Genetic Disequilibria and the Interpretation of Population Genetic Structure in Daphnia

Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 668

______

LARS M. BERG

ACTA UNIVERSITATIS UPSALIENSIS UPPSALA 2001 Dissertation for the Degree of Doctor of Philosophy in Conservation Biology presented at Uppsala University in 2001.

ABSTRACT

Berg, L. M. 2001. Genetic disequilibria and the interpretation of population genetic structure in Daphnia. Acta Universitatis Upsaliensis. Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 668. 36 pp. Uppsala. ISBN 91-554-5152-7.

Understanding the processes that shape the spatial distribution of genetic variation within species is central to the evolutionary study of diversification and demography. Neutral genetic variation reflects past demographic events as well as current demographic characteristics of populations, and the correct interpretation of genetic data requires that the relative impact of these forces can be identified. Details of breeding systems can affect the genetic structure through effects on effective migration rate or on effective population size. Restrictions in recombination rate lead to associations between neutral marker genes and genes under natural selection. Although the effects on genetic structure can be substantial, the process will often be difficult to tell apart from stochastic effects of history or genetic drift, which may suggest erroneous conclusions about demography. In cyclically parthenogenetic freshwater invertebrates, which alternate between sexual and asexual reproduction, demographic fluctuations and reliance on diapausing eggs for dispersal enhances neutral genetic differentiation as well as effects of selection on associated genes. Although genetic founder effects are expected to be profound and long-lasting in these species, genetic hitch-hiking may reduce initial strong differentiation rapidly if better adapted genes are introduced by mutation or immigration. Fluctuating environmental conditions have been suggested to generate rapid shifts in the frequencies of clones during the asexual phase. In the presence of egg banks resting in sediments, genetic diversity is stabilised and the importance of migration for differentiation is reduced. Studies of unstable and young populations of cyclically parthenogenetic Daphnia pulex showed substantial variation for important fitness traits, within as well as between populations, despite hypothes- ised recent founder effects. Neutral markers indicated genetic equilibrium, but changes in clonal composition during asexuality disrupted the genetic structure in a manner compatible with local adaptation and exclusion of immigrants. This illustrates that the forces affecting sexual progeny may be markedly different from those shaping the structure among asexual individuals.

Key words: Breeding system, linkage disequilibrium, dispersal, genetic structure, Daphnia pulex

Lars Berg, Department of Conservation Biology and Genetics, Uppsala University, Norbyvägen 18D, SE-752 36 Uppsala, Sweden

ISSN 1104-232X ISBN 91-554-5152-7

Printed in Sweden by Uppsala University, Tryck & Medier, Uppsala 2001 Preface

This thesis is based on the following papers, which will be referred to in the text by their Roman numerals:

I Berg LM & Lascoux M (2000) Neutral genetic differentiation in an island model with cyclical parthenogenesis. Journal of Evolutionary Biology 13: 488-494.

II Berg LM (submitted) Selection and differentiation in cyclically parthenogenetic species. I. Hitch-hiking with sweeping allele.

III Berg LM (submitted) Selection and differentiation in cyclically parthenogenetic species. II. Fluctuating selection and egg banks.

IV Berg LM, Pálsson S & Lascoux M (2001) Fitness and sexual response to population density in Daphnia pulex. Freshwater Biology 46: 667-677.

V Berg LM, Lascoux M & Bengtsson J (submitted) Local genetic structure for microsatellites as indicator of demographic and selective processes in Daphnia pulex.

Reprints were made with permission from Blackwell Scientific Publishers © Contents

Sammanfattning Teori...... 1 Empiri...... 3 Slutsatser...... 4

Introduction The origin of genetic structure...... 7 Measuring genetic structure...... 8 Interpreting genetic structure...... 9 Daphnia as a model organism...... 12 Aims and outline of the thesis...... 13

Theoretical studies Neutral differentiation with cyclical parthenogenesis (I)...... 14 Selection and differentiation in Daphnia (II, III)...... 16

Empirical studies Geographic structure for fitness traits in Daphnia pulex (IV)...... 22 Geographic structure for neutral genes in Daphnia pulex (V)...... 25

Conclusions...... 29

References...... 30

Acknowledgements...... 36 “Problems in science are sometimes made easier by adding complications” Daniel C Dennett

Dedicated with love to Gösta Sædén (1917-1994), my first and best example of the scientific mind

Sammanfattning

Ekologiska tolkningar av populationsgenetiken hos hinnkräftor

DET ÄR en allmän uppfattning att Charles området molekylärekologi, som studerar Darwins idé om det naturliga urvalet revolu- sambanden mellan genetik och bl a demo- tionerade biologin. Detta stämmer förvisso, grafi. men är bara halva sanningen. Kanske ännu viktigare var hans insikt om betydelsen av den naturliga variation som finns mellan in- Teori divider inom arter, för utan variation kan För att gå från genetik till demografi krävs inget urval ske. Medan biologer före Darwin modeller som på ett förenklat sätt beskriver brukade se de avvikande exemplaren som hur förhållandet dem emellan borde se ut i anomalier och brott mot naturens ordning, olika situationer. Problemet med enkla vet- ser man numera variationen som en förut- enskapliga modeller är dock att det kan vara sättning för livets utveckling. Uppkomsten svårt att bedöma deras tillförlitlighet, efter- av nya arter förklaras allmänt med att växt- som även dåliga modeller kan ge resultat och djurbestånd delas upp i mindre grupper, som ser ut att vara vettiga, fastän de är varpå urvalet och slumpeffekter får dessa att missvisande. Inom populationsgenetiken är bli mer och mer olika varandra. I processen det t ex vanligt att man utgår från att olika omvandlas genetiskt betingad variation inom krafter är i jämvikt med varandra, så att man grupper till variation mellan grupper och – i kan bortse från svårkontrollerade historiska förlängningen – mellan arter. skeden. Samtidigt blir ekvationerna ofta lätt- Att förstå de processer som bestämmer are att lösa. fördelningen av genetisk variation i tid och När man så beskriver hur utbyte av rum är alltså grundläggande för evolutions- individer mellan populationer påverkar deras biologin. Här bidrar populationsgenetiska populationsgenetiska struktur, är det vanligt studier med en väsentlig del, nämligen med att man tillgriper modeller som bygger på en förståelsen av de fenomen som styr varia- jämvikt mellan de krafter som gör populationen hos specifika gener inom populationerna genetiskt olika och de som håller tioner. Man studerar ofta noggrant utvalda ihop dem. Det är framför allt förändringar i gener som fungerar som “markörer” för olika arvsanlagen, d v s mutationer, och slump- egenskaper. De flesta markörer kan även an- mässig förlust av varianter – så kallad gene- vändas till att indikera släktskap mellan tisk drift – som leder till genetisk olikhet, individer, vilket är en av populationsgene- medan likhet främjas av spridning mellan tikens viktigaste tillämpningar. Teorin kring populationer. Vidare kan naturligt urval släktskapsmarkörer handlar i grunden om verka åtskiljande eller sammanhållande, be- enkel folkräkning, eller demografi med ett roende på hur miljöfaktorerna varierar annat ord. Det kan gälla att t ex ta reda på mellan populationerna. Populationer kan hur många individer det finns, hur bra de är alltså vara lika varandra antingen för att de på att föröka sig och hur de rör sig i land- regelbundet utbyter individer, eller för att de skapet, kunskaper som bl a är eftertraktade utbytt individer tills helt nyligen, där inom naturvårdsarbetet. Dessutom hjälper “nyligen” i princip kan betyda allt från förra sådana insikter ekologer att bättre förstå de fältsäsongen till tiden för senaste inlandsisen, processer som ligger bakom arters utbred- beroende på evolutionsdynamiken hos den ning och samverkan med andra arter. De gen man studerar. Med en förenklad tolk- senaste årens snabba utveckling inom DNA- ningsmodell finns det alltså en viss risk för tekniken har gett upphov till vetenskaps- att slutsatsen man drar beror mer på hur

1 modellen ser ut än på hur verkligheten är könsumgänge. Kloning innebär att av- beskaffad. komman blir genetiskt identisk med för- Enkla modeller har dock en viktig äldern, alltså att alla gener nedärvs till- fördel framför de mer verklighetstrogna sammans och fullständig koppling uppstår. alternativen, nämligen att de ger mer Hinnkräftan Daphnia pulex (se Figur 1 allmängiltiga insikter. Det gäller att göra i avhandlingen) lever i vattenpölar och modellen tillräckligt enkel för att förstå, dammar, där den kan bilda stora lokala be- samtidigt som den bör vara tillräckligt stånd på kort tid. Så länge miljön är gynnsam komplicerad för att i allt väsentligt återspegla sker förökningen genom kloning, men när verkligheten. Men valet av modell beror hösten kommer, eller om t ex pölen är på väg också på om målet är att rätt beskriva ett att torka ut eller har blivit överbefolkad, sker särskilt system eller om det snarare gäller att könlig förökning och viloägg produceras. få mer övergripande förståelse. Dessa kan överleva tills miljön blir bättre I min avhandling undersöker jag hur igen, eller kan spridas med vind, vatten eller faktorer utanför de rent genetiska process- landdjur till andra pölar. Man tror att den erna kan påverka fördelningen av genetisk könliga förökningen, som ju medför en variation i tid och rum. Tonvikten ligger på blandning av gener från olika individer, leder jämviktsantaganden och effekter av av- till bredare överlevnadsmöjligheter hos avvikelser från dessa. Förutom historiska komman, som ju inte kan styra mycket över tillfälligheter studeras påverkan av det naturliga urvalet. Detta kan tyckas vara ett vilken miljö den kommer att kläckas i. Kön- självklart problemområde för evolutions- lig förökning är också viktigt för att förhind- biologiska studier, men så är inte fallet. ra slumpmässig förlust av viktig genetisk Sambandet mellan genetik och variation. demografi studeras enklast om man under- Den snabba tillväxten hos popula- söker variationen i genetiska markörer som tioner av Daphnia och andra liknande arter inte påverkas av urval, så kallad neutral innebär att pölar många gånger bebos av variation. I och för sig är det nog så att alla klonala avkomlingar från ett fåtal individer. gener påverkas mer eller mindre av urval, Som följd av detta finns det begränsat med men vissa gör det i så låg grad att man med genetisk variation, trots att individantalet är trygghet kan bortse från det. Orsaken är att stort. Det här har troligen samband med att gener nedärvs mer eller mindre oberoende av bestånden en gång grundats av ytterst få in- varandra genom den könliga förökningens divider. Genom sådana “flaskhalseffekter” mekanismer, så att krafter som verkar på t ex kan det uppstå stora genetiska skillnader genetiska cancerfaktorer inte nödvändigtvis mellan pölar, även om de ligger nära varand- påverkar spridningen av gener för exempel- ra, och detta trots att djurens spridningsför- vis hårfärg. När gener nedärvs helt obero- måga troligen är god. Eftersom populationer- ende av varandra säger man att det råder na är väldigt stora tar det sedan lång tid in- kopplingsjämvikt. Avvikelser från kopplings- nan jämvikt inträder mellan effekterna av jämvikt kan leda till att mönstret hos den spridning och genetisk drift. neutrala variationen blir förvridet så att den I artikel I härleds dessa samband demografiska informationen blir missvis- matematiskt i en förenklad modell för gen- ande. Då gäller det dels att upptäcka detta, etisk struktur, där alla populationer antas dels att förstå hur informationen bäst ska utväxla individer i samma utsträckning. Vi omtolkas. påvisar också att den icke-könliga fortplant- Min forskning syftar till att förstå hur ningen inte i sig påverkar utvecklingen populationsgenetiska jämvikter uppkommer nämnvärt, och inte heller parning mellan och bevaras, samt hur jämviktsavvikelser kan individer inom samma klon förväntas ha tolkas demografiskt. Jag har valt att studera någon större betydelse. Detta arbete visar att en djurgrupp där kopplingsjämvikt förhind- klassiska modeller kan användas på Daphnia ras genom att den mesta fortplantningen sker – d v s om variationen verkligen är neutral – genom så kallad “kloning”, d v s utan men att vissa ganska enkla justeringar be-

2 höver göras. Man måste t ex ta hänsyn till att varianter kan tillföras senare, genom invand- populationernas storlek varierar inom år, och ring från andra områden. att spridning oftast sker endast en gång per Anledningen till att de genetiska år, d v s under viloäggsfasen. skillnaderna är stora kan vara dels historiska Problemet med kopplingsjämvikt och slumpeffekter (se artikel I) och dels att naturligt urval tas upp i de två följande artik- kopplingen mellan gener hos Daphnia leder larna (II och III), där jag använder dator- till att neutrala markörer troligen ständigt simuleringar för att undersöka påverkan av påverkas av naturligt urval på andra gener. naturligt urval på variationen hos neutrala Man har visat att urvalet många gånger är gener. I det första fallet handlar det om en ny inkonsekvent över tid, vilket gör att frek- genetisk variant, eller mutantgen, som vensen av olika varianter ökar och minskar gynnas av urvalet och därigenom sprider sig med stor hastighet. Sådana snabba för- från population till population och till slut tar ändringar får lätt följder för den genetiska över hela systemet. De gener som är kopp- strukturen i stort, och det är lätt att varianter lade till en sådan lyckosam variant kan så att går förlorade genom slumpeffekter. Å andra säga “lifta” med denna under urvalsprocess- sidan har andra studier visat att viloäggen en, vilket kan orsaka viss förvirring hos den kan utgöra en reservoar för genetisk va- som studerar neutral variation. Om kopp- riation, och att kraftiga förändringar under lingen är stark kan samma variant lifta den aktiva fasen t o m kan underlätta be- genom alla populationer, vilket leder till varandet av variation med hjälp av en sådan genetisk homogenisering. Om kopplingen “fröbank”. däremot är svag så kan den inte lifta alls och Artikel III visar att så är fallet även ingen påverkan sker. Men någonstans där- hos Daphnia, men att detta inte förhindrar att emellan finns en möjlighet att olika neutrala populationerna blir genetiskt olika varandra varianter kan lifta i olika populationer, något när urvalet förändras hastigt. Fröbankar som som förstås kan innebära kraftig genetisk effektivt bevarar variation tycks göra så att differentiering. spridningen mellan populationer saknar be- I Daphnia kan sådan här påverkan bli tydelse för den genetiska strukturen. In- extra stark eftersom hela genuppsättningen är tressant nog inställer sig jämvikt mycket kopplad under den klonala fasen. Alla de snabbt i dessa system, vilket är bra från mo- gener man studerar kan alltså bli påverkade lekylärekologisk synvinkel. Å andra sidan av samma händelse. Förutom att bekräfta tvingas man till försiktighet i tolkningen av denna väntade effekt så visar jag att möjlig- den genetiska strukturen, eftersom de starka heten till ökad differentiering dessutom slumpeffekterna gör enskilda iakttagelser förstärks ytterligare av det samband mellan mycket osäkra. spridning och könlig förökning, som kommer av att viloäggen står för det mesta av sprid- Empiri ningen. Efter denna teorigenomgång följer nu en Trots detta så argumenterar jag för att studie av ett riktigt Daphnia-bestånd. Jag har liftande varianter oftast minskar genetiska valt ett område vid norra upplandskusten skillnader mellan populationer, eftersom som heter Ängskär, där Daphnia före- dessa skillnader oftast är stora av andra skäl kommer rikligt i hällkar längs med vatten- redan innan sådana här händelser inträffar. linjen. Därifrån hämtade jag, med hjälp av Endast gener som är extra starkt kopplade till Martin Lascoux m fl, exemplar av arten gynnade varianter kan förväntas genomgå Daphnia pulex under somrarna 1997-1999. ytterligare differentiering till följd av Dessa analyserade jag sedan med avseende liftnings-processer. Dessutom sker starkt på genetisk variation i fem neutrala släkt- gynnsamma mutationer tämligen sällan. skapsmarkörer. Om området vet vi att de Möjligen kan man tänka sig att sådana här nuvarande populationerna inte kan vara äldre effekter uppstår oftare i unga populationer, än mellan 500 och 1000 år, eftersom land- eftersom det är troligt att de första kolonisa- höjningen inte frilade klipporna förrän då. törerna inte är perfekt anpassade till den lo- Vidare vet vi att pölarna ligger mycket nära kala miljön. Då är det möjligt att gynnade varandra, och i vissa fall tidvis kan ha dir-

3 ektkontakt med varandra. Vi har alltså ett tydlig rumslig struktur som såg ut att åter- system som kan tänkas uppvisa historiska spegla det geografiska avståndet mellan effekter, men där det är möjligt att det sker pölarna. Detta är, även det, en aning para- mycket spridning av Daphnia mellan pölar. doxalt, eftersom man borde förvänta sig att Att där dessutom finns en glasskiosk gör inte populationerna i så fall skulle vara mer lika saken sämre, även om det tyvärr medför en varandra genomsnittligt räknat. Vi bestämde viss antropogen (mänsklig) störning av de oss för att försöka ta reda på om, och hur, det processer vi är intresserade av.1 naturliga urvalet kan ha påverkat mönstret För att förstå hur naturligt urval kan och eventuellt förstärkt skillnaderna. påverka neutrala gener behövde vi veta något Med vetskap om att det finns variation om den genetiska dynamiken hos egenskaper för urvalet att verka på kan man ställa upp ett som utsätts för starkt urval. Detta studeras i antal hypoteser för hur den genetiska struk- artikel IV, medan artikel V undersöker den turen hos neutrala gener kan påverkas av genetiska strukturen hos neutrala markörer. detta. Vi hade två huvudhypoteser: för det Vi bestämde den genetiska variationen för ett första lokal anpassning, som kan leda till att antal egenskaper som är viktiga för över- individer som nyligen spridit sig mellan levnad och fortplantning, som t ex avväg- pölar klarar sig sämre än de infödda. Detta ningen mellan satsning på könlös och könlig förväntas ge genetiska skillnader som sam- förökning under hård konkurrens om föda, varierar med miljöskillnader. För det andra och hur denna variation fördelade sig inom kan starkt urval leda till divergens hos och mellan populationer. Det visade sig att markörerna p g a slumpartade kopplings- det fanns påtaglig variation, detta trots att effekter. Denna hypotes förutsäger att populationerna hade litet genetisk variation i genetisk skillnad samvarierar med hur ex- neutrala markörer. En stor del av variationen trema miljöförhållandena är, alternativt hur i fortplantningsegenskaper fanns inom popu- mycket miljön förändras under året. lationer (d v s inom pölar). Det här antyder För att försöka välja mellan dessa att urval som verkar genom konkurrens- hypoteser jämförde vi hur miljön och den förmågan bör kunna leda till genetiska genetiska strukturen förändrades inom och förändringar, inom och möjligen mellan mellan år. Allt tydde på att den genetiska populationer. strukturen kunde förutsägas utifrån vetskap Det paradoxala i att det fanns stor om de miljömässiga skillnaderna, något som variation för egenskaperna, trots att de är alltså tyder på att lokal anpassning har min- viktiga för överlevnad och fortplantning, kan skat överlevnaden för de individer som eventuellt förklaras med att miljön förändras flyttat sig mellan pölar. Intressant nog tyck- så ofta och så mycket att urvalet får minskad tes det finnas skillnader mellan påverkan på effekt. Daphnia-populationerna var mycket den könliga fasens avkomma jämfört med instabila, vilket tyder på att såväl mängden den könlösa fasen. Ett generellt olikhets- konkurrenter som födotillgången varierar index, som innefattade ett stort antal miljö- kraftigt inom året. Samtidigt varierade variabler, samvarierade med den förstnämn- många andra miljöfaktorer kraftigt, t ex da, medan den sistnämnda samvarierade med salthalt, pH och temperatur. ett fåtal faktorer, bl a parasitförekomst. Man De neutrala markörerna visade tecken kan eventuellt förklara detta med att de kön- på påverkan av historiska slumpeffekter i ligt producerade viloäggen härstammar från området som helhet. Samtidigt fanns det fler än en säsong, samtidigt som vissa av tecken på att skillnaderna mellan pölarna miljöfaktorerna varierar mellan år. Man kan närmar sig en jämviktsnivå. De genetiska då kanske förvänta sig att ett generellt index skillnaderna var stora, men det fanns en ska vara mer koncist mellan år, jämfört med enstaka variabler. 1 En Daphnia-population dog ut under studie- perioden, troligen till följd av att dess vattenpöl Slutsatser plötsligt hyste småfisk. Även en viss antropogen Kopplingen mellan demografi och genetik störning på forskaren personligen kunde noteras, innehåller en betydande historisk kompo- framför allt orsakad av juvenila individer. nent, som kan förstärkas ytterligare om de

4 första grundarna av växt- och djurbestånd har av kontroverserna kring tolkningen av den en förmåga att snabbt monopolisera livs- moderna människans tidiga historia. Forsk- utrymmet. Det finns dock metoder för att ningen kring den genetiska strukturen hos upptäcka sådana historiska effekter, så att Daphnia ger ett tydligt exempel på att det är man ändå kan göra vettiga tolkningar av den riskabelt att dra slutsatser utan att ha goda genetiska strukturen. Samma sak gäller i historiska och ekologiska insikter om det princip för påverkan mellan gener och effek- studerade systemet. Starka urvalseffekter kan ter av naturligt urval på neutrala markörer. i vissa fall, särskilt när variationen bevaras i Ett problem är dock att historiska demograf- fröbankar, göra att spridningen mellan popu- iska händelser kan lämna likadana spår som lationer saknar betydelse för den genetiska urvalet. En neutral genvariant som liftar med strukturen. Fröbankar verkar stabiliserande en annan gen utplånar t ex genetisk variation på förändringar mellan säsonger, men inom på ungefär samma sätt som om populationen en och samma säsong kan urval och slump- hade genomgått en snabb tillväxt. Att jäm- effekter ha stor betydelse. För att bättre för- föra mönstren för olika markörer kan vara ett stå evolutionsdynamiken hos Daphnia är det sätt att komma åt det problemet. För Daph- troligen nödvändigt att studera eventuella nia och andra arter med en stor andel könlös fröbankar noggrannare. Viktig historisk kun- förökning tillkommer komplikationen att skap finns att hämta där, och genom att stora delar av genuppsättningen påverkas på kläcka gamla ägg på laboratorier kan man i samma gång, precis som vid populations- efterhand studera hur utvecklingen av olika tillväxt. egenskaper har gått till, samt hur detta på- Det framförs ofta kritik mot försöken verkat neutrala markörer. Här ligger en källa att dra långtgående slutsatser om historia och till kunskap som inte enbart är till glädje för demografi enbart med stöd av genetiska data. populationsgenetikens praktiska tillämpare, Att slutsatserna varierar beroende på val av utan också för alla de som är intresserade av markör och av analysmodell kan man t ex se livets utveckling i stort.

5 6 Introduction

IT IS common wisdom that Charles Darwin however, we need to know something about revolutionised biology by introducing the the forces behind the variation, and how va- idea of natural selection. Although certainly riation has been quantified. correct, this notion neglects the importance of Darwin's most fundamental, and in a way The origin of genetic structure more revolutionary premise: the recognition In population genetics, the concept of genetic of natural variation within species (e.g. Lev- variation refers to the variability present in ins & Lewontin 1985, Dennett 1995). While specific locations in the genome, and which earlier biologists tended to view variants as e.g. segregate during meiosis to form homo- anomalies in the scheme of nature, modern zygotes or heterozygotes at the loci in ques- evolutionary biologists virtually vacuum- tion. All genetic variation traces its origin to clean the Earth in search for deviating indiv- errors during DNA replication, or mutations, iduals. Individual variation is now viewed as which are generally considered to be random the very source of species variation. In turn, with respect to their expressed effects in or- evolution is seen as a process of accumul- ganisms. The mechanistic and probabilistic ation of changes in the genetic material, and laws that govern the segregation, spread, and the origin of species is explained by separ- ultimate fate of genetic variants, or alleles, ation of populations which adapt to different within populations, are well known in prin- conditions and undergo different stochastic ciple for one-locus systems (e.g. Crow & genetic events. Kimura 1970). In the absence of natural sel- Thus, understanding the distribution of ection, the rate of allele substitution in popul- genetic variation in space and time is a key ations equals the mutation rate, µ, per gene- aspect of evolutionary biology. At the same ration, but is independent of the population time, if correctly interpreted, measures of size, N (Kimura 1968). This is because the genetic variation can yield important insights substitution rate depends on two terms: the into past and present demography and ecol- new mutations, which occur at a rate 2Nµ, ogy of species. The recent explosion of gen- and the loss of most of them through the etic marker technology has fostered the dev- stochastic element of the reproductive pro- elopment of molecular ecology, which deals cesses, known as genetic drift, which occurs with the relationship between genetic and ec- at rate 1/[2N]. ological diversity. This subject is central for Since not all gametes contribute equal- many projects in nature conservation, e.g. ly to the next generation, the average related- when one wants information on the size of ness of individuals tends to increase over the mating population of a threatened speci- generations. This process may be referred to es, or on the exchange of breeding individ- as genealogical coalescence, and it means uals between populations. However, such that all alleles at a given locus trace their interpretations are based on models that origin to a common ancestor. If the popul- make many simplifying assumptions. Much ation is large the time to the most recent effort is being put into exploring the effects common ancestor will be long, while if the of lifting these assumptions, and finding out population size is small it will be short. In which assumptions are violated in specific the first case a large number of mutations populations. The present thesis contributes to will have had time to accumulate in popul- this literature, which will be reviewed briefly ations, whereas in the second case, the vari- in this first section. Before we can treat the ation will be more limited. Hence, two genes matter of interpretations of genetic variation, will be identical by descent if they trace back

7 to the same common ancestor, and if neither Two main classes of descriptors can has mutated. Note that two genes can be be identified as pair-wise measures and coal- identical by state but not by descent if there escent-based measures. Pair-wise measures are recurrent mutations. The rate of coales- quantify the average genetic similarity of ga- cence, or in other words the rate of genetic metes. In contrast, coalescent-based methods drift, defines the effective population size, Ne, describe the relatedness of whole samples by and the amount of standing variation within characterising the genealogy that is most populations is highly dependent on the pro- likely to relate them to each other. In both duct Neµ. cases, inferences about relatedness are drawn Populations that are isolated from each with the help of theory that relates identity other become genetically differentiated with by state with identity by descent, which re- time, through mutation and genetic drift, but quires knowledge about the mutation pro- migration between populations can retain si- cess. The conceptual relationship between milarity. These are the main determinants of the two has been clarified by Slatkin (1991). the spatial distribution of neutral genetic var- “Traditional” summary measures based on iation in populations, henceforth referred to pair-wise comparisons, such as heterozyg- as genetic structure, in one-locus systems. osity and pair-wise DNA sequence diver- When more than locus is of interest, the rate gence, ignore much of the information in of crossing over during meiosis, or more genetic data. Consequently, coalescent-based generally, the rate of recombination between methods may appear inherently superior to loci, determines their degree of statistical in- those. However, their advantage might not dependence, which is important when infer- always justify the computing time that they ences are drawn about the genome as a generally demand. The main reason is that whole from studies of a limited number of population genetic processes do not all occur loci (e.g. Weir 1996). Another potentially on the same time scale and taking into ac- important force is of course natural selection. count the deeper part of the genealogy may Genes under selection are not governed only not in some cases add much information by the probabilistic laws just outlined, but al- (Nordborg 2001, Rousset 2001). This will be so by the spatial and temporal distribution of the case if one is interested in inferring selective pressures, which may profoundly demography on a local geographic scale. affect their genetic structure. Furthermore, The most commonly used descriptors because of non-independence among loci, in studies of the impacts of migration and ge- the combined influence of selection and re- netic drift are, in one way or another, related combination can have important consequen- to Wright's fixation indices, also called F- ces also for the structure of selectively neut- statistics (Wright 1951). The F-statistics are ral genetic variation, as will be treated below simple functions of the probability of identi- and in later sections. ty of randomly sampled pairs of genes. For instance, let Q2 be the identity in state of two Measuring genetic structure genes picked in the same population and Q3 While the behaviour of one-locus systems in be the same probability for two genes picked randomly mating populations is well under- in different populations. Then the well- stood, some controversy remains in the des- known FST parameter of genetic differentia- cription of genetic structure. A first distinc- tion is defined as: tion can be made between model-based and purely descriptive statistics, such as spatial Q − Q autocorrelation of allele frequencies. Recent F = 2 3 . (1) ST − work has shown that autocorrelation can in- 1 Q3 deed be accounted for by population genetic models, facilitating the interpretation of gen- Estimation of F-statistics can be done within etic patterns especially in continuously dist- an analysis of variance framework, which ributed species (Hardy & Vekemans 1999). partitions the genetic variation into compon- However, only model-based descriptors will ents between gametes within individuals, be treated here. between individuals within populations, and

8 between populations within larger constel- tical by descent. By contrasting relatedness lations of species distributions, in an in prin- with information on geography, ecology, or ciple unlimited hierarchy (Cockerham 1969, historical species distributions, we should in Weir & Cockerham 1984). The structure principle be able to make inferences about within continuously distributed species can the forces responsible for present distribution be described with similar measures, by con- patterns and population sizes. In short, the trasting genetic identity against distance genetic structure carries knowledge of ecol- (Malécot 1975). ogical relevance. However, things are not Ratios of these variance components really this simple. First, we normally have quantify the genetic structure in measures data from a single point in time and there are such as FST and its equivalents (viz. the var- many random processes involved, so the iance between populations divided by total estimations are subject to considerable error. variance in group). These ratios gained wide- Secondly, the mutation process has to be un- spread use partly because they were consid- derstood if we are to infer identity by descent ered to be independent of the overall diver- from identity by state. Finally, the studied sity. However, this is strictly true only for bi- genes need to be neutral so that only demo- allelic loci, and the recent advent of DNA graphy and mutation have to be considered. sequencing and highly polymorphic genetic The genetic laws are much simpler and more markers has generated some confusion about universal than the action of natural selection, the meaning of FST-values. Since they do not which depends on many contingent factors specify the identity of alleles, these measures that are difficult, if not impossible, to recon- can seriously underestimate genetic differ- struct after the fact. It should be emphasised ences between populations that are highly that neutrality is a rather strong assumption, variable, but share few alleles (Wright 1978, as it can be argued that all nucleotides are in Hedrick 1999). This phenomenon causes some way influenced by selection (Gillespie particular concern when comparisons are 2000). made between loci or populations with dif- In passing, we should note that neutral ferent mutation rates or breeding systems variation can also function as a crude indi- (Charlesworth et al. 1997, Pamilo et al. cator of the amount of genetic variation for 1999). traits and genes under selection, although the As an alternative to relative measures exact relationship between the two is contro- of differentiation, absolute variance compon- versial (Hedrick & Miller 1992, Frankham ents can be used, but these suffer from sim- 1996). This issue is central in the field of ilar problems. Methods based on allele iden- conservation biology, where there is concern tity, such as Slatkin's (1985a) private allele about the evolutionary potential of threaten- method, coalescent-based methods, or as- ed species in the face of evolutionary change signment of individuals to populations based (e.g. Lynch & Lande 1993, Bürger & Lynch on their allele frequencies (Waser & Stro- 1995), or about the dangers related to in- beck 1998, Pritchard et al. 2000), make use breeding in small populations (Hedrick 1994, of the improved resolution offered by highly Lynch et al. 1995). As Ne decreases, the rate variable markers. Statistical development of of loss is increased, even for alleles that are these methods should therefore be encourag- being favoured by selection. Therefore, one ed in the future. might say that the accuracy of Ne as a general health indicator is inversely related Interpreting genetic structure to its estimated value. On the other hand, it According to the coalescent model just de- has been argued that populations small en- scribed, all genes in a given species are rel- ough to lose important beneficial alleles are ated through their genealogy, and the stan- even more acutely threatened by ding variation at neutral genes reflects the demographic stochasticity (Lande 1988). length of the genealogy and the mutation Yet, there is no need to impose a dichotomy process. If a given mutation has occurred on this issue. Genetic and demographic once and only once, then we may conclude threats should often interact in population about gametes carrying it that they are iden- extinction (Mills & Smouse 1994, Bürger &

9 Lynch 1995), thus both need to be taken genising and differentiating forces. The is- seriously. A large body of data suggests that land model of migration (Wright 1931, Ne is typically much smaller than the census 1951) is the simplest of these. It ignores also size of populations, which can be explained the geographic structure expected if dispersal by breeding system, population fluctuations, is limited in space, and assumes that all pop- or skewed mating success among individu- ulations in a system have the same rate of als, among other things (Frankham 1995). gene flow between them. The striking result This knowledge is important in the planning is that just a few migrating individuals can of conservation efforts. maintain similarity among populations to As explained above, isolated popul- such a degree that large samples are needed ations become differentiated at neutral loci to detect the differentiation. Equally striking, by the action of mutation and genetic drift. this result is independent of the local popul- Thus, if populations are genetically dissimil- ation sizes, since FST at equilibrium depends ar, this can be explained by isolation of their on the product Nem, where m is the rate of pools of breeding individuals, and only by migration between populations. Thus, the this. On the other hand, the reverse implica- model can be used to estimate the number of tion does not hold. Populations that are sim- migrants between populations. Recent work ilar can be so either because their breeding has shown that these general conclusions of pools overlap, or because they became isol- Wright's hold for systems with altered mat- ated from each other only recently. Alter- ing systems, and sexual biases in ploidy and natively, if the mutation mechanism allows the amount and timing of dispersal, if Ne and the creation of the same allele more than m are corrected for such deviations (Prout once, populations can be similar because 1981, Maruyama & Tachida 1992, Berg et they happen to have undergone the same al. 1998). Allowing mutations which create mutations. But if such congruent evolution new unique alleles alters the parameter of can be controlled for, the above statement differentiation into Ne(m+µ) (Latter 1973), suggests that inferences can be drawn about where µ is the mutation rate per gamete and historical events or about recent rates of ex- generation at a locus. In most relevant cases, change of breeders, or gene flow, among however, µ can be ignored because m will be populations, simply based on allele frequen- much larger. The equilibrium between local cy differences between populations. genetic drift and migration is attained at a The present thesis focuses on the link rate dependent on the larger of 1/m and 2Ne, between gene flow and dispersal, as revealed which means that strong differentiation re- by the genetic structure of populations. In quires a very long time to develop when pop- this line of research, historical effects are of- ulations are large, for a given number of mi- ten viewed as a nuisance that must be con- grants. trolled for. Conversely, current gene flow is Later work showed that a regular spa- a complicating factor in studies of the hist- tial structure in dispersal, described by the orical relationship between populations or stepping stone model, induces a geographic species. Both of these fields face the problem structure with regard to pair-wise FST-values of separating effects of history from those of between populations, in that nearby popula- current gene flow (Felsenstein 1982). We tions tend to be less differentiated from each shall return to this problem below, but first other than distant populations (Kimura & we need to consider in more detail the mod- Weiss 1964). This phenomenon is known as els that aspire to relate genetic differentiation isolation by distance. In fact, when limited to gene flow. The literature in this field has dispersal range is expected in nature, the de- been extensively reviewed by Slatkin gree of isolation by distance can be regarded (1985b, 1994) and others, and the following as an indicator of whether drift-migration- treatment is not intended to be complete. mutation equilibrium has been attained The first models of genetic differen- (Slatkin 1993). A similar phenomenon is ex- tiation between partly overlapping popul- pected in continuously distributed species, ations circumvented the problem of history where the concept of local populations has to by assuming an equilibrium between homo-

10 be abandoned. Here, the average relative ge- riers to dispersal in the landscape, leading to netic variation between individuals depends differentiated migration rates, or effects of only on the average dispersal distance and historical events such as recent population the population density, provided that disper- colonisations or expansions (Ibrahim et al. sal is equally likely in all directions and the 1996, Le Corre & Kremer 1998). The use of population distribution is uniform (Malécot several types of markers that differ in the 1975). rate of evolutionary change can help Few, if any, species are uniformly resolving such instances, but the success of distributed in space. On the other hand, many this approach appears to be mixed (Bossart species cannot easily be divided into local & Prowell 1998). Alternatively, methods that populations either. The direct link between search for the combination of population- dispersal between populations and FST at eq- specific parameters of dispersal or age that uilibrium hinges on the assumption of ran- maximise the likelihood of obtaining the data dom mating within populations. In other at hand can improve the understanding of words, populations should be in Hardy- particular study systems. The detailed struc- Weinberg equilibrium, meaning that geno- ture of populations far from migration-drift types occur at frequencies equal to the proba- equilibrium can be studied by analysing the bility of drawing their combination of alleles distribution of allele frequencies within pop- at random from the pools of gametes. Refine- ulations, since recent demographic disturb- ments of the spatial models to account for ances are expected to affect rare alleles dis- inbreeding in local populations (Maruyama proportionally (Maruyama & Fuerst 1985). & Tachida 1992, Tachida & Yoshimaru Here it is especially important to analyse 1996) has helped to clarify the interpretation, several independent loci, both because the but the problem often remains to define the stochasticity is enhanced in small popula- populations in the first place. The need for a tions and because selection on linked loci model of isolation by distance with arbitrary can make the pattern for particular marker distinctiveness of populations has only re- loci very similar to those caused by demo- cently begun to be fulfilled. Rousset (1997) graphic perturbations (e.g. Schlötterer et al. showed that the slope of a regression of 1997, see below). estimated FST/(1- FST) on geographic distance In the case of local colonisation ev- can be used to estimate the product of pop- ents, the theoretical understanding has im- ulation density and average dispersal dist- proved by assuming an equilibrium at the ance, independently of the scale at which global level in spite of local departures from samples are taken. However, knowledge of drift-migration-mutation equilibrium. This the scale of sampling relative to that of indi- theory draws on the ecological understand- vidual dispersal is necessary for an inter- ing of systems of populations, known as pretation in terms of the movement of indi- metapopulations, that persist through a bal- viduals, and this knowledge can be difficult ance between the number of local extinctions to obtain. and recolonisations (see e.g. Hanski 1999). In practice, pair-wise measures of There can exist a global genetic equilibrium genetic differentiation do hardly ever relate in such systems, where the average differen- perfectly linearly with distance, neither is tiation will most often be increased due to differentiation the same for all pairs of pop- stochasticity during local founder events ulations in systems with equal dispersal. (Wade & McCauley 1988, Whitlock & Rather, there is usually some scatter to the McCauley 1990). The system as a whole is data (e.g. Bossart & Prowell 1998). The also likely to lose genetic variation faster reasons for this are manifold, the most fun- than an equivalent system without local ex- damental one being the inherent stochasticity tinctions and recolonisations (Whitlock & of the genealogical processes themselves Barton 1997). An important feature in these (e.g. Nichols 1996). This uncertainty can be systems is the degree of correspondence be- ameliorated by analysing many independent- tween dispersal between extant populations, ly segregating loci. Other uncertainties are and dispersal that results in colonisation of more problematic, such as undetected bar- new habitat. For example, in plants there are

11 many cases where dispersal occurs mainly through pollen. Since pollen cannot colonise new habitats this leads to a large difference between number of migrants and number of colonisers. As a consequence, differentiation is enhanced by founder effects. When dispersal rates are similar in the two sexes there is instead a possibility for reduced differentiation, if colonisation entails more population mixing than migration does (Whitlock & McCauley 1990). Our final factor which may cause deviations from expected relationships between migration and differentiation is one that has only recently begun to receive consideration. It is increasingly being understood that natural selection may affect differentiation in un- expected ways, when loci under selection do not segregate independently from the neutral loci under study. Such associations are called linkage disequilibrium, a phenomenon that arises for instance when mutations are new, Figure 1. Adult specimen of Daphnia pulex, or when populations have recently been mix- carrying (at least) three parthenogenetic eggs ed after some time of separation. High rates (courtesy of Rowe & Hebert 1999). of crossing over during meiosis reduces linkage disequilibrium, but the phenomenon can remain for an appreciable amount of time Daphnia as a model organism even when loci are located on separate chro- Daphnia (Crustacea: Cladocera, see Figure mosomes. Essentially, two different classes 1) are freshwater invertebrates that exhibit of mechanism have been suggested for inter- several characteristics which may violate the ference between selection and the spatial assumptions of classical models of popula- structure of neutral genetic diversity. On the tion genetic structure. Perhaps most impor- one hand, local adaptation might reduce the tantly, asexual reproduction is common. The effective gene flow at neutral loci in large majority of populations of Daphnia are regions of genomes (Charlesworth et al. thought to reproduce by cyclical partheno- 1997). On the other hand, associations bet- genesis, a breeding system characterised by ween selected and neutral loci may affect the alternation of asexual and sexual reprod- effective population size in regions of close uction. In permanent habitats the asexual linkage, through genetic hitch-hiking phase may extend for several years, while in (Maynard Smith & Haigh 1974, Kaplan et al. intermittent habitats clonal lines typically 1989) or background selection (Charlesworth must engage in sexual reproduction at least et al. 1993). Such linkage effects may de- once every year in order to contribute to the crease or increase genetic differentiation, next generation of asexuals. During the asex- depending on whether local effective popul- ual phase populations have an impressive ation size is increased or decreased. In addi- capacity for growth. In contrast, the sexual tion, if selection determines the relative suc- phase is initiated by the production of males, cess of dispersers, linkage disequilibrium which subsequently fertilize diapausing may in some cases affect migration rates, eggs, called ephippia. These can remain thus complicating the prediction of its con- viable for several decades and consequently sequences for genetic differentiation (Pamilo form veritable egg banks (e.g. Cáceres et al. 1999, Schierup et al. 2000). 1998). Ephippia are also thought to be easily

12 dispersed between separate water bodies by Meester et al. 2002), so evolution caused by animals and by wind, in contrast to asexually selection may often occur at a rate comparab- produced progeny which have much smaller le or faster than that caused by stochastic de- prospects for migration between water bodi- mographic effects. To further complicate es (cf. Boileau & Hebert 1991). This alter- things, selection during the asexual phase is ation between the active and passive stage, expected to affect all loci similarly, which together with reliance on passive dispersal of makes it more difficult to discern effects of ephippia, creates rather special opportunities demographic history from those of selection. for genetic differentiation, which can be Thus, these species are particularly interest- greatly affected by founder events (Boileau ing for developing and testing theory on the et al. 1992, Boileau & Taylor 1994) and effects of various types of genetic disequi- long-distance migration (Weider & Hobæk libria on population genetic structure. 1997, Weider et al. 1999). In addition, prolonged parthenogenetic reproduction may Aims and outline of the thesis considerably enhance linkage disequilibrium, If the study of cyclically parthenogenetic leading to associations between neutral gen- freshwater invertebrates is to enhance our etic markers and genes under selection that understanding of the processes behind pop- disturbs the pattern for neutral genes (Lynch ulation genetic structure, the existing theory & Spitze 1994, Vanoverbeke & De Meester needs to be modified to account for the pecu- 1997, Pálsson 2001). Prolonged asexuality liarities of their systems of breeding and dis- may also increase inbreeding by facilitating persal. The first part of this thesis aims to do mating within clones, which should enhance just this, both in the realm of strictly neutral relative genetic differentiation between pop- variation and in that of natural selection. In ulations (Lynch et al. 1999). the second part I aim to apply this know- In light of this, it is not surprising to ledge to a convenient natural system: cyclic- find that populations of Daphnia and other ally parthenogenetic Daphnia inhabiting a freshwater invertebrates often exhibit a recently emerged rockpool habitat which is patchy genetic structure, i.e. characterised by characterised by instability. The work will a lack of isolation by distance, as well as shed light on the interpretation of genetic departures from Hardy-Weinberg equilibri- structure, but since natural selection takes an um within populations (e.g. De Meester inherent part in it, there will also be conclu- 1996). The interpretation of these geograph- sions relevant for the general understanding ic genetic patterns is a particularly challeng- of evolution in spatially structured environ- ing test of theories relating genetic structure ments. Thus, although the motivation for this to history or gene flow, since the neutral pat- thesis lies in the practical applications of terns are likely to be further disturbed by sel- population genetics, I hope that the findings I ection. Several lines of evidence indicate that present will be considered as relevant to the Daphnia can evolve rapidly in response to general study of evolutionary biology as environmental challenges (reviewed by De well.

13 Theoretical studies

ONE MAJOR limitation to the use of classical may easily be further strengthened by comp- models for inferring gene flow is their re- etitive exclusion of secondary colonists (De liance upon an equilibrium between homo- Meester 1996, De Meester et al. 2002), while genising and differentiating forces. The reali- the structure of more mature systems may sation that natural systems are often far from often be determined by random associations this equilibrium inspired the development of between genetic markers and loci under sel- more temporally realistic models, such as ection (Lynch & Spitze 1994, Vanoverbeke metapopulation models (Whitlock & & De Meester 1997). However, in no case McCauley 1990) and range expansion mod- has the null model of neutrality and drift- els (Ibrahim et al. 1996, Le Corre & Kremer mutation-migration equilibrium been explic- 1998). The latter of these may explain the itly stated. Needless to say, the most fruitful often patchy genetic structure found in speci- approach to these questions is detailed study es with long generation times, and in relativ- of the putative selective and historic factors ely young habitats, such as those influenced themselves. Yet, to the extent that departure by pleistocene glaciation. The genetic struc- from the null model is being taken as indica- ture can for a long time be influenced by init- tion of the action of this or that factor, it is ial founder effects when, e.g. as in many tree instructive to investigate the properties of species, the initial founders are able to ex- this null model, before going any further into clude secondary colonisers, especially when investigation or speculation. Paper I is an dispersal occurs over long distances (e.g. Le attempt to clarify the neutral expectation of Corre et al. 1997). An intriguing variation to differentiation in cyclically parthenogenetic, this theme is found among pond-dwelling in- and other freshwater invertebrates. Some of vertebrates, whose strong capacity for popul- the issues regarding selection at linked loci ation growth at the same time facilitates col- are dealt with in papers II and III below. onisation by extremely few founders, and rapid attainment of large population size. Neutral differentiation with cyclical parthe- The logical consequence of this was pointed nogenesis (I) out by Boileau et al. (1992): if the typical Two potentially important demographic feat- number of colonisers is low (say, 1-10 indi- ures of pond-dwelling invertebrates are their viduals), and if there is an approximate cor- fluctuations in population size and their re- respondence between the number of colo- liance on resting eggs for dispersal (e.g. nisers and the number of migrants among Boileau & Hebert 1991). Paper I examines extant populations, founder effects might be the effect that this may have on strictly neut- appreciable for many thousands of genera- ral genetic variation in cyclically partheno- tions. This is because the rate of attainment genetic species, and whether there are any of equilibrium is proportional to the ratio of potential effects of the asexual mode of population size to number of migrants, and reproduction, e.g. due to mating within clon- this ratio of course increases as population al lines (clonal self-fertilization). The treat- size increases, provided that the number of ment adhered to most assumptions of migrants is constant. Wright's island model, apart from the speci- As seen in the previous section, the fic features of interest. One important excep- idea of long-lasting founder effects in pond- tion was that immigrants were regarded as dwelling invertebrates has been invoked in originating from an infinite number of popu- several cases. For cyclically parthenogenetic lations of equal size, rather than from one in- species such as Daphnia spp., founder effects finitely large population. This was called for

14 because the inbreeding coefficient, F, of im- Clonal selfing can be modelled by migrants during the asexual phase has an im- substituting F with Rousset's (1996) expres- pact on the recursions, and it would be unin- sion for the inbreeding coefficient under sel- formative to assume that these were com- fing, and solving for the equilibrium case. pletely non-inbred. Populations were assum- The result: ed to be monoecious, although cyclical parthenogenesis involves the production of 1 fˆ ≈ (4) males and females at the onset of sexual 0  s  reproduction. Monoecy is a good approxim- 1+ 4N ()m + µ 1−  e  2  ation insofar as sex ratios of the final clutch do not differ substantially from one. In case of clonal specialisation in sex ratio, this is equivalent to the previous treatment by should be accounted for by using the modi- Maruyama & Tachida (1992) for sexual pop- fied expressions from Berg et al. (1998). ulations. Equation 4 applies when individu- Letting c signify the number of asexuals mate within clones more often than ex- al generations during cycles the recursion for pected by chance, and all mating occurs sim- the kinship coefficient, f, at the start of a cyc- ultaneously at the end of the cycle. However, le t becomes as follows if we ignore the we should also consider another putative higher order terms of 1/N: cause of clonal selfing in Daphnia, i.e. clonal specialisation in the timing of sexual reproduction. The selfing probability in each bout c ()c−i+1 of sexual reproduction is then given by the a ∏b j + ()c+1 c−1 inverse of the number of sexually active a a Ft−1 j=i+1 f ≈ + ∑ clones. But since each such bout must be 0,t 2N 2N c i=0 i weighted by its relative contribution to the c ()c+1 pool of sexual resting eggs, this number a ∏b F − − j t 1 cancels out, and the overall selfing probabili- c 1 j=i+1 + ∑ (2) ty is given by the number of sexual bouts di- i=0 2Ni vided by the total number of ephippia pro- c  c  duced. In practically all conceivable cases, + ()c+1  − 1  a ∏bi 1 ∑  f0,t−1 this ratio will be very small. i=0  i=0 Ni  Thus, under the assumptions of large N, small m and µ, and moderate c, asexual where phases and fluctuations in demographic para- a = (1-µ)2 meters do not affect the classical results for 2 bi = (1-mi) . neutral genes. However, care needs to be taken of the timing of genetic sampling. The If we also neglect the higher order terms of equilibrium result is only valid at the start of m and µ, and assume moderate c, we can cycles, and late sampling will show strong solve this for the equilibrium expectation: deviations from this expectation when demographic parameters fluctuate substantially during cycles. Caution is needed also concer- ˆ ≈ 1 f0 , (3) ning the possible violation of the assumption 1 + 4N ()m + µ e of moderate c. The model is constructed to apply to intermittent populations, and these where migration rate is averaged over the will rarely violate this assumption. However, cycle and the effective size is given by the cycles extending over several years most cer- harmonic mean over the cycle, as expected. tainly will. In the presence of migrating ˆ Note that f0 is equivalent to FST if the num- asexual offspring this would lead to a higher ber of local populations is large, because rate of increase of f than predicted from then Q3 is negligible compared to Q2 (see Equation 2, because such migration would Equation 1). not reduce the inbreeding coefficient during

15 of hundreds, in which case we may dismiss 0.4 genetic drift as a relevant agent, except for 0.35 bottleneck effects in very young populations. Population fluctuations, on the other 0.3 hand, are well documented, but may easily 0.25 be confounded by continued hatching during f 0.2 the asexual phase. Yet, if we stick to the as- 0.15 sumption of synchronous hatching, which is 0.1 conservative with regard to the question of elevated differentiation, and allow for the 0.05 possibility of massive migration (>>10 mig- 0 rants per cycle), it is clear that strong neutral 1 1.5 2 2.5 3 subdivision at equilibrium requires repeated R severe bottlenecks, high levels of selfing, or both. Even though drastic reductions in population size are often reported, especially in Figure 2. Kinship coefficient, f, at migration- mutation-drift equilibrium, as a function of popula- small water bodies (Korpelainen 1984, 1986, tion growth rate R within seasons, and length of Bengtsson 1988), these population crashes asexual season c, in an island model with cyclical are associated with sexual reproduction, parthenogenesis. Number of migrating eggs N0mc which rescues the genetic variation and saves = 5. Within-season migration rate, mutation, and clonal selfing rate are set equal to zero. ——— : it for future seasons. At the same time, high length of the asexual phase c = 1, — — — : c = 5, levels of clonal selfing due to temporal sub------: c = 10. structuring is not supported by the evidence from natural populations of Daphnia, because the production of males and ephippia does not appear to occur simultaneously cycles. We should also note that long cycles (Ferrari & Hebert 1982, Hobæk & Larsson would potentially raise the impact of muta- 1990, Kleiven et al. 1992, Yampolsky 1992, tion over migration, and especially for fast De Meester & Vanoverbeke 1999). And mutating loci the interpretation of differenti- even if it does, this is not expected to have a ation in terms of gene flow might be con- large impact, as shown here. Yet, we should founded by mutation. point out that also less extreme inbreeding The results demonstrate that strong has a similar effect, and e.g. spatial substruc- differentiation in cyclically parthenogenetic turing within populations could easily inflate (as well as strictly sexual) pond-dwelling in- genetic differentiation between populations. vertebrates can indeed be compatible with the null model of neutral genes at equilibrium, even at migration rates on the order of Selection and differentiation in Daphnia (II, magnitude assumed by Boileau et al. (1992). III) At the same time, our understanding of the Having improved our understanding of the process of attainment of equilibrium is en- expectations for neutral genes, we now turn hanced by Equation 2. As exemplified in to effects of selection in cyclically partheno- Figure 2, the decisive parameters are the genetic species. I mentioned above that neut- number of migrating sexually produced eggs, ral genes may easily be affected by selection the population growth rate after hatching, on other loci in these species, because of the and the number of asexual generations lack of recombination during the asexual during cycles. Little is known about the first phase. Cases have been made in the literature two of these parameters in natural systems. for effects of background selection (Pálsson The evidence regarding dispersal ability is 2001), genetic hitch-hiking (Lynch & Spitze non-conclusive (Jenkins & Underwood 1994, Vanoverbeke & De Meester 1997), 1998), and we lack direct estimates of dis- and rapid local adaptation which excludes persal between water bodies. It may well be immigrants, thus affecting gene flow bet- that the relevant number is often of the order ween populations (De Meester et al. 2002).

16 The idea of hitch-hiking is intriguing but has ferent populations and thus increased differ- received very little formal treatment in the entiation. An approximate treatment of the case of cyclical parthenogenesis. The most haploid and sexual case was given by Slatkin extreme form of hitch-hiking occurs when a & Wiehe (1998). Interestingly, they showed favoured mutation increases in frequency that there may be a window of opportunity from one or a few copies to fixation, erasing for differentiation with intermediate recomb- most of the neutral variation in regions of ination and migration rate. In cyclical par- low recombination rate (e.g. Schlötterer & thenogens that rely on sexually produced res- Wiehe 1999). A similar effect can be anti- ting eggs for dispersal, the coupling of sex cipated when favoured alleles are introduced with dispersal, in excluding the possibility of by immigration, which may be quite com- migration before recombination, reduces the mon e.g. in recently established populations. likelihood of spreading the initial associa- In organisms like Daphnia it could have es- tion. Thus the potential for increased differ- pecially drastic effects, since in these the entiation is enhanced compared to the all- whole genome might be affected by selection sexual case. However, this does not neces- on just a few loci, which increases the rate at sarily mean that differentiation will be en- which sweeps may affect a given locus. hanced regardless of the genetic structure While the process of reduced neutral prior to the sweep. For example, if popula- variation within populations during selective tions are already differentiated due to drift or sweeps is well understood (Maynard Smith selection, the same sweep that would in- & Haigh 1974, Kaplan et al. 1989), the effect crease differentiation if occurring in genetic- on genetic differentiation among intercon- ally identical populations may instead make nected populations is less clear. Clearly, if populations more similar. loci are tightly linked one should expect The role of selective sweeps in cyclic- sweeps to erase variation between as well as ally parthenogenetic species is investigated within populations. However, if there is a in paper II. A simulation model with two chance that the initial association between populations was created, which adhered to the neutral and selected locus is broken in the assumptions of paper I. Differentiation at migrating individuals, the result may be the neutral locus was quantified by hitch-hiking of different neutral alleles in dif- GST=[HT−HS]/HT (Nei 1987). At a second

0.8

0.6

G ST 0.4

0.2

0 0.015 0.03 0.06 0.125 0.25 0.5 1

M e

Figure 3. G ST at a neutral locus before and after introduction of a favoured allele at selected locus at migration-drift equilibrium, for different effective number of migrants M e. Selection coefficient s = 0.25, dominance coefficient h = 0.5, population size N = 3000. length of asexual phase c= 9, — — — c = 19, ◊: value at migration-drift equilibrium, ❑: value at the time of global fixation at selected locus.

17 habitats, where sexual reproduction most of- (a) 20 ten occur at least once every year. When the asexual season was long (c=19) the result of sweep events did not de- pend on the level of differentiation prior to the sweep. With shorter seasons the associ- Grel 10 ation between loci was not strong enough to reduce strong differentiation, although an effect was still seen. Differentiation was negatively related to the migration rate, as expected (Figure 3). Thus, if we believe that mig- 0 ration rate is seldom low in Daphnia, and in 0 0.1 0.2 0.3 0.4 0.5 light of the often strong differentiation found s which may be due to other effects selection, it is suggested that sweeps will rarely increase differentiation in Daphnia, but may 0.2 (b) often decrease it. The above results were obtained with 0 very strong selection (s = 0.25). For small s 0.1 0.2 0.3 0.4 0.5 there was a threshold effect, meaning that selection on the order of r/s = 1 did not affect -0.2 neutral loci, but that at least an order of mag- Grel nitude stronger selection was required (Fig- -0.4 ure 4a). However, this threshold was only apparent when the sweep occurred at Hardy- -0.6 Weinberg equilibrium. When introducing the favoured allele at migration-drift equilibri- -0.8 um, on the other hand, an appreciable effect ≈ s was seen already at r/s 1 (Figure 4b), which is the theoretically expected limit for Figure 4. Relative effect of selection strength, s, sweep effects. The lack of threshold effect on neutral differentiation, quantified as when the sweep occurred at equilibrium sug- Grel=[GST(fix)−GST(lost)]/GST(lost), at the time of gests that even mild selection and slow global fixation at selected locus, when effective sweeps may often reduce differentiation in –5 migration rate me = 2x10 , length of the asexual nature, provided that other selective forces phase c = 19, and other parameters take values do not interfere too much with the selective as in Figure 3. Favoured allele introduced a) at sweep. Reductions are especially likely fol- global Hardy-Weinberg equilibrium, b) at migration drift equilibrium. lowing population foundation, since differentiation is likely to be strong while the first colonisers may not carry the alleles which would be optimal in their new habitat, thus locus, initially fixed for allele A, a new mut- opening for subsequent sweeps of alleles ation, a, affecting viability was introduced in introduced by immigrants. The plausibility one copy. This locus showed an additive fit- of this conclusion depends on the ability of ness effect, so that AA individuals had fitness bottlenecked populations to adapt rapidly to 1 - s, whereas Aa and aa gave fitness 1 - 0.5s new habitats. Based on the finding that cyc- and 1, respectively. The two loci segregated lical parthenogenesis should enhance the rate freely at sexual reproduction, but recombina- of adaptive evolution because of efficient se- tion rate was restricted depending on the lection among multi-locus genotypes during number of asexual generations between the asexual phase (Lynch & Gabriel 1983), sexual bouts, according to the relationship De Meester et al. (2002) argued that local r=1/[2c]. The values of c were chosen so as adaptation should be expected to further con- to be relevant for populations in intermittent solidate monopolisation of new habitats by

18 the first founders. This process would streng- tion. This procedure makes the study pheno- then differentiation caused by founder menological rather than causal, insofar as effects, while the present results suggest that conclusions are drawn about neutral hitch- initial differentiation may easily be erased if hiking loci in e.g. Daphnia. Yet, considering selection pressures are consistent over time the current lack of detailed knowledge about and space. Further investigations into these the genes involved, it can be deemed less matters, empirical as well as conceptual, will speculative to study the apparent pattern, be welcome. than to build models based on insufficient premises. Efforts were made to make the Rapid shifts in genotype frequencies selective force similar to the pattern observ- have been known in Daphnia and other cyc- ed by Lynch (1987): values for the intrinsic lically parthenogenetic organisms since long, rate of increase, r, of each genotype were and seem to be the rule especially in permanent habitats with extended periods of 0.5 asexuality (Hebert 1987). Such fluctuations were interpreted in terms of selection by 0.4 Lynch (1987), in a survey of available long- term studies. He found that selection pres- 0.3 H sures appeared to fluctuate erratically over 0 time, but were on average close to zero. If 0.2 fluctuations are uncorrelated between popul- (a) ations, he suggested, they may produce dras- 0.1 tically enhanced differentiation, ultimately 0 causing fixation of different alleles in different populations and, one has to conclude, 0 600 1200 1800 2400 long-lasting differentiation. cycle Low spatial autocorrelation is expected if the respective habitats are dissimilar, or if selection is acting on other loci than those studied, meaning that the latter are merely 0.5 hitch-hiking due to random associations created during asexual reproduction in each 0.4 population (Carvalho 1988, Lynch & Spitze 1994, Vanoverbeke & De Meester 1997). On 0.3 the other hand, the presence of large egg H 0 banks in these species is expected to counter- 0.2 act fixations efficiently, especially when se- (b) lective forces differ between seasons (Ellner 0.1 & Hairston 1994, Hedrick 1995). Thus, a useful description of the process should al- 0 low for the genetic structure to be continual- 0 600 1200 1800 2400 ly created and upheld, rather than having cycle been created by random fixations at some time in the past. In paper III I investigate the con- Figure 5. Observed heterozygosity at hatching of sequences of spatially uncorrelated fluctu- sexually produced eggs, as a function of time ations in selection pressures on genetic dif- since global Hardy-Weinberg equilibrium, when ferentiation in systems with two populations, the realised variance in selection coefficients and the role of egg banks as a temporal stabi- within cycles Var(S) ≈ 0.2, and the length of the liser of the genetic structure. The study was asexual phase c = 9. Egg bank contribution to active population (%): : 0, — — — : 10, based on the simulation model from paper II, ------: 20, — - — - — : 30, — - - — - - : 90. a) with the main difference that genetic struct- deep bank (d = 1.1), number of migrants M = 10, ure was followed at the locus under selec- b) shallow bank (d = 2), M = 10.

19 sampled from a normal distribution with fluctuations in selection. At the beginning of mean = 0 and variance σ 2 , and the selection each growing season, the hatching of eggs r from the bank resets the scene for selection coefficient, S, was calculated as the differ- to act upon, and as long as all genotypes are ence in r between a genotype and that of all present in higher than negligible numbers the other genotypes considered as a group. Valu- recruitment to the egg bank is mainly deter- σ 2 es of r were chosen so as to yield varian- mined by selection. Immigration may of ces in S over time which were comparable course add to the stability early in seasons, with the ones observed by Lynch (1987). but its effect is unimportant when the egg The results conformed to previous bank is of sufficient magnitude to maintain studies, in that fluctuating selection main- variation by itself. Still, high migration rates tained genetic variation indefinitely in the may of course affect the structure in the presence of an egg bank of sufficient depth longer term, by facilitating the spread of new or magnitude (Figure 5). This happens be- alleles, but this is a slightly different story. cause the random action of selection causes Paper III thus demonstrates that ap- recruitment of resting eggs laid down in the egg bank to vary between cycles, so that the 0.6 (a) bank always harbours all genetic variants. The effect is similar to what is seen when ge- 0.5 nerations overlap and selection is acting dif- 0.4 ferently on different age groups. The flip side of this observation is that the egg bank G ST 0.3 appeared necessary for preventing rapid fixation in the presence of fluctuating selection 0.2 of the magnitude observed by Lynch (1987). 0.1 This suggests that egg banks are crucial for the understanding of genetic diversity in the 0 populations and species that exhibit them. 0306090 Mirroring the stability in genetic vari- bank contribution (%) ation within populations, average differentiation over replicates was remarkably stable over time, if measured at time intervals cor- 0.6 (b) responding to the length of the cycle (e.g. after sexual reproduction). On the other 0.5 hand, differentiation varied considerably within cycles in the presence of egg banks, 0.4 typically being lower early in the season than 0.3 late. There was also considerable variation G ST over replicates of diversity measures, especi- 0.2 ally early in seasons. This inherent variance means that a large number of independently 0.1 segregating markers may be required in empirical studies of the phenomenon. 0 Perhaps the most striking finding was 0 306090 that differentiation appeared to be independ- bank contribution (%) ent of migration rates when the egg bank was deep or large enough to maintain variation Figure 6. GST after sexual reproduction at indefinitely (Figure 6). This can be attributed differentiation equilibrium, as a function of the to the observation that, in each replicate pop- proportion of hatching eggs emerging from egg bank, when the realised variance in selection ulation, genotype frequencies late in the sea- ≈ son were essentially independent of early coefficients within cycles Var( S) 0.2. : number of migrants M = 10, — — — : M = 0.5, ◊: frequencies. Instead, they were determined length of asexual phase c = 9, : c = 19. a) deep by the stochastic outcome of cumulative bank (d = 1.1), b) shallow bank (d = 2).

20 parent fluctuating selection on neutral mar- rary to the apparent fluctuating selection due kers can lead to the sort of migration-inde- to random hitch-hiking, local adaptation is pendent differentiation patterns that are sug- reduced by small population size. Thus, the gested in many cyclically parthenogenetic two phenomena may be expected to occur in species. To the extent that the phenomenon is different natural systems. Finally, we should caused by random associations between mar- note that clonal successions need not be in- kers and genes under selection, it will occur terpreted in terms of selection among compe- mainly in small populations, where such ting genotypes. Alternatively, clones may be associations may easily develop (Hebert specialised for maximal performance during 1974a, b, Mort & Wolf 1985). The monopol- different phases of seasons (Carvalho 1988). isation hypothesis and the hypothesis of local Notably, subdivision of this sort may also adaptation (De Meester et al. 2002), men- enhance genetic differentiation between pop- tioned above, is not necessarily in conflict ulations, leading to observed inconsistency with the present hypothesis. However, cont- in gene flow estimates.

21 Empirical studies

IT SHOULD be clear from the above treatment number of sexual progeny produced. When that historical contingencies and natural sel- the habitat deteriorates, or competition for ection can have profound effects on patterns food increases, a slow sexual response en- of genetic structure for neutral genes within tails a greater risk of clonal death. On the and between populations. It should also be other hand, too quick a response may lead to clear that there are several reasons for ex- missed opportunities for asexual reproduc- pecting these forces to be especially strong in tion, and thus a decreased eventual contribu- cyclically parthenogenetic species. This is tion to the pool of ephippia. Therefore, varia- due to the occurrence of enhanced association between populations for this character is tions between loci, and to the vivid popula- interpreted in terms of local adaptation (e.g. tion dynamics, especially when coupled with Deng 1997), while variation within popula- reliance on sexually produced diapausing tions requires another type of explanation, eggs for dispersal. In the rest of this thesis I e.g. inconsistent selective pressures, geno- shall attempt to apply this knowledge in the type by environment interactions (Deng description of the genetic structure of a nat- 1996), or recurrent mildly deleterious muta- ural system which exhibits several of the tions (cf. Lynch et al. 1998). features covered by the theory just discussed: While most previous treatments have a system of Daphnia-populations inhabiting used deterministic arguments for the mainte- small and environmentally unstable pools in nance of variation in this important fitness an area which is young on an evolutionary character, paper IV investigates the matter time scale, and in which historical effects from the standpoint of stochastic effects. The may enter both because the whole system sexual response to population density and was founded recently, and because popula- competition for food was investigated in a tions may regularly be re-founded following system where stochastic effects may be im- recurrent extinctions. Paper IV examines the portant, because the populations are compar- variation in important fitness characters, ably young, small, and unstable. The study while in paper V the effects of fitness vari- system is a large number of populations of ation and history on neutral genetic structure Daphnia pulex inhabiting small rockpools are explored. which lie in close proximity to the uninhabit- able Baltic Sea, in Ängskär, about 100 km Geographic structure for fitness traits in north of Uppsala on the eastern coast of Daphnia pulex (IV) Sweden (Figure 7). The area was covered by In cyclically parthenogenetic species the sea water until approximately 500-1000 reproductive success of clonal lines is influ- years before present, when it emerged in the enced by a balance between asexual and land-uplift following the Pleistocene glacia- sexual reproduction. In Daphnia, the choice tion. Pools and ponds in Ängskär constitute between the two types of reproduction is de- relatively short-lived habitats, since they are termined by an environmental response, eventually overgrown in the succession pro- which is triggered by signs of habitat deter- cess. Thus, continued existence requires the ioration or crowding. The individual varia- colonisation of newly created habitat, a phe- tion for this response typically exhibits a ge- nomenon which causes freshwater species in netic component (e.g. Deng 1996). The sexu- the area as a whole to exhibit metapopulation al response is a central component to the fit- dynamics (cf. Sjögren-Gulve & Berg 1999). ness of clones, since the timing of the switch We know then that the system of populations to sexuality to a large extent determines the could be expected to exhibit departures from

22 migration-mutation-drift equilibrium, if drift ness characters were examined with analysis is indeed important in this system. At the of variance. Since all individuals from a same time, the pools lie in close proximity to clone are genetically identical the variance one another and some of them may tempor- among clones yield estimates of the genetic arily be inter-connected due to overflow, variance for studied traits. meaning that dispersal capabilities should be The results showed substantial genetic good. variation for the sexual response, both within Six pools were sampled in Ängskär and between populations in Ängskär (Table (Figure 7). Clonal lines were cultivated in the 1). In line with deterministic expectations, laboratory, and subsequently subjected to the variation between populations, as quanti- three levels of population density in an ex- fied by QST (Lynch & Spitze 1994), was not periment under controlled conditions. For larger within Ängskär than at the large scale. comparison, two populations from a larger There was thus no indication that the pattern distance (Uppsala, Sweden, and Crail, Scot- in Ängskär was especially affected by histo- land) were also sampled and included in the ry, although it is unclear how strong a signal experiment. Effects of clone, sample origin, would be required to exclude this possibility and population density on a number of fit- with certainty. Thus, even though Ängskär

Figure 7. Map of study area in Ängskär, Sweden. Dots indicate the locations of unsampled, and squares that of sampled rockpools. Pools indicated in bold letters were included in the study in paper IV, while underlined letters indicate pools subjected to detailed environmental study in paper V.

23 Table 1. Summary of a) GLM and b) ANOVA for reproductive traits of samples of six Daphnia pulex populations in Ängskär, Sweden, subjected to different population density and food competition treatments. Significance levels for Χ2-test (a) and F-test (b), after adjustment for unbalanced design: *: p<0.05, **: p<0.01, ***: p<0.001. —: factor/effect not treated. Trait nomenclature: eb: proportion of units producing ephippia, so: sex ratio in final clutch (males/total), eq: mean number of ephippia produced per female, Rc: asexual reproductive rate per female andday, ri: intrinsic growth rate of units, males included. a)

Effect\Factor eb so

df deviance df deviance pool 5 17.10** 5 195.99*** density 2 25.94*** 2 19.70*** density x pool 10 10.91 10 47.78*** residual 142 60.30 58 486.65

Covariates: adult sex ratio 1 2.85 — — adult mortality 1 0.0022 — —

eq Rc ri

df MS F MS F MS F pool 5 0.25 11.29*** 0.92 3.92** 0.030 2.06 clone in pool 24 0.15 6.44*** 0.53 2.29** 0.028 1.96* density 2 0.70 30.90*** 106.36 455.48*** 11.54 795.51*** density x pool 10 0.15 6.44*** 0.37 1.57 0.0096 0.66 density x clone in pool 43 0.078 3.47*** 0.42 1.81* 0.017 1.18 residual 124 0.022 — 0.23 — 0.015 —

Covariates: adult sex ratio 1 0.062 2.75 21.63 92.64*** 2.77 191.28*** adult mortality 1 0.0032 0.14 0.40 1.73 0.048 3.34

24 Table 2. Pearson product-moment correlations with previous work showing that the two between a) reproductive traits, and b) the traits types of responses may be genetically un- underlying intrinsic growth rate. Upper right coupled (Ferrari & Hebert, 1982; Hobæk & triangle: Ängskär data, lower left triangle: all data. Significance levels: *: p<0.05, **: p<0.01, ***: Larsson, 1990; Yampolsky, 1992; De Meest- p<0.001. See Table 1 for trait nomenclature. er & Vanoverbeke, 1999), thus promoting out-crossing and reducing the genetic dif- a) ferentiation for neutral genes (see paper I).

eq so rI Rc Geographic structure for neutral genes in eq — -0.22 -0.24** -0.29*** Daphnia pulex (V) so 0.11 — -0.07 -0.07 As hinted on above, the balance between

ri -0.31*** -0.08 — 0.90*** asexual, viviparous reproduction, and sexual production of diapausing offspring is central Rc -0.32*** -0.08 0.91*** — to the clonal composition in pelagic populations of Daphnia. The finding of genetic b) variation within and between populations for Rc Mmx these traits (IV) suggests the presence of

Rc — 0.16 -0.29*** dynamics in genetic diversity which is to some extent caused by competition for re- M -0.17** —0.03 sources. In the final paper (V), the dynamics mx -0.32*** 0.08 — in neutral genetic diversity is studied over a three year period in 17 populations in Äng- skär (Figure 7). Intraspecific competition, appears unstable, either bottlenecks are not however, is only one factor among many that severe enough to produce an enhanced gen- could exert a selective pressure. By contrast- etic structure, or there are strong forces re- ing measures of genetic diversity in five mic- storing variation in fitness traits, such as rosatellite markers against relevant environ- mutation or immigration. mental parameters and demographic data, we Production of ephippia was negatively attempted to draw inferences about putative correlated with intrinsic growth rate of clon- selective processes and historical effects. Of es (Table 2), but the significant differentia- particular interest, apart from effects of intra- tion for the former was not mirrored in a sig- specific competition, is the impact of the nificant effect of population on the latter microsporidian endoparasite Larssonia sp., (Table 1). Positive correlations between traits which has been shown to incur severe fec- related to clonal growth rate suggest that the undity reductions in D. pulex in Ängskär clonal variation for growth rate may partially (Bengtsson & Ebert 1998), and that of inter- reflect fitness effects which are less condi- specific competition with the congener D. tional on environment (cf. Lynch et al. longispina, which has been shown in other 1998). Recent work by Pálsson (2001) sug- studies to have large consequences on the gests that cyclical parthenogenesis can en- population dynamics of D. pulex (Bengtsson hance genetic loads of deleterious mutations, 1989, 1993). especially in small populations. If this ap- There are at least three hypotheses for plies to the Daphnia populations in Ängskär, how natural selection might affect the clonal it may decrease differentiation for neutral composition within populations, which in genes due to increased variation within pop- turn generate quite different predictions ulations (Pálsson 2001), or enhanced effect- about the genetic structure between popul- ive immigration thanks to heterosis effects in ations. Firstly, random genetic hitch-hiking, crosses between populations (Whitlock et al. caused by associations that develop during 1998, C Haag, J Hottinger, M Riek & D the asexual phases between selected loci and Ebert unpublished manuscript). neutral markers (Vanoverbeke & De Meester Finally, we note that production of 1997, II, III), may increase or decrease gen- males was not correlated with production of etic differentiation, depending on the identity ephippia (Table 2). This finding is in line of hitch-hiking alleles. Secondly, local adap-

25 tation may lead to exclusion of recent immi- sence of alleles that could only be explained grants (De Meester et al., 2002). This pro- by recent bottlenecks or rare instances of cess may be denoted “non-random hitch-hik- long-distance migration. In spite of this, ing”, because it is always predicted to reduce when samples were taken shortly after hatch- differentiation when present, as opposed to ing of sexually produced eggs in 1997, a sig- random hitch-hiking. Finally, associative nificant pattern of isolation by distance overdominance or heterosis in recently bott- among populations was detected with princi- lenecked populations may lead to an in- pal component analysis, assignment test, and crease in observed heterozygosity and en- Wright's FST (Figure 8). This indicates a con- hanced gene flow, and consequently to de- cordance between migration and gene flow creased differentiation during the season as predicted at migration-drift equilibrium (Pálsson 2001). A preliminary scheme for (Slatkin 1993, Rousset 1997). discriminating between these hypotheses is However, the interpretation of isola- given by the following predictions. If the sel- tion by distance is problematic, since it may ective pressure is proportional to the magni- be confounded by spatial autocorrelation of tude of environmental variables, as is the environmental characteristics, which was case with parasites or density of competitors, also found to be significant. Close correla- the hypothesis of random hitch-hiking pre- tion between pair-wise genetic and environ- dicts an association between pair-wise differ- mental distance as summarised by principal entiation and mean values of important en- component analysis suggests that local adap- vironmental variables. The local adaptation tation is as valid an interpretation as migra- hypothesis, on the other hand, predicts asso- tion-drift equilibrium (Table 3). Since the ciations between environmental and genetic pattern was observed before selection could distance. The third hypothesis, finally, pre- have had a strong effect in the asexual phase, dicts heterozygosity-fitness correlations (see the explanation must be sought either in cu- e.g. David 1998), and, albeit quite loosely, mulative effects of local adaptation over associations between differentiation and the several generations, or in differing hatch- size and age of populations. This final pre- ability between immigrants and residents. diction was not considered in the present Significant neutral genetic differences bet- study. ween diapausing and hatched rotifers Historical effects were suggested by (Gómez & Carvalho 2000) suggest that the the low genetic diversity compared to pre- genetic structure of cyclical parthenogens vious studies in a larger part of the Baltic may be affected by such differential hatch- area (Pálsson 2000), and a conspicuous ab- ability.

97-1 97-2

1 1

0.8 0.8

0.6 0.6

F ST 0.4 F ST 0.4 0.2 0.2

0 0 0100200 0100200 -0.2 -0.2 meters meters

Figure 8. Pairwise F ST as a function of distance in meters early (97 -1), and late (97-2) in the season o f 1997. 97-1: r2 = 0.152, b = 0.00147, p<0.01 (Mantel test), 97-2: r2 = 0.00051, b = 0.009.

26 Table 3. Analysis of association between genetic ferentiation overall. A similar pattern has and environmental distances. a) rank correlation been previously reported for D. magna in coefficients for pairwise differences in each variable contrasted against genetic distance between Central Europe, and was attributed to ran- populations, tested by Mantel test (Raymond & dom hitch-hiking (Vanoverbeke & De Rousset 1995), b) r2 from linear regression of gen- Meester 1997). However, in the present etic distance on pairwise environmental differ- study the evidence indicated local adapt- ence, for variables found significant with Mantel 1 ation. Although the summation of environ- test. Value obtained when controlling for effect of geographic distance. * p<0.05, ** p<0.01, *** mental differences was a poor predictor of p<0.001, no multiple test correction. differentiation late in the season, a number of its constituent variables now had a large ex- a) planatory power (Table 3). Notably, preval- 97-1 97-2 ence of Larssonia sp. was among the most powerful factors, indicating that populations of D. pulex were to some degree locally ad- PCA ordination 0.483* 0.113 apted to the parasite, and that immigrants depth 0.478* 0.698** from populations that lacked the parasite re- organic content 0.012 0.513* produced less than residents in the asexual cv salinity -0.114 0.583* phase due to higher susceptibility to parasite infections. This hypothesis remains to be ex- parasite prevalence 0.576* 0.608* perimentally tested. Local adaptation of Daphnia to particular strains of parasites geographic distance 0.425 0.065 have been demonstrated elsewhere (Ebert 1994), but does not seem to be ubiquitous (Ebert et al. 1998). However, adaptation to b) the presence or absence of parasites should 97-1 97-2 be more feasible. Observations of heterozygote excess PCA ordination 0.0301 — late in 1997 are compatible with the hypoth- esised effects of associative overdominance depth 0.188* 0.545*** or heterosis due to recent mixing of bottle- organic content — 0.264* necked populations. This is in contrast with cv salinity — 0.373** most studies on intermittent populations of parasite prevalence 0.1581 0.335**1 Daphnia, which typically exhibit heterozygote deficiencies (Hebert 1987). The instability and small size of rockpool populations is geographic distance 0.241* 0.009 expected to enhance the processes leading to heterozygote advantage, so the direction of the evidence in Ängskär is in line with this The genetic pattern in samples taken hypothesis. However, the evidence was late in the 1997 season was strikingly differ- statistically insecure, and was not strengthen- ent from that of the hatching populations in ed by studies relating genetic variation 1997-1999. There were several cases of Har- within individuals to fecundity. These dy-Weinberg and linkage disequilibrium, correlations were absent or, in one case, with indications of excess of heterozygotes negative. within populations. The clonal composition Rapid adaptation in cyclically parthen- of some populations showed significant ogenetic freshwater invertebrates has been over-representation of particular clones, reported for quantitative traits (Hairston et while overall clonal diversity had decreased. al. 1999, Cousyn et al. 2001). In a recent re- Reductions in diversity within populations view of the topic, De Meester et al. (2002) were associated with high differentiation bet- lay forward compelling arguments for this ween populations, leading to a drastically re- process as an important determinant of the duced isolation by distance pattern (Figure genetic structure of these species. To the ex- 8), and a tendency towards increased dif- tent that local adaptation can occur rapidly in

27 recently established, bottlenecked pop- tecture is often characterised by epistatic and ulations, young systems may show enhanced dominance variance (Lynch & Deng 1994), and randomised genetic structure, not only may indeed be well equipped for adapting to because of the founder effects (cf. Boileau et new habitats after colonisation. Alter- al. 1992, I), but more so because of rapid natively, newly established populations may monopolisation of the habitat by the first be prone to invasion by immigrant clones colonisers (De Meester 1996, Okamura & which happen to be better adapted to the Freeland 2001). As noted in the previous particular habitat, possibly reducing dif- section, this hypothesis hinges on the ability ferentiation drastically (II). The empirical of founders to maintain heritable variation in studies of this thesis (IV, V) lends cor- quantitative characters, which is expected to relational evidence to the idea that local be better than the maintenance of variation adaptation may be efficient also in young for single genes. The conversion of non- systems, and with comparably small additive and epistatic genetic variance into population sizes, thanks to either mainten- additive variance observed after bottlenecks ance of variation during bottlenecks, or to (e.g. Bryant & Meffert 1995), suggests that rapid recovery of variation through mutation cyclical parthenogens, whose genetic archi- (cf. Lynch et al. 1998).

28 Conclusions

THE LINK between migration and gene flow the controversies regarding the history of often contains a significant historical com- modern humans (e.g. Hey 1997, Reich & ponent. This may be particularly important Goldstein 1998, Harpending et al. 1998). As when the first colonisers of a habitat have the shown in the present thesis, we have a par- ability to quickly monopolise the resources. ticularly strong case against such simplified On the other hand, there are a number of treatments in freshwater invertebrates such methods for detecting such historical effects, as Daphnia. The presented work contributes although they may require large amounts of to the understanding of these systems. For data or computing time. Therefore, historical example, we have seen that strong selection, effects probably often go undetected, al- in conjunction with egg banks, can make the though they may be essential for the inter- genetic structure essentially independent of pretation of the genetic structure. Likewise, migration. Egg banks stabilise the genetic effects of selection on linked loci can be de- variation and mitigate evolutionary change tected, but one problem is that these may of- between seasons, but in Daphnia changes ten be similar to the ones caused by demo- can be rapid and profound within the graphic history. Comparisons among differ- asexual, active phase. Thorough analysis of ent loci can reveal whether disturbances are egg banks appear to be the single most similar throughout the genome or locus- important requirement for a sufficient specific, the latter implying selection rather understanding of the evolutionary dynamics than demography. With Daphnia and other in these species. Egg banks contain vital species with substantial asexual reproduc- historic information, which make these tion, there is then the additional problem that systems quite unique among model systems selective effects may affect larger parts of in evolutionary biology. By hatching eggs the genome at once, thus making this distinc- laid down decades or centuries ago, the tion more difficult. evolution of complex characters can be Critical voices have been raised studied in retrospect, while the effect of against making far-reaching ecological in- selection on neutral markers may at the same ferences based solely on genetic data (e.g. time be revealed. Thus, here is a source of Nichols 1996). Conclusions sometimes vary interesting information for applied pop- with the genetic marker and with the choice ulation genetics as well as more fundamental of analytical model, as is e.g. exemplified by evolutionary biological studies.

29 References

Bengtsson J (1988) Life histories, Bürger R & Lynch M (1995) Evolution and interspecific competition and regional extinction in a changing environment: a distribution of three rockpool Daphnia quantitative-genetic analysis. Evolution 49: species. Comprehensive Summaries of 151-163. Uppsala Dissertations from the Faculty of Science and Technology 142. Uppsala. Bryant EH & Meffert LM (1995) An analysis of selectional response in relation to a population bottleneck. Evolution 49: 626- Bengtsson J (1989) Interspecific competition 634. increases local extinction rate in a metapopulation system. Nature 340: 713- Cáceres CE (1998) Interspecific variation in 715. the abundance, production, and emergence of Daphnia diapausing eggs. Ecology 79: Bengtsson J (1993) Interspecific competition 1699-1710. and determinants of extinction in experimental populations of three rockpool Carvalho GR (1988) Differences in the Daphnia species. OIKOS 67: 451-464. frequency and fecundity of PGI-marked genotypes in a natural population of Daphnia Bengtsson J & Ebert D (1998) Distributions magna Straus (Crustacea: Cladocera). and impacts of microparasites on Daphnia in Functional Ecology 2: 453-462. a rockpool metapopulation. Oecologia 115:213-221. Charlesworth B, Morgan MT & Charlesworth D (1993) The effect of Berg LM, Lascoux M & Pamilo P (1998) deleterious mutations on neutral molecular The infinite island model with sex- variation. Genetics 134: 1289-1303. differentiated gene flow. Heredity 81: 63-68. Charlesworth B, Nordborg M, & Boileau MG & Hebert PDN (1991) Genetic Charlesworth D (1997) The effects of local consequences of passive dispersal in pond- selection, balanced polymorphism and dwelling copepods. Evolution 45: 721-733. background selection on equilibrium patterns of genetic diversity in subdivided Boileau MG & Taylor BE (1994) Chance populations. Genetical Research, Cambridge events, habitat age, and the genetic structure 70: 155-174. of pond populations. Archive für Hydrobiologie 132: 191-202. Cockerham CC (1969) Variance of gene frequencies. Evolution 23: 72-84. Boileau MG, Hebert PDN & Schwartz SS (1992) Non-equilibrium gene frequency div- Cousyn C, De Meester L, Colbourne JK, ergence: persistent founder effects in natural Brendonck L, Verschuren D & Volckaert F populations. Journal of Evolutionary Biology (2001) Rapid, local adaptation of 5: 25-39. zooplankton behavior to changes in predation pressure in the absence of neutral Bossart JL & Prowell DP (1998) Genetic genetic changes. Proceedings of the National estimates of population structure and gene Academy of Sciences USA 98: 6256-6260. flow: limitations, lessons, and new directions. Trends in Ecology and Evolution Crow JF & Kimura M (1970) An 13: 202-206. Introduction to Population Genetics Theory.

30 Harper & Row, New York. Felsenstein J (1982) How can we infer geography and history from gene David P (1998) Heterozygosity-fitness frequencies? Journal of Theoretical Biology correlations: new perspectives on old 96: 9-20. problems. Heredity 80: 531-537. Ferrari DC & Hebert PDN (1982) The induction of sexual reproduction in Daphnia De Meester L (1996) Local genetic magna: Genetic differences between arctic differentiation and adaptation in freshwater and temperate populations. Canadian zooplankton populations: Patterns and Journal of Zoology. 60: 2143-2148. processes. Écoscience 3: 385-399. Frankham R (1995) Effective population size De Meester L & Vanoverbeke J (1999) An / adult populations size ratios in wildlife - a uncoupling of male and sexual egg review. Genetical Research, Cambridge 66: production leads to reduced inbreeding in the 95-107. cyclical parthenogen Daphnia. Proceedings of the Royal Society of London, series B 266: Frankham R (1996) Relationship of genetic 2471-2477. variation to population size in wildlife. Conservation Biology 10: 1500-1508. De Meester L, Gómez A, Okamura B & Schwenk K (2002) Dispersal, Gillespie JH (2000) Genetic drift in an monopolisation and (the lack of) gene flow infinite population: the pseudohitchhiking in aquatic organisms. Acta Oecologia 00: model. Genetics 155: 909-919. 000-000. Gómez A & Carvalho GR (2000) Sex, Deng H-W (1996) Environmental and parthenogenesis and genetic structure of genetic control of sexual reproduction in rotifers: microsatellite analysis of Daphnia. Heredity 76: 449-458. contemporary and resting egg bank populations. Molecular Ecology 9: 203-214. Deng H-W (1997) Photoperiodic response of sexual reproduction in the Daphnia pulex Hairston NG Jr, Lampert W, Cáceres CE, group is reversed in two distinct habitats. Holtmeier CL, Weider LJ, Gaedke U, Limnology and Oceanography 42: 609-611. Fischer JM, Fox JA & Post DM (1999) Rapid evolution revealed by dormant eggs. Dennett CD (1995) Darwin's Dangerous Nature 401: 446. Idea. Simon & Schuster, New York. Hanski I (1999) Metapopulation Ecology. Ebert D (1994) Virulence and local adapt- Oxford University Press, Oxford. ation of a horisontally transmitted parasite. Science 265: 1084-1086. Hardy OJ & Vekemans X (1999) Isolation by distance in a continuous population: Ebert D, Zschokke-Rohringer CD & Carius reconciliation between spatial autocorrel- HJ (1998) Within- and between-population ation analysis and population genetics mod- variation for resistance of Daphnia magna to els. Heredity 83: 145-154. the bacterial endoparasite Pasteuria ramosa. Proceedings of the Royal Society of London, Harpending HC, Batzer MA, Gurven M, series B 265: 2127-2134. Jorde LB, Rogers AR & Sherry ST (1998) Genetic traces of ancient demography. Ellner S & Hairston NG Jr (1994) Role of Proceedings of the National Academy of overlapping generations in maintaining Sciences USA 95: 1961-1967. genetic variation in a fluctuating environment. The American Naturalist 143: Hebert PDN (1974a) Enzyme variability in 403-417. natural populations of Daphnia magna. II.

31 Genotypic frequencies in permanent Kimura M (1968) Evolutionary rate at the populations. Genetics 77: 323-334. molecular level. Nature 217: 624-626.

Hebert PDN (1974b) Enzyme variability in Kimura M & Weiss GH (1964) The stepping natural populations of Daphnia magna. III. stone model of population structure and the Genotypic frequencies in intermittent decrease of genetic correlation with distance. populations. Genetics 77: 335-341. Genetics 49: 561-576. Hebert PDN (1987) Genetics of Daphnia. In: Daphnia (H Peters & R De Bernardi eds.), Kleiven OT, Larsson P & Hobæk A (1992) pp. 439-460. Istituto Italiano di Idrobiologia, Sexual reproduction in Daphnia magna Pallanza. requires three stimuli. OIKOS 65: 197-206.

Hedrick PW (1994) Purging inbreeding Korpelainen H (1984) Genic differentiation depression and the probability of extinction: of Daphnia magna populations. Hereditas full-sib mating. Heredity 73: 363-372. 101: 209-216.

Hedrick PW (1995) Genetic polymorphism Korpelainen H (1986) Temporal changes in in a temporally varying environment: effects the genetic structure of Daphnia magna of delayed germination or diapause. Heredity populations. Heredity 57: 5-14. 75: 164-170.

Hedrick PW (1999) Highly variable loci and Lande R (1988) Genetics and demography in their interpretation in evolution and biological conservation. Science 241: 1455- conservation. Evolution 53: 313-318. 1459.

Hedrick PW & Miller PS (1992) Latter BDH (1973) The island model of Conservation genetics: techniques and population differentiation: a general solution. fundamentals. Ecological Applications 2: 30- Genetics 73: 147-157. 46. Le Corre V & Kremer A (1998) Cumulative Hey J (1997) Mitochondrial and nuclear effects of founding events during genes present conflicting portraits of human colonisation on genetic diversity and origins. Molecular Biology and Evolution14: differentiation in an island and stepping- 166-172. stone model. Journal of Evolutionary Biology 11: 495-512. Hobæk A & Larsson P (1990) Sex determination in Daphnia magna. Ecology Le Corre V, Machon N, Petit RJ & Kremer 71: 2255-2268. A (1997) Colonization with long-distance Ibrahim KM, Nichols RA & Hewitt GM seed dispersal and genetic structure of (1996) Spatial patterns of genetic variation maternally inherited genes in forest trees: a generated by different forms of dispersal simulation study. Genetical Research, during range expansion. Heredity 77: 282- Cambridge 69: 117-125. 291. Levins R & Lewontin R (1985) The Jenkins DG & Underwood MO (1998) Dialectical Biologist. Harvard University Zooplankton may not disperse readily in Press, Cambridge. wind, rain, or waterfowl. Hydrobiologia 387: 15-21. Lynch M (1987) The consequences of Kaplan NL, Hudson RR & Langley CH fluctuating selection for isozyme (1989) The 'hitch-hiking effect' revisited. polymorphisms in Daphnia. Genetics 115: Genetics 123: 887-899. 657-669.

32 Lynch M & Deng H-W (1994) Genetic Maynard Smith J & Haigh J (1974) The slippage in response to sex. The American hitch-hiking effect of a favorable gene. Naturalist 122: 745-764. Genetical Research, Cambridge 23: 23-35.

Lynch M & Gabriel W (1983) Phenotypic Mills LS & Smouse PE (1994) Demographic evolution and parthenogenesis. The consequences of inbreeding in remnant American Naturalist 122: 745-764. populations. The American Naturalist 144: 412-431. Lynch M & Lande R (1993) Evolution and extinction in response to environmental Mort MA & Wolf HG (1985) Enzyme change. In: Biotic Interactions and Global variability in large-lake Daphnia Change (PM Kareiva, JG Kingsolver & RB populations. Heredity 55: 27-36. Huey eds.), pp. 234-250. Sinauer Associates, Sunderland. Nei M (1987) Molecular Evolutionary Genetics. Columbia University Press, New Lynch M & Spitze K (1994) Evolutionary York. genetics of Daphnia. In: Ecological Genetics (LE Read ed.), pp. 109-128. Princeton Nichols RA (1996) Assessing relatedness University Press, Princeton. and evolutionary divergence: why the genetic evidence alone is insufficient. In: Lynch M, Conery R & Bürger R (1995) Molecular genetic approaches in Mutation accumulation and the extinction of conservation (TB Smith & RK Wayne eds.), small populations. The American Naturalist pp 365-379. Oxford University Press, New 146: 489-518. York.

Lynch M, Latta L, Hicks J & Giorganni M Nordborg M (2001) Coalescent theory. In: (1998) Mutation, selection, and the Handbook of Statistical Genetics (DJ maintenance of life-history variation in a Balding, MJ Bishop & C Cannings eds.), pp natural population. Evolution 52: 727-733. 179-211. John Wiley & Sons, Chichester.

Lynch M, Pfrender M, Spitze K, Lehman N, Okamura B & Freeland J (2001) Gene flow Hicks J, Allen D, Latta L, Ottene M, Bogue and the evolutionary ecology of passively F & Colbourne JK (1999) The quantitative dispersing aquatic invertebrates. In: and molecular genetic architecture of a Dispersal (J Bullock ed.), pp. 000-000. subdivided species. Evolution 53: 100-110. Blackwell Scientific Publications, London.

Malécot G (1975) Heterozygosity and Pamilo P, Pálsson S & Savolainen O (1999) relationship in regularly subdivided Deleterious mutations can reduce populations. Theoretical Population Biology differentiation in small, subdivided 8: 212-241. populations. Hereditas 130: 257-264. Pálsson S (2000) Microsatellite variation in Maruyama K & Tachida H (1992) Genetic Daphnia pulex from both sides of the Baltic variability and geographical structure in Sea. Molecular Ecology 9: 1075-1088. partially selfing populations. Japanese Journal of Genetics 67: 39-51. Pálsson S (2001) The effects of deleterious mutations in cyclically parthenogenetic org- Maruyama T & Fuerst PA (1985) Population anisms. Journal of Theoretical Biology 208: bottlenecks and non-equilibrium models in 201-214. population genetics. II. Number of alleles in a small population that was formed by a Pritchard JK, Stephens MS & Donelly P recent bottleneck, Genetics 111: 675-689. (2000) Inference of population structure

33 using multilocus genotype data. Genetics polymorphism at microsatellite loci in 155: 945-959. natural Drosophila melanogaster populations. Genetics 146: 309-320. Prout T (1981) A note on the island model with sex dependent migration. Theoretical Sjögren-Gulve P & Berg LM (1999) and Applied Genetics 59: 327-332. Allozyme variation as a demographic predictor at high latitudes: the moor frog and Raymond M & Rousset F (1995) GENEPOP the pool frog at 60°N. Hereditas 130: 317- (version 1.2): population genetics software 323. for exact tests and ecumenicism. Journal of Heredity 86:248-249. Slatkin M (1985a) Rare alleles as indicators of gene flow. Evolution 39: 53-65. Reich DE & Goldstein DB (1998) Genetic evidence for a paleolithic human population Slatkin M (1985b) Gene flow in natural expansion in Africa. Proceedings of the populations. Annual Review of Ecology and National Academy of Sciences USA 95: Systematics 16: 393-430. 8119-8123. Slatkin M (1991) Inbreeding coefficients and Rousset F (1996) Equilibrium values of coalescence times. Genetical Research, measures of population subdivision for Cambridge 58: 167-175 stepwise mutation processes. Genetics 142: 1357-1362. Slatkin M (1993) Isolation by distance in equilibrium and non-equilibrium Rousset F (1997) Genetic differentiation and populations. Evolution 47: 264-279. estimation of gene flow from F-statistics under isolation by distance. Genetics Slatkin M (1994) Gene flow and population 145:1219-1228. structure. In: Ecological Genetics (LE Read, ed.), pp. 3-17. Princeton University Press, Rousset F (2001) Inferences from spatial Princeton. population genetics. In: Handbook of Statistical Genetics (DJ Balding, MJ Bishop Slatkin M & Wiehe T (1998) Genetic hitch- & C Cannings eds.), pp 239-269. John Wiley hiking in a subdivided population. Genet. & Sons, Chichester. Res. 71: 155-160.

Rowe CL & Hebert PDN (1999) Cladoceran Tachida H & Yoshimaru H (1996) Genetic website. University of Guelph. diversity in partially selfing populations with http://www.cladocera.uoguelph.ca/ the stepping stone structure. Heredity 77: 469-475. Schierup MH, Charlesworth D & Vekemans Takahata N & Nei M (1984) FST and GST X (2000) The effect of hitch-hiking on genes statistics in the finite island model. Genetics linked to a balanced polymorphism in a sub- 107: 501-504. divided population. Genetical Research, Cambridge 76: 63-73. Vanoverbeke J & De Meester L (1997) Among-populational genetic differentiation Schlötterer C & Wiehe T (1999) in the cyclical parthenogen Daphnia magna Microsatellites, a neutral marker to infer (Crustacea, Anomopoda) and its relation to selective sweeps. In: (DB Goldstein & C geographic distance and clonal diversity. Schlötterer eds.), Microsatellites. Evolution Hydrobiologia 360: 135-142. and Application, pp 49-65. Oxford University Press, Oxford. Wade MJ & McCauley DE (1988) Extinction and recolonisation: their effects Schlötterer C, Vogl C & Tautz D (1997) on the genetic differentiation of local Polymorphism and locus-specific effects on populations. Evolution 42: 995-1005.

34 Waser PM & Strobeck C (1998) Genetic Whitlock MC & McCauley DE (1990) Some signatures of interpopulation dispersal. population genetic consequences of colony Trends in Ecology and Evolution 13: 43-44. formation and extinction: genetic correlations within founding groups. Weider LJ & Hobæk A (1997) Postglacial Evolution 44: 1717-1724. dispersal, glacial refugia, and clonal structure in Russian/Siberian populations of the arctic Whitlock M, Ingvarsson P & Hatfield T Daphnia pulex complex. Heredity 78: 363- (1998) Local drift load and the heterosis of 372. interconnected populations. Heredity 84: 452-457. Weider LJ, Hobæk A, Colbourne JK, Crease TJ, Dufresne F & Hebert PDN (1999) Wright S (1931) Evolution in Mendelian Holarctic phylogeography of an asexual populations. Genetics 16: 97-159. species complex. I. Mitochondrial DNA variation in Arctic Daphnia. Evolution 53:777-792. Wright S (1951) The genetical structure of populations. Annals of Eugenics 15: 323- Weir BS (1996) Genetic Data Analysis II. 354. Sinauer Associates, Sunderland. Wright S (1978) Evolution and the genetics Weir BS & Cockerham CC (1984) of populations. Vol. 4. Variability within and Estimating F-statistics for the analysis of among natural populations. University of population structure. Evolution 38: 1358- Chicago Press, Chicago. 1370. Yampolsky LY (1992) Genetic variation in Whitlock MC & Barton NH (1997) The the sexual reproduction rate within a effective size of a subdivided population. population of a cyclical parthenogen, Genetics 146: 427-441. Daphnia magna. Evolution 46: 833-837

35 Acknowledgements

First of all I want to thank those who im- John Colbourne generously shared proved this text by means of careful scrutiny: some of his microsatellite primers with me. Martin Lascoux, Cilla Ahlström-Rapaport, Niklas Gyllenstrand and Mats Isaksson help- David Lerdell, and Helena Berg. Thanks also ed me when I got stuck in the realm of bio- to Jacob Höglund for providing sufficient chemistry, and their presence certainly made computer power to generate swarms of the time in the lab more interesting. Cia Daphnia, and to Martin Carlsson for his Olsson and Kerstin Santesson ran the lab in a willingness to do so. wonderfully smooth way. Rose-Marie I am forever indebted to Martin Andersson was the best boss I think I may Lascoux, at first for all his relentless support ever have. Other practical help is acknow- while I was struggling to find the right ques- ledged in the papers, but what the heck, tions and the financial means to ask them, thanks again everyone! and later, for all his efforts to help me answer Special thanks go out to Lena Clap- completely different questions. Pekka Pamilo ham and the undergraduate students I have is the man without whom all this hard work taught. It lifts you up to know that you are would never have taken place. He and Martin actually fulfilling a purpose once in a while. converted me, quite unexpectedly, from a And last but not least, all the boys and conservation biologist into a population girls at the department of Conservation bio- geneticist. If that meant selling my soul, at logy and Genetics and at our neighbour de- least it might have rendered me a Ph D. Per partments at EBC and the Genetic Center. In Sjögren Gulve stimulated my early devel- the next life I will be less of a family man opment, and his encouragement has been an and go to more parties. To the as yet ungra- important factor in my career decisions. I duated subset of my colleagues I just want to also send a thought to Urban Gullberg who say: remember the immortal words of the once invested much time and money in me, Swedish poet Robert Broberg: “Målet är and kept me going, but never got much in ingenting, vägen är all-la-la-la-la-la-lallt”! return. Very important persons are also found Many thanks also to my other co- in the VSVB on-line community: Stefan authors, Snæbjörn Pálsson and Jan Krüger, David Lerdell, and Jakob Sintring. Bengtsson. Snæbjörn never gave up teaching You inspired me, and distracted me, and thus me all I didn't know that I needed to know, helped me reach some of my goals. even at times when I didn't want to know. He This work would never have been also had the good taste to introduce me to done in time without the love and understan- Daphnia pulex, and to do most of the primer ding of my family: Helena and Maja. Maybe optimisation work for me to parasitise on. I now I will be able to explain to you what I have difficulties imagining what this thesis have been doing all this time. Just read this would look like without his influence. book! (Vilken bra godnattsaga...)