Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A clone together : exploring the causes and consequences of range divergence between sexual and asexual… Hersh, Evan Whitney 2020

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2020_november_hersh_evan.pdf [ 11.83MB ]
Metadata
JSON: 24-1.0392003.json
JSON-LD: 24-1.0392003-ld.json
RDF/XML (Pretty): 24-1.0392003-rdf.xml
RDF/JSON: 24-1.0392003-rdf.json
Turtle: 24-1.0392003-turtle.txt
N-Triples: 24-1.0392003-rdf-ntriples.txt
Original Record: 24-1.0392003-source.json
Full Text
24-1.0392003-fulltext.txt
Citation
24-1.0392003.ris

Full Text

A Clone Together: Exploring the Causes andConsequences of Range Divergence BetweenSexual and Asexual Easter DaisiesbyEvan Whitney HershB.A., University of California Santa Cruz, 2009M.Sc., Lund University, 2012a thesis submitted in partial fulfillmentof the requirements for the degree ofDoctor of Philosophyinthe faculty of graduate and postdoctoralstudies(Botany)The University of British Columbia(Vancouver)June 2020c© Evan Whitney Hersh, 2020The following individuals certify that they have read, and recommend to the Facultyof Graduate and Postdoctoral Studies for acceptance, the thesis entitled:A Clone Together: Exploring the Causes and Consequences ofRange Divergence Between Sexual and Asexual Easter Daisiessubmitted by Evan Whitney Hersh in partial fulfillment of the requirements forthe degree of Doctor of Philosophy in Botany.Examining Committee:Jeannette Whitton, BotanySupervisorAmy Angert, BotanySupervisory Committee MemberMike Whitlock, ZoologyUniversity ExaminerDarren Irwin, ZoologyUniversity ExaminerMarcel Dorken, Trent UniversityExternal ExaminerAdditional Supervisory Committee Members:Sally Otto, ZoologySupervisory Committee MemberLoren Rieseberg, BotanySupervisory Committee MemberDolph Schluter, ZoologySupervisory Committee MemberiiAbstractSexual and asexual organisms exhibit a wide variety of biological differences thatcan impact their ecological and evolutionary trajectories. One result of these differ-ences is that closely related sexual and asexual taxa often exhibit range divergence,with asexuals typically having larger geographic ranges and being found at higherlatitudes and elevations. This pattern, termed “geographical parthenogenesis”, hasbeen documented in numerous plant and animal systems and a variety of potentialmechanisms have been proposed. Hypotheses relate to differences in reproductiveassurance (asexuals can reproduce without mates, while most sexuals require mates),genetic consequences of sexuality vs asexuality, selection on clonal lineages, ecologicalimpacts of sexuality and asexuality, and demographic differences between reproduc-tive modes. In this thesis, I explore several of these potential drivers of geographicalparthenogenesis in Townsendia hookeri, a subalpine perennial flowering plant in theAsteraceae that has diploid sexual and polyploid apomictic (reproducing asexuallythrough seed) forms with divergent but overlapping ranges. Population genomic anal-yses of apomicts in T. hookeri revealed largely monoclonal populations and geograph-ically widespread clones, suggesting that apomictic range expansion may have beenaided by “general-purpose genotypes” that can withstand an array of environmentalconditions. Results from a large-scale, multi-year reciprocal transplant garden ex-periment showed that sexual populations had comparable performance when plantediiiinto the apomictic range as into their own, but that fitness of apomictic individualsgenerally declined in sexual regions. This provides evidence that while sexuals arelikely limited by dispersal (they cannot reach suitable habitat outside of their cur-rent range), apomicts are not well adapted to the ecological conditions (biotic and/orabiotic) in the sexual range. When comparing early life history traits between sex-uals and apomicts, apomicts were found to have increased germination success andimproved seed dispersal traits in comparison to sexuals. These traits are expectedto have given apomicts a colonization advantage, which (along with reproductiveassurance) has likely contributed to their increased range size. Overall, the work pre-sented in this thesis highlights the intricate nature of geographical parthenogenesisin Townsendia hookeri, and underscores the need to investigate complex biologicalphenomena using a diverse suite of approaches.ivLay SummaryMost organisms reproduce sexually (with sperm and egg from two mates fusing toform a new individual), but many reproduce asexually without mates (forming clonesof themselves). In nature, we often find that closely related sexuals and asexuals arefound in different places, with asexuals usually having broader geographic ranges thansexuals. Biologists have several ideas for why this happens, but because organismsare complicated and diverse, no single explanation adequately addresses the patternas a whole. In this thesis, I investigate this pattern in Townsend’s Easter daisies(Townsendia hookeri), a flowering plant with sexual and asexual forms that havedifferent ranges. Using several approaches, including DNA sequencing and field ex-periments, I found evidence that the spread of asexual Easter daisies was likely aidedby particularly successful clones, seed traits that help them start new populations,and their ability to reproduce without mates. In other words: “it’s complicated”.vPrefaceAll of the work presented in this thesis was conducted in the field or in the BiodiversityResearch Centre at the University of British Columbia, Point Grey campus.I was the lead investigator for all chapters and was responsible for conceptualiza-tion, data collection, data analysis, and manuscript writing. Several undergraduatesand colleagues contributed to the data collection. Chris Lee, a PhD student in theWhitton Lab, collected plant material that was sequenced for Chapter 2. JaimeGrimm and Katya Hernandez (undergraduates) as well as Adam Wilkinson, RachelWilson, Erin Warkman, Ryan Cologne, Lynn Riedel, and Bianca Hersh (researchassistants) contributed to the collection of data presented in Chapter 3. AlbertoRuiz-Larrera, an undergraduate in the Whitton Lab, contributed seed dispersal traitdata that was presented in Chapter 4. Amy Angert, Sally Otto, Loren Rieseberg, andDolph Schluter contributed manuscript edits and helped with the interpretation ofresults for all chapters. Jeannette Whitton was my PhD supervisor, and was involvedat a fundamental level in all aspects of this thesis work, including conceptualization,data collection, interpretation of results, and manuscript edits.viTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 The history of geographical parthenogenesisand its proposed explanations . . . . . . . . . . . . . . . . . . . . . . 21.1.1 Uniparental reproduction . . . . . . . . . . . . . . . . . . . . . 31.1.2 Polyploidy and hybridization correlate with parthenogenesis . 51.1.3 Selection on clonal lineages . . . . . . . . . . . . . . . . . . . . 61.1.4 Ecological impacts on the benefits of sex vs asexuality . . . . . 81.2 Geographical parthenogenesis in Townsendia hookeri . . . . . . . . . 10vii1.3 Thesis outline and objectives . . . . . . . . . . . . . . . . . . . . . . . 132 Clonal Population Genomic Structure of Polyploid Apomictic EasterDaisies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.2 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . 202.2.1 Study system . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.2.2 Population sampling . . . . . . . . . . . . . . . . . . . . . . . 212.2.3 ddRAD library preparation . . . . . . . . . . . . . . . . . . . 222.2.4 De novo assembly and SNP calling . . . . . . . . . . . . . . . 232.2.5 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 242.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272.3.1 Clone structure . . . . . . . . . . . . . . . . . . . . . . . . . . 272.3.2 Genetic structure of sexual populations and relationships withapomictic lineages . . . . . . . . . . . . . . . . . . . . . . . . . 292.3.3 Genetic diversity of sexuals and apomicts . . . . . . . . . . . . 322.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322.4.1 Clone identification . . . . . . . . . . . . . . . . . . . . . . . . 322.4.2 Clone structure in Townsendia hookeri . . . . . . . . . . . . . 352.4.3 Origins and spread of apomictic lineages . . . . . . . . . . . . 372.4.4 Beyond frozen niches and general-purpose genotypes . . . . . 383 Investigating Drivers of Geographical Parthenogenesis in Townsendiahookeri Using a Reciprocal Transplant Experiment . . . . . . . . . 483.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483.2 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . 543.2.1 Reciprocal transplant experimental design . . . . . . . . . . . 54viii3.2.2 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 573.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593.3.1 Establishment success . . . . . . . . . . . . . . . . . . . . . . 593.3.2 Reciprocal transplant experiment . . . . . . . . . . . . . . . . 603.3.3 Site characteristics . . . . . . . . . . . . . . . . . . . . . . . . 623.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 623.4.1 Sexual range expansion is limited by dispersal . . . . . . . . . 633.4.2 Habitat is not suitable for apomicts beyond their southern rangelimit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663.4.3 Drivers of geographical parthenogenesis in Townsendia hookeri 684 Differences in Early Life History Traits between Diploid Sexualand Polyploid Apomictic Easter Daisies . . . . . . . . . . . . . . . . 854.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 854.2 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . 894.2.1 Study system . . . . . . . . . . . . . . . . . . . . . . . . . . . 894.2.2 Seed collection . . . . . . . . . . . . . . . . . . . . . . . . . . 904.2.3 Seed dispersal traits and terminal velocity measurements . . . 914.2.4 Germination in the lab . . . . . . . . . . . . . . . . . . . . . . 914.2.5 Seedling traits . . . . . . . . . . . . . . . . . . . . . . . . . . . 924.2.6 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 934.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 944.3.1 Seed dispersal traits and terminal velocity . . . . . . . . . . . 944.3.2 Germination traits . . . . . . . . . . . . . . . . . . . . . . . . 954.3.3 Seedling traits . . . . . . . . . . . . . . . . . . . . . . . . . . . 974.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97ix5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118xList of TablesTable 2.1 Population information (Chapter 2) . . . . . . . . . . . . . . . . . 46Table 2.2 Pairwise Fst between all populations . . . . . . . . . . . . . . . . . 47Table 3.1 Garden and source population information (Chapter 3) . . . . . . 80Table 3.2 Likelihood ratio test statistics for establishment success . . . . . . 81Table 3.3 Summary statistics for establishment success. . . . . . . . . . . . . 81Table 3.4 Likelihood ratio test statistics for ASTER models . . . . . . . . . 82Table 3.5 Raw data for the reciprocal transplant garden experiment. . . . . . 82Table 3.6 Likelihood ratio test statistics for leaf number models . . . . . . . 83Table 3.7 Likelihood ratio test statistics for leaf length models . . . . . . . . 84Table 3.8 Likelihood ratio test statistics for seed set models . . . . . . . . . . 84Table 4.1 Population information (Chapter 4) . . . . . . . . . . . . . . . . . 105Table 4.2 Relationships between seed traits and terminal velocity. . . . . . . 105Table 4.3 Likelihood ratio test statistics for seed traits . . . . . . . . . . . . 106Table 4.4 Summary statistics for terminal velocity . . . . . . . . . . . . . . . 106Table 4.5 Likelihood ratio test statistics for germination and seedling traits . 107Table 4.6 Summary statistics for germination success . . . . . . . . . . . . . 107Table 4.7 Summary statistics for germination speed . . . . . . . . . . . . . . 108Table 4.8 Summary statistics for seedling survival . . . . . . . . . . . . . . . 108xiTable 4.9 Summary statistics for seedling leaf number . . . . . . . . . . . . . 109Table 4.10 Summary statistics for seedling leaf length . . . . . . . . . . . . . 109xiiList of FiguresFigure 1.1 Townsendia hookeri growing in its natural habitat. . . . . . . . . 11Figure 2.1 Genetic distance histograms . . . . . . . . . . . . . . . . . . . . . 25Figure 2.2 Map of sampled populations and clonal structure . . . . . . . . . 28Figure 2.3 K -means clustering results - apomictic individuals . . . . . . . . . 29Figure 2.4 Neighbor-joining tree - all individuals . . . . . . . . . . . . . . . . 30Figure 2.5 DAPC - all individuals . . . . . . . . . . . . . . . . . . . . . . . . 31Figure 2.6 Observed heterozygosity - all individuals . . . . . . . . . . . . . . 33Figure 2.7 Allele balance histograms to confirm ploidy . . . . . . . . . . . . 42Figure 2.8 Neighbor-joining tree - all populations . . . . . . . . . . . . . . . 43Figure 2.9 Neighbor-joining tree - sexual populations . . . . . . . . . . . . . 44Figure 2.10 DAPC plots for K values 9-12 from K -means clustering analysis . 45Figure 3.1 Map of source populations and garden sites . . . . . . . . . . . . 55Figure 3.2 Establishment success and ASTER model results . . . . . . . . . 60Figure 3.3 Hierarchical structure of ASTER models . . . . . . . . . . . . . . 72Figure 3.4 Random sampling results of establishment and bud production data 73Figure 3.5 Predicted leaf number by year . . . . . . . . . . . . . . . . . . . . 74Figure 3.6 Predicted leaf length by year . . . . . . . . . . . . . . . . . . . . . 75Figure 3.7 Seed set results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76xiiiFigure 3.8 Root to shoot volume ratio results . . . . . . . . . . . . . . . . . 77Figure 3.9 Vegetation cover in each garden region . . . . . . . . . . . . . . . 78Figure 3.10 Climate data for each garden . . . . . . . . . . . . . . . . . . . . 79Figure 4.1 Seed dispersal traits . . . . . . . . . . . . . . . . . . . . . . . . . 95Figure 4.2 Germination and seedling traits . . . . . . . . . . . . . . . . . . . 96Figure 4.3 Seed dispersal traits by mating system . . . . . . . . . . . . . . . 103Figure 4.4 Terminal velocity, germination, and seedling traits by population 104xivAcknowledgmentsI would first like to thank my family (Sid, Lynne, Jocie, Bianca, and Larry Hersh) fortheir love and support throughout my time at UBC. I give my sincere and heartfeltthanks to my PhD supervisor, Jeannette Whitton, for her guidance, mentorship, andcrucial contributions to this thesis and my growth as a scientist. In addition, I greatlyappreciate the feedback, support, and guidance given by my committee members:Amy Angert, Sally Otto, Loren Rieseberg, and Dolph Schluter.My garden experiment would not have happened without the generosity of theBlake family, Lynn Riedel, and Daniel Tinker, who donated their land (or land theywere responsible for) to the cause of science. I owe my undying gratitude to AdamWilkinson, who helped me get my experiment off the ground and taught me everythingthere is to know about “the rules of the road”, as well as Julia Costa and MarcoTodesco, who patiently and effectively gave me a crash course on molecular biology.I must also thank Jamie Fenneman, Chris Lee, and Shona Ellis for being the bestdarned teaching team this side of the Georgia Strait, and for making me a betterbotanist and educator.I would be remiss without extending my thanks to my friends and colleagues,including (but not limited to): Jacob Templeton, Chris Prioleau, Evan Wilson andthe rest of my Santa Cruz family (you know who you are), The Pacific Spirit LoungeSociety, The Lovely Group of Guys, and Tha Doughboiz.xvChapter 1IntroductionAs humans and scientists, we are obsessed with exploring, documenting, and (hope-fully) understanding the natural world. Life on Earth presents many of the mostcomplex systems - and most tantalizing mysteries - for us to unravel. In the past200 years alone we’ve achieved incredible insight into biological processes large andsmall, but each discovery spawns a cascade of new questions that unfurl infinitely inall directions, whether at the microscopic scale of genomes or from the planet-wideperspective of interlocking ecosystems. Our individual investigations often focus onone species at a time, one question at a time, and what we continue to learn is thateverything is connected - genes and morphology, individuals and populations, pastand future. Grappling with this dizzying complexity and diversity is the great charge(and challenge) of ecology and evolutionary biology.Active areas of research in ecology and evolution reflect this endeavor to connectthe large with the small, and use known patterns to inform what we can expect in thefuture. Recent advancements in sequencing technology along with rapidly reducingcosts have given rise to the widespread use of genomics, which has allowed researchersto investigate population genetic processes at greater depth and more easily tie them1to ecological questions. Similarly, the increase in publicly available climate data hasallowed biologists to gain deeper and broader insight than previously possible intothe interface between biotic and abiotic forces. In addition, the increased use of pro-gramming, particularly open-source languages like R and Python, has made advancedstatistical analyses and bioinformatics widely available to anyone who is willing toput in the time and effort to learn. All of these tools have helped push the scope andbreadth of ecology and evolution research programs that aim to address big ques-tions with granular data. The work presented in this thesis focuses on investigatinga pattern that sits at the intersection of several important research areas, includingthe evolution of sex, drivers of species range limits, local adaptation, the evolutionaryimpacts of polyploidy, and the co-evolution of mating system and dispersal traits.Geographical parthenogenesis (GP), as the name implies, describes an often-foundmotif where closely related sexual and asexual (parthenogenic) taxa inhabit differentgeographic ranges. The emerging question is simple: why? It might seem straightfor-ward at first, but patterns of GP are the result of multiple interacting ecological andevolutionary forces that are bound to differ depending on the unique biological con-texts in which the pattern manifests. The fact that so many potential drivers coincidecan make GP research challenging, but also extremely rewarding in the insights thatwe gain into each of the processes involved.1.1 The history of geographical parthenogenesisand its proposed explanationsThe term “geographical parthenogenesis” was first introduced by Albert Vandel in the1920’s, who was investigating geographic trends in certain species of arthropods wheremales were more rare at higher latitudes. He realized that these latitudinal trendswere due to an increased prevalence of obligately parthenogenic forms of the same2species, which could be applied to similar patterns found in other groups. He definedGP as the phenomenon where sexual and asexual forms of the same species occupydistinct (though potentially overlapping) geographic ranges. Several decades later,cases of GP were reviewed in animals by Glesener & Tilman (1978) and in plants byBierzychudek (1985), from which some refinements of the pattern emerged. Moderndefinitions of GP propose that asexuals generally have broader distributions, occurat higher latitudes, and are found at higher elevations than their sexual progenitors(Tilquin & Kokko 2016). Along with these, several models imply (sometimes indi-rectly) that parthenogens are also found more frequently in “marginal” or disturbedhabitats (Vrijenhoek & Parker 2009). A number of potential mechanisms have beenproposed for GP, and while many of them overlap in some aspects, they tend to focuson different ecological, evolutionary, and/or genetic components that may affect theranges of closely related sexuals and asexuals.1.1.1 Uniparental reproductionParthenogenesis is a derived trait that is found in all major groups of eukaryotes, andis generally defined as a form of asexual reproduction where a zygote is formed froman unfertilized female gamete (Tilquin & Kokko 2016). Parthenogens are capable ofuniparental reproduction, meaning that (in most cases) they have full reproductiveassurance and can proliferate without mates. Parthenogenic organisms are thereforeassumed to have a demographic advantage (all else being equal) over those that re-produce sexually, because sexuals are largely dependent on mates (Bell 1982). Thisadvantage forms the basis for one of the most obvious and intuitive potential expla-nations for GP. Parthenogens with full reproductive assurance are expected to have acolonization advantage over sexuals, because it only takes a single propagule to estab-lish a new population, whereas sexuals require two propagules and successful mating3in order to successfully establish (Baker 1955). Uniparental reproduction is clearlyadvantageous during colonization, but also in other contexts where mate-limitation islikely (i.e., low density populations or, for plants, areas with low pollinator availabil-ity; Gascoigne et al. 2009), for example in highly disturbed environments or previouslyglaciated areas (Ho¨randl 2009). This implies that unless dispersal/demographic bar-riers are too high to overcome, sexuals should be able to catch up given enough time,which points to other drivers interacting with uniparentality in maintaining stablerange divergence between sexuals and asexuals.Support for the effects of reproductive assurance can be gleaned from “Baker’sLaw” (Baker 1955), which predicts that self-compatible plants will be more likely tohave larger ranges than those that are not capable of selfing (Pannell et al. 2015).The Baker’s Law literature is vast, and a recent large-scale study suggests that selfingspecies do have larger ranges than outcrossing species (Grossenbacher et al. 2015),but comparisons of selfing and parthenogenesis should be made with caution. Selfingresults from the fusion of two gametes and thus will lead to an increase in homozygos-ity. The transition to selfing is therefore often associated with inbreeding depression,while parthenogens avoid inbreeding depression as a result of frozen levels of het-erozygosity (Haag & Ebert 2004). We would therefore expect different long-termtrajectories of selfing and parthenogenic populations due to the two reproductivestrategies having different genetic consequences. While inbreeding depression canlead to decreased fitness and potential extirpation in selfing populations, selfers alsohave the ability to purge deleterious mutations (Schemske & Lande 1985) that areexpected to accumulate in asexuals and hinder their long-term persistence (Muller1964).The evolution of dispersal and mating system traits is related to the idea of Baker’sLaw effects, but is curiously rare in the GP literature. Given that parthenogens often4have larger ranges than sexuals, it is possible that they may also possess traits (beyondreproductive assurance) that benefit dispersal and/or colonization ability. Besides afew recent examples (see Coughlan et al. 2014 and Chrtek et al. 2018), differences indispersal ability per se between sexuals and asexuals have rarely been explored in aGP context (but see O’Connell & Eckert 2001 who investigated differences betweensexual and apomictic diaspores in Antennaria parlinii). This is an important openresearch area within GP, as dispersal in time and space can both aid range expansionsand allow escape from pathogens or environmental instability (Wilson 2011).1.1.2 Polyploidy and hybridization correlate withparthenogenesisGiven that parthenogens originate from an incredibly diverse range of taxa, it isno surprise that it functions differently depending on the evolutionary context inwhich it arose. Still, many parthenogens share common features; for example, mostparthenogens are polyploids of hybrid origin (Bengtsson 2009). Because of this associ-ation, some proposed hypotheses for GP focus on the effects that ploidy and hybridityhave on driving patterns of range divergence. Polyploidy itself, as well as many ofthe traits associated with polyploidy, have been hypothesized to bestow advantages,including larger cells, increased genetic expression, protection from deleterious muta-tions, and increased evolutionary potential (Comai 2005). Polyploidy has also beenlinked with increased invasive ability (Te Beest et al. 2012), though some large-scalestudies suggest that polyploids do not consistently show range shifts relative to theirdiploid progenitors (Martin & Husband 2009). Similar to Baker’s Law, the litera-ture concerning the ecological and evolutionary impacts of polyploidy is expansive,and several reviews and meta-analyses have been published in recent years that oftenappear to take opposing stances; for example, Glennon et al. (2014) found that poly-5ploids do not consistently show niche shifts in comparison to diploids, while Baniagaet al. (2020) propose that polyploids show faster rates of niche evolution. Becausepolyploidy is so pervasive (especially in plants; Otto & Whitton 2000), the predictedeffects of polyploidy on patterns of GP will differ depending on the evolutionarycontext in each system. For example, genome duplication can occur within a singlespecies (autopolyploidy) or as a result of hybridization between two close relatives(allopolyploidy), and the latter will be influenced by the interacting effects of bothhybridization and polyploidy.Viewed on its own, hybridization between two sexual species is often deleterious(Johnson 2010), but in some cases can result in hybrid vigour (Rieseberg et al. 1999;Baack & Rieseberg 2007; Chen 2013). The infusion of genetic diversity from hy-bridization may aid plants during colonization scenarios, and such benefits would beconserved by parthenogenesis (Kearney 2005). However, because most of the well-studied GP systems are allopolyploids (Bengtsson 2009), it can be difficult to disen-tangle the relative contributions of reproductive assurance, polyploidy, and hybridityin driving the pattern. These potential combinatory effects highlight the importanceof viewing GP as the result of an interacting suite of traits (especially given thatthey are commonly associated) as opposed to attempting to identify the single mostimportant component in driving the pattern across a diverse array of taxa.1.1.3 Selection on clonal lineagesSome models for GP focus on the ways in which selection acts differently on sexualsand asexuals, and how different types of selective pressures can affect clonal diversity.As new clonal lineages emerge, those with a favorable set of attributes will succeedand be able to establish and spread. Depending on the frequency of origins of newclones and their fitness, a variable number of clonal lineages may become established,6and those that do will be protected (by parthenogenesis) from the breakdown ofbeneficial allele combinations that would otherwise occur in sexual populations (Lynch1984). This idea lead to the formulation of two opposing (but not mutually exclusive)hypotheses: the general-purpose genotype hypothesis (Baker 1965; Parker et al. 1977)and the frozen niche variation hypothesis (Vrijenhoek 1979; evidence for both reviewedin Vrijenhoek & Parker 2009).The general-purpose genotype hypothesis predicts that selection on clones inhighly variable environments will give rise to lineages that are tolerant of a wide-array of environmental conditions. Populations consisting of these “general-purpose”clonal lineages are expected to be better able to persist in abiotically unpredictableareas than more specialized sexual genotypes, but would be outcompeted by sexuals(or specialized clones) in parts of the range that are stable enough for populationsto become locally adapted (Tilquin & Kokko 2016). This pattern is in line with thedescriptions of Baker (1965), who likened parthenogens to “weedy” species that areexcluded to the range margins where they are safe from competitors.The frozen niche variation hypothesis instead predicts that a diverse array ofspecialized clones will be better able to partition the available niche space than apopulation of sexuals, because sex can pull divergent phenotypes back to the mean ofthe distribution (Weeks 1993). This hypothesis assumes that a variety of clones willarise within populations, and competition (either among clones or with sexual pro-genitors) will eliminate clonal lineages whose niches overlap, leaving only specializedclones that are “frozen” in their respective niche spaces. While the general-purposegenotype hypothesis is concerned with fitness fluctuations in time (i.e., due to vari-able environmental conditions), the frozen niche variation hypothesis focuses moreon fitness variation in space (i.e., heterogenous resource availability within habitats;Vrijenhoek & Parker 2009).7Broadly speaking, the general-purpose genotype hypothesis predicts the estab-lishment of a few geographically widespread clones, while the frozen niche variationhypothesis predicts the establishment of a diverse array of local, specialized clones.More specific predictions about clonal diversity within and among populations aresomewhat unclear, however. For example, it seems possible that marginal popu-lations may consist of multiple general-purpose genotypes, and a specialized clonemay still become widespread if it is able to disperse to distant locations that ex-hibit similar niche characteristics. In addition, both patterns may become apparentwithin a system if clones in different parts of the range are subject to varying selec-tive pressures (Kenny 1996). Like with other models of GP, it remains important toevaluate system-specific parameters (i.e., population genetic diversity, mode of clonalorigin, and niche characteristics) in order to assess how selection has influenced thegeographic spread of clonal lineages.1.1.4 Ecological impacts on the benefits of sex vs asexualityAnother realm of GP theory relates to how the benefits of sex/asex vary in differentecological conditions. The two most prominent candidates here are the Red Queenhypothesis and the Tangled Bank hypothesis. The Red Queen hypothesis, which wasverbally presented by Glesener & Tilman (1978), suggests that sexuals will be betterable to adapt in the face of rapidly changing biotic pressures (e.g. pathogens andcompetition) than asexuals due to the benefits of genetic recombination. Sexuals willtherefore be able to persist in more ecologically complex areas, while parthenogenswill be excluded to more marginal habitats (e.g. higher elevations and latitudes)where biotic interactions are assumed to be less intense (Louthan et al. 2015). TheTangled Bank hypothesis focuses more on competition over shared resources, andpredicts that the phenotypic diversity of sexuals will allow them to better parse re-8sources in heterogeneously structured environments than a homogeneous set of clones(Bell 1982). Similar to the Red Queen hypothesis, the Tangled Bank expects thatparthenogens will be favoured in environments that are unpredictable over space andtime, but their demographic advantage over sexuals will vanish in areas of greaterdiversity (in this case resource diversity as opposed to biotic diversity).As with all models, there are important assumptions to keep in mind about boththe Red Queen and Tangled Bank models. The Red Queen hypothesis assumes thelong-held belief that biotic interactions decrease in intensity and frequency with lati-tude, but that has been challenged by recent large-scale reviews (Moles et al. 2011b;Hargreaves et al. 2020). In addition, both the Red Queen and the Tangled Bankmodels assume that asexuals exhibit reduced phenotypic diversity in comparison tosexuals (Tilquin & Kokko 2016), but as explained in the previous section, competitionamong clones may lead to a diverse array of clonal lineages that may exhibit greaterdiversity (at least in terms of niche breadth) than local sexuals. The processes andpredictions of these two models also bear a conspicuous resemblance to those of thegeneral-purpose genotype and frozen niche variation hypotheses. For example, theRed Queen and general-purpose genotype hypotheses both predict that parthenogenswill be relegated to marginal environments. The Tangled Bank and frozen niche vari-ation hypotheses both discuss the ability to partition niche space, but the TangledBank assumes that sexuals will have the advantage in this context while the frozenniche expects parthenogens to have the advantage (assuming the diversity of clones ishigh enough). Taken together, the overlap among models of GP (and in some casesdissonant expectations) underscores the need to take a comprehensive approach toinvestigations of the pattern in each system, as the drivers described above may com-pound or counteract each other in ways that are difficult to predict without sufficientorganismal context.91.2 Geographical parthenogenesis in TownsendiahookeriSome of the most well-developed GP systems are in plants. Plants provide excellentmodel systems for GP in part due to their great diversity of ploidies and reproduc-tive modes. Angiosperms display a wide range of reproductive strategies, includingvegetative reproduction, selfing, and apomixis (parthenogenic seed production), all ofwhich can vary among and even within species. Polyploidy is also quite common inflowering plants (Otto & Whitton 2000; Jiao et al. 2011), with some genera exhibitinga range of ploidies, for example, North American Crepis range from diploid to as highas decaploid (Sears & Whitton 2016). This widespread variation, coupled with theimmense diversity of flowering plants in general (∼300,000 species; Christenhusz &Byng 2016), provides all but limitless contexts in which to test the myriad potentialdrivers of GP. This thesis employs one of the classic plant GP systems included inthe seminal review of Bierzychudek (1985), Townsendia hookeri (Townsend’s EasterDaisies; figure 1.1), to explore several of the most prominent theories described above.Townsendia is a charismatic genus in the sunflower family (Asteraceae). Firstdescribed as a monotypic genus by Hooker (1840), Townsendia expanded over thedecades with the botanical work of Gray (1888), Larsen (1927), and Beaman (1957b).Beaman’s work constitutes the first comprehensive investigation of the genus from anevolutionary perspective, and incorporates field studies, crossing experiments, herbar-ium specimens, and important cytological work. Beaman was the first to extensivelycharacterize ploidy and mating system variation in Townsendia, finding widespreadpolyploidy and asexual seed production via apomixis amongst many of the speciesin the genus (Beaman 1954). In addition, he was responsible for clarifying confusingtaxonomy concerning T. exscapa and T. sericea (they were included together in a10mixed herbarium voucher), which resulted in him renaming T. sericea to T. hookeriin honour of William Hooker (Lee 2015).Figure 1.1: Townsendia hookeri growing in its natural habitat.As with a number of other species in Townsendia, Beaman (1957b) found thatT. hookeri comprised diploid and polyploid forms, with the diploids being obligatelyoutcrossing (self-incompatible) and the polyploids reproducing via autonomous ga-metophytic apomixis (producing seed asexually without the need for pollen). He alsoinferred that polyploids were of autopolyploid origin, and that parthenogenic em-bryo development was precocious, suggesting that apomixis is obligate without muchchance for sexually produced ovules. Through his field and herbarium work, Beaman11was the first to identify that the divergent ranges of sexuals and apomicts in T. hook-eri exhibit a classic pattern of GP. T. hookeri is a long-lived perennial, and its twoforms are macro-morphologically indistinguishable despite having different ploidies.The species as a whole occurs primarily along subalpine zones surrounding the RockyMountains of North America, with a small disjunct range in the Yukon Territory. Pop-ulations typically consist of scattered individuals over somewhat open areas, such asrocky outcrops and hogback ridges, and often co-occur with succulents (Coryphanthaand Sedum spp.), Phlox spp., sagebrush (Artemisia spp.), and other tough peren-nials. While sexual populations are situated along Colorado and Wyoming’s FrontRange, apomictic populations are distributed from southern Wyoming to as far northas British Columbia. Though the sexual and apomictic ranges overlap, no mixedpopulations had been identified before the work of this thesis began (see Chapter 2).Since Beaman’s (1957b) monograph, Townsendia was not subject to much conceptualor experimental work (aside from the addition of new species) until it was revitalizedby Jeannette Whitton and her lab members in the early 2000’s.Building on the foundation laid down by Beaman, the Whitton lab has investigatedpatterns of GP in Townsendia hookeri using several approaches. Using chloroplast se-quence data from populations across the species range, Thompson & Whitton (2006)confirmed that apomixis has originated multiple times and subsequently spread fromsexual populations that were likely isolated to glacial refugia, reflecting a pattern thatis seen in many apomictic plants (Brochmann et al. 2003). Utilizing crossing exper-iments, Garani (2014) confirmed that pollen from apomicts can successfully fertilizesexual ovules, and that these “heterospecific” crosses can negatively impact sexualseed set. Furthermore, flow cytometric analyses of the offspring from these crossessuggest that both diploids and polyploids were produced, which (along with the neg-ative impact on seed set) points to the potential role of asymmetrical reproductive12interference in limiting sexual range expansion (Kyogoku 2015). Most recently, Lee(2015) used rangewide occurrence records and environmental niche modeling to char-acterize the niches of sexual and apomictic populations in T. hookeri. Their nichemodels predict that while the occurrence of apomicts is largely in agreement withtheir current distribution, suitable habitat for sexuals exists within the apomicticrange that they are not inhabiting. These results suggest that dispersal limitationmay play a role in driving patterns of GP in the system.1.3 Thesis outline and objectivesThe work presented in this thesis continues the excellent work achieved by my lab-mates and further explores the drivers of geographical parthenogenesis in Townsendiahookeri. I evaluate several potential mechanisms for range divergence between sexualand apomictic forms, guided by the rich realms of theory discussed above. Ratherthan focusing on identifying the single most important driver of GP, I take particularcare throughout to consider the ways in which all of these processes may interactto influence the distributional patterns we see today. My hope is that the followingchapters will inform the greater biological community beyond the view of Townsendiahookeri, and help inspire more integrative approaches to investigating complex eco-logical and evolutionary processes.In Chapter 2, I use novel population genomic techniques to analyze the structure ofseveral sexual and apomictic populations across the range of Townsendia hookeri. Inparticular, assessing the diversity of clones within and among apomictic populations(e.g. the number of clonal lineages per population and the occurrence of widespreadclonal lineages that are found in multiple populations) helps us draw conclusions abouthow different selective pressures may have shaped the historical spread of apomictsin the system.13In Chapter 3, I employ modern reciprocal transplant experimental design to ex-plore potential drivers of range divergence between sexuals and apomicts. Dispersallimitation is an important putative factor for limiting sexual ranges in a GP context,while apomicts may be limited more by ecological/biotic factors. Transplanting sex-uals and apomicts into each others’ ranges and assessing their fitness allows us tocompare within vs beyond the range performance for both mating types, which aidsin evaluating the factors that limit their respective ranges.In Chapter 4, I investigate differences in early life history traits between sexualsand apomicts that may contribute to patterns of GP. Given that apomicts have amuch broader geographic range than sexuals, it is possible that they possess traitsimportant for colonization (i.e., improved germination success and dispersal abilityof seeds in comparison to sexuals), though these may trade-off with traits that areimportant for establishment and competition.14Chapter 2Clonal Population GenomicStructure of Polyploid ApomicticEaster Daisies2.1 IntroductionSex is prevalent in nature despite its costs (i.e. recombination load, the cost ofproducing males, the two-fold cost of sex), while asexuality is rare despite its de-mographic and genetic advantages (Bell 1982; West et al. 1999). A widely acceptedexplanation for this paradox is that meiotic recombination provides benefits to adap-tive potential that eventually overcome the costs of sex, while the genetic inflexibilityof asexuals limits their response to selection and results in them being “evolution-ary dead-ends”. However, in systems that have closely related sexuals and asexuals,asexuals often have larger ranges (i.e. geographical parthenogenesis; Vandel 1928;Glesener & Tilman 1978; Bierzychudek 1985). This indicates that asexuals have theability to establish on the landscape and out-perform their sexual progenitors, at15least in the short term (though there are examples of long-lived parthenogenic lin-eages; Schon et al. 1998; Welch & Meselson 2000). Systems that show patterns ofgeographical parthenogenesis (GP) are therefore especially valuable for determiningthe conditions that allow asexuals to flourish where sexuals are not found.The primary explanations for patterns of GP focus to varying degrees on the de-mographic benefits of uniparentality (Baker’s Law), ecological differentiation betweensexuals and asexuals, the genetic consequences of parthenogenesis, and differences inhow sexuals and parthenogens respond to selection (Tilquin & Kokko 2016). Distin-guishing among these explanations requires teasing apart the ecological and geneticcomponents contributing to the range advantage of asexuals. For example, contem-porary clone structure and diversity can give indications of how selection has actedon clonal lineages. We might expect that surviving asexual lineages succeed becausethey perpetuate trait combinations that are well suited to their environments; anequally fit sexual genotype would be subject to recombinational breakdown of theirbeneficial suite of alleles. The idea that asexuals are “protected” from this geneticcost of sex (Lynch 1984) has given rise to two opposing (but not mutually exclusive)hypotheses concerning the success of parthenogens in GP contexts: frozen niche vari-ation and general-purpose genotypes (Vrijenhoek & Parker 2009). The frozen nichevariation hypothesis asserts that as multiple clonal lineages arise from a diverse sex-ual background, selection favours clones that best partition the available niche space,resulting in the fixation of an array of specialized clones that have reduced niche over-lap with already established sexuals and asexuals (Vrijenhoek 1979). On the otherhand, the general-purpose genotype hypothesis states that selection in a fluctuatingenvironment will fix a clonal genotype that is successful in many different conditions,an idea first introduced by Baker (1965) when describing traits associated with weedyplant species.16Though these predictions do not appear to be explicit, researchers often interpretthe presence of widespread clones (and generally low clonal diversity) as evidence forgeneral-purpose genotypes, and high clonal diversity (both within and among popu-lations) as evidence for frozen niche variation. There is mixed empirical support forthese hypotheses in the literature, however (reviewed in Vrijenhoek & Parker 2009).Providing direct support for either hypothesis is difficult, as one must show both popu-lation genetic (i.e., presence of either widespread or partitioned clones) and ecological(i.e., clones with wide vs narrow niches) support. To make matters more complicated,there is no widely agreed-upon definition of a “clone”, and researchers use differentclonal concepts and make different interpretations depending on the resolution of themarker system used (Martens et al. 2009). The fact that both widespread clones andpolyclonal populations have been found indicates that these two hypotheses are notcompeting hypotheses that preclude each other, but instead represent two extremes ofa continuum along which asexuals fall depending on each system’s unique attributes.Frequency of asexual origins, proximity to sexual progenitors, dependence on spermor pollen, and the competitive regime facing newly arisen asexuals are all thoughtto be important factors shaping the diversity and spread of parthenogenic lineages(Vrijenhoek & Parker 2009).The frozen niche variation model focuses on the importance of direct competitionwith sexual progenitors (and highly similar clones) in influencing niche diversification,and hence requires a varying array of clonal genotypes for selection to act on (Rough-garden 1974; Vrijenhoek 1979; Bell 1982). For this reason, frozen niche variation isoften associated with the need for recurring origins of asexual lineages from extantsexual progenitors in order to efficiently partition the available niche space and with-stand strong competition (though the generation of too many clones is predicted tolead to exclusion of sexuals; Weeks 1993). The general-purpose genotype model, on17the other hand, is associated with parthenogens that escape competition with theirsexual progenitors; none of the widespread clones reviewed by Vrijenhoek & Parker(2009) required co-occurring sexuals (i.e., were sperm dependent). Instead, thesewidespread clones are typically considered weak competitors, shunted into “marginal”habitats where biotic interactions are generally less intense (Baker 1965), but whereconditions vary over time. In these environments, genotypes that can survive in arange of conditions may be able to persist, whereas highly specialized clones mightbe eliminated under fluctuating conditions (Parker et al. 1977). Under models of GP,while the generation of variable clonal genotypes is still required, the assumption isthat only a few successfully establish and become widespread. Given that GP sys-tems often comprise a broad geographic range (with sexuals inhabiting only a smallportion), it stands to reason that both patterns (i.e., the existence of both widespreadgeneralist and narrow specialist clones) may be evident in different parts of the rangedepending on the competitive regime and proximity to sexual populations.Other genetic attributes are thought to play important roles in patterns of GP,including polyploidy, hybridization, and levels of heterozygosity (Bierzychudek 1985;Haag & Ebert 2004; Kearney 2005). Polyploidy has been linked to rapid diversifi-cation and the production of genotypes that may have increased invasive potential(Comai 2005; Te Beest et al. 2012). In general, benefits conferred by polyploidyor hybridization will be conserved by asexual reproduction, suggesting that theseprocesses may work in concert to generate patterns of GP. Similarly, asexuals arehypothesized to be protected from inbreeding depression, maladapted gene flow, andloss of heterozygosity (Haag & Ebert 2004), all of which (when combined with thedemographic benefits of uniparentality) may contribute to the distributional successof asexual lineages.In plant GP systems, the majority of parthenogens are polyploid apomicts (pro-18ducing seeds asexually) often with hybrid ancestry (Whitton et al. 2008). To date,few studies of GP in plants have investigated the structure and diversity of apomic-tic populations, but instead focused on comparing the ranges/niches of sexuals andapomicts, the distribution of apomictic cytotypes, or deeper phylogenetic origins ofapomictic lineages. Existing evidence, however, seems to provide greater support forfrozen niche variation than general purpose genotypes. Literature surveys of clonaldiversity in plants (though mainly from vegetative parthenogens) indicate that mostasexual populations are polyclonal and that widespread clones are rare (Ellstrand& Roose 1987; Wide´n et al. 1994; Horandl & Paun 2007; Silvertown 2008). Silver-town (2008) found that while apomicts did have lower clonal diversity than vegetativeclones, they rarely consisted of monoclonal populations. Similarly, surveys of clonaldiversity in model plant GP systems such as Taraxacum (Lyman & Ellstrand 1984;Menken et al. 1995; Van Der Hulst et al. 2003) and Ranunculus (Cosendai et al.2013) suggest that populations are rarely if ever monoclonal, but rather have remark-ably high levels of genotypic diversity. Support for general-purpose genotypes, onthe other hand, has largely come from ecological studies showing that apomicts havea broader niche breadth than sexual progenitors (e.g. in Antennaria; Bierzychudek1989), though Coughlan et al. (2017) interpret the presence of widespread clones (butnot monoclonal populations) in Crataegus as evidence for general-purpose genotypes.Regardless of the trends found in the few well-studied plant GP systems, more evi-dence is required before generalizations can be made about the diversity and spreadof apomictic lineages.Townsendia hookeri (Asteraceae) is a perennial flowering plant species with diploidsexual and polyploid apomictic forms that display a classic pattern of GP (Bierzy-chudek 1985). While the apomictic range dwarfs that of the sexuals, the two formscome into contact in a small region of range overlap (Lee 2015). Sexuals are obligately19outcrossing, and apomicts set seed autonomously (i.e., they are non-pseudogamousand therefore do not require pollination to initiate embryo development). As a re-sult, they can colonize new sites without the need for mates or pollinator services(Beaman 1957a; Garani 2014). As part of a broader research program that aims todetermine the conditions under which apomicts established a distinct range along-side their sexual progenitors, we used double-digest restriction-site associated DNA(ddRAD) sequencing to explore the genetic structure of sexual and apomictic pop-ulations sampled throughout the range in order to address the following questions:(1) Is there evidence for widespread clones with monoclonal populations or an ar-ray of clones with polyclonal populations (or both)? (2) What are the relationshipsbetween sexual and apomictic populations, and can we infer apomictic origins fromsexual populations/regions? (3) Does genetic diversity differ between sexuals andapomicts? We use our findings to explore the predictions of frozen niche variationand general-purpose genotype hypotheses in the context of how selective pressureshave shaped clonal diversity in Townsendia hookeri.2.2 Materials and methods2.2.1 Study systemTownsendia hookeri consists of sexual individuals that are self-incompatible (obli-gately outcrossing) and apomicts that set seed autonomously (non-pseudogamous).Sexual populations have a smaller range than apomicts and primarily occur betweenBoulder, CO and Laramie, WY, while apomictic populations range from southernWY along the eastern Rocky Mountains to British Columbia (Lee 2015). A smalldisjunct distribution of diploid-sexual and polyploid-apomictic populations occurs inthe Yukon Territory (Thompson & Whitton 2006; Garani 2014). The majority of20apomictic populations are triploid, with the only known tetraploid populations oc-curring in the Yukon territory. Before this study, no mixed populations containingboth sexual and apomictic individuals had been detected, though there is a rangeof overlap (centered around Laramie, WY) where both types of populations can befound. The genome size is estimated to be quite large in T. hookeri (diploids ≈ 7.3gb). While not much was known previously about the species’ population geneticstructure, phylogenetic analysis of plastid haplotype variation indicates a minimumof four origins of apomixis (Thompson & Whitton 2006).2.2.2 Population samplingWe sampled leaf tissue from 27 populations across the range of T. hookeri, including12 populations previously identified as sexual diploids and 15 identified as apomicticpolyploid populations (Table 2.1). We sampled populations most densely aroundthe core of the species range (centered around Laramie, WY), but also includedpopulations on the periphery in ND, BC, SK, and YT. Because of our focus ondescribing the structure and distribution of clones, we aimed to sample 5 individualsper apomictic population (except for populations L41 and S03, for which only 4individuals were available), and 3 individuals per sexual population, for a total of 114individuals. We sampled individuals somewhat haphazardly within populations, butmade sure to spread our sampling spatially across each site. We dried field-collectedleaf tissue using silica gel, and then stored it in a -80◦C freezer. The ploidy levelof populations used in this study was either known previously or confirmed usingflow cytometry (Lee 2015, unpublished). Mixed-ploidy populations were previouslythought not to occur, but flow cytometric analyses of one of the populations usedin this study (L62; previously characterized as diploid only) revealed that it harborspolyploids at low frequencies. Ploidy assignments were verified using SNP data (see21Section 2.2.5 “Data analysis” below).2.2.3 ddRAD library preparationWe extracted DNA from frozen leaf tissue using a modified version of the protocolof Murray & Thompson (1980). We assessed the quality of DNA using a NanoDropspectrophotometer and DNA quantity using a Qubit 2.0 flourometer (Thermo FisherScientific, Waltham, MA, USA). We used this DNA to make one ddRAD (a.k.a.genotype-by-sequencing) library containing 114 individuals. We included DNA fromeach polyploid-apomictic individual twice in the library in order to increase the se-quencing coverage, as higher read depth is required to confidently identify variants inpolyploids (Dufresne et al. 2014); this had the added benefit of providing replicateswhich allowed us to estimate sequencing and genotyping error rates (see Section 2.2.5).We created the PstI-MspI ddRAD library using a modified version of the protocol de-scribed in Poland et al. (2012). In summary, we first digested the DNA with HF-PstIand MspI at 37◦C for 5 hours, then ligated barcoded adapters and common adaptersto digested DNA at 22◦C for 3.5 hours. Following ligation, we cleaned and concen-trated samples using SPRI magnetic beads, then amplified by PCR using KAPA HiFiHotstart master mix (Kapa Biosystems, Wilmington, MA, USA). We then pooled allsamples together into a single library using normalized concentrations, and selectedDNA fragments between 300-450 bp using gel size selection. We checked the qualityof the completed library using qPCR and Bioanalyzer (Agilent, Santa Clara, CA,USA) before sequencing on one lane of Illumina HiSeq 2500 paired-end 125bp plat-form at Ge´nome Que´bec (Montre´al, Que´bec, Canada). The sequencing resulted in∼270 million reads with an average quality score of 35.222.2.4 De novo assembly and SNP callingWe de-multiplexed the raw reads, concatenated the polyploid-apomictic replicate files,and performed de novo assembly and SNP calling with dDocent v2.7.8 (Puritz et al.2014). dDocent is a bioinformatic pipeline that combines several existing softwarepackages and is designed specifically for efficient assembly and SNP calling of paired-end RAD data in non-model organisms. Notably, dDocent uses freebayes (Garrison &Marth 2012) to call SNPs, which is capable of processing both diploid and polyploiddata. In order to reduce the potential for recently derived paralogs to influencethe quality of the assembly, we only included previously identified diploid-sexualindividuals when constructing the reference catalogue. We also excluded the YTsexual population (C59) from the assembly, because preliminary analyses showedit to be highly divergent from the other sexual populations. We used a referenceoptimization script provided by Puritz (2019) to choose assembly parameters, usingonly reads that had a depth of at least 4 within individuals and were found in at least 4individuals, and a clustering similarity of 90%. After creating the reference catalogue,we used it to call SNPs on all individuals, outputting the results in a variant call file(VCF). We re-ran the pipeline using un-concatenated polyploid-apomict files in orderto estimate error rates between replicate samples.We filtered the resulting VCFs using bcftools (Li et al. 2009) and vcflib (Garrison2019) following the dDocent SNP filtering tutorial (Puritz 2019). First, we filteredSNPs to a minimum quality score of 30, minor allele count of 3, minor allele frequencyof 5%, and only kept SNPs genotyped in 95% of the individuals. We also filteredSNPs with an average depth below 20 (set high to ensure confident SNP calling inpolyploids) and above 132.5 (exluding paralogs and multicopy loci). We performedfurther filtering steps to remove SNPs likely to be the result of sequencing errors,23paralogs, multicopy loci or artifacts of library preparation. These include: filteringbased on allele balance (removed loci for which the less common SNP variant wasbelow a frequency of 0.125 and above 0.875), removing SNPs found on both forwardand reverse reads, filtering based on ratio of mapping qualities between referenceand alternate alleles (below 0.9 and above 1.05), and filtering based on ratio of locusquality score and depth (removing any locus that has a quality score below 1/4 ofthe depth). We removed indels and other complex variants, and only kept bi-allelicSNPs. Our filtering steps resulted in a final VCF containing 16,573 SNPs.2.2.5 Data analysisWe analyzed clone structure in the SNP dataset using the packages poppr v.2.8.3(Kamvar et al. 2014, 2015) and adegenet v.2.1.2 (Jombart 2008; Jombart & Ahmed2011) in R v.3.6.1 (R Core team 2019). We imported the VCF into R using the vcfRpackage (Knaus & Gru¨nwald 2016, 2017), and converted the data into a “genind”object (used by adegenet and poppr), which allows individuals to be coded asdifferent ploidy levels. We calculated pairwise prevosti distances (Prevosti et al.1975) between each individual and used poppr’s mlg.filter function to determine thenumber of multi-locus genotypes (MLGs) present in the dataset. This is accomplishedby visualizing the distribution of pairwise genetic distances and choosing a distancethreshold indicated by a gap in the distribution; we chose a distance cutoff of 0.1,which grouped individuals into the same MLG (i.e., clone) if they had a geneticdistance of 0.1 or less (Figure 2.1A). This cutoff is somewhat higher than the averagepairwise distance between apomictic replicate samples (0.027), which served as ourestimate of genotyping error rate. Error rates between replicates were largely below0.05, but four replicate pairs had especially high error rates (between 0.075 and 0.10).As discussed below (Section 2.3.1), whether we exclude these samples as outliers and24Figure 2.1: Density plots of pairwise prevosti distances (A) between all apomicticindividuals and (B) between all sexual individuals. In panel (A), the red lineindicates the mean distance between replicates (our estimation of genotyping errorrate) and the black line indicates our genetic distance threshold for MLGdesignations (where any individuals with a genetic distance less than 0.1 weregrouped into the same MLG).use a lower cutoff, or include these samples and set a higher cutoff, does not changeour interpretation of the number of MLGs.We used discriminant analysis of principal components (DAPC; Jombart et al.2010) in adegenet to investigate patterns of genetic structure and compare clus-tering of apomictic individuals to the distance-based MLG designations. First, we25estimated the number of groups within the apomicts using K -means clustering, whichuses principal components analysis (PCA) and then determines clusters using dis-criminant analysis without prior population membership. To accomplish this, weused the find.clusters algorithm to identify clusters and explored a range of K -values(corresponding to the number of clusters) that had the lowest Bayesian informationcriterion (BIC). We also used DAPC with prior population membership to exploregenetic diversity between all populations using the dapc function. We used the xval-Dapc function to select the optimal number of principal components (PCs) to includein this analysis.To further explore the relationships between sexual and apomictic populations, webuilt neighbor-joining trees of individuals and populations. We used prevosti distanceto construct a tree displaying all individuals, and Nei’s distance (Nei 1972) to showrelationships among populations. We used poppr’s aboot function to generate treeswith 1000 bootstraps each and used the ape package (Paradis & Schliep 2019) tovisualize.We used the hierfstat package (Goudet 2005) to calculate overall and pairwiseFst values for sexual populations and tested for structure between populations usingthe G-statistic test with 1000 bootstraps. We used poppr to calculate observed het-erozygosity and a Kruskal-Wallis test to test for differences in observed heterozygositybetween mating systems.We confirmed the ploidy of individuals using a method described by Knaus &Gru¨nwald (2018), where distributions of allele balances of heterozygotes are comparedto expected patterns at each ploidy level. A peak in an allele balance histogramat 0.5 is expected for diploids, while peaks at 0.33/0.66 are expected for triploidsand 0.25/0.5/0.75 for tetraploids. One individual that was originally thought to bepolyploid (L45 1, in an otherwise polyploid population in the apomictic part of the26range) was identified as diploid from allele balance plots (Figure 2.7).2.3 Results2.3.1 Clone structureGenetic distance thresholds delineated 10 MLGs amongst the 73 apomictic individualsin the dataset. Most populations were inferred to be monoclonal, with just twopopulations (L39 and L16) having representatives from two MLGs (Figure 2.2). Twoclonal genotypes were widespread, each occurring in multiple, geographically clusteredpopulations. The most commonly detected MLG was spread across 7 populationsin central WY and MT (shown in blue on Figure 2.2) and comprised 33 sampledindividuals. Mixed populations (consisting of multiple MLGs or both sexual andapomictic individuals) were mostly located in the region where the ranges of sexualsand apomicts overlap. Four individuals had higher within-MLG pairwise distancesthan the others (see the small distribution between ∼ 0.05 and 0.1 in Figure 2.1A)due to having higher error rates. Excluding these samples as outliers and setting theMLG cutoff to 0.055 resulted in the same number of MLGs as the current setting of0.1.K -means clustering analyses largely support the distance-based MLG designa-tions, with apomictic individuals from the same MLG being clustered, and generallyshowing the same geographically structured groups (Figure 2.3C). Notably, the twowidespread clones (blue and sky blue) were recovered under values of K from 8-11(at K=12 individual L16-3 is assigned its own group). Posterior membership proba-bilities were mostly 1.0, except for the large widespread MLG (blue, K=9-12). Thismay indicate sub-structure within that clonal lineage, though it more likely reflectsthat those values of K were too high and that individuals were being assigned to two27Figure 2.2: Map displaying populations sampled and the clonal structure ofapomictic populations in T. hookeri. Circles represent apomictic populations,squares represent mixed sexual-apomictic populations, and triangles representsexual populations. Colored pies indicate the MLG designation based on geneticdistance thresholds (MLG cutoff = 0.1). White shading in pies L45 and L62indicate sexual individuals in mixed populations.groups that were not substantially different from one another (see overlap of blue,lilac, and purple groups when K=9-12; Figure 2.10).DAPC of all individuals showed 7 distinct clusters of apomictic individuals, andoverall depicted a similar pattern to the genetic distance and K -means results (Fig-ure 2.5). Individuals from the two widespread clones formed distinct groups, andall other monoclonal populations were well-differentiated from each other. Individ-uals from populations designated as polyclonal by genetic distance thresholds weregrouped together in the DAPC, which suggests that these clonal lineages share a28Figure 2.3: K -means clustering results for apomictic populations. (A) Plot of BICvalues under a range of K values. (B) DAPC showing differentiation of clusterswhen K=8. (C) Plots of posterior membership probabilities of group assignmentsfor K=8-12. Individuals are grouped by population, and shared colors indicateindividuals with the same group membership.recent evolutionary history. L62-5 (an apomictic individual from a majority-sexualpopulation) grouped closely with the individuals in population L39, despite beingmuch closer geographically to populations L06, L16, and L17 (Figure 2.2).2.3.2 Genetic structure of sexual populations andrelationships with apomictic lineagesSexual individuals generally had greater pairwise genetic distances between individ-uals within populations than we observed between apomicts within MLGs, the one29Figure 2.4: Neighbor-joining tree of all sexual and apomictic individuals based onpairwise prevosti distances. Apomictic individuals are represented by circles whichare colored by their MLG designations. Sexuals are represented by triangles whichare colored according to whether the populations are in Colorado, near Laramie,WY., or the Yukon Territory. Bootstrap values are displayed for the nodes along thebackbone of the tree.30Figure 2.5: Discriminant analysis of principal components (DAPC) of all sexualand apomictic individuals. Linear disciminants 1 & 2 are plotted. Circles representapomictic individuals, triangles represent sexual individuals. Individuals are coloredby their MLG designations (apomicts) or their geographic location (sexuals).exception involving the disjunct sexual population found in YT (C59), which hadlower pairwise distances (∼ 0.05-0.07; Figure 2.1B). Sexual populations were differ-entiated into 3 groups geographically situated around Boulder, CO (black), Laramie,WY (gray), and YT (gold) (Figure 2.2, Figure 2.4, Figure 2.5). Pairwise Fst amongsexual populations averaged 0.166 and the G-statistic test indicated significant pop-ulation structure (P = 0.01). Pairwise Fst values ranged between 0.087 and 0.371(Figure 2.9, Table 2.2), with the higher values being between the YT (C59) and thepopulations in the core of the range.31Neighbor-joining trees suggest at least two origins of apomictic lineages, with onebeing most closely related to the lineage represented by the YT sexual populationand the other emerging from the Laramie sexual clade (Figure 2.4, Figure 2.8). Theperipheral apomictic populations found in Canada (and the YT sexual population)were well-differentiated from the others but most closely related to the Boulder sexuals(albeit with lower bootstrap support; 81.7). The clonal lineages in the apomicticinterior range were more closely related to the sexuals found in Laramie.2.3.3 Genetic diversity of sexuals and apomictsAverage observed heterozygosity was higher amongst apomictic individuals (0.425)than amongst sexual individuals (0.215; χ2=76.831, P=2.2x10-16, Figure 2.6). YTsexual individuals (C59) had the lowest observed heterozygosities, while individualswith the highest values were part of MLGs spread across multiple clonal lineages.2.4 Discussion2.4.1 Clone identificationThe identification of clonal lineages requires careful consideration of several factors,including the marker system used, the number of markers, the use of distance/similar-ity cutoffs, and what clonal concept is being employed (Martens et al. 2009). Markersystems vary in sensitivity, such that the type of marker used (e.g., dominant vscodominant) and their rates of evolution (e.g., AFLPs vs microsatellites) can have alarge impact on the estimation of clonal diversity (Horandl & Paun 2007). In addition,if clones are being defined as individuals with 100% identity in whatever marker sys-tem is used, then the number of clonal genotypes inferred is expected to rise with thenumber of markers scored. Given some probability for scoring errors (Douhovnikoff32Figure 2.6: Boxplots of observed heterozygosity in sexual and apomicticindividuals. Jittered points represent individual means, and points with error bars(to the right) indicate means (± s.d.) for each mating system.& Dodd 2003), the possibility of overestimating the number of clones will also in-crease with the number of markers used (Arnaud-Haond et al. 2007). For this reason,Arnaud-Haond et al. (2007) recommend using cutoffs to allow small amounts of ge-netic distance when designating clones; this is done by creating histograms of pairwisegenetic distances between individuals and using a “hump” in the distribution (repre-senting individuals with very low distances) to guide the cutoff. The use of cutoffsfor clone identification is crucial when using high-throughput sequencing (HTS) ap-33proaches such as ddRAD, because the elevated number of markers (and associatederror-rates) would cause every individual to be assigned a unique MLG without thecareful application of distance thresholds. However, we note that in taking this ap-proach, we will likely group together clones that are differentiated by a small numberof somatic mutations. As a result, MLGs as we define them might be best thought ofas clonal lines (sometimes referred to as multi-locus lineages) rather than strict clones.With that in mind, our replicate sequencing of apomictic individuals helped us verifywhich humps were due to sequencing artifacts or genotyping errors and which onesrepresent true genetic differences.Due to the variation in marker systems and statistical methodology used, it isdifficult to compare measures of clonal diversity between studies. The fact that someresearchers use a cutoff approach while others do not highlights a definitional prob-lem, in that some use a “molecular” clonal concept where clones are defined as groupsof individuals with 100% sequence similarity, while others use a “phylogenetic” clonalconcept where a clone is defined as a monophyletic cluster of individuals that aregenetically very similar (see Martens et al. 2009 for a review of clonal concepts).Somatic mutations are common in plants (Horandl & Paun 2007), which must beconsidered when analyzing asexual populations using modern marker systems thatgenerate thousands of SNPs. To our knowledge, the use of HTS to investigate thepopulation genetic structure of clonal plants has been limited, and so far does notappear to have been applied to apomictic plants (but see Bock et al. 2018 who usedddRAD to identify clones in vegetatively propogating Helianthus tuberosus using sim-ilar cut-off methods). Given the volume of data and potential for somatic mutationsand genotyping error rates, researchers should strongly consider the use of appropriatecutoffs when using HTS to investigate clone structure. Our study shows that ddRADcan provide the resolution to identify clones with a degree of certainty that may not34be possible with other marker systems (see Figure 2.1A which shows a multimodaldistribution of pairwise distances with a clear distinction between humps). The costof ddRAD and other HTS technologies continues to decrease, allowing the generationof very rich datasets for non-model organisms (Matz 2018). For those with access toa reference genome, HTS allows the investigation of in-depth patterns of genome evo-lution that are of interest to those studying apomictic plants (e.g., the accumulationof mutations in Boechera; Lovell et al. 2017).2.4.2 Clone structure in Townsendia hookeriAll of our analyses point to the presence of widespread clones and largely monoclonalpopulations in apomictic T. hookeri, a pattern that appears to be rare in asexual plantpopulations. Consistent with expectations under general-purpose genotype models,we found that 13 out of 15 populations were monoclonal, and that over 60% of oursampled apomictic individuals belonged to one of two widespread clones. Perhapsmost striking is our finding of one geographically widespread clone (found in 7 outof 15 populations and 33 out of 73 individuals), which was the only clonal lineagedetected in all populations sampled in Montana and northern Wyoming. While thesample sizes from apomictic populations were relatively small (5 individuals) anddenser sampling may have unveiled additional clonal lineages, the general pattern oflow intrapopulation diversity (as well as the broad spatial patterns) are likely to berobust to additional sampling.Townsendia hookeri has multiple characteristics that have been hypothesized tobe associated with conditions that would favor the general-purpose genotype model.Apomicts are autonomous and have no dependence on pollen or pollinators, attributeswhich have likely allowed them to disperse long distances and escape competitionwith their sexual progenitors. Evidence provided by Beaman (1957a) indicates that35embryo development is precocious in Townsendia, which suggests that apomixis isobligate and that the potential for sexually produced ovules is low. All else beingequal, obligate apomicts are expected to have lower clonal diversity than facultativeapomicts (Horandl & Paun 2007). In addition, the apomictic range of T. hookeri isconsistent with high abiotic stress, low competition (populations are mainly foundon rocky soils with sparse vegetation), and small population sizes in comparison tosexuals, all of which is in line with Baker’s (1965) assertion that general-purposegenotypes are likely to be weak competitors excluded to “marginal” environments.Our finding of broadly distributed clones stands in stark contrast with what hasbeen found in other apomictic species. While monoclonal populations have been foundin several well-studied apomicts, most populations have been found to be polyclonal,with the vast majority of genotypes being restricted to a single site (Horandl & Paun2007). As discussed above, some of this incongruence may be due to the adoption ofdifferent clonal concepts and variation in marker systems, but these studies remainour only point of comparison. In a similar study to our own, Cosendai et al. (2013)found that almost every sampled apomictic individual of Ranunculus kuepferi had aunique MLG. Although the two systems have many similarities (both have apomicticautopolyploids and self-incompatible sexuals with comparatively small ranges), thereare some important differences, most notably that R. kuepferi is a pseudogamousfacultative apomict (with partial sexuality found in ∼ 1/3 of sampled seeds; Cosendai& Ho¨randl 2010). Cosendai et al. (2013) suggest that the high clonal diversity inR. kuepferi is the result of facultative sexuality and multiple long-distance dispersalevents. Similarly, occasional sex is believed to have led to the high genotypic diversityobserved within some apomictic microspecies of Taraxacum (Van Der Hulst et al.2003), though there appears to be considerable variation in clonal diversity betweenmicrospecies (Majesky´ et al. 2015). These studies highlight how the rates of sex36(which are linked to the frequency with which new clonal lineages are generated) canimpact the clonal diversity of apomictic plants.2.4.3 Origins and spread of apomictic lineagesDiploid-sexual individuals clustered into 3 groups geographically, corresponding topopulations situated around Boulder, CO, Laramie, WY, and YT. Our neighborjoining tree (Figure 2.4) indicates that apomictic lineages in the core of the range(WY, MT, and ND) were most closely related to the sexuals found near Boulderand Laramie, while some of the apomictic lineages found in the range periphery (SK,BC, and YT) were most closely related to the YT sexuals. These results suggest atleast two deep origins of apomictic lineages in Townsendia hookeri, with clones orig-inating in YT spreading south and those originating in Boulder/Laramie spreadingnorth. This is consistent with the previously proposed scenario involving post-glacialdispersal of apomictic lineages originating from sexual populations found in glacialrefugia (Thompson & Whitton 2006), a pattern that is often associated with GP(Bierzychudek 1985; Kearney 2005). Apomicts were likely able to colonize previouslyglaciated habitat relatively quickly due to reproductive assurance (i.e., Baker’s Laweffects) and improved dispersal potential (Chapter 4), while sexual expansion wasarrested due to dispersal limitation and/or being excluded from habitat already col-onized by apomicts (Mogie 1992; Hewitt 2004; also see evidence for asymmetricalreproductive interference between sexuals and apomicts in T. hookeri; Garani 2014).Demographic bottlenecks that would likely have occurred as the wave of colonizationproceeded would not have led to genetic bottlenecks in apomicts, given that apomictsare protected from loss of heterozygosity (Figure 2.6). In contrast, for colonizing sex-uals, bottlenecks could easily have resulted in inbreeding, with the expectation (givena history of obligate outcrossing) of at least moderate inbreeding depression (Haag37& Ebert 2004). The disjunct sexual population found in YT (C59) had the lowestlevels of observed heterozygosity, which may indicate a history of inbreeding causedby genetic bottlenecks or the breakdown of self-incompatiblity mechanisms (Ho¨randl2010).It may not be possible to tie the origins of particular apomictic lineages to sexualpopulations given that the progenitor sexual populations may no longer exist or maynot have been sampled in this study. In our neighbor-joining tree (Figure 2.4), theonly instance of sexuals and apomicts grouping together involves the lone diploid-sexual individual (diploidy confirmed via allele balance plot; figure 2.7) found inpopulation L45, an otherwise apomictic population in a part of the range wheresexuals are normally not found. This putative sexual individual shares brancheswith the peripheral apomictic lineages - a somewhat paradoxical result given that itappears to be quite distantly related to the apomictic individuals found in the samepopulation. K -means clustering analysis (Figure 2.3) grouped apomictic lineagesfound in BC and SK together (K=8, populations C43 and S03), as well as lineagesfrom populations L16, L39, and L62 (all from polyclonal populations found nearthe core of the range; K=9-10, light green), which may indicate common origins.The latter were also tightly grouped in DAPC analyses (Figure 2.5) and overlappedwith sexual individuals from Laramie, which suggests these apomicts originated fromLaramie sexual populations.2.4.4 Beyond frozen niches and general-purpose genotypesVarious features of lineages with parthenogenic organisms may influence their diver-sity and geographic structure, including the frequency of origins, proximity to sexualancestors, dependence on sperm, and the competitive regimes that clonal lineagesface in time and space. Vrijenhoek & Parker (2009) point out that portraying frozen38niche variation and general-purpose genotypes as mutually exclusive hypotheses failsto acknowledge that the two models focus on different sources of fitness variation:the frozen niche variation hypothesis is often considered in the context of spatial fit-ness variation, while general-purpose genotype models focus on fitness fluctuations intime. This is likely an oversimplification - fitness can vary in both space and time -which underscores the need to move away from the false dichotomy of general-purposegenotypes vs frozen niche variation and towards more nuanced models of geographicalparthenogenesis (and asexual diversity in general).The predictions that general-purpose genotype and frozen niche variation hypothe-ses make about the population genetic structure of clones are unclear, at least in partbecause the two terms are inconsistently applied. Support for these hypotheses tendsto come in the form of either ecological or population genetic evidence, but rarelyboth. To add to the confusion, the term “general-purpose genotype” was first used todescribe the characteristics of weedy plant species (Baker 1965), so this term is alsoused in non-parthenogenic contexts. The existence of higher-than-expected clonal di-versity in many systems is invoked as evidence against general-purpose genotypes andsupport for frozen niche variation, even when the results are not consistent with eithermodel. Cosendai et al. (2013) state that, while their findings are more consistent withfrozen niche variation than general-purpose genotypes, the fact that apomicts in Ra-nunculus kuepferi do not show clonal population structure (i.e. every individual wasa different clone) suggests that ecological niches are not being “frozen” by specializedapomictic lineages. This begs the question: what are the specific predictions thatfrozen niche variation makes about the population genetic structure of asexuals? Itseems that either multiple clones per population or monoclonal populations (but nowidespread clones) have been interpreted as consistent with frozen niche variation.This is consistent with the definition of the model, provided that each clone occupies39a different niche. With the niche breadth lens, a widespread clone that occupies anarrow but broadly available niche would also fit the frozen niche variation model,but in the absence of the evaluation of niche breadth this pattern would likely beerroneously interpreted as a general-purpose genotype (Vrijenhoek & Parker 2009).Despite the opacity of the general-purpose genotype vs frozen niche variation dis-cussion, it can nonetheless be helpful to think about the conditions that are expectedto give rise to each pattern. In T. hookeri, the patterns of clonal diversity found inthe apomictic range indicate that a small number of lineages were successfully ableto establish and spread (and/or that clones originate infrequently). The existence ofa very widespread clone spread across several monoclonal populations (colored blue;Figure 2.2) is consistent with the hypothesis that this region was colonized by a singlesuccessful genotype. On the other hand, we found greater clonal diversity near thecore sexual range, which could suggest that in this region, clones are successful overnarrower conditions. Given the proximity of sexual populations, this distributionpattern also supports an origin of many of the apomictic lineages in this southern re-gion of their range. Thus we see patterns of diversity consistent with general-purposegenotypes in one part of the range and (at least as assessed by clonal diversity) frozenniche variation in the other. In the north, the lack of clonal diversity could reflecta history of strong selection excluding genotypes incapable of persisting in the harshconditions that characterize their habitat. Such “hard selection” is expected to leadto the evolution of a broad niche width, resulting in genotypes that can perseverein varying abiotic conditions but are sensitive to biotic pressures (Kenny 1996). Inthe south, less variable environments with more intense competition may have led tothe the fixation of multiple clones, which would be predicted to have comparativelynarrow niches.Patterns of geographical parthenogenesis are the result of complex ecological and40evolutionary dynamics between sexuals and their asexual descendants. General-purpose genotype and frozen niche variation models are complimentary frameworksthat, when taken together, provide predictions about how clonal diversity is shapedby the frequency of origins and the subsequent selective pressures that clonal lineagesface. Characterizing the genetic structure of clonality provides necessary insight intothe history of asexual spread, but these results can only be fully appreciated whenplaced within a context that includes information about the ecological performanceand dispersal potential of sexual and asexual forms.41Figure 2.7: Histograms showing allele balance of heterozygous loci. A) IndividualL45-1; the peak centered around 1/2 indicates that this individual is diploid. B)Individual L45-2; the peaks centered around 1/3 and 2/3 indicate that thisindividual is triploid.42Figure 2.8: Neighbor-joining tree based on pairwise Nei’s distance between allpopulations.43Figure 2.9: Neighbor-joining tree based on pairwise Nei’s distance between sexualpopulations.44Figure 2.10: DAPC plots showing differentiation of groups under different valuesof K from K -means clustering analysis of apomictic populations. When K=9-12,blue, lilac, and purple groups overlap, indicating that these groups are not welldifferentiated.45Table 2.1: List of Townsendia hookeri populations sampled.Mating System Population Ploidy State/Province Latitude Longitude Number sampledSexual B53 2 CO 39.8913 -105.2655 3B60 2 CO 40.0352 -105.3001 3B42 2 CO 40.1070 -105.2836 3B46 2 CO 40.5384 -105.1335 3B49 2 CO 40.8403 -105.3214 3Mixed L62 2/3 WY 41.1140 -105.5118 4/1L05 2 WY 41.1372 -105.2916 3L08 2 WY 41.1529 -105.3578 3L10 2 WY 41.1919 -105.3967 3L11 2 WY 41.2400 -105.4342 3L12 2 WY 41.2529 -105.4062 3C59 2 YT 62.1235 -136.2575 3Apomictic L06 3 WY 41.1430 -106.0375 5L16 3 WY 41.3065 -105.5216 5L17 3 WY 41.3864 -105.4735 5Mixed L39 2/3 WY 42.7833 -105.9000 1/4L41 3 WY 42.8087 -105.3778 4L45 3 WY 43.2566 -107.2798 5C87 3 MT 44.9827 -109.2657 5C86 3 MT 45.6619 -110.4752 5C88 3 MT 45.9108 -109.9078 5C85 3 MT 46.6505 -111.7200 5C23 3 ND 46.8788 -103.6716 5C27 3 MT 47.5500 -112.4667 5C43 3 BC 50.6057 -116.0595 5S03 3 SK 50.6526 -107.9437 4SM 4 YT 61.2461 -135.4625 546Table 2.2: Pairwise Fst values between diploid-sexual populations of Townsendiahookeri.B42 B46 B49 B53 B60 C59 L05 L08 L10 L11 L12 L13 L45 L62B42B46 0.165B49 0.203 0.180B53 0.132 0.161 0.199B60 0.156 0.188 0.226 0.146C59 0.346 0.328 0.343 0.337 0.371L05 0.192 0.171 0.159 0.189 0.217 0.319L08 0.177 0.155 0.148 0.177 0.204 0.306 0.126L10 0.167 0.146 0.139 0.166 0.191 0.295 0.121 0.111L11 0.163 0.141 0.139 0.160 0.189 0.297 0.119 0.106 0.102L12 0.176 0.152 0.145 0.173 0.199 0.301 0.125 0.119 0.111 0.105L13 0.177 0.156 0.143 0.175 0.201 0.302 0.127 0.115 0.109 0.105 0.107L45 0.272 0.257 0.257 0.265 0.295 0.293 0.242 0.229 0.219 0.208 0.230 0.229L62 0.152 0.133 0.125 0.153 0.173 0.272 0.111 0.098 0.094 0.087 0.094 0.096 0.18547Chapter 3Investigating Drivers ofGeographical Parthenogenesis inTownsendia hookeri Using aReciprocal Transplant Experiment3.1 IntroductionClosely related sexual and asexual taxa often have disparate ranges, with asexualstending to be more geographically widespread and/or found at higher elevations andlatitudes than related sexuals (Bierzychudek 1985). This pattern, termed geographi-cal parthenogenesis (GP), has been found in numerous plant and animal systems, andits proposed explanations interface with several open areas of ecology and evolutionresearch. These include: the nature of sex, the evolutionary impacts of polyploidy,clone selection, dispersal limitation, and range limits / local adaptation (see Tilquin& Kokko 2016 for an excellent review of GP). Most models of GP propose that the48effects of parthenogenesis (or its correlates, e.g., polyploidy) influence the ecology ofasexuals in a way that allows them to expand their range beyond that of the sexuals,or alternately, that the benefits of sexual recombination allow sexuals to persist inareas where asexuals cannot. Given that range shifts between close relatives occurfrequently (Weber & Strauss 2016), it is not surprising that patterns of GP may bethe result of several different interacting mechanisms that require explorations of bothgenetic and ecological attributes of sexuals and asexuals in each system.The defining feature of parthenogenic organisms is their ability to reproduce with-out mates (i.e., reproductive assurance), and this is often invoked as one of the primaryexplanations for GP. First described by Baker (1955) in the context of self-compatibleplant species, “Baker’s law” predicts that plants capable of uniparental reproduction(i.e., have reproductive assurance) will be better able to colonize new habitats thanplants dependent on mates, because only a single propagule is required to establish anew population (see Pannell et al. 2015 for a modern review of Baker’s law). Baker’slaw provides an intuitive explanation for GP, but it is not without its caveats. All elsebeing equal, reproductive assurance would provide an asexual range advantage in theshort term, but sexuals should be able to catch up given enough time. Asexual rangeadvantage would only persist in the long term if sexuals suffer from mate limitation(Gascoigne et al. 2009), for example in habitats characterized by high mortality andlow pollinator-availability. Another possibility is that asexuals occupy the sexuals’available niche space after a wave of early colonization, preventing sexual range ex-pansion through competitive exclusion or reproductive interference (Britton & Mogie2001; Kyogoku 2015; Hersh et al. 2016).While Baker’s law emphasizes advantages conferred by parthenogenesis as theprimary explanation for range disparity between sexuals and asexuals, some mod-els propose that sexual recombination gives sexuals an adaptive advantage, allowing49them to respond to selective pressures more effectively than asexuals (who are as-sumed to suffer from reduced adaptive potential; Tilquin & Kokko 2016). The RedQueen hypothesis proposes that asexuals escape to “marginal” habitats where bioticinteractions are less intense, whereas sexuals are able to adapt to constantly evolvingpressures such as pathogens and interspecific competition (Glesener & Tilman 1978).It is important to note that the Red Queen hypothesis still supposes a demographicadvantage to asexuality, and predicts that parthenogens will have the advantage inareas that are absent of parasites, predators, and competitors. The Red Queen hy-pothesis points towards asexuals being less suited to (and likely to be excluded from)environments with intense or frequent biotic interactions but reaping the benefits oftheir demographic advantage in habitats where abiotic stress is the dominant selectivepressure. Sexuals are expected to have higher fitness than asexuals in areas subjectedto frequent biotic pressures, and may also exhibit adaptations (e.g., resistance to her-bivores or increased plant size) depending on which biotic interactions are influencingselection.Implicit within the definition of GP is the fact that sexuals and asexuals have dif-ferent ranges, and therefore different range limits. Given that they occupy differentranges, an obvious question to ask is whether sexuals and asexuals are adapted todifferent niches. Ecological niche models provide one way to assess whether speciesrange limits coincide with their niche limits (reviewed in Lee-Yaw et al. 2016), butideally, the best method to test range limits is by using reciprocal transplant experi-ments. Species’ range limits (RL) are largely thought to be a reflection of their nichelimits (NL) (Sexton et al. 2009), and recent surveys of reciprocal transplant studiesindicate that fitness does often decline when species are transplanted beyond theirrange (Hargreaves et al. 2014; Lee-Yaw et al. 2016). When fitness does not decline be-yond the range (i.e. RL 6= NL), this is often seen as evidence for dispersal limitation50(Pulliam 2000). When local populations have higher fitness than populations fromother regions (“home-site advantage”), this provides some evidence for local adapta-tion (Kawecki 2008), though strict definitions of local adaptation require reciprocalhome site advantage (i.e., local populations have the highest fitness in every site andperform better at home than away; Blanquart et al. 2013). The framework of testingrange limits through reciprocal transplant experiments is particularly well suited forinvestigations of GP, because it allows us to simultaneously evaluate two importantalternatives: (1) that sexuals are dispersal limited (evidenced by comparable perfor-mance when planted within and beyond their range, supporting Baker’s law), and(2) that the two mating types are ecologically adapted to their respective ranges,evidenced by a decline in fitness (due to abiotic and/or biotic factors) when eithertype is planted beyond their range. While ecological niche models have recently beenapplied to investigate patterns of GP (Lee 2015; Kirchheimer et al. 2018), few haveapplied modern reciprocal transplant experimental design within a GP context.In addition to a review of transplant experiments beyond the species range, Har-greaves et al. (2014) provide an excellent framework to guide the design of reciprocaltransplant experiments and interpret their results. The authors propose several rec-ommendations for effective experimental design: (1) include both core and range edgesites and source populations in order to determine whether populations are locallyadapted to different parts of the range. (2) Include multiple transplant sites beyondthe range to detect gradients in habitat quality. (3) Estimate lifetime fitness insteadof single life-stage components in order to get the best indication of whether popula-tions are self-sustaining. (4) Conduct experiments under natural conditions in orderto assess the impacts of biotic interactions on fitness.Most broadly, results from transplant experiments can be sorted into two cate-gories: RL = NL (decline in fitness beyond the range) or RL 6= NL (fitness does51not decline beyond the range or fitness declines within the range; Hargreaves et al.2014). Ideally, interpretations of whether range limits and niche limits coincide aresupported by lifetime fitness estimates, which allow researchers to assess whetherpopulations are self-sustaining (λ ≥ 1). However, because lifetime fitness estimatesare often difficult to obtain, many rely on comparisons of relative fitness within andbetween sites. In cases where RL < NL, that could be due to dispersal limitation- i.e., suitable habitat exists outside of the current range where populations can beself-sustaining, but they have not been able to colonize these areas. Cases where RL> NL, for example at a range edge site, could indicate that the range edge comprisessink populations being maintained by dispersal from populations within the rangeinterior. It is important to note that this framework is somewhat (perhaps necessar-ily) simplistic and that niche limitation can interact with and contribute to dispersallimitation. For example, declines in fitness driven by niche limitation will contributeto dispersal limitation by reducing mate availability and seed set. These effects areimportant to keep in mind in a GP context, as they are more likely to affect sexualsthan asexuals due to having comparatively reduced reproductive assurance.Townsendia hookeri (Asteraceae) is a long-lived perennial plant species that dis-plays a classic pattern of GP (Bierzychudek 1985). It has two forms: diploid sexualsthat are self-incompatible, and polyploid autonomous apomicts that produce seedsasexually without the need for pollen. The apomictic range is much larger than thesexual range and there is a range of overlap where populations of both types canbe found. Given that sexuals are obligately outcrossing and that apomicts have fullreproductive assurance, it is possible that sexual range expansion was limited by dis-persal. Sexual dispersal limitation may be compounded by environmental and/orreproductive barriers; for example, it is possible that apomicts colonized new habitatfirst, and then prevented subsequent sexual range expansion through reproductive52interference. Apomictics have the potential to reduce seed set and produce apomicticoffspring when pollinating sexual flowers, but apomicts are unaffected by pollen pro-duced by sexuals (Garani 2014). On the other hand, the distinctness of the two rangesindicates a clear possibility that sexuals and apomicts are ecologically differentiated,which is supported by ENMs suggesting significant niche divergence between the twoforms (Lee 2015).These patterns provide an excellent backdrop to test for local adaptation anddispersal limitation using reciprocal transplant experiments. The ranges of the twoforms are generally situated from south (sexual range) to north (apomictic range),with overlap in the middle. This allows us to establish a mirrored version of the classicreciprocal transplant design, where interior sites for one mating type can act as withinrange for itself and beyond the range sites for the other, with sites in the overlap zonerepresenting range edge sites. We used this design, following the recommendationsprovided by Hargreaves et al. (2014) as closely as possible, to address the followingquestions: (1) Do the ranges of sexuals and apomicts coincide with their ecologicalniches (RL = NL), such that there is a decline in fitness when planted into therange of the opposing type? (2) Does either mating type show evidence of dispersallimitation (RL < NL), as indicated by comparable performance when planted in siteswithin and beyond the range? While it is possible that both sexuals and apomictsmay be dispersal limited, we expect this to be more likely in sexuals due to beingmate and pollinator dependent. (3) Are there any trait differences that may indicatethat sexuals and apomicts adopt different strategies in response to biotic stress (e.g.,competition)?533.2 Materials and methods3.2.1 Reciprocal transplant experimental designIn 2013 and 2014, we collected achenes (hereafter referred to as “seeds”) from 6 sexualand 6 apomictic source populations distributed throughout the core of the range ofT. hookeri (Figure 3.1; Table 3.1). We chose three source populations from each offour regions that correspond to the interior and edge of the geographic ranges of thesexual and apomictic forms: the interior sexual range (Ss, located in the southern-most portion of the range), the sexual “edge” where the sexual and apomictic rangesoverlap (SOs), the apomictic “edge” in the zone of overlap (AOs), and the interiorapomictic range (As, located in the north). We chose 11 maternal plants (hereafterreferred to as “moms”) from each source population, and 32 viable seeds (based oncriteria described in Hersh Ch. 4) from each mom and germinated them in agar-filledpetri dishes. We grew the resulting seedlings in growth chambers, and hardened themin the greenhouse before transplanting them into field sites (see Chapter 4 for detailedmethods on germination and greenhouse experiments).In early September of 2014, we reciprocally transplanted seedlings from all momsand populations into two gardens (separated by ∼ 2 km) in each of the four regionsdescribed above over the course of two weeks (garden regions labelled: Sg, SOg, AOg,Ag; Figure 1). The placement of the gardens allowed us to evaluate the performanceof populations when planted in the range interior, range edge, just beyond the range,and far beyond the range of each mating type as recommended by Hargreaves et al.(2014). We randomized the seedlings and planted them in rows of 10, with 10 cmspacing between plants and a small path down the middle of each plot to facilitatedata collection. We transplanted 250-350 seedlings into each garden, with the largergardens being located in the overlap regions (we anticipated higher mortality in the54Figure 3.1: Map of source populations and garden sites. Circles represent sourcepopulations and square represent garden sites. Points are coloured based on rangelocation: A = apomictic range (blue), AO = apomictic overlap range (sky blue), SO= sexual overlap range (orange), S = sexual range (red).overlap regions based on observations from multiple field seasons), for a total of 2,531plants (Table 3.5). We chose garden sites based on proximity to known naturalpopulations and qualititative attributes based on our experience with the habitatscontaining T. hookeri in each part of the range (i.e., typically rocky soils with sparsevegetation, but higher vegetation cover in Sg which is characteristic of populationsfound there). Natural individuals were found within 500 m of each garden site, whichsupports the suitability of the sites. We planted seedlings directly without removingvegetation or conditioning the soil in order to expose the plants to conditions thatclosely reflect the area’s natural ecology. Aside from watering the seedlings rightafter planting, we did not install irrigation or provide any supplemental water to thegardens.55At each garden site, we installed a small “establishment” plot with two seeds fromeach of the families represented in the gardens in order to capture the germinationand early seedling establishment life history stages (totalling 480 seeds per gardenregion; Table 3.3). Each establishment plot consisted of a small plastic grid, with theseeds randomly placed on the soil in each cell of the grid (one seed per cell). Beforeinstalling each plot, we cleared the soil of vegetation and scraped away the top layer ofsoil in order to reduce competition with the natural seed bank. We covered each plotwith thrip-proof mesh to protect the seeds from predation and from being scatteredby strong winds that are typical in sites where T. hookeri is found.We returned to the garden sites every spring from 2015-2019 to collect data onsurvival, growth (number of leaves, leaf length - measuring one haphazardly chosenrepresentative leaf per individual) and reproduction (the number of inflorescences,hereafter referred to as “buds”) produced per individual. We also checked the estab-lishment plots to score the presence of juvenile individuals in 2015 and 2016 (though95% of recruits emerged in the first year). The first flower buds were seen in 2016.Our permitting requirements included the need to mitigate the release of alien seedand pollen in the sexual range. Therefore, to address this in a consistent manner,we removed flower buds before they opened at all garden sites. In order to quantifypotential differences in seed set between mating types, in 2017 we obtained permis-sion to allow individuals in the gardens to flower. We bagged the apomicts before theflowers opened, and the sexuals after allowing enough time for pollination to occur.We also collected mature seed heads from natural individuals in source populations tocomplement the seeds collected from the gardens, in case sexuals in the gardens werepollen limited (due to having few compatible individuals concurrently flowering).We assessed vegetation cover in our gardens in 2017. We used 1x1 meter quadratsto estimate percent vegetation cover at each site. At the end of the experiment (2019),56we harvested surviving individuals and estimated plant size by water displacement.We immersed below ground parts in a graduated cylinder filled with water, and thenused a graduated pipette to estimate below ground volume. A similar approach wasused for whole plants, allowing us to estimate an index of above vs below-groundbiomass.3.2.2 Data analysisWe analyzed establishment success in the field using generalized linear mixed models(GLMMs; logit link - binomial) in R version 3.6.1 (R Core team 2019), includingsource region, garden region, and their interaction as fixed effects, and originallyincluding garden, population, and mom (nested within population) as random effects.After full mixed models did not converge, we dropped the random effects of populationand mom, which allowed the models to converge successfully. Damage to one of theestablishment plots installed in Ag led to a data gathering error which resulted in apartial loss of data; this reduced the sample size and likely contributed to slightlylarger predicted confidence intervals in that garden region.Given that we collected data from our reciprocal transplant experiment over sev-eral years and that lifetime bud production was strongly zero inflated, we used ASTERmodels (aster package, reaster() function; Geyer et al. 2007, 2013) in R to analyzeour garden data. Our ASTER models incorporate survival, flowering, and number ofbuds produced over the course of the experiment, and are hierarchically structuredso that, for example, survival in year 5 is contingent upon that individual survivingin years 1-4 as well (see ASTER model diagram; Figure 3.3). We modeled survivaland flowering as Bernoulli variables (coded as 0 or 1), and yearly bud productionas zero-truncated Poisson variables. Individuals did not become reproductive untilyear 2, so we removed the year 1 flowering and bud production nodes from our model57structure. We tested the fixed effects of source region, garden region, and their in-teraction, and originally included garden, population, and mom as random effects.The random effects of garden and population resulted in model singularities (randomeffect variances estimated as zero), and were dropped from the model. We visualizedbud production over 5 years (our estimate of lifetime fitness) by plotting predictedvalues with 95% confidence intervals from fixed effects models (there is no predict()function for random effects ASTER models).Because we were not able to incorporate our seedling establishment data intoour ASTER models (because they measure different individuals), we used a randomsampling approach to combine the two datasets. We separately randomly sampledeach dataset 1000 times and calculated the mean establishment success and meannumber of buds produced per individual planted over the course of the experiment ineach source region and garden region. We multiplied these values for each randomly-sampled dataset and calculated means and 95% confidence intervals.We analyzed leaf length and leaf number in the gardens using mixed effects mod-els. We tested the effects of source region, garden region, and their interaction as fixedeffects and originally included garden, population, and mom (nested within popula-tion) as random effects, modelling the data from each year separately. When modelsdid not converge, we removed random effects until models converged. We modeledleaf length using linear mixed effects models (LMMs) and leaf number using GLMMs(log link - poisson).We analyzed root to shoot volume ratio of individuals surviving to year 5 usingLMMs, including mating system as a fixed effect and population as a random effect.We analyzed proportion of viable seed and total number of seeds per bud (collectedin 2017) in the gardens and source populations using LMMs, including mating systemas a fixed effect and population as a random effect. We analyzed percent vegetation58cover in the gardens using LMMs, with garden region as a fixed effect and garden(nested within region) as a random effect.All mixed models (besides ASTER) were performed using the lme4 package(Bates et al. 2015) in R. We tested the effects of all models using likelihood ratiotests and visualized predicted means and confidence intervals using the ggeffectspackage (Lu¨decke 2018) in R. We inferred statistical differences between groups whenboth the likelihood ratio test P-values were below 0.05 and confidence intervals frommodel predictions were largely non-overlapping.In order to compare mean annual temperature (MAT) and mean annual precipi-tation (MAP) during experiment years to historical averages, we pulled climate datafrom 1901-2018 at each garden site using ClimateNA v.5.10 (Wang et al. 2016).3.3 Results3.3.1 Establishment successEstablishment success varied among source regions (labelled Ss, SOs, AOs, and As)and garden regions (labelled Sg, SOg, AOg, and Ag), and there was an interactionbetween these two effects (Table 3.2). Despite statistical support for an interactionbased on LRTs, visualization of confidence intervals suggests a lack of difference inestablishment success among source regions within garden regions (Figure 3.2A).Overall, populations did not have higher establishment success when planted intotheir own region, and therefore our results are not consistent with local adaptationat early life stages. For all source regions, establishment success was highest in AOg(averaging 77%) and lowest in Sg (averaging 14%; Table 3.3).59Figure 3.2: A) Predicted establishment success by source region and garden fromGLMMs. B) Predicted estimates of lifetime bud production by source region andgarden region from ASTER models. Error bars represent 95% confidence intervals.3.3.2 Reciprocal transplant experimentASTER model analyses showed that lifetime bud production differed among gardenregions and that there was an interaction between source region and garden region,but no effect of source region overall (Table 3.4). While the fitness of apomicts (Asand AOs) decreased when transplanted into the sexual range (Sg and SOg), sexualfitness (Ss and SOs) was similar or increased when transplanted into the apomicticrange (Ag and AOg; Figure 3.2B). Some source regions show a pattern that is (at leastpartially) consistent with local adaptation. For example, As had the highest fitnessin its own region (though performance in Ag did not differ among source regions) andhad low fitness in other garden regions. Ss performed well and had the highest fitnessof all source regions in Sg, but it had equally high (if not higher) fitness in Ag, whichis the region farthest away from its own source region. AOs performed best in theapomictic regions, had lower fitness in sexual regions, and was the source region with60the most consistent performance across sites. SOS, on the other hand, had low fitnessin the sexual regions and considerably higher fitness in the apomictic garden regions.Overall, fitness was lowest in SOg and highest in Ag (Table 3.5).Combining establishment and bud production per individual planted (using therandom sampling approach) shows similar patterns to lifetime bud production fromASTER models, but with wider confidence intervals and less pronounced differences inperformance between source regions (Figure 3.4). Most notably, when establishmentand bud production are combined, the fitness of Ss was considerably reduced in Sgand Ag due to low establishment success in both regions.Leaf number and leaf length were affected by source region, garden region, andtheir interaction, but these effects differed in significance and magnitude by year(Table 3.6, Table 3.7). However, plant size was largely comparable among sourceregions within garden regions and years (Figure 3.5, Figure 3.6). Leaf number tendedto be higher in Ag (and higher amongst apomicts within that garden region), whileleaf length was consistently high in Sg with individuals from Ss often having thelongest leaves.The total number of seeds produced per bud did not differ by mating system ineither the gardens or in source populations. Nonetheless, apomicts tended to havea higher proportion of viable seeds within gardens and source populations (Table3.8, Figure 3.7), though this effect was weak in source populations (as indicated byoverlap in confidence intervals and slightly higher P -value from likelihood ratio tests(P=0.0716)). In the plants surviving to the end of the experiment (year 5), apomictshad a higher root to shoot volume ratio than sexuals (df=4, χ2 = 3.883, P = 0.0488;Figure 3.8), though partial overlap in predicted confidence intervals indicate that thisis not a particularly strong effect.613.3.3 Site characteristicsSg had considerably higher vegetation cover than the other garden regions, which didnot differ from each other (df=6, χ2 = 22.424, P = 5.3 x 10-5; Figure 3.9). Whencomparing mean annual temperature and mean annual precipitation during experi-ment years to historical averages, experiment years largely fall within the historicaldistributions for both climate variables (Figure 3.10). Mean annual temperature washigher in Sg gardens (SS1 and SS2) both historically and during the experiment, whilemean annual precipitation was similar across sites.3.4 DiscussionThe results from our reciprocal transplant experiment show that, in comparison totheir home ranges, the fitness of sexuals was similar or increased when transplantedinto the apomictic range (indicating dispersal limitation), but the fitness of apomictslargely decreased when transplanted into the sexual range (indicating that their rangelimits coincide with their niche limits). Differences in fitness appear to be drivenprimarily by survival and reproduction, as establishment success did not differ amongsource regions within garden regions. Source populations from the sexual interiorhad markedly higher performance in their home range than populations from otherregions, which indicates that they are locally adapted. The sexual interior gardensalso had considerably warmer temperatures and higher vegetation cover than theother garden regions, which hints at the potential influence of biotic interactions inlimiting the fitness of populations from other regions. Overall, these patterns areconsistent with Baker’s law effects limiting the sexuals’ northern range limit andapomicts’ being constrained by unsuitable habitat beyond their southern range limitin Townsendia hookeri, though there are several alternative interpretations which we62discuss below.3.4.1 Sexual range expansion is limited by dispersalASTER model analyses incorporating survival and bud production over 5 years indi-cate that sexual populations from both the sexual range interior (Ss) and the sexualrange edge (SOs) perform as well or better when transplanted beyond their range aswhen planted into their home regions (Figure 3.2B). Ss had equally high fitness in itshome region (Sg) as in Ag, despite Ag being more than 600 km north of the sexualrange edge. SOs performed best when transplanted into the apomictic range andsomewhat surprisingly had the lowest fitness in its home region. This suggests that,at least over the years of our study, suitable conditions exist beyond the current rangeof sexuals that they have not been able to colonize (RL<NL). This type of patternis typically interpreted as evidence for dispersal limitation (Hargreaves et al. 2014).Given that in Townsendia hookeri sexuals are self-incompatible and that apomictsare autonomous with full reproductive assurance, it seems plausible that mate orpollen limitation could play an important role in limiting sexual range expansion.Apomicts produced a higher proportion of viable seeds in both the garden experi-ment and in natural populations (Figure 3.7). In particular, the reduced seed set ofsexuals in the gardens (which likely had fewer concurrently flowering individuals thanwould be found in natural populations) suggests that recruitment is likely to suffer insexual populations when there is a low density of potential mates (Gascoigne et al.2009). The fact that sexuals and apomicts produced a similar number of seeds overallindicates that sexuals are capable of comparable seed set to apomicts in good years(i.e., given sufficient mates and pollinator services), though increased germinationsuccess under lab conditions and improved dispersal architecture (Chapter 4) maystill give apomicts a colonization advantage even when seed set is equal.63While species range limits often coincide with their niche limits (Hargreaves et al.2014; Lee-Yaw et al. 2016), there are numerous examples from transplant studieswhere fitness does not decline beyond the range edge. Samis et al. (2016) found thatin almost 1/3rd of the within- vs beyond-range transplant studies reviewed by Harg-reaves et al. (2014), fitness actually increased beyond the range limit. Inferences aboutdispersal limitation are dependent on mating system and life-history characteristics,however. For example, Stevens & Emery (2015) found that asexual gametophytes ofthe fern Vittaria appalachiana had similar or better performance when transplantedbeyond the range as compared to within. The authors interpreted their results asevidence of dispersal limitation because asexual fern gametophytes have no methodof long-distance dispersal (which, in ferns, is primarily accomplished by sexually pro-duced spores). Similarly, Samis et al. (2016) found that fitness of the dune plantCamissoniopsis cheiranthifolia continually increased when transplanted toward andbeyond the species range limit. In this case, the authors acknowledged that theirresults were in line with dispersal limitation, but the mechanism was unclear be-cause C. cheiranthifolia is self-compatible and population density did not decreaseat the northern range edge (i.e., dispersal limitation was not likely driven by matelimitation). Instead, dispersal limitation may have been driven by metapopulationdynamics (Holt & Keitt 2000); C. cheiranthifolia seeds lack dispersal architecture andmay suffer from reduced dispersal between habitat patches at the range edge (Samis& Eckert 2009). These studies highlight how mating system, dispersal mechanism,life history, and demographics can all play important roles in delimiting where speciesare found on the landscape.The comparatively increased colonization potential of asexuals due to reproduc-tive assurance (Baker’s law effects) is often considered one of the primary drivers ofpatterns of GP (Tilquin & Kokko 2016). Supporting evidence for this mechanism64comes in part from an over-abundance of apomictic taxa in previously glaciated habi-tats (Brochmann et al. 2003), which points to rapid asexual colonization from glacialrefugia (Ho¨randl 2009). GP systems are complicated, however, and the importanceof reproductive assurance in giving apomicts a colonization advantage over sexualswill vary in each system depending on the interacting suite of traits possessed by eachmating type. For example, many apomicts require pollen to fertilize the endosperm inorder to set seed (pseudogamy), while others set seed autonomously (Whitton et al.2008). Pseudogamous apomicts that are self-incompatible would not be expected tohave the same colonization advantage as an autonomous apomict, because they arestill dependent on pollen from another individual and would require two individualsto establish a new population. Apomicts in T. hookeri set seed autonomously whilesexuals are self-incompatible, making Baker’s law effects more likely to be importantin this system than in others where there may be less of a discrepancy in reproduc-tive assurance between mating types. In the Boechera agamic complex, for example,sexual progenitors are self-compatible and populations are highly selfing. Sexuals inBoechera therefore have full reproductive assurance and are at less of a demographicdisadvantage in comparison to apomicts, which may help explain why the genus doesnot show a strong pattern of GP (Mau et al. 2015). This is a rare case, however, asthe sexual progenitors of most apomicts are self-incompatible (Asker & Jerling 1992),and pseudogamous apomicts often benefit from the breakdown of self-incompatibility(so that self-pollen can trigger endosperm development; Ho¨randl 2010), which pointsto Baker’s law effects as being a plausible driver in most plant GP systems.653.4.2 Habitat is not suitable for apomicts beyond theirsouthern range limitIn contrast with sexuals, the fitness of apomicts generally decreased when trans-planted beyond their range (Figure 3.2B) indicating that the apomicts’ southernrange limit and niche limit coincide (RL=NL). This suggests that their southernrange limit is caused by a gradient in habitat quality (Hargreaves et al. 2014), whichcould mean that a shift in abiotic conditions, biotic interactions, or some combinationof these is important in limiting the range (Gaston 2003; Case et al. 2005; Normandet al. 2009). While we do not have lifetime fitness estimates and therefore cannot becertain that apomictic populations would not be self-sustaining in the sexual range(i.e., λ may still be greater than 1), apomicts had comparatively lower fitness thanlocal sexual populations in the sexual interior (Sg), which provides additional sup-port that apomicts are not well adapted to this region. One possible cause for thisreduction in fitness is that apomicts are simply not well adapted to the environmental(abiotic) conditions in the southern portion of their range. This pattern is consis-tent with environmental niche models, which predicted that, despite overlap in theenvironmental niches of the two mating types, the apomicts’ niche did not extendfully into the sexual range (Lee 2015). As for individual abiotic factors that impacttheir respective niches, Lee (2015) found that temperature variables were more im-portant for sexuals and precipitation variables were more important for apomicts.Populations from the sexual interior had the highest fitness in their own region (Sg),which also had historically warmer temperatures (Figure 3.10) than the other gardenregions. This provides some evidence that sexuals are better suited to warm climatesthan apomicts, and may indicate that the southern range limits of apomicts is limitedby temperature (and/or other factors that correlate with temperature).Temperature is linked with several other factors, including many biotic interac-66tions (Burnside et al. 2014). Biotic interactions have traditionally been thought tobe more intense at lower latitudes (Louthan et al. 2015), though it should be notedthat some large-scale analyses have challenged this assumption; Moles et al. (2011a)found that herbivory and plant defenses are not greater at lower latitudes, and Har-greaves et al. (2020) found that biotic interactions may affect fitness but often fail todrive local adaptation. In the case of T. hookeri, the manner in which biotic inter-actions are thought to contribute to southern range limits is consistent with how theRed Queen hypothesis is thought to contribute to patterns of GP, allowing sexuals(through their superior ability to adapt to biotic pressures) to persist in areas whereasexuals are excluded. Apomictic fitness was lowest in the southernmost garden re-gion (sexual interior; Sg), where vegetation cover was considerably higher (Figure 3.9)and temperatures were warmest (SS1 and SS2; Figure 3.10). Populations from thesexual interior (Ss) had considerably higher performance in their home region (Sg)than populations from other regions (Figure 3.2B), a pattern indicative of local adap-tation (Kawecki 2008). In addition, all source regions had longer leaves in Sg than inother garden regions, and populations from Ss had the longest leaves in that regionin most years (Figure 3.6). While this increase in leaf size among all source regions islikely a response to higher temperatures in the sexual interior (Peppe et al. 2011), thefact that local source populations tended to have the longest leaves and the highestfitness indicates that they may be better adapted to respond to competition thanpopulations originating from areas with lower vegetation cover (Novoplansky 2009).By transplanting seedlings directly into the plots without removing local vegeta-tion, we were able to detect potential biotic effects (i.e., low apomictic fitness and highvegetation cover in Sg) which would have been obscured had we standardized siteslike most transplant studies have done historically (Hargreaves et al. 2014). However,our experiment was not explicitly designed to test for the effects of biotic interac-67tions, and we are therefore somewhat limited in what conclusions we can draw abouthow/which biotic interactions affected fitness, or indeed, whether abiotic factors alonewould produce the same results. Studies that have explicitly investigated Red Queeneffects within the context of GP in plants provide mixed empirical support. Usingpopulation surveys and greenhouse experiments in Taraxacum officinale, Verhoeven& Biere (2013) found that soil pathogens and seed-eating weevils were more commonin southern (largely sexual) populations than in northern apomictic populations. Onthe other hand, Herman et al. (2017) investigated seed predator intensity and vegeta-tion density in Hieracium alpinum but found no differences in the prevalence of eitherbiotic pressure between sexual and apomictic populations. Given that the Red Queenhypothesis is consistently invoked in the conversation surrounding the evolution andubiquity of sex, investigations of the relative effects of biotic interactions in more GPsystems will provide important insight into this potential mechanism.3.4.3 Drivers of geographical parthenogenesis inTownsendia hookeriResults from our reciprocal transplant garden experiment show that the fitness ofapomicts decreased when transplanted into the sexual range, but sexual fitness wascomparable (or increased) when transplanted into the apomictic range (Figure 3.2B).This suggests that while apomicts have largely been able to occupy their availableniche space (RL=NL), sexual range expansion is likely limited by dispersal and es-tablishment (RL< NL) in Townsendia hookeri. These results corroborate what wasfound by previous niche models, which indicated that the apomictic range was suit-able for both mating types, but the sexual range was suitable only for sexuals (Lee2015). Our results are in line with apomicts being excluded from the sexual rangedue to ecological differentiation, possibly driven by their inability to persist in the68face of increased competition beyond their southern range limit.Our results provide evidence for Baker’s law effects (i.e. dispersal limitation dueto lack of reproductive assurance) limiting the sexuals expansion northward and RedQueen effects (i.e. lack of adaptation to biotic pressures) preventing the apomictsfrom moving south, but other processes are likely at play as well. Sexual popula-tions had quite low fitness in the gardens placed at their northern range edge (SOg),which may indicate that this area exists outside of the sexual niche (RL<NL). Thisinhospitable part of the range could act as an environmental barrier preventing sex-uals from dispersing to northern regions that are more suitable (AOg and Ag; Figure3.2B). Interestingly, populations from the sexual edge (SOs) performed as well as (ifnot better than) the apomictic edge populations (AOs) in the apomictic edge gardenregion (AOg). This indicates that the sexual edge populations are well poised to ex-pand their range further into the apomictic edge region (as predicted by Hargreaveset al. 2014), but they apparently have not done so despite being in relatively closeproximity (SOg and AOg are only separated by ∼30 km). Given that mixed sexual-apomictic populations are exceedingly rare (Chapter 2), it is possible that asymmetricreproductive interference from apomicts (i.e., apomictic pollen reducing sexual seedset and/or siring apomictic offspring; Garani 2014), contributes to limiting the sex-uals’ expansion, and may even cause the sexual range to shrink over time (Britton& Mogie 2001). Populations from the apomictic edge (AOs) performed best in thesexual overlap region (SOg), which suggests that they may be well-situated to invadethe sexual range. In fact, apomicts were detected at low frequencies in a populationsituated at the sexual range edge (population L62; Chapter 2), which points to thepossibility that a “frozen” apomictic lineage has emerged that is able to persist inthis otherwise unsuitable environment.We have emphasized how biotic interactions may be excluding apomicts from the69southern portion of the range, and this is likely influenced by the fact that sexualsand apomicts appear to embody different ecological strategies. While sexuals tendedto have longer leaves (particularly in the sexual interior; Figure 3.6), apomicts tendedto produce more leaves (Figure 3.5), which hints at the possibility of trade-offs in re-source allocation to vegetation vs reproduction - each leaf is typically associated withan axillary bud which has the potential to become a flower, so plants with more leaveshave a larger “bud bank” (Kleiman & Aarssen 2007). In addition, apomicts appearto devote relatively more energy to root growth than shoot growth in comparisonto sexuals (Figure 3.8). Changes in allocation to root mass are typically associatedwith reduced water or nutrient availability, but can also be positively correlated withwind intensity (Poorter et al. 2012), which may indicate that apomicts are betteradapted to open or disturbed environments (Reynolds & Pacala 1993). These pat-terns echo the predictions from various hypotheses (general-purpose genotypes, Lynch1984; asexuals as “weedy” species, Baker 1965; establishment vs dispersal tradeoffs,discussed in Hersh Ch. 4) which liken apomicts to being “r-selected” taxa that canbest express their demographic advantages in marginal environments that are domi-nated more by environmental stochasticity than biotic pressures. This highlights howthe poor establishment performance in the sexual interior (Sg, Figure 3.2A) mitigatesthe apomicts’ demographic strength, which likely contributes significantly to theirinability to invade the sexual range.Many investigations into GP focus either on the geographic success of asexuals orthe adaptive benefits of sexuality, but in reality these alternatives are opposite sidesof the same coin. Tilquin & Kokko (2016) sagely wrote that any model for GP shouldaddress: “why sex (here) and why asex (there)”. In this study we have harnessed thepower of modern reciprocal transplant experimental design and adapted it to inves-tigate the performance of both sexuals and apomicts across the species range, which70has enabled us to address both sexual and asexual aspects of the GP question. In fact,this “mirrored” reciprocal transplant design can be used to address similar questionsabout dispersal limitation and local adaptation in any pair of sister species with para-patric ranges. Although GP is housed within a complex sexual-asexual framework,the questions it addresses transcend the idiosyncrasies of the (often understudied)systems the work is done in and can inform contemporary conversations surroundingplant mating systems, range limits, dispersal, and the dynamic interplay of ecologicaland evolutionary processes.71Figure 3.3: Hierarchical structure of ASTER models used to analyze lifetimefitness in Townsedia hookeri over the course of a 5-year reciprocal transplantexperiment. Survival is contingent upon survival in previous years, and the numberof buds produced is contingent upon flowering (that year) and survival in previousyears. Because no individuals flowered in year 1, we removed those variables fromthe model.72Figure 3.4: Means and 95% confidence intervals of establishment success x budsproduced per individual calculated by randomly sampling each dataset 1000 times.73Figure 3.5: Predicted leaf number by source region and garden region in each yearof the reciprocal transplant garden experiment. Points represent means from modelpredictions, and error bars represent 95% confidence intervals.74Figure 3.6: Predicted leaf length by source region and garden region in each yearof the reciprocal transplant garden experiment. Points represent means from modelpredictions, and error bars represent 95% confidence intervals75Figure 3.7: Seed set (number of seeds per bud and proportion viable) in reciprocaltransplant gardens and source populations by mating system. Boxplots with jitteredpoints represent values for each bud, and points with confidence intervals (to theright) indicate model predictions.76Figure 3.8: Root to shoot volume ratio of plants surviving to the end (year 5) ofreciprocal transplant garden experiment. Boxplots with jittered points representvalues for each individual measured, and points with confidence intervals (to theright) indicate model predictions.77Figure 3.9: Predicted % vegetation cover in each garden region in the reciprocaltransplant garden experiment.78Figure 3.10: Mean annual temperature (MAT) and log mean annual precipitation(MAP) at each garden site. Violin plots represent historical distribution of the twoclimate variables from 1901 - 2012, and colored points represent values duringreciprocal transplant experiment years.79Table 3.1: Source population and garden locations. For mating system, S = sexualand A = apomictic.Site Region Mating system Latitude LongitudeSource Populations B53 S S 39.89244 -105.26918B42 Ss S 40.10705 -105.28365B46 Ss S 40.54588 -105.13292B49 SOs S 40.84053 -105.33504L11 SOs S 41.24026 -105.43345L12 SOs S 41.25297 -105.40644L06 AOs A 41.14204 -106.04387L16 AOs A 41.30268 -105.52930L17 AOs A 41.38666 -105.47291C86 As A 45.66205 -110.47504C85 As A 46.65055 -111.72013C27 As A 47.54790 -112.47641Gardens SS1 Sg 40.07072 -105.28236SS2 Sg 40.06680 -105.24803SO1 SOg 41.11396 -105.51121SO2 SOg 41.10375 -105.50114AO1 AOg 41.38728 -105.48304AO2 AOg 41.38809 -105.47413AA1 Ag 45.91201 -109.90557AA2 Ag 45.88509 -109.8689780Table 3.2: Likelihood ratio test statistics for the fixed effects of source region,garden region, and their interaction on establishment success.Effect df χ2 PSource region 8 13.745 0.0033Garden region 8 10.320 0.0160Source region x Garden region 17 18.395 0.0309Table 3.3: Summary statistics for establishment success.Garden region Source region Number of seeds Mean establishment successSg Ss 120 0.133SOs 120 0.083AOs 120 0.200As 120 0.133SOg Ss 120 0.467SOs 120 0.433AOs 118 0.424As 120 0.583AOg Ss 120 0.717SOs 120 0.667AOs 120 0.800As 120 0.900Ag* Ss 76 0.342SOs 88 0.523AOs 88 0.341As 84 0.548* Damage to one of the establishment plots installed in this garden region lead to a datagathering error which resulted in a loss of data.81Table 3.4: Likelihood ratio test statistics for the fixed effects of source region,garden region, and their interaction on lifetime bud production from ASTERmodels.Effect df χ2 PSource region 19 2.558 0.465Garden region 19 50.68 5.7 x 10-11Source Region x Garden Region 28 31.664 0.0002Table 3.5: Raw data for the reciprocal transplant garden experiment.Garden region Source region # planted # surviving year 5 # of flowering individuals total # of buds producedSg Ss 128 19 19 50SOs 132 8 6 10AOs 144 9 9 20As 165 10 6 13SOg Ss 153 1 2 4SOs 161 1 2 5AOs 180 5 6 23As 208 1 5 7AOg Ss 152 3 6 10SOs 157 14 18 47AOs 176 23 28 46As 205 5 14 26Ag Ss 128 8 11 56SOs 129 5 9 39AOs 141 6 16 48As 162 6 12 5982Table 3.6: Likelihood ratio test statistics for leaf number in the reciprocaltransplant garden experiment. Tests were performed separately for each year.Effect df χ2 PYear 1 Source region 9 13.809 0.0032Garden region 9 10.921 0.0122Source Region x Garden Region 18 25.309 0.0027Year 2 Source region 9 4.579 0.2054Garden region 9 7.353 0.0615Source Region x Garden Region 18 117.78 2.2 x 10-16Year 3 Source region 10 4.6 0.2036Garden region 10 8.705 0.0335Source Region x Garden Region 19 76.955 6.5 x 10-13Year 4 Source region 10 1.213 0.75Garden region 10 11.045 0.0115Source Region x Garden Region 19 207.84 2.2 x 10-16Year 5 Source region 9 2.0424 0.5637Garden region 9 12.134 0.007Source Region x Garden Region 18 189.38 2.283Table 3.7: Likelihood ratio test statistics for leaf length in the reciprocaltransplant garden experiment. Tests were performed separately for each year.Effect df χ2 PYear 1 Source region 10 16.405 0.0009Garden region 10 19.559 0.0002Source Region x Garden Region 19 23.687 0.0048Year 2 Source region 10 14.931 0.0019Garden region 10 25.42 1.3 x 10-5Source Region x Garden Region 19 65.338 1.2 x 10-10Year 3 Source region 10 19.697 0.0002Garden region 10 29.491 1.8 x 10-6Source Region x Garden Region 19 29.686 0.0005Year 4 Source region 9 30.468 1.1 x 10-6Garden region 9 29.252 2.0 x 10-6Source Region x Garden Region 18 4.586 0.8688Year 5 Source region 9 19.064 0.0003Garden region 9 7.36 0.0613Source Region x Garden Region 18 9.978 0.3523Table 3.8: Likelihood ratio test statistics for seed set by mating system inreciprocal transplant gardens and source populations.Seed trait (effect of mating system) df χ2 PSeed per bud (natural) 4 0.2103 0.6465Proportion viable seed (natural) 4 3.247 0.0716Seed per bud (garden) 4 2.229 0.1355Proportion viable seed (garden) 4 7.761 0.005484Chapter 4Differences in Early Life HistoryTraits between Diploid Sexual andPolyploid Apomictic Easter Daisies4.1 IntroductionGeographical parthenogenesis (GP) describes a pattern in which closely related sexualand asexual taxa exhibit differences in geographic distributions, with asexuals oftenhaving larger ranges and occurring at higher latitudes and elevations than their sexualancestors (Stebbins 1940; Glesener & Tilman 1978; Bierzychudek 1985). Despitethe consistency of the pattern, the contexts in which GP occurs vary wildly acrosstaxa, encompassing several forms of asexuality and origins of parthenogenesis (for athorough review of GP in plants and animals, see Tilquin & Kokko 2016). While thespecific causes of the pattern may be context-dependent in each system, it remainsimportant to investigate the common conditions that allow both sexual and asexualreproduction to be maintained despite, on the one hand, the costs of sex and, on the85other, the prevalence and presumed benefits of sex in nature.Apomictic complexes in flowering plants provide classical examples of GP. Asex-uality in these systems typically comes in the form of apomixis (asexual reproduc-tion through seeds), and the vast majority of apomictic plants are polyploids thatare often of hybrid origin (i.e. allopolyploids) (Ho¨randl 2006; Whitton et al. 2008).Because polyploidy and hybridization can both affect plant species’ ranges and areoften coincident with apomixis, disentangling their influences on patterns of GP hasbeen challenging. One of the central questions around the existence of GP involvesdistinguishing between the relative contributions of enhanced colonization potentialthat apomicts derive from uniparental reproduction (“Baker’s Law” effects; Baker1955) and the ecological and trait shifts that further enhance dispersal and estab-lishment. Reproductive assurance of apomicts can contribute to range expansion,but this also means that individuals have to disperse to new areas and successfullybecome established. As a result, traits that enhance dispersal and the likelihood ofestablishment are expected to be associated with range expansion (Chuang & Peter-son 2016). While uniparental reproduction can give asexuals a head start on rangeexpansion (e.g. post-glaciation; Kearney 2005), it alone cannot explain why sexualswould not catch up; therefore, characterizing variation in other dispersal/establish-ment traits between mating types may help clarify whether sexuals would be capableof matching the ranges of their asexual counterparts given enough time. To date,few studies have considered changes in dispersal and establishment traits (other thanuniparental reproduction) when investigating GP (but see Coughlan et al. 2014 andChrtek et al. 2018).It might seem that given a pattern in which apomicts have established a rangebeyond their sexual relatives, they would necessarily possess traits that favoured col-onization. However, traits that enhance plant dispersal ability are often thought86to trade off with traits promoting establishment (e.g., seed number vs seed size)(Coomes & Grubb 2003). In wind-dispersed species of Asteraceae, dispersal poten-tial depends on the diaspore (seed) mass and the size of the dispersal architecture(pappus) (Matlack 1987). Several studies have demonstrated an interaction betweenseed mass, pappus size, and drop time, where small seeds with a long pappus havea lower terminal velocity (staying aloft longer) than large seeds with a short pap-pus (O’Connell & Eckert 2001; Soons & Heil 2002; Gravuer et al. 2003; Riba et al.2005). On the other hand, larger seeds often have higher germination success andmore rapid seedling growth (Dolan 1984; Westoby et al. 1996; Turnbull et al. 2004).This suggests a trade-off in seed traits, where capacity for long distance dispersalcomes at the cost of reduced establishment success. Additionally, some researchershave reported a negative relationship between seed size and germination speed, wheresmaller seeds tend to germinate more rapidly than larger seeds (Grime et al. 1981;Hendrix 1984; Bu et al. 2016, but see also McKersie et al. 1981; Piper 1986; Eriksson1999 who found no relationship). The combination of these seed trade-offs suggestthe possiblity of contrasting dispersal syndromes: one with large, slow-germinatingseeds that have high germination success but low dispersal potential, and one withsmall seeds capable of rapid germination and long distance dispersal but with reducedgermination success.These divergent “establishment” and “dispersal” oriented life histories may con-tribute to patterns of GP if closely related sexuals and asexuals have divergent strate-gies, but inferring the role of selection in generating any observed differences is com-plicated by the association of apomixis with polyploidy and hybrid ancestry. Whilepolyploidy is typically invoked to explain phenotypic shifts in GP contexts, the ef-fects of apomixis may also contribute to differences in dispersal and establishmenttraits. Because of this, there is some tension in what mechanisms and outcomes87would be expected when considering both GP and life history trade-off theory. Poly-ploidy alone may result in a shift to either an establishment strategy (e.g., via theproduction of larger seeds that improve germination success in new environments) ora dispersal strategy (e.g., by producing seeds with a larger pappus:seed ratio that arebetter able to achieve long distance dispersal). While allopolyploids are historicallyconsidered to experience more extreme transformations due to heterosis (Rieseberget al. 1999; Chen 2010; Paun et al. 2011), there is some evidence that autopolyploids(originating from a single parent species) can also experience phenotypic shifts thatmay significantly impact niche divergence (Paterson 2005; Ramsey 2011; Spoelhofet al. 2017).Apomixis is not thought to result in phenotypic shifts directly, but it may indi-rectly contribute to differences in dispersal and establishment traits through cloneselection. For example, high-dispersal sexual genotypes may be “frozen” by apomixisand, if successful, spread across the landscape into previously uncolonized regions(Lynch 1984; Vrijenhoek & Parker 2009). Similarly, clonal lineages may be underselection to escape to marginal habitats where biological interactions are less intense,favoring genotypes with strong colonization potential (i.e., Red Queen hypothesis;Asker & Jerling 1992). Having high dispersal ability may also be favored in marginalenvironments with high levels of mortality and environmental disturbance, allowingclones to maintain metapopulations in the face of repeated local extinctions whileavoiding the deleterious effects of inbreeding depression (due to frequent genetic bot-tlenecks) that would occur in sexual metapopulations (Haag & Ebert 2004).Townsendia hookeri provides an excellent system in which to test for differences inearly life history traits within the context of GP. It is a wind-dispersed species withdiploid outcrossing and autopolyploid apomictic forms that exhibit a classic pat-tern of GP (Bierzychudek 1985). T. hookeri provides a broad geographic range over88which to examine differences in dispersal and establishment traits between sexualsand apomicts. In this study, we ask the following questions: (1) Are there differencesin dispersal and/or establishment traits between the two forms of T. hookeri? (2)If there are differences, are these consistent with the existence of opposing strategies(establishment vs dispersal) as predicted by life history trade-off theory? (3) Overall,do apomicts show evidence of enhanced colonization ability relative to their sexualprogenitors? To address these questions, we conducted lab and greenhouse experi-ments to test for differences in early life history traits (seed mass, pappus morphology,terminal velocity of diaspores, germination success and speed, and seedling growth)between diploid sexual and polyploid apomictic forms of T. hookeri using naturalpopulations sampled throughout their ranges.4.2 Materials and methods4.2.1 Study systemTownsendia hookeri is a diminutive perennial member of the sunflower family (Aster-aceae) with two forms: diploids that reproduce sexually and autopolyploids (mainlytriploid) that reproduce via gametophytic apomixis (Beaman 1957a). Sexual individ-uals are self-incompatible, while apomicts set seed autonomously without the needfor pollen to fertilize the endosperm. T. hookeri is assumed to exhibit a generalistpollination syndrome like most species of Asteraceae (Mani & Saravanan 1999), anda variety of pollinators (including bees, flies, and beetles) have been seen visitingTownsendia species in the field (personal observation; Tepedino et al. 2004). Despitediffering in ploidy and reproductive mode, the two forms are morphologically indis-tinguishable in the field. Polyploids can be identified by pollen staining (polyploidsproduce larger pollen grains than diploids and have much lower pollen viability) or by89using flow cytometry to estimate genome size; to date, information from more than90 populations throughout the range reveal a nearly perfect association between lowpollen viability and high genome size (Thompson & Whitton 2006; Garani 2014; Lee2015), suggesting that diploids are sexual and polyploids are apomictic (Thompson& Whitton 2006; Thompson et al. 2008).Sexual populations have a much smaller range than the apomicts and primarilyoccur between Boulder, CO and Laramie, WY (Lee 2015). Apomictic populationsextend from southern WY along the eastern side of the Rocky Mountians as far northas British Columbia. A small number of diploid-sexual and polyploid-apomict popu-lations can also be found in a disjunct distribution in the Yukon territory (Thompson& Whitton 2006; Garani 2014).4.2.2 Seed collectionIn order to compare diaspore traits between reproductive modes, we used wind-dispersed diaspores (technically achenes, but from now on referred to as “seeds”) col-lected between 2008 and 2013 from five diploid sexual and seven polyploid apomicticpopulations from across T. hookeri’s range, including northern populations in BritishColumbia and the Yukon territory (Table 4.1). For the germination and greenhouseexperiments, we collected seeds from six diploid sexual populations and six polyploidapomict populations in the spring of 2013 and 2014 (Table 4.1). Seeds were stored atroom temperature in paper coin envelopes. The ploidy level of each population usedin this study was assessed previously by flow cytometry (Lee 2015).904.2.3 Seed dispersal traits and terminal velocitymeasurementsWe selected two seeds per maternal plant from up to ten moms per population, usingonly filled and darkly coloured seeds (traits that indicate viability; Garani 2014)with an intact pappus. In order to restore the pappus to a comparably open state(relative to the somewhat flattened state that resulted from storage in the collectionenvelopes), we placed each seed on wet filter paper in a sealed petri dish for ∼24hours, then removed them and allowed them to air-dry for another 24 hours. At thispoint the pappus had achieved a more regular form consistent with what is seen atthe time of field collection. For each seed, we estimated the mass using an analyticalbalance, recorded the number of bristles, and measured the length of two bristles fromthe center-apex using an ocular micrometer. We measured the angle of attack, themaximum angle across the open pappus bristles centered on their point of attachmentto the seed proper, using a protractor.We estimated terminal velocity by dropping individual seeds (=diapores) down aclear 120 cm long plexiglass tube. Seeds were dropped by holding the pappus withtweezers and releasing. We recorded the bottom 50 cm with a video camera (shootingat 30 frames per second), thus allowing the seeds to reach terminal velocity over thefirst 70 cm of the drop. We calculated the terminal velocity of each seed by dividingthe drop time by drop distance (50 cm). We took great care throughout not todamage or disturb the structure of the pappus.4.2.4 Germination in the labWe chose 11 maternal plants (hereafter known as “moms”) from each population, and32 “viable” seeds (based on the same criteria as above) from each mom. We assessedgermination in agar-filled petri dishes (hereafter referred to as “plates”). Each plate91was divided into four quadrants, and four seeds from a single mom were randomlyassigned to each quadrant. Seeds were arranged within quadrants to avoid contact.Due to the large number of seeds, we filled the plates over the course of three days(beginning on June 18, 2014). We stacked the plates on the lab bench at roomtemperature (21◦C) away from direct sunlight, and re-ordered the stacks every 3-4days. We checked the plates every 1-2 days until no new seeds had germinated for 10days, and scored germination as successful based on the emergence of both a radicleand a pair of cotyledons.4.2.5 Seedling traitsWe transplanted seeds from agar into racks of cone-tainer pots (Proptek - Watsonville,California) 2-3 days after germination over a period of two weeks. We filled the potswith a well-draining soil mixture (4 potting soil: 2 sand: 1 perlite) and placed theseedling racks in growth chambers (Conviron, various models). Growth chamberswere set for 12h light (20◦C) / 12h dark (10◦C) cycles. We hand-watered seedlingsdaily in the week following transplanting, and every two days after that. We rotatedthe racks between chambers once per week to account for potential differences amonggrowth chamber models. After most seedlings reached a height of at least 2 cm (earlyJuly 2013), we hardened the seedlings in a greenhouse at the UBC farm. The green-house environment was regulated by Argus Controls (12h light - 20◦C / 12h dark- 12◦C), and the seedlings were bottom-watered on flood tables. Because we sub-sequently transplanted the seedlings into a common garden experiment, we broughtthem outside two weeks before transplanting to harden them to UV light (3-5 hoursper day). Just prior to moving to field sites, we censused all seedlings for survivaland estimated plant size using two leaf measurements: total number of leaves, andthe leaf length of one haphazardly chosen representative leaf per seedling. Thus, our92measurement of seedling size reflects the size achieved from the day of germinationuntil the census approximately two months later. While not every seed germinated onthe same day (and therefore did not have equal time to grow until the census), 75%of seeds that germinated did so within a five day period, so we expect the variationin total growth time had minimal effects on our growth estimates.4.2.6 Data analysisWe analyzed differences in seed traits (terminal velocity, angle of attack, bristle length,number of bristles, and seed mass) between mating systems using mixed effects mod-els. Terminal velocity, angle of attack, and bristle length were analyzed using linearmixed effect models (LMMs), and number of bristles was analyzed used generalizedlinear mixed effects models (GLMMs) with a log link function for poisson data. Mat-ing system was a fixed factor; population and mom (nested within population) wereincluded as random factors. We used linear models to test the effects of each seed traiton terminal velocity separately. Traits were centered and scaled (using the scale()function in R), and included as a fixed effect in each model without random effects.We used GLMMs to analyze differences in germination success (logit link - bino-mial) and germination speed (log link - poisson, due to speed being represented byday counts) between mating systems. Mating system was a fixed factor; plate, pop-ulation, and mom (nested within population) were included as random factors. Wealso used GLMMs to analyze differences between mating systems for seedling survival(logit link - binomial) and leaf number (log link - poisson), and LMMs for leaf length.Mating system was a fixed factor; rack, population, and mom (nested within popu-lation) were included as random factors. Including rack as a random effect resultedin model singularities for the leaf length and leaf number models, so we removedit from those analyses. Because populations were the main level of replication, we93used subsequent mixed models to determine whether there were differences betweenpopulations, with population as a fixed effect and the same random factors as above.We performed the mixed-effect analyses using the lmer and glmer functions im-plemented in the lme4 package (Bates et al. 2015) and fixed effect analyses using thelm function in R version 3.6.1 (R Core team 2019). We evaluated the strength ofeffects by using a combination of likelihood ratio tests (LRTs; comparing fully fit-ted model to a model with the tested term removed) and model predictions using theggeffects package (Lu¨decke 2018). Statistical differences between groups were inferredwhen both the LRT P-values were below 0.05 and confidence intervals from modelpredictions were largely non-overlapping.4.3 Results4.3.1 Seed dispersal traits and terminal velocityVariation in terminal velocity was associated with variation in each of the four seedtraits, with bristle length and angle of attack having the strongest effects (Table 4.2;Figure 4.1). Apomictic seeds had longer pappus bristles and a wider angle of attackthan sexual seeds, but there were no differences in seed mass or bristle number betweenmating systems (Table 4.3; Figure 4.3). Terminal velocity differed between matingsystems (Table 4.3; Figure 4.1A) and populations (df=14, χ2=31.697, P=7.74x10-08),with apomicts having lower terminal velocity (0.91 cm/s) than sexuals (1.38 cm/s;Table 4.4), indicating increased dispersal ability of apomicts over sexuals. Despitedifferences in trait mean intercepts, the slopes of traits and terminal velocity did notdiffer between sexuals and apomicts (Figure 4.1), suggesting that the relationshipsbetween traits and terminal velocity do not differ between mating types.94Figure 4.1: Seed dispersal traits. A) Terminal velocity by mating system.Boxplots provide summaries of the data, with jittered points representing individualseeds. To the right of the boxplots, points and error bars represent predicted meansand their 95% confidence intervals. B-E) The relationship between terminal velocityand measured seed dispersal traits by mating system. The slopes are shown for datafrom sexuals and apomicts separately.4.3.2 Germination traitsGermination success differed between mating systems and among populations (Table4.5; Figure 4.2A, Figure 4.4B), with apomicts having much higher germination success(78%) than sexuals (59%; Table 4.6). Germination speed (mean days to germination)did not differ between mating systems (Figure 4.2B), but there were differences amongpopulations, with three of the sexual populations germinating slower than the ma-95Figure 4.2: Germination and seedling traits by mating system. Boxplots providesummaries of the data, with jittered points representing averages for each mom foreach trait. To the right of the boxplots, points and error bars represent predictedmeans and their 95% confidence intervals.jority of the apomictic populations (B53, B42, and B49; Table 4.5, Table 4.7; Figure4.4C). Apomicts had lower variance than sexuals in both germination traits (Table4.7, Table 4.6; Figure 4.2A&B).964.3.3 Seedling traitsSeedling survival was high overall (>90% with the exception of population L06),as expected for greenhouse-raised plants. Seedling survival differed between matingsystems and among populations, with apomicts having slightly lower survival (92.4%)than sexuals (95.7%; Table 4.5, Table 4.8; Figure 4.2C, Figure 4.4D). While LRTsdemonstrate that mating system affected survival, there was partial overlap in themodel prediction confidence intervals, indicating that this is not a particularly strongeffect. Neither leaf length or number of leaves differed between mating systems, butboth differed among populations (Table 4.5, Table 4.9, Table 4.10; Figure 4.2D&E,Figure 4.4E&F).4.4 DiscussionWe found that the seeds of apomicts had both lower terminal velocity and highergermination success than sexual seeds, indicating that apomicts have a distinct ad-vantage in dispersal ability and colonization success. This advantage is offset by lowersurvival of apomicts at the seedling stage, but the survival difference is much smallerthan the differences found in terminal velocity and germination success. In combi-nation with reproductive assurance conferred by apomixis, these differences in earlylife history traits likely contribute to the apomicts having a much broader range thansexuals in T. hookeri. Interestingly, our results are largely not in line with what wouldbe expected under life history trade-off theory, as the apomicts’ dispersal advantagecoincided with increased germination success as opposed to a decrease (though thecost to seedling survival is in the expected direction).As seen in other wind-dispersed diaspores (O’Connell & Eckert 2001; Soons & Heil2002; Gravuer et al. 2003; Riba et al. 2005), terminal velocity was affected by all of97the seed traits measured, with longer bristles and wider angle of attack contributingthe most to reducing drop times. These two dispersal traits were also the only ones todiffer between sexuals and apomicts. The apomicts’ increased pappus volume resultedin a ∼50% decrease in terminal velocity, indicating a considerable dispersal advantagein comparison to sexuals. Though seed mass did not differ by mating system (Figure4.3D, Table 4.4), we did observe a positive relationship between mass and terminalvelocity (Figure 4.1E) as found in other systems (Greene & Johnson 1993; Greene& Quesada 2005). The fact that there were no appreciable differences in seed massbetween mating types indicates that a large increase in pappus volume comes witha relatively small cost to maternal investment in T. hookeri, at least on a per-seedbasis.Apomicts had much higher germination success than sexuals, once again sug-gesting increased colonization potential. Life history trade-off theory predicts thatthis improvement would be associated with an increase in seed mass, but that wasnot the case in our study, as we found no appreciable differences in seed mass be-tween sexuals and apomicts. While germination speed did not differ overall betweenmating types, three sexual populations took longer than others to germinate, indi-cating that apomicts were more consistent in their ability to germinate rapidly thansexuals. It is important to note that rapid and increased germination success is notalways advantageous, as delayed germination or dormancy can provide a temporal es-cape from unfavourable environmental conditions (Finch-Savage & Leubner-Metzger2006). However, we have seen very little evidence that T. hookeri forms a seed bankor displays consistent dormancy (95% of established seedlings germinated in the firstyear in Chapter 3), which suggests that germination success is driven primarily byseed viability. As these seeds were collected in nature, we cannot discount the pos-sibility that environmental and/or genetic maternal effects contributed to the trait98differences that we found (Roach & Wulff 1987), but it is notable that apomictsshowed considerably less variance than sexuals in both germination traits. This find-ing suggests that germination performance is influenced at least to some extent bygenotype. Following this logic, it is not surprising that we found more variabilityamong outcrossing populations than among clonal populations when germinating un-der uniform and benign conditions.Seedling survival was quite high in the greenhouse experiment for both matingsystems, but slightly lower for apomicts. This is the one result from this study thatdiminishes the colonization advantage of apomicts and lends partial support to atrade-off between dispersal and competitive ability. On the other hand, we foundno differences in leaf number or leaf length between mating systems at the seedlingstage, so whatever contributed to the reduction in survival did not seem to affectgrowth rates (which we might expect if apomicts had reduced maternal investment).While the differences in seedling survival were quite small, these effects might beamplified under field conditions. If so, the reduced seedling survival of apomictsmight counteract their germination advantage in nature, which could help explainwhy we found no overall differences in seedling establishment plots (which incorporategermination success and seedling survival) in the field (Chapter 3).As is true of most GP studies, we are hampered in our power to disentanglethe effects of ploidy and apomixis in T. hookeri, though in our case we can dis-regard the effect of hybridization because apomicts are autopolyploids (Thompson& Whitton 2006). While we cannot discount the possibility that polyploidy hasaffected the anatomy and physiology of apomicts as found in some polyploids (Ram-sey & Schemske 2002; Nuismer & Cunningham 2005; Cohen et al. 2013; Elia´sˇova´ &Mu¨nzbergova´ 2014; Gao et al. 2016), until now, no consistent morphological differ-ences between diploids and polyploids have been documented in T. hookeri (Beaman991957a; Reveal 1970). The differences that we document in two seed traits (angle ofattack and bristle length) are the first indication of macro-morphological differencesbetween the mating types. In general, polyploids that benefit from increased germi-nation success also have larger seeds (Beaulieu et al. 2007; Haouala et al. 2009; Hahnet al. 2013), so it is surprising that polyploid T. hookeri have such a large germina-tion advantage without any obvious differences in seed size. The interaction betweenploidy, seed size, and germination performance may be complex, however. Bretag-nolle (1995) compared seeds of equal mass between diploid and polyploid Dactylisglomerata and found that polyploid seeds both germinated faster and had highergermination success than diploid seeds, indicating that the effects of ploidy on ger-mination may not be simply due to increased seed size. However, it is also importantto point out that despite the long recognized importance of polyploidy in the diver-sification of angiosperms, there remain few studies of life history differences betweencytotypes in nature or in common gardens from which we might attempt to extractgeneral predictions. Additionally, some broad-scale studies indicate that polyploidsdo not consistently show niche shifts (Glennon et al. 2014) or range shifts (Martin& Husband 2009) in comparison to related diploids (though see Baniaga et al. 2020and Prentis et al. 2008 for opposing trends), which suggests that other traits besidesgenome duplication may play a role in patterns of GP.The differences in colonization potential found in this study may be more easilyexplained by the joint effects of apomixis and polyploidy (as opposed to the effects ofpolyploidy alone). Assuming that apomicts originated from sexual populations andsubsequently spread, selection may have favored those clonal lineages with enhancedcolonization ability. In fact, given that our study necessarily surveys clones that havesuccessfully established, it is in some ways unsurprising that we find enhanced colo-nization and establishment traits. We expect that as clones arise, those that succeed100in colonizing new areas will be the lineages that have a successful suite of traits,protected (by apomixis) from the action of recombination, which would breakdownbeneficial allele combinations (Lynch 1984). This is consistent with our finding thatapomicts had reduced variance in germination speed, germination success, and ter-minal velocity in comparison to sexuals (Table 4.7, Table 4.6, Table 4.4). While thisreduced variance gives apomicts improved colonization trait performance on average,their values largely lie within the range of sexual traits. This pattern is particularlynoticeable in the germination traits; some sexual populations/families have the abil-ity to germinate rapidly and with high success, but sexuals have considerably morevariance than apomicts in both traits. This interpretation is in line with the fact thatapomictic populations and regions are largely monoclonal (Chapter 2), as we wouldexpect less variation among individuals within a clone than among siblings from anoutcrossing mom.An open question is whether reproductive assurance alone is enough to accountfor range differences found in plant GP systems. Chrtek et al. (2018) investigateddiaspore differences within a GP context in Hieracium alpinum, which, like T. hook-eri, also has self-incompatible diploids and autonomously apomictic autopolyploids.While they did find slight differences in terminal velocity favoring apomicts, theyalso found that sexuals had much higher germination success than apomicts, whichis opposite to what we found. This highlights that dynamics can differ significantlyeven between otherwise very similar systems. Regardless of the mechanism in othersystems, it is clear that in T. hookeri apomicts have early life history traits thatgive them an advantage in dispersal and colonization potential. This advantage iscompounded by the fact that apomicts already benefit from full reproductive assur-ance, and that sexuals are self-incompatible and entirely dependent on local mateavailability and pollinator services. This study provides some of the only evidence101thus far that dispersal ability can differ between sexual and asexual propagules ina GP context (but also see O’Connell & Eckert 2001; Coughlan et al. 2014). Thisis interesting, because dispersal can provide an escape (in the absence of sexual re-combination) from biotic stressors (i.e., Red Queen hypothesis; Glesener & Tilman1978; Judson 1997; Hartfield & Keightley 2012) and environmental instability (Haag& Ebert 2004). Comparisons of dispersal and colonization traits in other systems willprovide much needed context, and may provide an alternative (or complementary)explanation to patterns of GP driven by reproductive assurance.102Figure 4.3: Seed dispersal traits by mating system. Boxplots and jitter representindividual seeds, while points and error bars (right) represent predicted means with95% confidence intervals.103Figure 4.4: Terminal velocity, germination, and seedling traits by population.Points and error bars represent predicted means with 95% confidence intervals.Within each reproductive type, populations are arranged from south (on the left) tonorth.104Table 4.1: List of populations used in this study. ‘Experiment’ indicates theexperiment in which each population was used; S = seed, G = germination /greenhouse.Mating system Population Ploidy State/Province Latitude Longitude ExperimentSexual B53 2 CO 39.8913 -105.2655 GB42 2 CO 40.1070 -105.2836 S,GB46 2 CO 40.5384 -105.1335 S,GB49 2 CO 40.8403 -105.3214 S,GL11 2 WY 41.2400 -105.4342 GL12 2 WY 41.2528 -105.4062 S,GC59 2 YT 62.1235 -136.2575 SApomictic B52 3 CO 40.9410 -106.0499 SL06 3 WY 41.1430 -106.0375 GL16 3 WY 41.3064 -105.5216 S,GL17 3 WY 41.3864 -105.4735 GL18 3 WY 41.5478 -110.5527 SL20 3 WY 41.6802 -110.6119 SL40 3 WY 42.7973 -105.8814 SL41 3 WY 42.8087 -105.3778 SC86 3 MT 45.6619 -110.4752 GC21 3 MT 46.3329 -111.5563 SC85 3 MT 46.6505 -111.7200 GC27 3 MT 47.5500 -112.4667 GC43 3 BC 50.6057 -116.0595 SC54 4 YT 60.8360 -135.9143 STable 4.2: Relationships between seed traits and terminal velocity.Trait df Estimate (slope) R2 F PAngle of attack (scaled) 1 -0.2943 0.4112 122.2 2.00E-16Bristle length (scaled) 1 -0.2042 0.198 43.21 5.44E-10Number of bristles (scaled) 1 -0.0885 0.0372 6.753 0.0102Mass (scaled) 1 0.1352 0.0868 16.63 6.88E-05105Table 4.3: Likelihood ratio test statistics for the fixed effect of mating system onseed traits.Trait (seed) Source of variation df χ2 PTerminal velocity Mating system 5 12.163 0.0005Angle of attack 5 8.7348 0.0031Bristle length 5 6.3478 0.0118Number of bristles 4 2.1093 0.1464Weight 5 0.0247 0.8751Table 4.4: Terminal velocity - summary statistics by mating system andpopulation.Mating system Population n (seeds) Raw mean (cm/s) SE Predicted means (cm/s) Lower CI Upper CISexual 90 1.3800 0.0503 1.3581 1.2014 1.5147Apomictic 87 0.9120 0.0311 0.8975 0.7514 1.0436Sexual B42 28 1.3974 0.0883 1.3974 1.2242 1.5706B46 20 1.1968 0.0602 1.1968 0.9919 1.4018B49 20 1.6873 0.1228 1.6873 1.4823 1.8922L12 14 1.2901 0.1354 1.2901 1.0451 1.5351C59 8 1.1208 0.1146 1.1208 0.7967 1.4449Apomictic B52 20 0.9550 0.0485 0.9550 0.7500 1.1600L16 20 1.0943 0.0722 1.0943 0.8893 1.2992L40 20 0.7792 0.0765 0.7792 0.5743 0.9842L41 10 0.8438 0.0654 0.8438 0.5539 1.1337C21 6 0.9131 0.0273 0.9131 0.5389 1.2874C43 5 0.6100 0.0463 0.6241 0.2343 1.0139C54 6 0.9613 0.0660 0.9613 0.5871 1.3356106Table 4.5: Likelihood ratio test statistics for the separate fixed effects of matingsystem and population on germination and seedling traits.Trait (germination/seedling) Source of variation df χ2 PGermination success Mating system 5 6.0279 0.01408Germination speed 5 1.1727 0.2789Seedling survival 5 5.8538 0.0155Seedling leaf number 4 0.8434 0.3584Seedling leaf length 5 1.3304 0.2487Germination success Population 14 68.26 2.61x10-10Germination speed 14 123.82 2.20x10-16Seedling survival 14 22.121 0.0235Seedling leaf number 13 31.833 0.0008Seedling leaf length 14 39.701 4.03x10-05Table 4.6: Germination success - summary statistics by mating system andpopulation.Mating system Population n (seeds) Raw mean SE Predicted means Lower CI Upper CISexual 2112 0.5885 0.0107 0.6067 0.4720 0.7269Apomictic 2112 0.7822 0.0090 0.8219 0.7273 0.8887Sexual B53 352 0.3125 0.0247 0.2355 0.1385 0.3711B42 352 0.7074 0.0243 0.7357 0.5977 0.8391B46 352 0.6960 0.0246 0.7351 0.5964 0.8390B49 352 0.5142 0.0267 0.5210 0.3689 0.6694L11 352 0.5284 0.0266 0.5207 0.3654 0.6722L12 352 0.7727 0.0224 0.8283 0.7158 0.9023Apomictic L06 352 0.7813 0.0221 0.8166 0.7012 0.8942L16 352 0.7330 0.0236 0.7518 0.6179 0.8501L17 352 0.7244 0.0238 0.7466 0.6112 0.8467C86 352 0.7869 0.0219 0.8343 0.7238 0.9063C85 352 0.8182 0.0206 0.8706 0.7751 0.9293C27 352 0.8494 0.0191 0.8749 0.7842 0.9308107Table 4.7: Germination speed - Summary statistics by mating system andpopulation.Mating system Population n (seeds) Raw mean (days) SE Predicted means (days) Lower CI Upper CISexual 1204 6.8900 0.1010 6.7509 5.8104 7.8436Apomictic 1607 6.0800 0.0489 5.9938 5.1632 6.9579Sexual B53 104 8.1731 0.2848 8.0632 7.1891 9.0437B42 237 9.5316 0.3233 9.2823 8.5150 10.1188B46 240 5.8292 0.1283 5.7032 5.2020 6.2526B49 171 8.3918 0.2298 8.5059 7.7525 9.3325L11 183 4.9945 0.1591 4.9095 4.4141 5.4606L12 269 5.3532 0.1127 5.3279 4.8608 5.8398Apomictic L06 263 5.8555 0.1270 5.8174 5.3152 6.3672L16 253 5.8340 0.1158 5.9009 5.3915 6.4584L17 250 5.2320 0.1117 5.1097 4.6571 5.6063C86 268 6.4701 0.1196 6.5273 5.9754 7.1301C85 278 6.2518 0.1056 6.1440 5.6206 6.7162C27 295 6.7017 0.1167 6.5922 6.0458 7.1879Table 4.8: Seedling Survival - summary statistics by mating system andpopulation.Mating system Population n (seedlings) Raw mean SE Predicted means Lower CI Upper CISexual 1297 0.957 0.00565 0.9692 0.9531 0.9798Apomictic 1595 0.924 0.00663 0.9424 0.9191 0.9594Sexual B53 127 0.9449 0.0203 0.9572 0.8954 0.9832B42 257 0.9728 0.0102 0.9805 0.9520 0.9922B46 247 0.9676 0.0113 0.9762 0.9447 0.9900B49 188 0.9255 0.0192 0.9394 0.8801 0.9704L11 213 0.9577 0.0138 0.9690 0.9292 0.9868L12 265 0.9585 0.0123 0.9702 0.9358 0.9865Apomictic L06 265 0.8642 0.0211 0.8854 0.8092 0.9337L16 248 0.9556 0.0131 0.9654 0.9266 0.9841L17 248 0.9234 0.0169 0.9418 0.8906 0.9698C86 276 0.9420 0.0141 0.9508 0.9055 0.9750C85 280 0.9429 0.0139 0.9541 0.9108 0.9769C27 278 0.9173 0.0166 0.9344 0.8822 0.9645108Table 4.9: Seedling leaf number - Summary statistics by mating system andpopulation.Mating system Population n (seedlings) Raw mean SE Predicted means Lower CI Upper CISexual 1142 6.7723 0.0846 6.7040 6.3836 7.0405Apomictic 1388 6.9561 0.0799 6.9240 6.6050 7.2578Sexual B53 104 6.3462 0.2478 6.3226 5.7460 6.9570B42 231 6.7576 0.2033 6.7345 6.2968 7.2026B46 227 6.3612 0.1609 6.3389 5.9186 6.7890B49 151 6.6755 0.2199 6.6346 6.1399 7.1692L11 189 6.8571 0.2144 6.8327 6.3620 7.3382L12 240 7.3542 0.1991 7.3254 6.8592 7.8234Apomictic L06 208 6.9327 0.2032 6.9293 6.4665 7.4253L16 222 6.4595 0.1820 6.4292 6.0027 6.8859L17 216 7.7037 0.2198 7.6707 7.1811 8.1936C86 253 6.7668 0.1968 6.7418 6.3130 7.1997C85 242 6.4504 0.1577 6.4432 6.0248 6.8906C27 247 7.4575 0.1997 7.4233 6.9624 7.9148Table 4.10: Seedling leaf length - Summary statistics by mating system andpopulation.Mating system Population n (seedlings) Raw mean (mm) SE Predicted means (mm) Lower CI Upper CISexual 1142 15.1226 0.2125 14.9640 13.0280 15.9000Apomictic 1388 15.7421 0.1837 15.6882 14.7771 16.5993Sexual B53 104 14.7212 0.6650 14.6596 13.0986 16.2206B42 231 16.9740 0.5694 16.9179 15.8158 18.0201B46 227 16.2467 0.4629 16.1825 15.0752 17.2899B49 151 13.5364 0.5304 13.5261 12.2383 14.8139L11 189 13.7196 0.4844 13.7504 12.5638 14.9370L12 240 14.5542 0.4022 14.5292 13.4308 15.6277Apomictic L06 208 14.8077 0.4448 14.8202 13.6725 15.9678L16 222 14.9189 0.4344 14.8963 13.7830 16.0096L17 216 15.1250 0.4504 15.1072 13.9823 16.2320C86 253 15.8379 0.4708 15.7929 14.7199 16.8660C85 242 16.5703 0.4223 16.5830 15.4975 17.6685C27 247 16.8988 0.4508 16.8808 15.8058 17.9559109Chapter 5ConclusionA multitude of factors interact to contribute to patterns of geographical partheno-genesis. These factors take a different emphasis depending on both the biologicalcontext (i.e., system-specific differences between sexuals and asexuals) and the theo-retical context (i.e., proposed explanations and their specific ecological/evolutionarylens) being explored. In my thesis, I have explored GP in Townsendia hookeri witha diversity of approaches in order to reflect the diversity of models that have beenadvanced to explain the pattern. Each chapter views GP in the system through adifferent lens, examining population genetic diversity (Chapter 2), ecological differ-entiation (Chapter 3), and life history traits (Chapter 4). I believe that this is aneffective approach to take in any system with any question, but it is particularlyimportant for complex biological phenomena such as GP.One of the primary goals of this thesis was to assess what I consider the “default”explanation for patterns of GP: that asexuals have larger ranges than sexuals dueto the colonization advantages imparted by reproductive assurance (“Baker’s Laweffects”). In Chapter 3, we found that habitat suitable for sexuals exists within theapomictic range, which provides considerable evidence that the sexual range is limited110by dispersal. This is underscored by the results of Chapter 4, which indicate thatapomicts have early life history traits that likely give them a colonization advantagein comparison to sexuals. The biology of T. hookeri also points to the likelihoodof Baker’s Law effects being a driver for GP. Sexuals and apomicts lie on polaropposite ends of the mating system spectrum, with sexuals being self-incompatible(thus dependent on mates and pollinator services) and apomicts being autonomous(thus having no dependence on mates or even pollen). Given these attributes, itwould be surprising if divergent colonization potential did not play a role in GP inT. hookeri. While most apomicts are pseudogamous (requiring pollen to fertilize theendosperm and set seed) and autonomous apomixis is relatively rare (but commonin Asteraceae), most pseudogamous apomicts are self-fertile and are therefore stillcapable of uniparental reproduction (Ho¨randl 2010). This, coupled with the fact thatthe sexual progenitors of most apomicts are self-incompatible (Asker & Jerling 1992),suggests that Baker’s Law effects are likely to play at least some role in most plantGP systems due to the inherent dichotomy between their reproductive modes.Baker’s law effects are unlikely to be the sole cause of GP in Townsendia hookeri,however. Given enough time, sexuals should be able to expand into suitable habitatin the apomictic range. The fact that they have not suggests that there are barriersat their northern range limit preventing, or at least greatly slowing their expansion.Habitat at the sexuals’ northern range edge appears to be unsuitable for sexuals (asevidenced by the generally poor performance in the SOg region; Chapter 3), whichcould mean that sexuals are unable to push past this region into more favorable envi-ronments further north in the apomictic range. However, natural sexual populationsin this region are some of the largest we have seen (personal observation), whichgoes against the interpretation that the habitat is unsuitable. Although it is possi-ble that these populations are demographic sinks being maintained by high dispersal111from the sexual interior (RL>NL; Hargreaves et al. 2014), it seems unlikely that sinkpopulations would be larger than sources if habitat is less suitable in sink sites.A more likely scenario is that suitable sites just beyond the sexual edge (AOg)that are occupied by apomicts can exclude sexuals through reproductive interference.Mixed sexual/apomictic populations are exceedingly rare (we have only detected twoinstances; Chapter 2), and apomicts appear capable of reducing sexual seed set (andeven producing apomictic offspring) when acting as pollen donors for sexual ovules(Garani 2014). Apomicts exhibit reduced pollen viability in comparison to sexuals(Garani 2014), which can favor stable coexistence between sexuals and apomicts(Britton & Mogie 2001), but even modest asymmetrical reproductive interferencecan, over time, lead to sexual extirpation in mixed populations (Mogie 2011). Giventhis, the sexual invasion of sites where apomicts have precedence seems unlikely.As it stands, apomicts seem well-primed to push southward into the sexual range;apomictic edge populations performed best in the sexual edge region (Chapter 3), andone of the natural populations that we studied in this region, although first identifiedas sexual, appears to harbor apomicts at low frequencies (population L62; Chapter 2;unpublished flow cytometry data). If apomicts are able to establish in this regionin sufficient numbers, they may displace sexuals over time, leading to sexual rangecontraction (Mogie 1992).Assuming apomicts have considerable colonization and demographic advantagesover sexuals, strong barriers must be in place that are preventing their southernexpansion. The poor performance of apomictic populations in the sexual interior (Sg;Chapter 3) suggests that habitat is not suitable for apomicts there, but the causeof the gradient in habitat quality is not entirely obvious. The sexual interior gardensites had higher vegetation cover and historically warmer temperatures than the othersites, suggesting that interspecific competition (and potentially other biotic factors)112is more intense in this region. In addition, populations from the sexual interior hadhigher fitness in their home region than any of the other source regions, and tendedto have longer leaves as well. These results fit quite well with the expectations of theRed Queen hypothesis, which predicts that sexuals will be able to adapt and persistin the face of biotic pressures (perhaps by growing longer leaves to compensate forincreased light competition?), while asexuals will be excluded due to their inabilityto adapt (Glesener & Tilman 1978). Our experiment was not explicitly designed totest for biotic interactions, but our data suggest that Red Queen effects make for apromising candidate theory warranting further investigation.Another interpretation is that apomicts are simply not well adapted to the envi-ronmental (abiotic) conditions in the southern portion of the range. This fits withthe predictions from ecological niche models, which suggest that the climatic nicheof apomicts does not extend into the southernmost (sexual interior) portion of therange (Lee 2015). Though there was some overlap in niche characteristics betweensexuals and apomicts in T. hookeri, Lee (2015) found that some important climaticvariables differed between regions occupied by the two types. While precipitationin the wettest quarter was important for both sexuals and apomicts, niche modelsput greater emphasis on temperature variables for sexuals and precipitation variablesfor apomicts. These results hint at potential environmental characteristics that areimportant in limiting the apomicts southern range, but additional experiments willbe required to pinpoint whether apomicts are affected by abiotic factors per se orinstead by ecological factors (including biotic factors) that correlate with climate.It is important to tie ecological and life history differences to patterns of clonaldiversity, because attributes benefitting range expansion are expected to be conservedwithin apomictic lineages (Lynch 1984). The results from Chapter 2 reveal largelymono-clonal populations, including two widespread clonal lineages - one of which113encompasses much of the apomictic interior range, and another that occurs at theapomicts’ southern range edge. In Chapter 4, we present evidence that apomicticseeds have improved germination success and seed dispersal architecture in compari-son to sexuals, and several of the populations investigated were found to be membersof the two widespread clonal lineages characterized in Chapter 2. Assuming that theseearly life history traits benefitting colonization were conserved within apomictic lin-eages, they likely contributed to the geographic expansion of apomictic populations.The pattern of a single widespread clone occupying much of the range is in linewith the general-purpose genotype hypothesis, which predicts that varying abioticconditions will select for the rare clonal lineage that is resilient to environmentalunpredictability (Vrijenhoek & Parker 2009). Improved colonization ability (alongwith reproductive assurance conferred by apomixis) is likely to be a boon to apomictssubjected to intense abiotic pressures, as they will be better able to re-establishpopulations and bounce back from low densities after the frequent disturbances thatare assumed to occur in “marginal” environments (Baker 1965). It is important tonote that while several of the proposed hypotheses for GP suggest that asexualswill dominate in “marginal” environments, Tilquin & Kokko (2016) point out that“marginality” is a vague concept, and is defined somewhat differently across theGP literature. That being said, our anecdotal experience with these plants in thefield (particularly in the apomictic range) certainly agrees with several of the definedfeatures of marginal habitats, including low amount and diversity of resources, lowpopulation productivity/density, and high habitat openness and vacancy.The pattern of clonal diversity is somewhat different in the southern portion ofthe range (around the southern apomictic edge), where we found a greater diversityof apomictic lineages (Chapter 2). These results are partially in line with the ex-pectations of the frozen niche variation hypothesis, which predicts that competitive114interactions will select for an array of specialized clonal lineages that partition theresource space in order to avoid niche overlap with sexual progenitors and other clones(Vrijenhoek & Parker 2009). If this is the case, this may help explain why we findprimarily monotypic sexual and apomictic populations in the overlap range despitebeing in relatively close proximity, though this would depend on the scale of envi-ronmental variation in this region. In addition, given that apomicts originate fromsexuals (Whitton et al. 2008), it is not surprising that we find increased apomicticdiversity closer to the sexual range.Discussions of the diversity of clones and frequency of origins brings to mind thelong-term dynamics and trajectories of sexuals and apomicts in T. hookeri. While thedemographic benefits of asexuality may be advantageous in the short-term, the accu-mulation of deleterious mutations (Muller 1964) and inability to adapt to a changingenvironment are predicted to lead to the demise of asexual lineages over the long-term.In fact, there are no known “ancient” apomictic taxa, and all apomicts appear to haveclosely-related sexual progenitors (van Dijk 2009). This highlights that apomicts arelikely dependent on their sexual progenitors for much-needed injections of geneticdiversity, whether it be from de novo origins of new clonal lineages directly fromsexuals, or via occasional hybridization between sexuals and apomicts.Thompson & Whitton (2006) found that apomixis likely originated multiple timesin T. hookeri, and that apomicts most likely spread from glacial refugia (in Coloradoand the Yukon territory) into their current distribution post-glaciation. Relation-ships between sexual populations and apomictic lineages (Figure 2.4, Chapter 2) arelargely consistent with this interpretation. Given the challenges to long-term persis-tence of apomictic lineages, we might predict cyclical extinctions of individual clonesand their periodic replacement by clones originating from sexual parts of the range.The widespread clone occupying large swaths of the apomictic range has proven to115be quite successful so far, but it would likely be vulnerable to major environmentalshifts or the advent of a new predator/pathogen. If this clonal genotype were to beextirpated, it would leave a large portion of the range unoccupied, which could takea long time to be recolonized considering there are no nearby sexual populations tosupply a replacement general-purpose genotype. It is somewhat ironic that apomictsare dependent on the existence of sexuals over the long run, but at the same timemay be contributing to sexual range contraction over time (Mogie 1992) via repro-ductive interference and “contagious” apomixis (Garani 2014). Whatever is limitingthe apomicts’ southern expansion (and potential decimation of sexual populations)may turn out to be the salvation of the system in the long run.It seems clear that no single proposed model for GP is sufficient to capture therange of possibilities across the diverse array of taxa in which the pattern occurs.Within plants alone, apomixis can take various forms and exists within strikinglydifferent evolutionary contexts. GP in Townsendia hookeri is set against a backdropof autonomous apomixis vs obligate outcrossing, autopolyploidy, rare co-occurrence,and putatively infrequent origins from sexual populations. We can use this informa-tion (along with carefully designed experiments) to make predictions about the causesof range divergence between sexuals and apomicts. In another system that is charac-terized by (for example) pseudogamous apomixis, allopolyploidy, mixed populations,and frequent origins through hybridization, we should expect very different dynamicsand mechanisms underlying GP. This is not to say that it is futile to attempt broadgeneralizations for complex eco-evolutionary phenomena, but rather that predictionsshould be made using a comprehensive compendium of theory and biological context,interspersed with plenty of caveats. Perhaps we would benefit from framing our hy-potheses within a nested set of “if-then” statements, much like a dichotomous key butfor concepts instead of species. Trying to speak the cryptic language of geographical116parthenogenesis can be maddening at times, but I prefer this reality to the alterna-tive. If biological phenomena were easy to explain, there would far less intrigue, andfar fewer labyrinthine systems to devote entire theses to exploring...and where wouldbe the fun in that?117BibliographyArnaud-Haond, S., Duarte, C.M., Alberto, F. & Serra˜o, E.A. 2007. Standardizingmethods to address clonality in population studies. Molecular Ecology 16:5115–5139. → page 33Asker, S. & Jerling, L. 1992. Apomixis in plants. CRC Press, Boca Raton. → pages65, 88, 111Baack, E.J. & Rieseberg, L.H. 2007. A genomic view of introgression and hybridspeciation. Current Opinion in Genetics and Development 17: 513–518. → page6Baker, H. 1965. Characteristics and modes of origin of weeds. In: The Genetics ofColonizing Species (H.G. Baker & G.L. Stebbins, eds), pp. 147–142. AcademicPress, New York. → pages 7, 16, 18, 36, 39, 70, 114Baker, H.G. 1955. Self-Compatibility and Establishment After ‘Long-Distance’Dispersal. Evolution 9: 347–349. → pages 4, 49, 86Baniaga, A.E., Marx, H.E., Arrigo, N. & Barker, M.S. 2020. Polyploid plants havefaster rates of multivariate niche differentiation than their diploid relatives.Ecology Letters 23: 68–78. → pages 6, 100Bates, D., Ma¨chler, M., Bolker, B. & Walker, S. 2015. Fitting Linear Mixed-EffectsModels Using lme4. Journal of Statistical Software 67. → pages 59, 94Beaman, J.H. 1954. Chromosome Numbers, Apomixis, and InterspecificHybridization in the Genus Townsendia. Madron˜o 12: 169–180. → page 10Beaman, J.H. 1957a. The systematics and evolution of Townsendia (Compositae).Contributions from the Gray Herbarium of Harvard University 183: 1–151. →pages 20, 35, 89, 99Beaman, J.H. 1957b. Townsendia. In: Flora of North America North of Mexico,1993rd edn, chap. Townsendia, pp. 3: 193–194. New York and Oxford. → pages10, 11, 12118Beaulieu, J.M., Moles, A.T., Leitch, I.J., Bennett, M.D., Dickie, J.B. & Knight,C.A. 2007. Correlated evolution of genome size and seed mass. New Phytologist173: 422–437. → page 100Bell, G. 1982. The Masterpiece of Nature: the Evolution and Genetics of Sexuality.Croom Helm, London. → pages 3, 9, 15, 17Bengtsson, B.O. 2009. Asex and evolution: A very large-scale overview. In: LostSex: The Evolutionary Biology of Parthenogenesis (I. Schon, K. Martens & P.J.van Dijk, eds), chap. 1, pp. 1–20. Springer, Dordrecht. → pages 5, 6Bierzychudek, P. 1985. Patterns in plant parthenogenesis. Experientia 41:1255–1264. → pages 3, 10, 15, 18, 19, 37, 48, 52, 85, 88Bierzychudek, P. 1989. Environmental sensitivity of sexual and apomicticAntennaria: do apomicts have general-purpose genotypes? Evolution 43:1456–1466. → page 19Blanquart, F., Kaltz, O., Nuismer, S.L. & Gandon, S. 2013. A practical guide tomeasuring local adaptation. Ecology Letters 16: 1195–1205. → page 51Bock, D.G., Kantar, M.B., Caseys, C., Matthey-Doret, R. & Rieseberg, L.H. 2018.Evolution of invasiveness by genetic accommodation. Nature Ecology andEvolution 2: 991–999. → page 34Bretagnolle, F. 1995. The Influence of Seed Size Variation on Seed Germination andSeedling Vigour in Diploid and Tetraploid Dactylis glomerata L. Annals ofBotany 76: 607–615. → page 100Britton & Mogie 2001. Poor male function favours the coexistence of sexual andasexual relatives. Ecology Letters 4: 116–121. → pages 49, 69, 112Brochmann, C., Gabrielsen, T.M., Nordal, I., Landvik, J.Y. & Elven, R. 2003.Glacial Survival or tabula rasa? The History of North Atlantic Biota Revisited.Taxon 52: 417–450. → pages 12, 65Bu, H.Y., Wang, X.J., Zhou, X.H., Qi, W., Liu, K., Ge, W.J., Xu, D.H. & Zhang,S.T. 2016. The ecological and evolutionary significance of seed shape and volumefor the germination of 383 species on the eastern Qinghai-Tibet plateau. FoliaGeobotanica 51: 333–341. → page 87Burnside, W.R., Erhardt, E.B., Hammond, S.T. & Brown, J.H. 2014. Rates ofbiotic interactions scale predictably with temperature despite variation. Oikos123: 1449–1456. → page 67119Case, T.J., Holt, R.D., McPeek, M.A. & Keitt, T.H. 2005. The community contextof species’ borders: Ecological and evolutionary perspectives. Oikos 108: 28–46.→ page 66Chen, Z.J. 2010. Molecular mechanisms of polyploidy and hybrid vigor. Trends inPlant Science 15: 57–71. → page 88Chen, Z.J. 2013. Genomic and epigenetic insights into the molecular bases ofheterosis. Nature Reviews Genetics 14: 471–482. → page 6Christenhusz, M.J. & Byng, J.W. 2016. The number of known plant species in theworld and its annual increase. Phytotaxa 261: 201–217. → page 10Chrtek, J., Hartmann, M., Mra´zova´, V., ZdvoÅŹa´k, P., Sˇtefa´nek, M. & Mra´z, P.2018. Seed traits, terminal velocity and germination in sexual diploid andapomictic triploid Hieracium alpinum (Asteraceae): Are apomicts betterdispersers? Flora: Morphology, Distribution, Functional Ecology of Plants 240:76–81. → pages 5, 86, 101Chuang, A. & Peterson, C.R. 2016. Expanding population edges: theories, traits,and trade-offs. Global Change Biology 22: 494–512. → page 86Cohen, H., Fait, A. & Tel-Zur, N. 2013. Morphological, cytological and metabolicconsequences of autopolyploidization in Hylocereus (Cactaceae) species. BMCPlant Biology 13: 1471–2229. → page 99Comai, L. 2005. The advantages and disadvantages of being polyploid. NatureReviews Genetics 6: 836–846. → pages 5, 18Coomes, D.A. & Grubb, P.J. 2003. Colonization, tolerance, competition andseed-size variation within functional groups. Trends in Ecology and Evolution 18:283–291. → page 87Cosendai, A. & Ho¨randl, E. 2010. Cytotype stability, facultative apomixis andgeographical parthenogenesis in Ranunculus kuepferi (Ranunculaceae). Annals ofbotany 105: 457–70. → page 36Cosendai, A.C., Wagner, J., Ladinig, U., Rosche, C. & Ho¨randl, E. 2013.Geographical parthenogenesis and population genetic structure in the alpinespecies Ranunculus kuepferi (Ranunculaceae). Heredity 110: 560–569. → pages19, 36, 39Coughlan, J.M., Han, S., Stefanovic´, S. & Dickinson, T.A. 2017. Widespreadgeneralist clones are associated with range and niche expansion in allopolyploidsof Pacific Northwest Hawthorns (Crataegus L.). Molecular Ecology 26:5484–5499. → page 19120Coughlan, J.M., Stefanovic´, S. & Dickinson, T.A. 2014. Relative resource allocationto dispersal and competition demonstrates the putative role of hybridity ingeographical parthenogenesis. Journal of Biogeography 41: 1603–1613. → pages5, 86, 102Dolan, R.W. 1984. The Effect of Seed Size and Maternal Source on Individual Sizein a Population of Ludwigia leptocarpa (Onagraceae). American Journal ofBotany 71: 1302–1307. → page 87Douhovnikoff, V. & Dodd, R.S. 2003. Intra-clonal variation and a similaritythreshold for identification of clones: Application to Salix exigua using AFLPmolecular markers. Theoretical and Applied Genetics 106: 1307–1315. → page 32Dufresne, F., Stift, M., Vergilino, R. & Mable, B.K. 2014. Recent progress andchallenges in population genetics of polyploid organisms: an overview of currentstate of the art molecular and statistical tools. Molecular Ecology 23: 40–69. →page 22Elia´sˇova´, A. & Mu¨nzbergova´, Z. 2014. Higher seed size and germination rate mayfavour autotetraploids of Vicia cracca L. (Fabaceae). Biological Journal of theLinnean Society 113: 57–73. → page 99Ellstrand, N.C. & Roose, M.L. 1987. Patterns of genotypic diversity in clonal plantspecies. American Journal of Botany 74: 123–131. → page 19Eriksson, O. 1999. Seed size variation and its effect on germination and seedlingperformance in the clonal herb Convallaria majalis. Acta Oecologica 20: 61–66.→ page 87Finch-Savage, W.E. & Leubner-Metzger, G. 2006. Seed dormancy and the control ofgermination. New Phytologist 171: 501–523. → page 98Gao, R., Wang, H., Dong, B., Yang, X., Chen, S., Jiang, J., Zhang, Z., Liu, C.,Zhao, N. & Chen, F. 2016. Morphological, Genome and Gene Expression Changesin Newly Induced Autopolyploid Chrysanthemum lavandulifolium (Fisch. exTrautv.) Makino. International Journal of Molecular Sciences 17: 1690. → page99Garani, A. 2014. Evidence for reproductive interference between sexual andapomictic populations of the Easter Daisy (Townsendia hookeri, Asteraceae).Master’s thesis, University of British Columbia, Vancouver. → pages12, 20, 37, 53, 69, 90, 91, 112, 116Garrison, E. 2019. vcflib; https://github.com/vcflib/vcflib. → page 23121Garrison, E. & Marth, G. 2012. Haplotype-based variant detection from short-readsequencing . → page 23Gascoigne, J., Berec, L., Gregory, S. & Courchamp, F. 2009. Dangerously fewliaisons: A review of mate-finding Allee effects. Population Ecology 51: 355–372.→ pages 4, 49, 63Gaston, K.J. 2003. The structure and dynamics of geographic ranges. OxfordUniversity Press, Oxford. → page 66Geyer, C.J., Ridley, C.E., Latta, R.G., Etterson, J.R. & Shaw, R.G. 2013. Localadaptation and genetic effects on fitness: Calculations for exponential familymodels with random effects. Annals of Applied Statistics 7: 1778–1795. → page57Geyer, C.J., Wagenius, S. & Shaw, R.G. 2007. Aster models for life history analysis.Biometrika 94: 415–426. → page 57Glennon, K.L., Ritchie, M.E. & Segraves, K.A. 2014. Evidence for sharedbroad-scale climatic niches of diploid and polyploid plants. Ecology Letters 17:574–582. → pages 5, 100Glesener, R.R.R. & Tilman, D. 1978. Sexuality and the Components ofEnvironmental Uncertainty: Clues from Geographic Parthenogenesis inTerrestrial Animals. The American Naturalist 112: 659–673. → pages3, 8, 15, 50, 85, 102, 113Goudet, J. 2005. HIERFSTAT, a package for R to compute and test hierarchicalF-statistics. Molecular Ecology Notes 5: 184–186. → page 26Gravuer, K., Von Wettberg, E.J. & Schmitt, J. 2003. Dispersal biology of Liatrisscariosa var. novae-angliae (Asteraceae), a rare New England grassland perennial.American Journal of Botany 90: 1159–1167. → pages 87, 97Gray, A. 1888. Synoptical Flora of North America: The Gamopetalae., 2nd edn.Smithsonian Institution, Washington. → page 10Greene, D.F. & Johnson, E.A. 1993. Seed Mass and Dispersal Capacity inWind-Dispersed Diaspores. Oikos 67: 69–74. → page 98Greene, D.F. & Quesada, M. 2005. Seed size, dispersal, and aerodynamic constraintswithin the Bombacaceae. American Journal of Botany 92: 998–1005. → page 98Grime, J.P., Mason, G., Curtis, A.V., Rodman, J. & Band, S.R. 1981. AComparative Study of Germination Characteristics in a Local Flora. The Journalof Ecology 69: 1017–1059. → page 87122Grossenbacher, D., Briscoe Runquist, R., Goldberg, E.E. & Brandvain, Y. 2015.Geographic range size is predicted by plant mating system. Ecology Letters 18:706–713. → page 4Haag, C.R. & Ebert, D. 2004. A new hypothesis to explain geographicparthenogenesis. annales zoologici fennici 41: 539–544. → pages 4, 18, 37, 88, 102Hahn, M.A., Lanz, T., Fasel, D. & Mu¨ller-Scha¨rer, H. 2013. Increased seed survivaland seedling emergence in a polyploid plant invader. American Journal of Botany100: 1555–1561. → page 100Haouala, R., Ouerghemmi, S., Tarchoune, A. & Boughanmi, N. 2009. Improvementof Trigonella maritima Delilee x. Poir. germination by polyploidization. PakistanJournal of Botany 41: 3001–3008. → page 100Hargreaves, A.L., Germain, R.M., Bontrager, M., Persi, J. & Angert, A.L. 2020.Local Adaptation to Biotic Interactions: A Meta-analysis across Latitudes. TheAmerican Naturalist 195: 000–000. → pages 9, 67Hargreaves, A.L., Samis, K.E. & Eckert, C.G. 2014. Are Species’ Range LimitsSimply Niche Limits Writ Large? A Review of Transplant Experiments beyondthe Range. The American Naturalist 183: 157–173. → pages50, 51, 52, 53, 54, 63, 64, 66, 67, 69, 112Hartfield, M. & Keightley, P.D. 2012. Current hypotheses for the evolution of sexand recombination. Integrative Zoology 7: 192–209. → page 102Hendrix, S.D. 1984. Variation in Seed Weight and Its Effects on Germination inPastinaca sativa L. (Umbelliferae). American Journal of Botany 71: 795–802. →page 87Herman, P., Zdvora´k, P., Hartmann, M., Chrtek, J., Sˇtefa´nek, M. & Mra´z, P. 2017.The Red Queen hypothesis and geographical parthenogenesis in the alpinehawkweed Hieracium alpinum (Asteraceae). Biological Journal of the LinneanSociety 122: 681–696. → page 68Hersh, E., Grimm, J. & Whitton, J. 2016. Attack of the clones: reproductiveinterference between sexuals and asexuals in the Crepis agamic complex. Ecologyand Evolution 6: 6473–6483. → page 49Hewitt, G.M. 2004. Genetic consequences of climatic oscillations in the Quaternary.Philosophical Transactions of the Royal Society B: Biological Sciences 359:183–195. → page 37Holt, R.D. & Keitt, T.H. 2000. Alternative causes for range limits: Ametapopulation perspective. Ecology Letters 3: 41–47. → page 64123Hooker, S.W.J. 1840. Flora boreali-americana: or the botany of the northern partsof British America. Treuttel and Wurtz, London. → page 10Ho¨randl, E. 2006. The complex causality of geographical parthenogenesis. NewPhytologist 171: 525–538. → page 86Ho¨randl, E. 2009. Geographical Parthenogenesis: Opportunities for Asexuality BT -Lost Sex. In: Lost Sex (I. Sch o n, K. Martens & P. Dijk, eds), pp. 161–186.Springer Netherlands, Dordrecht. → pages 4, 65Ho¨randl, E. 2010. The evolution of self-fertility in apomictic plants. Sexual PlantReproduction 23: 73–86. → pages 38, 65, 111Horandl, E. & Paun, O. 2007. Patterns and sources of genetic diversity in apomicticplants: implications for evolutionary potentials. Regnum Vegetabile pp. 169–194ST – Patterns and sources of genetic dive. → pages 19, 32, 34, 36Jiao, Y., Wickett, N.J., Ayyampalayam, S., Chanderbali, A.S., Landherr, L., Ralph,P.E., Tomsho, L.P., Hu, Y., Liang, H., Soltis, P.S., Soltis, D.E., Clifton, S.W.,Schlarbaum, S.E., Schuster, S.C., Ma, H., Leebens-Mack, J. & DePamphilis, C.W.2011. Ancestral polyploidy in seed plants and angiosperms. Nature 473: 97–100.→ page 10Johnson, N.A. 2010. Hybrid incompatibility genes: Remnants of a genomicbattlefield? Trends in Genetics 26: 317–325. → page 6Jombart, T. 2008. adegenet: a R package for the multivariate analysis of geneticmarkers. Bioinformatics 24: 1403–1405. → page 24Jombart, T. & Ahmed, I. 2011. adegenet 1.3-1: new tools for the analysis ofgenome-wide SNP data. Bioinformatics 27: 3070–3071. → page 24Jombart, T., Devillard, S. & Balloux, F. 2010. Discriminant analysis of principalcomponents: a new method for the analysis of genetically structured populations.BMC Genetics 11: 94. → page 25Judson, O.P. 1997. A model of asexuality and clonal diversity: Cloning the RedQueen. Journal of Theoretical Biology 186: 33–40. → page 102Kamvar, Z.N., Brooks, J.C. & Gru¨nwald, N.J. 2015. Novel R tools for analysis ofgenome-wide population genetic data with emphasis on clonality. Frontiers inGenetics 6: 1–10. → page 24Kamvar, Z.N., Tabima, J.F. & Gru¨nwald, N.J. 2014. Poppr : an R package forgenetic analysis of populations with clonal, partially clonal, and/or sexualreproduction. PeerJ 2: e281. → page 24124Kawecki, T.J. 2008. Adaptation to Marginal Habitats. Annual Review of Ecology,Evolution, and Systematics 39: 321–342. → pages 51, 67Kearney, M. 2005. Hybridization, glaciation and geographical parthenogenesis.Trends in Ecology & Evolution 20: 495–502. → pages 6, 18, 37, 86Kenny, N.T. 1996. A Test of the General-Purpose Genotype Hypothesis in Sexualand Asexual Erigeron Species. American Midland Naturalist 136: 1. → pages8, 40Kirchheimer, B., Wessely, J., Gattringer, A., Hu¨lber, K., Moser, D., Schinkel, C.C.,Appelhans, M., Klatt, S., Caccianiga, M., Dellinger, A., Guisan, A., Kuttner, M.,Lenoir, J., Maiorano, L., Nieto-Lugilde, D., Plutzar, C., Svenning, J.C., Willner,W., Ho¨randl, E. & Dullinger, S. 2018. Reconstructing geographicalparthenogenesis: effects of niche differentiation and reproductive mode onHolocene range expansion of an alpine plant. Ecology Letters 21: 392–401. →page 51Kleiman, D. & Aarssen, L.W. 2007. The leaf size/number trade-off in trees. Journalof Ecology 95: 376–382. → page 70Knaus, B. & Gru¨nwald, N. 2016. VcfR: a package to manipulate and visualize VCFformat data in R. vcfr: a package to manipulate and visualize variant call formatdata in R p. 041277. → page 24Knaus, B.J. & Gru¨nwald, N.J. 2017. vcfr: a package to manipulate and visualizevariant call format data in R. Molecular Ecology Resources 17: 44–53. → page 24Knaus, B.J. & Gru¨nwald, N.J. 2018. Inferring Variation in Copy Number UsingHigh Throughput Sequencing Data in R. Frontiers in Genetics 9: 1–12. → page26Kyogoku, D. 2015. Reproductive interference: ecological and evolutionaryconsequences of interspecific promiscuity. Population Ecology 57: 253–260. →pages 13, 49Larsen, E. 1927. A Revision of the Genus Townsendia. Annals of the MissouriBotanical Garden 14: 1–34. → page 10Lee, C. 2015. Diversification and evolution of Townsend’s daisies (Townsendia-Asteraceae): A phylogenetic and niche modelling perspective. Ph.D. thesis,University of British Columbia. → pages 11, 13, 19, 20, 21, 51, 53, 66, 68, 90, 113Lee-Yaw, J.A., Kharouba, H.M., Bontrager, M., Mahony, C., Csergo, A.M., Noreen,A.M., Li, Q., Schuster, R. & Angert, A.L. 2016. A synthesis of transplantexperiments and ecological niche models suggests that range limits are often nichelimits. Ecology Letters 19: 710–722. → pages 50, 64125Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G.,Abecasis, G. & Durbin, R. 2009. The Sequence Alignment/Map format andSAMtools. Bioinformatics 25: 2078–2079. → page 23Louthan, A.M., Doak, D.F. & Angert, A.L. 2015. Where and When do SpeciesInteractions Set Range Limits? Trends in Ecology and Evolution 30: 780–792. →pages 8, 67Lovell, J.T., Williamson, R.J., Wright, S.I., McKay, J.K. & Sharbel, T.F. 2017.Mutation Accumulation in an Asexual Relative of Arabidopsis. PLoS Genetics13: 1–14. → page 35Lu¨decke, D. 2018. ggeffects: Tidy Data Frames of Marginal Effects from RegressionModels. Journal of Open Source Software 3: 772. → pages 59, 94Lyman, J.C. & Ellstrand, N.C. 1984. Clonal diversity in Taraxacum officinale(Compositae), an apomict. Heredity 53: 1–10. → page 19Lynch, M. 1984. Destabilizing hybridization, general-purpose genotypes andgeographic parthenogenesis. Quarterly Review of Biology 59: 257–290. → pages7, 16, 70, 88, 101, 113Majesky´, Ä., Vasˇut, R.J. & Kitner, M. 2015. Genotypic diversity of apomicticmicrospecies of the Taraxacum scanicum group (Taraxacum sect.Erythrosperma). Plant Systematics and Evolution 301: 2105–2124. → page 36Mani, M.S. & Saravanan, J.M. 1999. Pollination Ecology and Evolution inCompositae (Asteraceae). Science Publishers, Enfield, New Hampshire. → page89Martens, K., Loxdale, H.D. & Scho¨n, I. 2009. The Elusive Clone - In Search of ItsTrue Nature and Identity. In: Lost Sex: The Evolutionary Biology ofParthenogenesis (I. Schon, K. Martens & P.J. van Dijk, eds). Springer,Dordrecht. → pages 17, 32, 34Martin, S.L. & Husband, B.C. 2009. Influence of phylogeny and ploidy on speciesranges of North American angiosperms. Journal of Ecology 97: 913–922. →pages 5, 100Matlack, G.R. 1987. Diaspore size, shape, and fall behaviour in wind-dispersedplant species. American Journal of Botany 74: 1150–1160. → page 87Matz, M.V. 2018. Fantastic Beasts and How To Sequence Them: EcologicalGenomics for Obscure Model Organisms. Trends in Genetics 34: 121–132. →page 35126Mau, M., Lovell, J.T., Corral, J.M., Kiefer, C., Koch, M.A., Aliyu, O.M. & Sharbel,T.F. 2015. Hybrid apomicts trapped in the ecological niches of their sexualancestors. Proceedings of the National Academy of Sciences of the United Statesof America 112: E2357–65. → page 65McKersie, B.D., Tomes, D.T. & Yamamoto, S. 1981. Effect of seed size ongermination, seedling vigor, electrolyte leakage, and establishment of bird’s-foottrefoil (Lotus corniculatus L.). Canadian Journal of Plant Science 61: 337–343.→ page 87Menken, S.B.J., Smit, E. & Nijs, H.J.C.M.D. 1995. Genetical Population Structurein Plants: Gene Flow between Diploid Sexual and Triploid Asexual Dandelions(Taraxacum Section Ruderalia). Evolution 49: 1108. → page 19Mogie, M. 1992. The Evolution of Asexual Reproduction in Plants, 1st edn.Chapman & Hall, London; New York. → pages 37, 112, 116Mogie, M. 2011. Pollen profile, spatial structure, and access to sex in asexualhermaphrodites. Biological Journal of the Linnean Society 103: 954–966. →page 112Moles, A.T., Bonser, S.P., Poore, A.G., Wallis, I.R. & Foley, W.J. 2011a. Assessingthe evidence for latitudinal gradients in plant defence and herbivory. FunctionalEcology 25: 380–388. → page 67Moles, A.T., Wallis, I.R., Foley, W.J., Warton, D.I., Stegen, J.C., Bisigato, A.J.,Cella-Pizarro, L., Clark, C.J., Cohen, P.S., Cornwell, W.K., Edwards, W., Ejrnæs,R., Gonzales-Ojeda, T., Graae, B.J., Hay, G., Lumbwe, F.C., Magan˜a-Rodr´ıguez,B., Moore, B.D., Peri, P.L., Poulsen, J.R., Veldtman, R., von Zeipel, H., Andrew,N.R., Boulter, S.L., Borer, E.T., Campo´n, F.F., Coll, M., Farji-Brener, A.G.,de Gabriel, J., Jurado, E., Kyhn, L.A., Low, B., Mulder, C.P., Reardon-Smith,K., Rodr´ıguez-Vela´zquez, J., Seabloom, E.W., Vesk, P.A., van Cauter, A.,Waldram, M.S., Zheng, Z., Blendinger, P.G., Enquist, B.J., Facelli, J.M., Knight,T., Majer, J.D., Mart´ınez-Ramos, M., McQuillan, P. & Prior, L.D. 2011b.Putting plant resistance traits on the map: A test of the idea that plants arebetter defended at lower latitudes. New Phytologist 191: 777–788. → page 9Muller, H.J. 1964. The relation of recombination to mutational advance. MutationResearch 106: 2–9. → pages 4, 115Murray, M.G. & Thompson, W.F. 1980. Rapid isolation of high molecular weightplant DNA. Nucleic Acids Research 8: 4321–4326. → page 22Nei, M. 1972. Genetic Distance between Populations. The American Naturalist106: 283–292. → page 26127Normand, S., Treier, U.A., Randin, C., Vittoz, P., Guisan, A. & Svenning, J.C.2009. Importance of abiotic stress as a range-limit determinant for Europeanplants: Insights from species responses to climatic gradients. Global Ecology andBiogeography 18: 437–449. → page 66Novoplansky, A. 2009. Picking battles wisely: Plant behaviour under competition.Plant, Cell and Environment 32: 726–741. → page 67Nuismer, S.L. & Cunningham, B.M. 2005. Selection for Phenotypic DivergenceBetween Diploid and Autotetraploid Heuchera Grossulariifolia. Evolution 59:1928–1935. → page 99O’Connell, L.M. & Eckert, C.G. 2001. Differentiation in reproductive strategybetween sexual and asexual populations of Antennaria parlinii ( Asteraceae ).Evolutionary Ecology Research 3: 311–330. → pages 5, 87, 97, 102Otto, S.P. & Whitton, J. 2000. Polyploid Incidence and Evolution. Annual reviewof genetics . → pages 6, 10Pannell, J.R., Auld, J.R., Brandvain, Y., Burd, M., Busch, J.W., Cheptou, P.O.,Conner, J.K., Goldberg, E.E., Grant, A.G., Grossenbacher, D.L., Hovick, S.M.,Igic, B., Kalisz, S., Petanidou, T., Randle, A.M., de Casas, R.R., Pauw, A.,Vamosi, J.C. & Winn, A.A. 2015. The scope of Baker’s law. New Phytologist208: 656–667. → pages 4, 49Paradis, E. & Schliep, K. 2019. Ape 5.0: An environment for modern phylogeneticsand evolutionary analyses in R. Bioinformatics 35: 526–528. → page 26Parker, E.D., Selander, R.K., Hudson, R.O. & Lester, L.J. 1977. Genetic Diversityin Colonizing Parthenogenetic Cockroaches. Evolution 31: 836. → pages 7, 18Paterson, A.H. 2005. Polyploidy, evolutionary opportunity, and crop adaptation.Genetica 123: 191–196. → page 88Paun, O., Bateman, R.M., Fay, M.F., Luna, J.A., Moat, J., Hedre´n, M. & Chase,M.W. 2011. Altered gene expression and ecological divergence in siblingallopolyploids of Dactylorhiza (Orchidaceae). BMC Evolutionary Biology 11:113. → page 88Peppe, D.J., Royer, D.L., Cariglino, B., Oliver, S.Y., Newman, S., Leight, E.,Enikolopov, G., Fernandez-Burgos, M., Herrera, F., Adams, J.M., Correa, E.,Currano, E.D., Erickson, J.M., Hinojosa, L.F., Hoganson, J.W., Iglesias, A.,Jaramillo, C.A., Johnson, K.R., Jordan, G.J., Kraft, N.J., Lovelock, E.C., Lusk,C.H., Niinemets, U¨., Pen˜uelas, J., Rapson, G., Wing, S.L. & Wright, I.J. 2011.Sensitivity of leaf size and shape to climate: Global patterns and paleoclimaticapplications. New Phytologist 190: 724–739. → page 67128Piper, J.K. 1986. Germination and growth of bird-dispersed plants: effects of seedsize and light on seedling vigor and biomass allocation. American Journal ofBotany 73: 959–965. → page 87Poland, J.A., Brown, P.J., Sorrells, M.E. & Jannink, J.L. 2012. Development ofhigh-density genetic maps for barley and wheat using a novel two-enzymegenotyping-by-sequencing approach. PLoS ONE 7. → page 22Poorter, H., Niklas, K.J., Reich, P.B., Oleksyn, J., Poot, P. & Mommer, L. 2012.Biomass allocation to leaves, stems and roots: Meta-analyses of interspecificvariation and environmental control. New Phytologist 193: 30–50. → page 70Prentis, P.J., Wilson, J.R., Dormontt, E.E., Richardson, D.M. & Lowe, A.J. 2008.Adaptive evolution in invasive species. Trends in Plant Science 13: 288–294. →page 100Prevosti, A., Ocan˜a, J. & Alonso, G. 1975. Distances between populations ofDrosophila subobscura, based on chromosome arrangement frequencies.Theoretical and Applied Genetics 45: 231–241. → page 24Pulliam, H.R. 2000. On the relationship between niche and distribution. EcologyLetters 3: 349–361. → page 51Puritz, J.B. 2019. ddocent; http://www.ddocent.com. → page 23Puritz, J.B., Hollenbeck, C.M. & Gold, J.R. 2014. dDocent: A RADseq,variant-calling pipeline designed for population genomics of non-model organisms.PeerJ 2014. → page 23R Core team 2019. R: A language and environment for statistical computing. →pages 24, 57, 94Ramsey, J. 2011. Polyploidy and ecological adaptation in wild yarrow. Proceedingsof the National Academy of Sciences 108: 7096–7101. → page 88Ramsey, J. & Schemske, D.W. 2002. Neopolyploidy in flowering plants. AnnualReview of Ecology and Systematics 33: 589–639. → page 99Reveal, J. 1970. A revision of the Utah species of Townsendia (Compositae). GreatBasin Naturalist 30: 23–52. → page 100Reynolds, H.L. & Pacala, S.W. 1993. An analytical treatment of root-to-shoot ratioand plant competition for soil nutrient and light. American Naturalist 141:51–70. → page 70129Riba, M., Mignot, A., Fre´ville, H., Colas, B., Imbert, E., Vile, D., Virevaire, M. &Olivieri, I. 2005. Variation in dispersal traits in a narrow-endemic plant species,Centaurea corymbosa pourret. (Asteraceae). Evolutionary Ecology 19: 241–254.→ pages 87, 97Rieseberg, L.H., Archer, M.A. & Wayne, R.K. 1999. Transgressive segregation,adaptation and speciation. Heredity 83: 363–372. → pages 6, 88Roach, D.A. & Wulff, R.D. 1987. Maternal Effects in Plants. Annual Review ofEcology and Systematics 18: 209–235. → page 99Roughgarden, J. 1974. Evolution of Niche Width. The American Naturalist 108:142–143. → page 17Samis, K.E. & Eckert, C.G. 2009. Ecological Correlates of Fitness across theNorthern Geographic Range Limit of a Pacific Coast Dune Plant. Ecology 90:3051–3061. → page 64Samis, K.E., Lo´pez-Villalobos, A. & Eckert, C.G. 2016. Strong geneticdifferentiation but not local adaptation toward the range limit of a coastal duneplant. Evolution 70: 2520–2536. → page 64Schemske, D.W. & Lande, R. 1985. The Evolution of Self-Fertilization andInbreeding Depression in Plants. II. Empirical Observations. Evolution 39: 41.→ page 4Schon, I., Butlin, R.K., Griffiths, H.I. & Martens, K. 1998. Slow molecular evolutionin an ancient asexual ostracod. Proceedings of the Royal Society B: BiologicalSciences 265: 235–242. → page 16Sears, C.J. & Whitton, J. 2016. A reexamination of the North American Crepisagamic complex and comparison with the findings of Babcock and Stebbins’classic biosystematic monograph. American Journal of Botany 103: 1289–1299.→ page 10Sexton, J.P., McIntyre, P.J., Angert, A.L. & Rice, K.J. 2009. Evolution and Ecologyof Species Range Limits. Annual Review of Ecology, Evolution, and Systematics40: 415–436. → page 50Silvertown, J. 2008. The Evolutionary Maintenance of Sexual Reproduction:Evidence from the Ecological Distribution of Asexual Reproduction in ClonalPlants. International Journal of Plant Sciences 169: 157–168. → page 19Soons, M. & Heil, G. 2002. Reduced colonization capacity in fragmentedpopulations of wind-dispersed grassland forbs. Journal of Ecology 90: 1033–1043.→ pages 87, 97130Spoelhof, J.P., Soltis, P.S. & Soltis, D.E. 2017. Pure polyploidy: Closing the gaps inautopolyploid research. Journal of Systematics and Evolution 55: 340–352. →page 88Stebbins, G.L. 1940. The significance of polyploidy in plant evolution. TheAmerican Naturalist 74: 54–66. → page 85Stevens, S.M. & Emery, N.C. 2015. Dispersal limitation and populationdifferentiation in performance beyond a northern range limit in an asexuallyreproducing fern. Diversity and Distributions 21: 1242–1253. → page 64Te Beest, M., Le Roux, J.J., Richardson, D.M., Brysting, A.K., Suda, J., Kubesˇova´,M. & Pysˇek, P. 2012. The more the better? The role of polyploidy in facilitatingplant invasions. Annals of Botany 109: 19–45. → pages 5, 18Tepedino, V.J., Sipes, S.D. & Griswold, T.L. 2004. Reproduction and demographyof Townsendia aprica (Asteraceae), a rare endemic of the Southern Utah Plateau.Western North American Naturalist 64: 465–470. → page 89Thompson, S.L., Choe, G., Ritland, K. & Whitton, J. 2008. Cryptic sex withinmale sterile polyploid populations of the Easter daisy, Townsendia hookeri.International Journal of Plant Sciences 169: 183–193. → page 90Thompson, S.L. & Whitton, J. 2006. Patterns of recurrent evolution and geographicparthenogenesis within apomictic polyploid Easter daisies (Townsendia hookeri).Molecular Ecology 15: 3389–3400. → pages 12, 20, 21, 37, 90, 99, 115Tilquin, A. & Kokko, H. 2016. What does the geography of parthenogenesis teachus about sex? Philosophical Transactions of the Royal Society B: BiologicalSciences 371: 2015.0538. → pages 3, 7, 9, 16, 48, 50, 64, 70, 85, 114Turnbull, L.A., Coomes, D., Hector, A. & Rees, M. 2004. Seed mass and thecompetition/colonization trade-off: competitive interactions and spatial patternsin a guild of annual plants. Journal of Ecology 92: 97–109. → page 87Van Der Hulst, R.G., Mes, T.H., Falque, M., Stam, P., Den Nijs, J.C. & Bachmann,K. 2003. Genetic structure of a population sample of apomictic dandelions.Heredity 90: 326–335. → pages 19, 36van Dijk, P.J. 2009. Apomixis: Basics for non-botanists. In: Lost Sex: TheEvolutionary Biology of Parthenogenesis (I. Scho¨n, K. Martens & P.J. van Dijk,eds), chap. 3, pp. 47–62. Springer, Dordrecht. → page 115Vandel, A. 1928. La parte´nogene`se ge´ographique. Contribution a` l’e´tude biologiqueet cytologique de la parte´nogene`se naturelle. Bulletin Biologique de la France etde la Belgique 62: 164–281. → page 15131Verhoeven, K.J.F. & Biere, A. 2013. Geographic parthenogenesis and plant-enemyinteractions in the common dandelion. BMC evolutionary biology 13: 23. → page68Vrijenhoek, R.C. 1979. Factors affecting clonal diversity and coexistence. Integrativeand Comparative Biology 19: 787–797. → pages 7, 16, 17Vrijenhoek, R.C. & Parker, E.D.J. 2009. Geographical Parthenogenesis: GeneralPurpose Genotypes and Frozen Niche Variation. In: Lost Sex: The EvolutionaryBiology of Parthenogenesis (I. Scho¨n, K. Martens & P. Dijk, eds), 6, chap. 6, pp.99–131. Springer Netherlands, Dordrecht. → pages3, 7, 16, 17, 18, 38, 40, 88, 114, 115Wang, T., Hamann, A., Spittlehouse, D. & Carroll, C. 2016. Locally downscaledand spatially customizable climate data for historical and future periods for NorthAmerica. PLoS ONE 11: 1–17. → page 59Weber, M.G. & Strauss, S.Y. 2016. Coexistence in Close Relatives: BeyondCompetition and Reproductive Isolation in Sister Taxa. Annual Review ofEcology, Evolution, and Systematics 47: 359–381. → page 49Weeks, S.C. 1993. The effects of recurrent clonal formation on clonal invasionpatterns and sexual persistence: a Monte Carlo simulation of the frozenniche-variation model. American Naturalist 141: 409–427. → pages 7, 17Welch, D.M. & Meselson, M. 2000. Evidence for the evolution of bdelloid rotiferswithout sexual reproduction or genetic exchange. Science 288: 1211–1215. →page 16West, S.A., Lively, C.M. & Read, A.F. 1999. A pluralist approach to sex andrecombination. Journal of Evolutionary Biology 12: 1003–1012. → page 15Westoby, M., Leishman, M. & Lord, J. 1996. Comparative ecology of seed size anddispersal. Phil. Trans. R. Soc. Lond. B 351: 1309–1318. → page 87Whitton, J., Sears, C.J., Baack, E.J. & Otto, S.P. 2008. The Dynamic Nature ofApomixis in the Angiosperms. International Journal of Plant Sciences 169:169–182. → pages 19, 65, 86, 115Wide´n, B., Cronberg, N. & Wide´n, M. 1994. Genotypic diversity, molecular markersand spatial distribution of genets in clonal plants, a literature survey. FoliaGeobotanica 29: 245–263. → page 19Wilson, C.G. 2011. Desiccation-tolerance in bdelloid rotifers facilitatesspatiotemporal escape from multiple species of parasitic fungi. Biological Journalof the Linnean Society 104: 564–574. → page 5132

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0392003/manifest

Comment

Related Items