columbia Twin Lakes Yukon Klusha Pond Yukon Feny Pond Northwest Temtories Srnall Frog Lake Northwest Temtories Three Forks Marsh Northwest Temtories Arctic Red River 2 Northwest Territories Crossley Lakes 15 Northwest Territories Old Crow 16 Yukon Old Crow 20 Yukon Eskimo Lakes 5 Northwest Temtories Eskimo Lakes 3 Northwest Temtories Great Slave Lake Northwest Territories Lac La Ronge Saskatchewan Amisk Lake Saskatchewan Reed Lake Manitoba Lake Wekusko Manitoba Lake William Manitoba Lake Manitoba Manitoba Lake of the Woods Ontario Quebec Pond Quebec Sprin@kld Lake Missouri Frank Lake Wisconsin Plum Lake Wisconsin US 46.00 89.50 Code Sample Site Region C L(N) L(W) Data

Soldier Lake Michigan Lake St. Clair Ontario Sparrow Lake Ontario Lake Couchiching On tario Prospect Lake Ontario Echo Lake Ontario Raven Lake Ontario Paint Lake Ontario Muskoka Lake Ontario Lake Rosseau Ontario Maple Lake Ontario Pine Lake Ontario Beech Lake Ontario Grass Lake Ontario Little Boshkong Lake Ontario Chemong Lake Ontario Lake Esson Ontario Silent Lake Ontario Coburg River Ontario Lake St. George Ontario Paudash Lake Ontario Millhaven Dock Ontario 1000 Islands Ontario Cape Breton Lake Nova Scotia Second Lake New Brunswick Magaguadavic Lake New Brunswick Patrick Lake Maine Gardener Lake Maine Berry Reservoir Georgia Alatoona Lake Georgia Santa Fe Lake Florida Dorr Lake Florida River Meuse Hastiere Lake Maarsseveen Utrecht Schoek Plon Appendix 2.2. Sumrnary of allele fiequencies in North American and European populations of Sida crystaZZina s.1. at five polymorphic allozyme loci. Alleies are labeled according to their mobility relative to the designated standard. Population codes are presented in Appendix 2.1.

Population Locus 02CD 04CD 05CD 06CD 07CD O8CD 09CD IOCD lZYM 13YM l4YM

Gpi Population Locus 15YM 16YM 17YM 18YM 19YM 20YM 22YM 24YM 26HB 27HB 28HB

Population Locus 55GL 56GL 57GL 60NA 61NA 64NA 65SE 66SE 67SE 68SE E3CH .Mpi (n) 0.92 1 .O0 1 .O8 pgm (n) 0.93 1 .O0 1 .O3 1 .O5 s.4 al In) 0.7 1 0.89 1 .O0 1 .O9 .Wh (n) 0.83 0.91 1 .O0 Gpi

0.55 0.80 1 .O0 1.16 0.89 1 .O4 Appendir 3.1. Location details and abbreviations for populations of A. stefanssoni

(3. i .1 ) and B. paludosa (3.1.2) included in the study. Populations are designated by both number and area of origin. Habitat parameters are denoted by: E (elevation in metres above sea level); D (depth of habitat in metres); C (conductivity of water in habitat in micro-siemens per centimetre); LA(latitude); Lo (longitude). Three dashes (-) indicate data not available. Asterisks (*) denote populations from which multiple individuais

were sequenced for preliminary distance analyses (see methods for details). 3.1.1 CODE E D C LA LO Area Site AH-! 188 Axel Heiberg Island Mokka Fiord 28 AH-2 194 Axe1 Heiberg Island Mokka Fiord 27 AH-3 IZI Axe1 Heiberg Island Flat Sound 18 AH4 82 Axel Heiberg Island Nansen Sound 15 AH-5* 3 Axe1 Heiberg Island Nansen Sound 30 BF-1 97 8aîXn lsland Enchsen Lakes 7 BF-2* 2.4 Baftln Island Broduer Peninsula 8 BF-3* 267 Baffin lsland Berlinguet Inlet 2 BT-1 49 Bathurst Island Freeman's Cove 6 BT-2 O Bathurst Island Polar Bear Pass BT-3 30 Bathurst Island Freeman's Cove 27 CW-I 91 Cornwallis Island Disappointment Bay 12 DV-1 O Devon Island TrueLove Lowlands 1 DV-2 O Devon Island TrueLove Lowlands 2 DV-3 61 Devon Island Owen Point 13 EL-I 448 Ellesmere Island Tanquary Fiord 10 EL-?* 45 Ellesmere lsland Tanquary Fiord 13 EL-3 1 Ellesmere Island Tanquary Fiord 13 EL4 434 Ellesmere Island Tanquary Fiord 5 EL-5 0.6 Ellesmere Island Sverdrup Pass 30 EL-6 448 Ellesmere Island Sawtooth Mountains 18 EL-7 127 Ellesmere lsland Fosheim Peninsula 32 EL.-8 91 Ellesmere Island Fosheim Peninsula 27 FB-l O Foxe Basin Bray Isiand 2 FB-2 O Foxe Basin Spicer Islands 1 FB-3 O Foxe Basin Spicer Islands 5 FB-I O Foxe Basin Spicer Islands 3 FB-5 76 Foxe Basin Iploolik Island 5 LC-I 18 Little Cornwallis Island Little Cornwallis 6 LW-i 76 Lowther Island Lowther lsland 4 MP-1 106 Melvitle Peninsufa Quaraite Lake 8 MP-2 20 MeIviIIe Peninsula Richard's Bay 54 MF-3 39 Melville Peninsula Richard's Bay 53 MP-4 112 Melville Peninsula Sarcpa Lake 17 MP-5 303 Melville Peninsula Sarcpa Lake 2 MP-6 182 Me IvilIe Peninsula Sarcpa Lake 1 SO-1 6. I Somerset isIand CwellBay 4 SO-2 54 165 73.74 95.32 Somerset fsland Aston Bay 2 1 3.1.2 Code E D C LA LO Area Site AH- I Axel Heiberg Island Mokka Fiord 28 AH-? Axel Heiberg Island Mokka Fiord 3 1 AH-3 Axel Heiberg Island Schei Peninsula 8 .4H4 Axel Heiberg Island Flat Sound 19 AH-5 Axel Heiberg Island Nansen Sound 8 A H-6 Auel Heiberg Island Nansen Sound 23 AH-7 Axel Heiberg Island Nansen Sound 30 RF-I * Baffin Island Iqaluit 2 B F-2 * Baffin Island Iqaluit 1 B F-3 Baffin lsland Longstaff Bluff 7 BF-I* Bafin Island Enchsen Lakes 7 BF-5 Baffin Island Broauer Peninsula 1 B F-6 Baffin Island Brodeur Peninsula 8 BF-7 Bafin Island Berlinguet fnlet 6 BF-8 Baffin Island Berlinguet Inlet 1 BF-9 Baffin Island Berlinguet Inlet 2 BF-IO Baffin Island Berlinguet Inlet 3 BT- I Bathurst Island Freeman's Cove 6 BT-2 Bathurst 1s land Freeman's Cove 16 BT-3 Bathurst Island Freeman's Cove ! BT-1 Bathurst Island Polar Bear Pass BT-5 Bathurst Island Bracebridge Inlet 2 CW-1 CornwaIlis Island Resolute 17 C W-2* Cornwallis Island Disappointment Bay 12 DV- I Devon Island TrueLove Lowlands 4 DV-3 Devon Island Viks Fiord 4 DV-3 * Devon Island Eden Point 8 DV4 Devon Island Owen Point 12 DV-5 Devon Island Owen Point 8 EL- I Ellesmere Isiand Cape Herschel 14 EL-2 Eilesmere Island Cape Herschel 12 EL-3 Ellesmere Island Tanquary Fiord 9 EL4* Ellesmere Island Tanquary Fiord 3 EL-5 Ellesmere Island Tanquary Fiord 6 E L-6 Ellesmere Island Fosheim Peninsula 7 Ellesmere Islrnd Fosheim Peninsula 25 Code E D C LA LO Are8 Site FB-I* O Foxe Basin Spicer Islands 2 FB-2 O Foxe Basin Spicer Islands 3 FB-3 O Foxe Basin Spicer Islands 5 FB-t O Foxe Basin Rowley Island 10 FB-5 O Foxe Basin Jens Munk Island 9 FB-6 7.6 Foxe Basin Igloolik Island 5 FB-7 O Foxe Basin lgloolik isIand 2 HB-l O Hudson Bay Belcher Islands JL-I - Jenny Lind Island Jemy Lind Island 10 LC-1 24 Little Comwalis Island Little Cornwallis 4 MB-l O Manitoba Churchill 1 MB-3 O Manitoba Churchill 2 MB-3 O Manitoba Churchill 3 ME34 O Manitoba Churchill 4 MF-I 106 Mellville Peninsula Quaruite Lake 8 MP-2* 213 Mellville Peninsula Sarcpa Lake 17 MP-3 10.6 Melville Peninsula Richard's Bay 75 MP-I 20 Meiville Peninsula Richard's Bay 54 so-I 54 Somerset Island Aston Bay 2 1 TK-1 O Tuktoyaknik Tareoknitok Pond TK-2 O Tuktoyakhdc Tuktayaktuk C ity TK-3 O Tuktoyaktu)c Telephone Pole Pond WY-l*3316 Wyoming Lost Pond WY-2 3181 Wyoming Towner Lake Pond SL-i* 1690 Slovakia Tatra Mountains GL-I - Greenland Uurnmannaq Island 78 GL-2 - Greenland Godhavn 48 GL-3 50 6933 53.53 Greenland Godhavn 2 1 Appendix 3.2. Summary of characteristics for the habitats occupied by A. stefanssoni and B. puludosa listed in Appendix 3.1 .

B. paludosa

iM ean Range Mean Range

Elevation n 1nlelre.S) 64 193.2 0-3316 Dep t h tmelres) 59 1 .S 0.3 - 4.8 Conductivity /us1crnj 46 467.4 40 - 5000 Latitude / " .VI 64 70.4 41.4 - 81.4 Longitude / " H'J 64 86.8 53.5 - 133.3 Ap pendix 33. Painvise sequence divergences for isolates of A. stefanssoni for both the cytochrome oxidase 1 (3.3.1) and 16s rDNA (3.3.2) mitochondnai genes. The estimates were corrected using the Kirnura (1 980) two-parameter mode1 of molecular evolution. LW-1 BT- 1 BT-2 EL4 FE2 E L-5 MP-1 LC-1 BT-3 AH-1 F&f F&5 E L-2 B F-3 BF-2 3flP-5 AH-5 MP-6 DV- 1 FB-3 EL-1 SO- 1 E L-8 FM EL4 E L-3 A H-4 MP-3 A H-3 DV-2 SO-2 cw-1 MP-2 A H-2 DV-3 MP-4 BF-1 EL-7 LW-1 BT- 1 BT-2 EL4 Fi3-2 EL-5 MP-1 LC- 1 BT-3 AH- 1 FB-1 FR5 E L-2 B F-3 BF-2 M P-5 A H-S MP-6 DV-1 FB-3 EL-1 SO- 1 E L-8 FB-l E L-6 EL3 AH4 iM P-3 A H-3 DV-2 SO-2 CW-1 MP-2 AH-2 DV-3 MPl BF-1 EL7 LW-1 BT- 1 BT-2 EL4 FB-2 E L-5 -MP-1 LC-1 BT-3 AH-1 FR1 FB-5 E L-2 B F-3 B F-2 iMP-5 AH-5 LMP-6 DV-1 FB-3 EL-1 SO- 1 EL-8 FM E L-6 E L-3 AH4 MP-3 AH-3 DV-2 $0-2 CW-1 MP-2 AH-2 DV-3 MP-4 BF-1 EL7 B F-3 AH-2 E L-5 so-2 MP-2 EU BF- 1 DV- 1 IM P-l A H-1 E L-7 Appendix 3.4. Paiwise sequence divergences for isolates of B. pafudosa for both the cytochrome oxidase 1 (3.4.1) and 16s rDNA (3.4.2) mitochondrial genes. The estimates were corrected using the Kimura (1 980) two-parameter mode! of molecular evolution.

DV-J BF-IO FE3 W\'-2 EL5 G1.J BF-7 Br-J CH'-2 BFJ BF8 FEI D V-5 F ES .\H-6 TK-t BF-9 FE6 m-5 SO- l BT-2 \f Eu OOtl FE7 O OIS RF-l O Il5 UT- f O 003 Cl.-1 O 027 BF-3 O Ol8 GIC2 1) O3 1 DF.4 O 008 SIPI O 058 ni-1 0018 w-7 0 032 Y P-3 O 031 DV-2 O006 BT-4 O O02 1.c-1 O 003 WB? 0014 FE7 O O05 .IL-l O 013 EL1 O 002 \SB-l O011 BF4 0 00s EL3 O 019 tH-5 O 029 TK-3 0 013 w-I 0 034 iH-2 O031 VP-2 0 029 tH4 O 029 EL2 0 003 \II-[ 0 079 w-3 O 034 EL-7 0 027 \l B3 O O14 EL4 0 019 Dt--l O 029 BF-2 O 125 WP-I 0 005 FIP4 O 018 HBI O OIJ FI34 O 018 BF-5 O 008 N\'-l O 118 EL4 O O05 3.4.1

DY-3 BF-IO Fr3-3 WY-2 EL.-5

Di'-3 BF-IO FE3 WY-? E L-5 GIA BF-7 Er-3 CW-2 BF-4 BF-8 FBI DV-5 t;s5 \IL6 îK- l BF-9 FM BT-5 U>-l Kr-2 vu FE? BF- l RT-I G L- 1 BF-3 c;L-L DV-4 SL-l rK-2 \H-7 \f P-3 DV-2 Er4 LC-1 $1 B2 FB7 .IL-l Elcl !¶BI BF-6 E 1.-3 \H-5 TK-3 (3'-1 \H-? \I P-2 \II4 El.-2 .w-1 \H-3 E 1.-7 ME3 €1.4 DV- I BF-2 VP-1 NfP-8 HE1 TB4 BF-5 %'Y-1 E [A 4 MP-1 MP-4 H&l FBJ BF-S W-1 EL4 DV-3 BF-IO FB3 \VI' - 2 E L-5 G L-3 BF-7 BT-3 (3-2 BF-4 El F-8 FEI DV-5 FBS AH-6 TK-1 BF-9 FE6 m-5 SO-1 BT- 2 \iW FEZ BF-l BT- 1 GL-1 BF-3 CL2 DV4 SL-I TL2 AH-7 \¶Pd DY-2 BT4 1.C- l Wb2 FB-7 JL-I EL-1 tt&l BF-6 E L-3 3.4.2

B F-8 !WB4 5V-3 FB-5 T K-2 WY- 1 ,MP-2 EL-1 EL-5 SL- 1 EL-7 AH4 MP-4 BF-1 DV-1 3.4.2

B F-8 MB-4 DV-3 FB-5 T K-2 WY- 1 MP-2 EL-1 E L-5 SL- 1 EL-7 AH4 MP-I BF-I DV- 1