THE ECOLOGY AND GENETICS OF AND IN DUNE

SUNFLOWERS

by

Katherine Lee Ostevik

H.B.Sc., The University of Toronto, 2008

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF

THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

in

THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES

(Botany)

THE UNIVERSITY OF BRITISH COLUMBIA

(Vancouver)

March 2016

© Katherine Lee Ostevik, 2016 Abstract

We can learn about the factors that promote and constrain speciation by comparing multiple instances of the of . It is particularly useful to compare systems with similar environmental transitions because is likely responsible for any evolutionary patterns that are consistently associated with ecological variation. In this thesis, I examine two cases of putatively similar recent or incipient in the sunflower genus . In each case, the divergence observed between geographically adjacent populations is associated with adaptation to sand dunes. In my first study, I comprehensively test for reproductive isolation between dune and non-dune ecotypes of H. petiolaris. Despite their recent divergence, I find that multiple reproductive barriers separate them, including post- pollination assortative in the form of pollen competition. In addition, I find that a striking difference in seed size between the ecotypes is a consequence of divergent natural selection, and that it leads to strong and extrinsic reproductive isolation via selection against immigrants and hybrids. I then broaden my study to include the dune endemic, H. neglectus, which is sister to typical H. petiolaris. I look for chromosomal rearrangements between H. neglectus and H. petiolaris, and find almost as many large translocations between them as between more distantly related sunflowers. Finally, I discover that larger seeds are associated with dune environments in both systems and that the genetic basis of that phenotypic evolution is partiality repeated. Taken together, these results suggest that dune adaption within H. petiolaris and between H. petiolaris and H. neglectus has similar consequences. However, it remains to be seen whether assortative mating and chromosomal evolution are unique to the evolution of dune H.

ii petiolaris and H. neglectus, respectively. Ultimately, understanding the similarities and differences between these systems will help answer the question - how predictable is speciation?

iii Preface

I was the primary force behind the design, execution, analysis and presentation of this research. Nevertheless, several others made meaningful contributions to the final product. My advisors, Loren Rieseberg and Sarah Otto, gave me intellectual guidance in the form of advice and feedback throughout, and several research assistants and technicians helped me cultivate and cross sunflowers, collect data in the field and greenhouse, and do basic lab work. Other exceptions are listed below:

Rose Andrew designed and carried out the test of fertility (chapter 2) and generated the backcrossed seeds (chapter 2), Matt Barbaur separated from wasps for pollinator identification (chapter 2), the USDA ARS Biology & Systematics

Laboratory identified bee (chapter 2), Greg Baute made the crosses used for linkage analysis (chapter 3), Chris Grassa cultivated H. annuus x H. petiolaris seedlings

(chapter 3) and Nadia Chaidir extracted H. annuus x H. petiolaris DNA (chapter 3).

I am preparing manuscripts that describe the work is chapters 3 and 4, and a version of chapter 2 has been submitted for publication:

Ostevik, KL, RL Andrew, SP Otto and LH Rieseberg. Multiple reproductive barriers separate recently diverged sunflower ecotypes.

Also, some of the ideas and paragraphs in chapter 1 are published in:

Ostevik, KL, BT Moyers, GL Owens and LH Rieseberg. 2012. Parallel ecological speciation in ? International Journal of Ecology 2012.

iv Table of Contents

Abstract ...... ii

Preface ...... iv

Table of Contents ...... v

List of Tables ...... ix

List of Figures ...... x

List of Abbreviations ...... xii

Acknowledgements ...... xiii

Dedication ...... xiv

Chapter 1: Introduction ...... 1

1.1 Background ...... 1

1.2 This thesis ...... 3

1.2.1 Study system ...... 4

1.2.2 Breakdown of chapters ...... 10

Chapter 2: Multiple reproductive barriers separate recently diverged sunflower ecotypes 14

2.1 Introduction ...... 14

2.2 Methods...... 16

2.2.1 Study system and seed collections ...... 16

2.2.2 Calculating barrier strength ...... 18

2.2.3 Reciprocal transplant ...... 20

2.2.4 Timelapse photography of flowering ...... 24

2.2.5 visitation and collections ...... 25

v 2.2.6 Pollen competition, seed set and pollination timing experiments ...... 27

2.2.7 Hybrid seed germination ...... 29

2.2.8 Pollen staining ...... 30

2.3 Results ...... 31

2.3.1 Reproductive barrier strengths ...... 31

2.3.2 Selection against immigrants ...... 34

2.3.3 Flowering time ...... 38

2.3.4 Pollinator assemblage ...... 40

2.3.5 Post-pollination assortative mating ...... 42

2.3.6 Intrinsic hybrid inviability ...... 43

2.3.7 Hybrid pollen sterility ...... 44

2.3.8 Selection against hybrids ...... 45

2.3.9 Total reproductive isolation ...... 46

2.4 Discussion ...... 46

2.5 Conclusion ...... 51

Chapter 3: Rapid chromosomal evolution between closely related sunflower species ...... 52

3.1 Introduction ...... 52

3.2 Methods...... 55

3.2.1 Study species and crosses ...... 55

3.2.2 Genotyping ...... 56

3.2.3 Genetic mapping ...... 57

3.2.4 Synteny Analysis ...... 57

3.3 Results ...... 58

vi 3.4 Discussion ...... 65

Chapter 4: Parallel genetic changes underlie adaptation to sand dunes in two sunflower species ...... 69

4.1 Introduction ...... 69

4.2 Methods ...... 71

4.2.1 Seed size of natural populations ...... 71

4.2.2 Seed size common garden ...... 72

4.2.3 Crossing and phenotyping ...... 72

4.2.4 Genotyping ...... 74

4.2.5 QTL mapping ...... 76

4.3 Results ...... 77

4.4 Discussion ...... 82

Chapter 5: Conclusion ...... 86

5.1 Discussion ...... 86

5.2 Strengths and limitations ...... 89

5.3 Future directions ...... 90

References ...... 92

Appendices ...... 92

Appendix A - Chapter 2 supplementary materials ...... 104

A.1 Seed size common garden methods and results ...... 104

A.2 2010 reciprocal transplant methods and results ...... 104

A.3 Supplementary tables ...... 108

Appendix B - GBS protocol ...... 122

vii B.1 Protocol summary ...... 122

B.2 Adaptor and primer sequences ...... 122

B.3 Protocol ...... 123

Appendix C - Chapter 3 supplementary figures ...... 127

Appendix D - Chapter 4 supplementary tables ...... 149

viii List of Tables

Table 2.1 Dune and non-dune seed weight from a common garden...... 17

Table 2.2 Experiments used to measure reproductive barriers and barrier strengths ...... 33

Table 3.1 Features of (PET) and H. neglectus (NEG) genetic maps ...... 63

Table 4.1 Helianthus petiolaris and H. neglectus F2 mapping population details ...... 74

Table 4.2 QTL regions identified ...... 79

Table A.1 List of the populations and sites used in chapter 2 ...... 108

Table A.2 Reproductive barriers included total RI calculations ...... 110

Table A.3 ASTER output from dune and non-dune plants ...... 111

Table A.4 ASTER output from all types ...... 112

Table A.5 ASTER output including flower number from dune and non-dune plants ...... 113

Table A.6 ASTER output including flower number from all plant types ...... 113

Table A.7 Insect collection details and locations ...... 115

Table A.8 Parameter estimates from GLMM of seedling emergence including all seed types .. 117

Table A.9 Parameter estimates from GLMM of seedling emergence including dune and non- dune subplots only ...... 118

Table A.10 Parameter estimates from LMMs for three difference proxies of fecundity...... 119

Table A.11 Aculeate caught visiting sunflowers ...... 120

Table A.12 Parameter estimates from a GLMM analysis of pollen competition data ...... 121

Table D.1 Description of the populations used in chapter 4 ...... 149

ix List of Figures

Figure 1.1 – Phylogeny of the annual sunflower clade ...... 5

Figure 1.2 – Range map for Helianthus petiolaris and H. neglectus ...... 6

Figure 1.3 – Photos of typical non-dune and dune ecotypes of H. petiolaris ...... 8

Figure 1.4 – Helianthus petiolaris ssp. fallax and H. neglectus plants in their habitats ...... 10

Figure 2.1 – Population map ...... 18

Figure 2.2 – Weight of seeds produced by plants that survived in the reciprocal transplant...... 22

Figure 2.3 – Mean seedling emergence versus mean seed weight for each type ...... 32

Figure 2.4 – Three proxies of fecundity for plants in the reciprocal transplant ...... 35

Figure 2.5 – The distributions of barrier strength estimates ...... 37

Figure 2.6 – Kaplan-Meyer estimates of survival curves during the reciprocal transplant ...... 39

Figure 2.7 – Flowering time distributions ...... 40

Figure 2.8 – PCA of pollinator communities ...... 41

Figure 2.9 – Siring bias ...... 42

Figure 2.10 – Viable pollen grains produced by plants of different cross types ...... 44

Figure 3.1 - The phylogenetic relationships between the species used in chapter 3 ...... 54

Figure 3.2 - Helianthus petiolaris and H. neglectus genetic maps ...... 59

Figure 3.3 - Cumulative frequency distributions of nearest-neighbor distances for all markers in the Helianthus petiolaris and the H. neglectus genetic maps ...... 60

Figure 3.4 - Helianthus petiolaris and H. neglectus genetic maps compared to the H. annuus reference genome ...... 62

Figure C.1 – PET1 and NEG1 compared to ANN1 ...... 127

x Figure C.2 – PET2 and NEG2 compared to ANN2 ...... 128

Figure C.3 – PET3 and NEG3 compared to ANN3 ...... 129

Figure C.4 – PET4/7 and NEG4 compared to ANN4 ...... 130

Figure C.5 – PET5 and NEG5 compared to ANN5 ...... 131

Figure C.6 – PET6 and NEG6/16 compared to ANN6 ...... 132

Figure C.7 – PET4/7 compared to ANN7 and PET7/4 compared to ANN4 ...... 133

Figure C.8 – PET7/4 and NEG7 compared to ANN7 ...... 134

Figure C.9 – PET8 and NEG8 compared to ANN8 ...... 135

Figure C.10 – PET9 and NEG9 compared to ANN9 ...... 136

Figure C.11 - PET10 and NEG10 compared to ANN10 ...... 137

Figure C.12 - PET11 and NEG11 compared to ANN11 ...... 138

Figure C.13 - PET12/16 and NEG12/15 compared to ANN12 ...... 139

Figure C.14 - PET12/16/17 and NEG16/12 compared to ANN12 ...... 140

Figure C.15 - PET13 and NEG13 compared to ANN13 ...... 141

Figure C.16 - PET14 and NEG14 compared to ANN14 ...... 142

Figure C.17 - PET15 and NEG15 compared to ANN15 ...... 143

Figure C.18 - PET12/16 and NEG6/16 compared to ANN16 ...... 144

Figure C.19 - PET16/12/17 and NEG16/12 compared to ANN16 ...... 145

Figure C.20 - PET17 and NEG17/16/12 compared to ANN17 ...... 146

Figure C.21 - NEG12/15 compared to ANN15 and PET12/16/17 compared to ANN17 ...... 147

Figure C.22 – Neg17/16/12 compared to ANN16 and ANN12 ...... 148

xi List of Abbreviations

ANN

ARG Helianthus argophyllus cM centiMorgan

DNA deoxyribonucleic acid

GLMM generalized linear mixed effect models

GSD Great Sand Dune National Park and Preserve

LMM linear mixed effect model

LG linkage group

MSSP Monahans Sandhills State Park

NEG Helianthus neglectus

NIV Helianthus niveus ssp. tephrodes

NSERC National Science and Engineering Research Council

PCA principal component analysis

PCR polymerase chain reaction

PET Helianthus petiolaris

QTL quantitative trait locus/loci

SNP single nucleotide polymorphism

USDA United States Department of Agriculture

WGS whole genome sequencing

xii Acknowledgements

I am deeply indebted to many people and organizations for their help and encouragement throughout this research. Many thanks to:

Loren Rieseberg and Sally Otto for the freedom, resources, and intellectual inspiration to explore any question that intrigued me;

Dolph Schluter and Jeannette Whitton for their thoughtful comments and insight;

Rose Andrew for her mentorship, generosity, and for introducing me to GSD;

The Rieseburglars who optimized GBS and created bioinformatic tools for sunflower;

RGSD for making light of the challenges and mishaps of graduate student life;

My labmates for moral support, fun, and education, especially, Greg, Hannes, Brook,

Jasmine, Greg and Kathryn;

The extremely supportive community at the Biodiversity Research Centre;

My very capable research assistants, Nicole Warren, Abigale Johnson, Corey Cheng,

Kashifa Hafeez, James Herndon, Nadia Chaidir, Sean Montgomery, and Kasey Moran;

Great Sand Dunes National Park and Preserve, Monahans Sandhills State Park, the

USDA, the Nature Conservancy, the USDA ARS Bee Biology & Systematics Laboratory and other helpful collaborators, including, Matt Barbour, Phyllis Bovin, Fred Bunch,

Terry Griswold, Andrew Valdez, Laura Marek, Lisa Donovan, and Chase Mason;

NSERC, Genome Canada, Genome BC, and the UBC Four Year Fellowship;

Heather, Arnie and Kari who taught me all the important things and who graciously tolerated the presence of a scientist in the family; and special thanks to Kieran for his contagious excitement about science (and for knowing so much about R).

xiii Dedication

To public funding for basic research

xiv Chapter 1: Introduction

1.1 Background

In the past, studies of speciation emphasized stochastic processes and the geographic distribution of diverging taxa (Coyne and Orr 2004). Over the last several decades, however, this focus has shifted drastically to the ecological mechanisms of speciation and most recently to the ways in which genomic architecture promotes and constrains species divergence. Consequently, it is now clear that much of biodiversity is the result of ecologically based divergent natural selection driving the evolution of reproductive barriers, i.e. ecological speciation (Schluter 2009, Nosil 2012). Furthermore, we know that a species’ genomic architecture can influence its evolutionary responses (Kim and

Rieseberg 1999, Via and Hawthorne 2002, Gavrilets 2004). An additional consequence of a focus on ecological mechanisms and genomic architecture is that it highlights the aspects of speciation that are most likely to be repeatable.

Cases of parallel speciation are evidence that species formation can be repeatable

(e.g., Rundle et al. 2000, Nosil et al. 2002, Boughman et al. 2005, Quesada 2007, Strecker et al. 2012). Parallel speciation is the process in which related lineages independently evolve similar traits that confer shared reproductive isolation from their ancestral populations (Schluter and Nagel 1995). It strongly implicates selection as the driver of the evolution of reproductive isolation because the same barriers are unlikely to arise independently by chance (Schluter and Nagel 1995). In addition, when those barriers repeatedly evolve in association with habitat changes, it suggests that ecologically mediated divergent natural selection was responsible. These are cases of parallel ecological speciation.

1 A defining characteristic of parallel ecological speciation is that ancestral and descendent groups of a species each represent a potentially compatible group (i.e., the independent populations would be reproductively compatible if brought together,

Schluter and Nagel 1995). However, it is also possible to have multiple ecological speciation events associated with repeated environmental transitions in which the ancestral or descendent types (or both) are made up of multiple reproductively isolated groups (Ostevik et al. 2012). In these cases, it is not clear that divergent natural selection alone is responsible for the evolution of reproductive isolation; however, they do provide an opportunity to explore how differences in initial conditions and stochastic processes influence speciation. For example, the scenario in which an ancestral species is a single compatible group but the descendent groups are incompatible with one another can be caused by mutation-order speciation, where the same selective pressure leads to different genetic changes in the multiple populations (Schluter 2009), or drift.

In either case, stochasticity influences the outcomes and presumably makes speciation less repeatable. Another scenario, called replicated ecological speciation, is made up of multiple distinct speciation events (i.e., there are multiple incompatible ancestral and descendant groups, Rosenblum and Harmon 2011). In this case, the initial conditions are not identical, so studying the similarities and differences between the replicate speciation events can help identify and predict the general patterns of speciation

(Roseblum and Harmon 2011).

One factor that could have a significant effect on replicate speciation events is genomic architecture. Examples include (1) chromosomal inversions, which can reduce recombination between locally adapted alleles and accumulate genetic incompatibilities between species (Navarro and Barton 2003, Kirkpatrick and Barton 2006); (2) ‘magic traits’, or traits under divergent natural selection that also contribute to non-random 2 mating, thereby facilitating the evolution of assortative mating in the presence of (Gavrilets 2004); (3) One versus two allele assortative mating mechanisms, which affect the ease with which assortative mating evolves in (Felsenstein 1981); and (4) gene duplications that can enable genetic incompatibles to accumulate between species (Bikard et al. 2009, Mizuta et al. 2010). The frequency of such genetic features during divergence and their association with progress towards speciation across replicated events are important unanswered questions.

The genetic basis of the traits under selection during replicated ecological speciation could also influence evolutionary outcomes. Although it is clear that organisms experiencing similar ecological conditions often display parallel phenotypic change (e.g., Jones et al. 1992, Pidgeon et al. 1997, Rundle et al. 2000, Rajakaruna et al.

2003, Rosenblum 2006, Østbye et al. 2006, Foster et al. 2007), it is not clear that the consequences of these changes will be repeatable. However, if the same genes underlie a parallel phenotype, subsequent evolutionary change will probably be more similar because those genes are more likely to have the same pleiotropic and epistatic effects.

For example, if several genes have the potential to cause particular phenotypes but only a subset have pleiotropic effects on assortative mating, then the particular genes fixed by natural selection would significantly impact the development of reproductive barriers. Therefore, understanding the repeatability of the genetic basis of adaptation will likely influence our ability to predict higher order processes, like speciation.

1.2 This thesis

This research is motivated by the question ‘is speciation predictable?’

3 I am particularly interested in the extent to which speciation is predictable for relatively long-lived species under natural conditions. However, these systems strongly limit our ability to observe, measure and manipulate speciation. For this reason, I approach the question more broadly by exploring the ecological and genetic factors associated with adaptation and the evolution of reproductive barriers, with extra interest in the repeatability of those processes. Specifically, I examine two cases of recent or incipient speciation in the genus Helianthus, which are occurring in putatively similar ways.

1.2.1 Study system

Helianthus is a genus that includes 49 sunflower species native to North America

(Timme et al. 2007). It includes the cultivated oilseed, H. annuus, and the tuber crop, H. tuberosus. Helianthus is also a model system for the genetics of adaptation and speciation. It is well known for hybridization, , ecological divergence, and three well-characterized and ecologically diverse homoploid hybrid species, H. anomalus, H. desrticola, and H. paradoxus. That said, the most important characteristic of the genus for the purposes of this thesis is that Helianthus species are frequently edaphic, or soil, specialists (Heiser 1969). Sister species tend to be found on different soil types suggesting that edaphic adaptation is important to speciation or species persistence in this genus (Kantar et al. 2015). Moreover, multiple sunflowers have colonized some soil types (e.g., sand dunes). This provides opportunities for studying parallel adaptation and replicated ecological speciation.

Sand dune adaptation is common among annual Helianthus species. The closely related Helianthus anomalus, H. niveus ssp. tephrodes, H. neglectus and H. petiolaris can all be found in sand dunes (Fig. 1.1, Heiser 1969, Andrew et al. 2012) and probably 4 experience similar ecological divergence. Although additional transitions to sand dune habitats would increase our ability to detect general patterns in the process, this thesis focuses on parallel sand dune adaptation in H. petiolaris and H. neglectus only. This is because it is important to be able to compare relatives, specifically sister taxa, when studying how species diverge and some of the non-dune relatives of dune Helianthus taxa are not optimal. H. anomalus is a homoploid hybrid species, which means it has two non-dune parent species, and the closet relative of H. niveus ssp. tephrodes, H. niveus ssp. canescens, is rare. During a trip to the Algodones Sand Dunes in California, I was unable to find any H. niveus ssp. canescens populations in the vicinity of the H. niveus ssp. tephrodes populations found there.

H. niveus tephrodes H. niveus canescens H. praecox H. debilis H. neglectus H. petiolaris (dune ecotype)

H. anomalus H. deserticola H. paradoxdus

H. annuus H. argophyllus H. exilis

Figure 1.1 – Phylogeny of the annual sunflower clade (Stephens et a. 2015). Species (or ecotypes) in red are those found in deep and active sand dunes. Branch lengths are not to scale.

5 This thesis focuses on dune adaptation between ecotypes within H. petiolaris and between H. petiolaris and H. neglectus. Both H. petiolaris and H. neglectus are self- incompatible, hermaphroditic, diploid, and annual. Helianthus petiolaris is a widespread sunflower species with a large effective population size (Strasburg and Rieseberg 2008).

Its large range is split into the subspecies Helianthus petiolaris ssp. petiolaris, commonly found in the southern Great Plains, and H. petiolaris ssp. fallax, found in more arid regions (Heiser 1961, Fig. 1.2). Helianthus neglectus has a small range limited to west

Texas and is endemic to deep sand dune habitats (Fig. 1.2).

H. neglectus H. petiolaris fallax H. petiolaris petiolaris Monahans Sandhills State Park Great Sand Dunes National Park and Preserve

Figure 1.2 – Range map for Helianthus petiolaris and H. neglectus (Rogers et al. 1982).

6 Across most of its range, H. petiolaris inhabits sandy soils. However, at Great

Sand Dunes National Park and Preserve (GSD, Fig. 1.2) in Colorado an ecotype of H. petiolaris ssp. fallax manages to survive in the impressive sand dunes found there (Fig.

1.3). These sand dunes are old (they started to form at least 130 thousand years ago), cover a relatively small area (~72 km2), and are extremely tall (up to 230 m, Madole et al. 2008). Not surprisingly, they make a harsh environment for plants. The surface of the sand can reach 65˚C, strong winds cover sedentary species in sand, and nutrients are limited. As such, only a few species of plants commonly grow in the dunes including Psoralidium lanceolatum (scurfpea), Achnatherum hymenoides (Indian ricegrass), and Redfieldia flexuosa (blowout grass). In contrast, the sand sheet that neighbours the dunes at GSD contains more soil, has higher nitrogen levels, is stabilized, and is much more densely vegetated. Small shrubs like Ericameria nauseosa (rubber rabbitbrush) are common on the sand sheet, and other plants, including Yucca glauca (narrowleaf yuccas), Opuntia polyacantha (prickly pear cactus), Hesperostipa comata (speargrass) and typical H. petiolaris ssp. fallax (non-dune ecotype), thrive there.

7

Figure 1.3 – Typical non-dune (left) and dune (right) ecotypes of H. petiolaris on the sand sheet (left) and in the dunes (right) at Great Sand Dunes National Park and Preserve in Colorado.

The dune ecotype of H. petiolaris ssp. fallax is distinct from its non-dune counterpart. For example, the dune plants have larger seeds, reduced branching and longer phyllaries (Andrew et al. 2013). Initially, these plants were hypothesized to be the result of hybridization been H. petiolaris and H. annuus much like the homoploid hybrid, H. anomalus. However, genetic analyses did not support simple hybrid ancestry

(Andrew et al. 2013). Instead, it seems that the dune ecotype started to diverge from typical H. petiolaris within the last 10,000 years (Andrew et al. 2013). There is evidence for isolation by adaptation and asymmetric migration between the ecotypes (more

8 migration into the non-dune habitat, Andrew et al. 2012). Consistent with expectations of adaptation with gene flow, there is considerable heterogeneity in levels of divergence between loci (Andrew et al. 2013). Furthermore, there are a few large regions of elevated divergence between the dune and non-dune ecotypes where most of the evidence for selective sweeps is found in the dunes (Andrew and Rieseberg

2103). Together these data suggest that there are likely reproductive barriers that separate the ecotypes.

Like the sunflowers at GSD, H. neglectus inhabits active sand dunes (centered around Monahans Sandhills State Park, MSSP, Fig. 1.2) while typical H. petiolaris ssp. fallax remains in nearby sandy soils (Fig. 1.3). Also like at GSD, nutrient levels are lower in the dunes at MSSP than in the surrounding soil (K. Ostevik, unpublished data).

There are a few subtle traits that differentiate the species. For example, H. neglectus tends to have more ovate leaves and thinner peduncles (Rogers et al. 1982). However, gene flow between H. neglectus and nearby H. petiolaris ssp. fallax is as strong as gene flow among different regions of H. petiolaris (Raduski et al. 2010). This has led to the species status of H. neglectus being questioned (Raduski et al. 2010). On the other hand, there are reproductive barriers that separate the species. Specifically, greenhouse crosses between H. petiolaris and H. neglectus are less compatible than intraspecific crosses and the resulting hybrids are somewhat sterile, having only 75‐80% pollen viability (Chandler et al. 1986, Heiser 1958). If H. neglectus is not yet a species, it seems to be well on its way. However, because of the smooth transition between H. neglectus and H. petiolaris ssp. fallax in the region surrounding MSSP (Raduski et al. 2010), I sometimes refer to the H. petiolaris ssp. fallax plants surrounding the dunes as “non-

9 dune H. neglectus”. I feel that this terminology clarifies which system of dune adaptation I am referring to.

Figure 1.4 – Helianthus petiolaris ssp. fallax plants along a roadside in Texas (left) and H. neglectus plants found in deep sand dunes at Monahans Sandhills State Park in Texas (right).

Overall, it appears that dune adaptation within H. petiolaris and between H. petiolaris and H. neglectus is comparable. Both dune systems have habitat transitions to lower nutrients and loose sand, and both pairs of sunflowers appear to be in the early stages of speciation. This is thus an excellent system for studying parallel adaptation to sand dunes and replicated ecological speciation.

1.2.2 Breakdown of chapters

In this thesis, I attempt to identify ecological and genetic characteristics important to speciation and adaptation in H. petiolaris and H. neglectus. I start by measuring reproductive barriers between dune and non-dune ecotypes of H. petiolaris at GSD.

Then, I make genetic maps of both species and identify chromosomal rearrangements that differentiate the species. Finally, I explore the role that large seeds play in dune

10 adaptation and determine whether the genetic basis of seed size is repeatable. I do these studies with the ultimate goal of being able to compare and contrast progress towards speciation in the two systems.

Chapter 2: Multiple reproductive barriers separate recently diverged sunflower ecotypes

Measuring reproductive barriers between groups of organisms is an effective way to determine the traits and mechanisms that impede gene flow. However, to understand the ecological and evolutionary factors that drive speciation, it is important to distinguish between the barriers that arise early in the speciation process and those that arise after speciation is largely complete. In this chapter, I comprehensively test for reproductive isolation between the dune and non-dune ecotypes of H. petiolaris. I find reproductive barriers acting at multiple stages of hybridization, including premating, postmating-prezygotic, and postzygotic barriers, despite their very recent divergence.

The barriers identified include selection against immigrants, a shift in pollinator assemblage, post-pollination assortative mating, and extrinsic selection against hybrids.

Together these data suggest that multiple barriers can be important for reducing gene flow even in the very earliest stages of speciation.

Chapter 3: Rapid chromosomal evolution between closely related sunflower species

Chromosomal rearrangements often differentiate species and are thought to facilitate speciation. In this chapter, I make high-density genetic maps for the two focal species of this thesis, typical H. petiolaris and H. neglectus. The H. petiolaris and H. 11 neglectus genetic maps are made up of 17 linkage groups that span 1329.6 and 1119.4 cM respectively. I then compare these maps to each other and to the domesticated sunflower, H. annuus, and map the chromosomal rearrangements that differentiate the species. I find 20 inversions and 5 translocations that differentiate each map from the H. annuus reference. Although most of these rearrangements are shared, there are several translocations unique to each wild species. In fact, there are nearly as many translocations between the very closely related H. petiolaris and H. neglectus as between the more distantly related H. petiolaris and H. annuus.

Chapter 4: Parallel genetic changes underlie adaptation to sand dunes in two sunflower species

The degree to which the same genes underlie traits that have evolved in parallel reveals potential constraints on the genetic basis of adaptation. In this chapter, I find that large seeds are associated with dune populations of H. neglectus, much like the dune ecotype of H. petiolaris and the dune sunflowers H. anomalus and H. niveus ssp. tephrodes. This parallel evolution of seed size strongly implicates natural selection as the driver of the trait differences. Furthermore, identifying the genes responsible for the parallel evolution provides an opportunity to quantify the repeatability of the genetic basis of adaptation. I find three QTL that underlie seed size phenotypes in H. petiolaris and two in H. neglectus. Two of the three QTL identified broadly overlap between the species indicating that the genetic basis of seed size is partially repeatable in this study.

These results further imply constraint on or unequal opportunity of the seed size genes available to natural selection in these systems and hint that any additional consequences of larger seeds are more likely to be the same. 12 Chapter 5: Conclusion

In the final chapter, I bring the results of the preceding chapters together and make some general conclusions about adaptation and speciation in these sand dune sunflowers. I discuss the strengths and limitations of this research and lay out possible future directions.

13 Chapter 2: Multiple reproductive barriers separate recently diverged sunflower ecotypes

2.1 Introduction

To understand speciation, we must understand how and why groups of organisms stop exchanging genes. A direct approach is to characterize the traits and mechanisms impeding gene flow between diverging or recently diverged groups of organisms.

Determining the order in which these reproductive barriers arise not only distinguishes between early- and late-acting mechanisms, but it also provides clues regarding the evolutionary processes underlying observed patterns. Reproductive barriers can arise via divergent natural selection, reinforcement, genetic conflict, mutation-order selection and drift, but the action and importance of these processes depend on specific ecological and evolutionary conditions. Therefore, identifying specific barriers and the contexts in which they arose (e.g., whether other barriers were restricting gene flow) can point to the ultimate causes of reproductive isolation (RI) between species.

Reproductive barriers are grouped into three categories: premating (e.g., flowering time), postmating-prezygotic (e.g., conspecific pollen precedence), and postzygotic (e.g., hybrid sterility). Many studies have found that prezygtoic barriers have stronger effects than postzygotic barriers among closely related taxa (Lowry et al.

2008a, Dell’Olivo et al. 2011, Sánchez-Guillén et al. 2012, Sobel and Streisfeld 2015), especially considering that prezygotic barriers act first. However, others have found that postzygotic barriers (Kozak et al. 2012) or both types of barriers are important for reducing gene flow (Sambatti et al. 2012, Scopece et al. 2013, Briscoe Runquist et al.

2014, Kao et al. 2015).

14 Observing the accumulation of reproductive barriers directly is rarely feasible.

An alternative is to measure many reproductive barriers in multiple systems throughout the stages of speciation and infer the likely sequence of events. Although there are excellent examples of multiple barriers measured between well-established species (e.g., McMillan et al. 1997, Ramsey 2003, Kay 2006, Martin and Willis 2007,

Matsubayashi and Katakura 2009, Dell’Olivo et al. 2011) and some at intermediate stages of speciation (e.g., Nosil 2007, Lowry et al. 2008b, Melo et al. 2014, Briscoe

Runquist et al. 2014, Sobel and Streisfeld 2015), the success of this method relies on numerous studies spread throughout the speciation continuum. In particular, we need more studies that identify and measure barriers at the earliest stage of divergence, as these barriers arguably drive the speciation process (rather than accumulate after speciation is mostly complete, Coyne and Orr 2004).

Incipient speciation within the prairie sunflower Helianthus petiolaris is a good system for measuring RI early in the speciation process. Divergence between typical H. petiolaris and a sand dune ecotype found in Colorado began less than 10,000 years ago

(Andrew et al. 2013). Gene flow between the two types is asymmetric (with fewer immigrants from the surrounding sand sheet into the sand dune environment) and too high for drift to cause widespread differentiation (Andrew et al. 2012, 2013).

Conversely, gene flow is too low to prevent some adaptive divergence, and consequently, there are a few large regions of genomic divergence that differentiate the ecotypes (Andrew and Rieseberg 2013). Finally, related work on additional species pairs within Helianthus (Rieseberg et al. 1995a, Sambatti and Rice 2006, Yatabe et al.

2007, Raduski et al. 2010, Sambatti et al. 2012, Renaut et al. 2013) means that this study will add breadth to our understanding of the stages of speciation in sunflowers.

15 In this paper, we comprehensively test for reproductive isolation between dune and non-dune ecotypes of H. petiolaris. We measure selection against immigrants and hybrids, flowering time, pollinator assemblages, post-pollination assortative mating, hybrid seed germination, and hybrid sterility. We ask whether reproductive barriers are limited to those associated with local adaptation or whether additional barriers separate these recently diverged ecotypes.

2.2 Methods

2.2.1 Study system and seed collections

Helianthus petiolaris is a widespread sunflower species that is annual, self-incompatible, and hermaphroditic. It is typically found in sandy soils across the central United States

(Heiser et al. 1969). However, populations of H. petiolaris inhabit active sand dunes at

Great Sand Dunes National Park and Preserve (GSD) in Colorado. These dunes are the tallest in North America and are a challenging environment for plant species due to shifting sand and low soil nutrients (Johnson 1968, Morenno-Casasola 1986, Andrew et al. 2012). Helianthus petiolaris plants found on dunes are morphologically distinct from typical H. petiolaris plants found on the vegetated sand sheet surrounds the dunes

(Andrew et al. 2012). For example, dune seeds are more than two times heavier than non-dune seeds, and these differences are maintained in a common garden (Appendix

A).

16 Table 2.1 Mean weight of the seeds produced by dune and non-dune plants grown in a common garden.

Ecotype N Seed weight (mg) 95% CI Dune 41 11.3 10.5-12.1 Non-dune 24 5.0 4.5-5.6

Throughout this study, we used seeds collected in 2008 and 2009 from natural dune and non-dune populations within GSD (Table A.1, Fig. 2.1), as well as F1 and backcrossed seeds generated in a greenhouse in 2009 and 2011. We generated two F1 seed lots by pooling crosses made between multiple dune and non-dune individuals from several populations. One F1 seed lot consisted of dune maternal plants crossed with pollen from non-dune plants (F1D), and the other used non-dune maternal plants and dune pollen (F1N). We also generated two types of backcrosses by pooling crosses made between maternal F1s (equal numbers of F1D and F1N) and either dune (BCD) or non-dune (BCN) pollen donors from several populations. The result was a single genetically variable seed lot for each of four hybrid types (see Table A.1 for the populations used).

17 0 2 4 6 8 10 kilometres Dunes Dune Populations Non-dune Populations N6 D10 D9 N2 D2 D8 D5 N7 D6 N8 D7 D3

D1 D13 D15 DRT D4 D17 N9 N3 N10 D16 D14 N1 N5 NRT D12 D11 N4

Figure 2.1 – Population map. See Table A.1 for the latitude and longitude coordinates. DRT= dune reciprocal transplant site, NRT= non-dune reciprocal transplant site.

2.2.2 Calculating barrier strength

Following Sobel and Chen (2014), we use a method for calculating reproductive isolation caused by a specific barrier that yields a simple linear relationship between the reduction of gene flow between groups and RI:

18 ! !" = 1 − 2 ∗ (1) ! + !

In general, H and C represent either the number of heterospecific and conspecific (prezygotic barriers) or the number of viable heterospecific and conspecific offspring (postzygotic barriers) given equal opportunities for both. In this paper, comparisons are between and within ecotypes instead of species. Using this equation, the RI values 1, 0 and -1 correspond to the proportions of heterospecific outcomes, 0,

0.5 and 1, respectively.

Where reproductive barriers are based on temporal or spatial co-occurrence

(e.g., flowering time or pollinator assemblages), the minimum value of RI is zero, corresponding to complete overlap. In this case, we use a comparable equation for RI that ranges from 0 to 1 and also yields a simple linear relationship between the reduction of gene flow between groups and RI:

! !" = 1 − (2) ! + !

In this case, S and U are the shared and unshared portions of occurrence (Sobel and

Chen 2014). When possible, we calculate the strength of RI that reduces gene flow from the sand sheet into the dunes (RIN->D) separately from the strength of RI that reduces gene flow from the dunes into the sand sheet (RID->N).

To calculate total reproductive isolation, we used Sobel and Chen’s (2014) equation 4E that considers shared and unshared space and time (barriers that affect co- occurrence) before other types of barriers. Estimates of total reproductive isolation

19 assume that reproductive barriers act sequentially. In plants, however, the initial source of genetic exchange can be from seed or pollen flow and some barriers affect only one source (Sambatti et al. 2012). For example, immigrant inviability, which is the reduced survival of foreign individuals (in this case seeds) in a habitat, only affects seed flow while differences in pollinator assemblages only affect pollen flow. For this reason, we calculated total reproductive isolation for pollen and seed flow separately.

2.2.3 Reciprocal transplant

We planted dune, non-dune, and hybrid seed into a site in the dunes and a site on the sand sheet (Table A.1, Fig. 2.1). The reciprocal transplant sites are 12 km apart, and the dune site has less vegetation cover and lower nitrogen availability (Andrew et al. 2012).

We used 10 seed types: seeds from three dune (D1-D3) and three non-dune (N1-N3) populations, F1 hybrid seeds with either dune (F1D) or non-dune (F1N) mothers, and backcrossed seeds with F1 mothers and either dune (BCD) or non-dune fathers (BCN).

After weighing groups of 10 seeds, we stored them at 4°C on wet filter paper for 4 weeks to mimic overwintering conditions and promote germination (Donovan et al.

2010).

At each site, we established 9 ring-shaped plots along each of 5 transects drawn through a natural population of sunflowers. We cleared each plot of vegetation, including any natural sunflower seedlings, and divided the plots into 16 equal subsections. We randomly assigned the 10 seed types to subsections and planted 10 seeds of the assigned type 5 cm deep in each subsection. To determine the number of naturally recruited seedlings in each plot (volunteers), we made 5 cm holes in the

20 remaining 6 subsections, later referred to as control subplots, but did not plant any seeds (see, e.g., Donovan et al. 2010).

We counted the number of seedlings in each subplot after six weeks. After an additional eight weeks, we recorded the number of seedlings that survived and measured their heights. Once the plants began flowering, we recorded the number of flower heads with open disc florets every 5-7 days and covered ripening flower heads to prevent the loss of shattered seed. Finally, as the plants senesced, we collected the ripe flower heads and counted and weighed the seeds they produced.

In our analysis of seedling emergence, we fit generalized linear mixed models

(GLMMs) to the number of seedlings that emerged in each subplot. These models assumed a Poisson distribution of error terms, included a term that accounts for frequent zero-valued observations (zero-inflation), and showed no evidence of additional overdispersion (glmmADMB function and package, Fournier et al. 2012,

Skaug et al. 2014). We fit models with environment, type (control, N1, N2, N3, BCN,

F1N, F1D, BCD, D1, D2, D3), and their interaction as fixed effects and plot as a random effect. We determined the significance of the fixed effects by comparing nested models using likelihood ratio tests. In addition, we looked for differences between the parental types alone (dune and non-dune) using the same analyses as above with the addition of population as random effect. We used R version 3.1.3 (R Core Team 2015) with the library plyr (Wickham 2011) for all statistical analyses.

Seed size is often primarily determined by maternal genotypes (Li and Li 2015).

Consequently, the hybrid cross type most genetically similar to a parental type might not be the most similar to that parent’s seed phenotype. For example, F1D seeds might be more similar in weight to D seeds (both having dune mothers) than are BCD seeds

21 (with F1 mothers), even though BCD seeds are more genetically similar to D seeds. We used ANOVA with a Welch correction for nonhomogeneity of variance and a post-hoc

Games-Howell test (posthocTGH function, userfriendlypackage, Peters 2015) to explore the relationship between seed type and weight. We also used GLMMs as described above to look for a relationship between seed weight and seedling emergence.

Although emergence in the control subplots was very rare (only 13 seedlings emerged in 540 control subplots), some plants surviving in the remaining subplots could represent natural recruitment (volunteers). We looked for volunteers by examining the weight of seeds that each plant produced (seed size differences are maintained in common gardens). All but three plants produced seeds that were the expected weight (Fig. 2.2) and our results were unaffected by whether these three plants were included or excluded in analyses (we present the former). We do not include the seedlings that emerged in control subplots in the remaining analyses.

16 likely volunteers 14

12

10 eight (mg)

w 8

Seed 6

4

2

Dune Non-dune environment environment

Figure 2.2 - Weight of seeds produced by control (gray triangles), dune (red squares), hybrid (purple circles), and non-dune (blue diamonds) plants that survived in the reciprocal transplant.

22

We used linear mixed models (lmer function, lme4 package, Bates et al. 2013a,b) to explore the effects of type, environment and their interaction on three proxies of fecundity: (1) plant height, (2) total number of flower heads produced, and (3) total number of seeds produced. Because of low emergence and survival during our reciprocal transplant, we grouped plants across the populations and hybrid cross types to make three composite types (dune, hybrid and non-dune) for this analysis. We used the mean trait value of surviving plants in each subplot as the response variable and included plot as a random effect. Again, we compared nested models with likelihood ratio tests to determine the significance of fixed effects.

ASTER is a likelihood method for fitting multiple fitness components with different probability distributions in a single model (Geyer 2007). We used ASTER to combine three sequential components of fitness: (1) emergence, (2) survival of emerged seedlings, and (3) number of seeds produced by surviving plants (aster function and package, Geyer 2013). We included plot and subplot as random effects and tested seed type, environment and their interaction as fixed effects using likelihood ratio tests. Like the emergence data, we repeated this analysis using only the parental types (dune and non-dune) and including a population random effect. Also, because sunflowers are hermaphroditic, we repeated this analysis using total number of flower heads in addition to total seed number to better account for both male and female contributions to fitness. However, the two analyses yielded very similar results (Tables A.3-A.6) so we present the data from the first analysis only.

Finally, we tested whether the day to first flower was affected by type, environment and their interaction for plants that survived in the reciprocal transplant by using an interval censored survival analysis (survfit and survreg functions, survival 23 package, Therneau and Grambsch 2000, Therneu 2013). As in the fecundity analyses, we grouped the pure populations and hybrid cross types into three composite types

(dune, hybrid and non-dune) and used mean values for each subplot as the response.

Survival was modeled using the logistic distribution and we included plot as a random effect.

We used the total number of seeds produced by each type in an environment

(scaled by the number of seeds added to the environment) and equation 1 to calculate the strength of selection against immigrants and hybrids in each environment. For example, to calculate RIN->D caused by selection against F1 hybrids, we used the number of seeds produced by hybrid plants in the dune environment divided by 0.2 as H (2/10 of the seed types were F1 hybrid) and the number of seeds produced by dune plants in the dune environment divided by 0.3 as C (3/10 of the seed types were D1-D3). Finally, for all barrier strength calculations, we used 10,000 bootstrap replicates to calculate 95% confidence intervals (Efron 1987). If a bootstrap replicate sampled 0 seeds for both H and C, we considered barrier strength to be 0, yielding a conservative confidence interval that was more likely to overlap 0.

2.2.4 Timelapse photography of flowering

To measure flowering time in natural populations, we positioned timelapse cameras in several dune and non-dune populations (Table A.1). Each camera photographed a patch of naturally occurring sunflowers daily throughout the flowering period. We used ImageJ (Schneider et al. 2012) to count the number of open flower heads visible in a static section of the population every day. When poor light conditions affected our ability to count flower heads (<2% of photos), we used the average number of open

24 flower heads on the adjacent days to estimate the missing data. We then added the number of open flowering heads across all sites in the same environment for each day.

Finally, we totaled the number of open flowers across all days to obtain the fraction of flowers that were open on any given day for each environment. This provided a temporal distribution of open flower heads for both the dune and non-dune habitats.

We used these data to determine the proportion of the two flowering distributions that were shared (S) and unshared (U) and then calculated RI using equation 2. Because we used density distributions and any overlap is an equal proportion of each distribution, these estimates of RI are necessarily symmetrical (RIN-

>D = RID<-N).

2.2.5 Insect visitation and collections

We were interested in whether the assemblages of potential pollinators differed between the dune and non-dune environments. We measured this rather than the visitation rates of one or a few pollinators because sunflowers do not have a single characteristic pollinator and are typically visited by many insect species. To assess these potential pollinator differences, we netted insects visiting the natural sunflowers surrounding the dune and non-dune reciprocal transplant sites approximately once per week (Table A.7). We only collected insects that we observed touching mature sexual parts of sunflowers to limit our study to the species most likely to be pollinators. We collected insects at the dune and non-dune sites during the same two-hour window on consecutive days and made the next pair of collections during a different time period in the following week. This sampling method allowed the collections at the two sites to be paired in time, but the paired collections to be spread throughout the flowering period

25 and time of day, with the goal of sampling as many potential pollinators as possible. To sample potential pollinator assemblages more broadly, we also positioned malaise traps and pan traps at several dune and non-dune populations (Table A.7). We sorted the collections into morphospecies and sent bees and non-parasitoid wasps (Aculeate specimens) to the USDA ARS Bee Biology & Systematics Laboratory for identification.

We tested whether potential pollinator assemblages differed between the environments after controlling for distance using partial Mantel tests. We reduced our data set to the most likely pollinators (, Lepidoptera, Diptera, and

Coleoptera) and calculated a Bray-Curtis dissimilarity index (Bray and Curtis 1957, mantel.partial and vegdist functions, vegan package, Oksanen et al. 2015). The effect size and significance of the partial Mantel test was not sensitive to the use of alternative dissimilarity indices (e.g., Mountford 1962).

To estimate RI caused by differences in pollinator assemblages, we identified species or morphospecies found in both environments using the data from all collection methods. Then we used the number of pollinators caught visiting the sunflowers that were shared between environments (S), the number of pollinators caught visiting the sunflowers that were found in a single environment (U), and equation 2 to calculate RI.

This component of RI measures the degree to which habitat isolation limits pollen flow between environments. It does not assess pollen flow between plants, even if they are different ecotypes, in the same environment. This measure of overlap in pollinator assemblages counts species as present or absent in each environment rather than accounting for pollinator abundance, so pollinators present in both environments but at very different densities did not contribute to the RI measured here.

26 2.2.6 Pollen competition, seed set and pollination timing experiments

We used plants grown from wild seed in the greenhouse to create three pollen types: bulked dune pollen, bulked non-dune pollen and a 50:50 mixture of the ecotypes.

Bulked pollen consisted of equal volumes of pollen from at least three plants from at least three populations within an ecotype. All plants used as dune and non-dune pollen donors were homozygous for alternate alleles of a genetic marker (PCO_08) strongly associated with ecotype. We made pollen mixtures by combining equal weights of these two bulked pollen pools. A test of this method using light microscopy showed that equal weights of pollen corresponded approximately to equal numbers of pollen grains. Marker PCO_08 is a cleaved amplified polymorphic sequence that was designed based on a SNP found on chromosome 5 (Andrew and Rieseberg 2013). The restriction enzyme, AvrII, will cleave DNA fragments with the allelic variant common on the dunes but will not cleave the variant common on the sand sheet.

To quantify post-pollination assortative mating, we covered immature flower heads with mesh to prevent natural pollination and hand pollinated each head with the mixed pollen on one day (pollen competition experiment). There was no need to emasculate flowers because H. petiolaris is self-incompatible (both for the dune and non-dune populations, pers. obs.). We did not pollinate plants with mixtures to which they had contributed. We then extracted DNA from 1-23 (mean=6.2) seeds produced and determined paternity of each seed with marker PCO_08. In total, we genotyped

1528 seeds and assayed 116 maternal plants from 11 dune and 8 non-dune populations

(Table A.1) for assortative mating.

We analyzed the data from the pollen competition experiment using logistic regression and assessed the significance of maternal ecotype as a fixed effect using likelihood ratio tests (glmer function, lme4 package, Bates et al. 2013a,b). The logistic 27 models used the binomial link function, included pollen mixture, maternal plant, and maternal population as random effects, and showed no evidence of overdispersion. The response variable was the ecotype of the individual that fathered each seed. When maternal plants were homozygous at marker PCO_08, determining the paternal ecotype of each seed was trivial. However, when mothers were heterozygous, we assumed they contributed each allele randomly and determined the expected number of offspring sired by each ecotype. In cases where the expected number of offspring was not a whole number (e.g., 1.5 dune and 3.5 non-dune), we randomly assigned split offspring to a single ecotype (2 dune and 3 non-dune or 1 dune and 4 non-dune).

Reassigning all of the split cases 1000 times showed that these assignments did not affect the direction or significance of the result.

For a subset (68) of the plants that received mixed pollen, we pollinated two additional flower heads, one with bulked dune and the other with bulked non-dune pollen (seed set experiment). In these cases, we pollinated 30 receptive styles and counted the number of seeds produced to determine whether the pollen types had biased success in the absence of direct competition. We simplified the resulting paired data by subtracting the number of seeds produced after non-dune pollination from those produced after dune pollination for each plant. Then we used one-sample t tests to determine whether the differences departed significantly from 0 for each ecotype. If dune and non-dune pollinations were equally successful, the differences should cluster around 0.

Finally, we pollinated several dune flower heads with both bulked pollen types in sequence (pollination timing experiment). We varied which pollen type was used first (D-N and N-D) and the time between pollinations (6 or 12 hours), for a total of four treatments. We tested whether the time between pollinations changed the siring 28 success of dune pollen when flowers were pollinated with dune pollen before non- dune pollen using logistic regression as described in the pollen competition analysis above. We repeated this analysis for the flowers in which non-dune pollen was applied before dune.

We used the data from the pollen competition experiment to calculate barrier strength by using the total number of hybrid (H) and parental (C) seeds produced after mixed pollinations and equation 1. We then used the seed set and pollination timing experiments to further interpret the source of this RI.

2.2.7 Hybrid seed germination

We used the seeds produced in the seed set experiment described above to test hybrid seed germination. We tested 3 seeds fathered by bulked dune pollen and 3 seeds fathered by bulked non-dune pollen from each maternal plant. Specifically, we scarified seeds, soaked them in a 100 mg/L solution of gibberillic acid for 1 hour to break dormancy (Chandler and Jan 1985), and put the seeds on wet filter paper. We put the seeds in the dark for 2 days, removed their seed coats, replaced the filter paper, and returned them to the dark. We moistened the filter paper and checked for germination

(radicles longer than 2 mm) every 2 days. After 4 days, 98% of the seeds had germinated. We terminated the experiment after 14 days. To explore the potential effects of seed dormancy, we repeated the basic experiment without a gibberillic acid treatment. The results of that experiment were similar (Appendix A).

We used generalized linear mixed models to determine whether germination was explained by maternal ecotype, paternal ecotype, and/or their interaction (glmer function, lme4 package, Bates et al. 2013a,b). Similarly, we compared models with and without the explanatory variable, cross type (hybrid versus pure). In both cases, we 29 modeled germination using a binomial link function and included maternal population and pollen mixture as random effects. The models showed no evidence of overdispersion. Finally, we used the proportion of hybrid seeds that germinated (H), the proportion of parental type seeds that germinated (C), and equation 1 to calculate barrier strength.

2.2.8 Pollen staining

We made crosses within and between dune and non-dune ecotypes, supplementing the non-dune parents collected within GSD with H. petiolaris individuals collected in New

Mexico by the US Department of Agriculture (USDA). We grew 8-12 progeny from those crosses under controlled greenhouse conditions and collected pollen they produced. Because pollen staining is a proxy for pollen viability (Nepi and Franchi

2000), we stained the pollen in 30% (w/v) sucrose and 0.1% (w/v) Thiazolyl blue tetrazolium bromide (MTT) stain for one day (24 hours, Chandler et al. 1986). We arcsine transformed the proportion of stained grains (180-1000 grains scored/plant) and determined whether cross type (within or between ecotype) significantly predicted pollen viability. Pollen viability did not vary strongly among non-dune cross types

! (GSD vs. USDA, !! =0.06, P=0.94); we therefore combined all non-dune parents for the purpose of calculating barrier strength. The average pollen viabilities from dune and non-dune parents (C) and from hybrid parents (H) were used in equation 1 to estimate reproductive isolation.

30

2.3 Results

We found that the weight of seed produced by a plant was strongly related to that plants’ origin (F5,308=2359, P<0.001, Fig. 2.3). F1 seeds produced by dune mothers (F1D) weighed roughly the same as dune seeds, and F1 seeds produced by non-dune mothers

(F1N) weighed roughly the same as non-dune seeds (Fig. 2.3). Both backcrosses (BCD and BCN whose mothers were F1) had intermediate seed weights, though seeds backcrossed to dune plants (BCD) were slightly heavier than seeds backcrossed to non- dune plants (BCN, Fig. 2.3). Moreover, all pairwise comparisons between the seed types were significantly different (all tdf>5.5 where df>125, all P<0.001) except between

D2 and F1D (t168=0.35, P=0.99).

2.3.1 Reproductive barrier strengths

We present the remaining results organized by the reproductive barriers measured in the approximate sequence in which they would act. Some of our experiments provided information about multiple barrier strengths, and some barrier strengths are informed by multiple experiments, as summarized in Table 2.2.

31 1.5 Sand sheet environment 1.0 F N D1 D2 0.5 1 BCD D3 N2 N1 F1D N3 BCN 0.0

3.0 Dune environment 2.5 2.0 D1 1.5 D2 D3 Estimated emergence 1.0 F N F D 1 BCD 1 N1 0.5 N3 N2 BCN 0.0 4 6 8 10 12 Seed weight (mg)

Figure 2.3 – Mean seedling emergence (estimated by GLMM) versus mean seed weight for each type.

! There is no environment-by-type interaction (!!=14.54, P=0.1497). The dotted line represents the upper confidence limit for emergence in the control subplots and the error bars are 95% confidence intervals. N1-N3: non-dune populations, F1N: F1 seeds with non-dune mothers, BCN: F1 plants backcrossed to non-dune plants, BCD: F1 plants backcrossed to dune plants, F1D: F1 plants with dune mothers, D1-D3: dune populations.

32 Table 2.2 Summary of the experiments used to measure reproductive barriers (and those used as supporting evidence) and the resulting barrier strengths. 95% confidence intervals are based on 10,000 bootstrap replicates.

Reproductive barrier Experiment(s) RIN->D RID->N Selection against immigrants Reciprocal transplant 0.89 (0.66-0.99) 0.76 (-1-0.98) Flowering time Timelapse photography 0.093 (0.09-0.11) 0.093 (0.09-0.11) (reciprocal transplant) Pollinator community Insect collections 0.55 (0.48-0.63) 0.36 (0.30-0.41) Post-pollination assortative Pollen competition 0.38 (0.31-0.44) 0.12 (0.05-0.19) mating (seed set & pollen timing) Intrinsic F1 hybrid inviability Hybrid seed germination 0.003 (-0.01-0.01) 0.006 (0-0.16) Selection against F1D hybrids Reciprocal transplant 0.44 (-0.02-0.79) 0.43 (-1-1) Selection against F1N hybrids Reciprocal transplant 0.99 (0.97-1) 1 (1) F1 hybrid pollen sterility Pollen staining 0.01 (-0.02-0.05) -0.03 (-0.08-0.01) Selection against BCD Reciprocal transplant 0.77 (0.48-0.92) 0.74 (-1-1) Selection against BCN Reciprocal transplant 0.94 (0.78-1) 0.55 (-1-1)

Total RI (seed flow) 0.999 (0.972-0.991) 0.992 (-1-1) Total RI (pollen flow) 0.986 (0.923-0.998) 1 (-1-1)

33 2.3.2 Selection against immigrants

During the reciprocal transplant, more seedlings emerged in all treatment subplots relative to the unseeded control subplots (Fig. 2.3, Table A.8). When we consider only dune and non-dune treatments, more seedlings emerged in the dune environment

! (!! =19.85, P<0.001) and more dune seedlings emerged than non-dune seedlings in both

! environments (!! =20.82, P<0.001, Fig. 2.3, Table A.9). However, we found no evidence

! of an environment-by-type interaction (!! =0.74, P=0.39). These results are consistent with an asymmetric reproductive barrier. In fact, while nearly all non-dune plants failed to survive in the dunes, gene flow could be facilitated from the dunes onto the sand sheet by the success of dune plants on the sand sheet.

On the other hand, non-dune plants had higher fecundity than dune plants in the non-dune environment. In contrast, there was little effect of ecotype on fecundity in the dune environment (Fig. 2.4, Table A.10). Consequently, we found significant

! type-by-environment interactions for each proxy of fecundity (plant height: !! =13.41,

! ! P=0.001; number of flower heads: !! =37.8, P<0.001; seed number: !! =6.97, P=0.03).

These differences in fecundity would impede gene flow from the dunes to the sand sheet but would have little to no effect in the other direction.

34 8 100 50

ers 40 80 w 6 o 30 60 4 20

Height (cm) 40 2

Number of fl 10 20 log(Number of seeds) 0 0 N H D N H D N H D N H D N H D N H D Sand sheet Dune Sand sheet Dune Sand sheet Dune

Figure 2.4 - Mean values of three proxies of fecundity (height, number of flowers and number of seeds) measured for non-dune (N, squares), hybrid (H, circles) and dune (D, triangles) plants grown in sand sheet (open symbols) and dune (solid symbols) environments. There are significant habitat-

! ! by-type interactions for each proxy (plant height: !!=13.41, P=0.001; flower number: !!=37.8, P<0.001;

! seed number: !!=6.97, P=0.03). The error bars represent 95% confidence intervals.

Although seedling emergence and fecundity did not show patterns of local adaptation individually when synthesized into a cumulative measure of fitness a pattern of local adaptation became apparent. This is seen as significant environment- by-type interactions in our ASTER models, in which each ecotype is the most fit in the

! environment with which it is associated (!! =1951.2, P<0.001, Table A.3). Moreover, we saw patterns consistent with local adaptation in additional pilot reciprocal transplants conducted in 2010 and 2011, though there was annual variation in which environment had higher overall seedling emergence (Appendix A).

Local adaptation resulted in substantial reproductive barriers between the ecotypes. Gene flow into the dune environment is strongly impeded by selection

35 against immigrants, particularly at the seedling emergence stage (RIN->D=0.89, Table 2.2,

Fig. 2.5). Selection against immigrants also hindered gene flow from the dune environment into the sand sheet, particularly at the fertility stage, although the confidence intervals estimated by bootstrapping were very broad (RID->N=0.76, Table

2.2, Fig. 2.5). This occurred because non-dune emergence was so low in the non-dune environment that there was substantial variation among bootstraps in whether any emerged plants were included.

36 Barriers that impede gene flow into the dunes (RIN->D) 1.0

0.5

0.0

Barriers that impede gene flow onto the sand sheet (RID->N) 1.0 ier Strength r 0.5 Bar 0.0

−0.5

−1.0 selection selection selection selection selection flowering pollinator assortative F F against 1 against against 1 against against time assemblage mating germination fertility immigrants F1D F1N BCD BCN

Figure 2.5 – The distributions of barrier strength estimates from 10,000 bootstrap replicates (“beanplots”). The thickness of each bean is proportional to the number of bootstrap replicates consistent with a given barrier strength. The barrier strength caused by differences in flowering time is symmetrical (the two beans are identical). Reproductive barrier strengths greater than zero (dashed line) retard gene flow and those less than zero facilitate gene flow.

37 2.3.3 Flowering time

There was no effect of environment, type, or an environment-by-type interaction on the

! ! day to first flower in the data from the reciprocal transplant (!! =1.40, P=0.24; !! =4.82,

! P=0.09; !!.!=1.81, P=0.26; Fig. 2.6). Nevertheless, imperfect overlap between dune and non-dune flowering distributions estimated from timelapse photography (Fig. 2.7) generated a barrier strength of 0.093 (Table 2.2, Fig. 2.5), with dune plants flowering slightly earlier than non-dune plants. This result must be treated cautiously because chance fluctuations in the measurement of flowering distributions could lead to an apparent RI, even if none were present. To assess this possibility, we repeatedly reestimated RI using two samples from a single flowering time distribution (either dune or non-dune). The resulting barrier strength approximately halved (0.035-0.053), indicating that the observed RI is about 0.044 higher than expected based on sampling effects alone. We used the original barrier strength (0.093) in subsequent analyses.

38 1.0

0.8

0.6

0.4 Dune 0.2 Hybrid

0.0 Non-dune 1.0

0.8

0.6

0.4

0.2 tion of plants with first flower r 0.0 Propo 1.0

0.8

0.6

0.4

0.2

0.0 220 230 240 250 260 Julian date

Figure 2.6 – Kaplan-Meyer estimates of inverted survival curves showing the cumulative proportion of plants that have started to flower (censored data) during the reciprocal transplant. Data were pooled across dune and non-dune environments, as there was no effect of environment or an environment-by-type interaction. The shaded areas represent 95% confidence intervals for each fit.

39 A 150 0 200 220 240 260 280

150 B 0 200 220 240 260 280

100 C ower heads ower

f 0 200 220 240 260 280

25 D 0

Number of 200 220 240 260 280 E 15 0 200 220 240 260 280 F 0.03

0.00

Flowering time Flowering 200 220 240 260 280 Julian date

Figure 2.7 – Flowering time distributions recorded from timelapse cameras. A-E: the number of flowering heads visible in timelapse photos from several populations (A: DRT - dune reciprocal transplant site, B: D12, C: NRT - non-dune reciprocal transplant site, D: N3, E: N1). Each line represents an individual camera. F: density distribution of flowering time for each ecotype.

40 2.3.4 Pollinator assemblage

There was a significant difference between potential pollinator assemblages collected in the dunes and on the sand sheet after accounting for the distances between sites

(r=0.35, P=0.001, Fig. 2.8). There were several likely pollinators (sampled visiting sunflowers at least 30% of the time) that showed particularly biased distributions

(Table A.11). For example, we collected 61 Microbemex monodonata specimens across 6 samples from the dune environment but only 2 individuals on the sand sheet.

Similarly, we collected 66 Lasioglossum sp. individuals across 8 samples from the sand sheet and a single individual in the dunes and 38 Perdita dolichocephala individuals across 8 sand sheet samples and 0 in the dunes. When we consider the proportion of pollinator visits made by insects found in both environments, the strength of RI from non-dune to dune is 0.55 and from dune to non-dune is 0.36 (Table 2.2, Fig. 2.5). 2 0 PC2 −2 −4 −6 −4 −2 0 2 4 6 PC1

Figure 2.8 – The principle components of potential pollinator communities sampled from dune (red symbols) and non-dune (blue symbols) sunflower populations. Circles = netted collections, squares = malaise traps, triangles = pan traps.

41 2.3.5 Post-pollination assortative mating

Following pollination by equal mixtures of dune and non-dune pollen, dune pollen sired 65-72% of seeds produced by dune plants, and non-dune pollen sired 52-59% of

! seeds produced by non-dune plants (!! =24.4, P<0.0001, Fig. 2.9, Table A.12). This assortative mating results in a barrier strength of 0.38 from non-dune to dune and 0.12 in the other direction (Table 2.2, Fig. 2.5).

Non-dune Dune 1.0 A B

0.8 y dune plants b 0.6 ing sired r 0.4

0.2 tion of offsp r 0.0

Propo Mothers Means

Figure 2.9 – Siring bias A) Mean siring bias for individual dune (black) and non-dune (grey) plants measured as the proportion of offspring sired by dune plants after mixed pollinations. Only plants that had at least 5 offspring genotyped are plotted. B) Mean siring bias for all dune (black) and non- dune (grey) plants. Error bars are 95% confidence intervals and the dashed lines represent the expected proportion of offspring sired by dune plants (0.5).

In addition, more seeds were produced when dune rather than non-dune pollen was used to pollinate dune plants (mean difference 95% CI=1.0-7.6, t40=2.64, P=0.01).

42 This was not true for non-dune plants that produced equivalent numbers of seeds with both pollen types (mean difference 95% CI=-7.4-3.8, t23=-0.66, P=0.51). Although this difference in siring success in the absence of competition would contribute to the siring bias seen after mixed pollinations of dune mothers, it is unlikely to explain the entire pattern. First, if we used the number of seeds produced by pure crosses to predict the proportion of seeds sired by dune fathers, the predicted bias would only be 54-59%, which is significantly below the observed bias. Second, the pollination timing experiment suggests that dune and non-dune pollen have different fertilization rates in dune styles. In particular, the proportion of seeds sired by non-dune fathers is increased when non-dune pollen is given a 12-hour head start relative to a 6-hour head

! start (!! =6.79, P=0.009), while the proportion of dune fathers is unaffected by a longer

! head start for dune pollen (!! =0.06, P=0.81). This suggests that 6 hours is enough time for dune pollen to fertilize all available ovules but not enough time for non-dune pollen to do the same. Therefore, non-dune pollen likely has a slower or more variable fertilization rate in dune styles. This could be due to several factors including different rates of pollen germination, elongation or acceptance.

2.3.6 Intrinsic hybrid inviability

We found that hybrid seeds germinated as well as parental seed types. Specifically, the factors maternal ecotype, paternal ecotype and/or their interaction do not improve

! ! model fits (maternal: !! =1.15, P=0.29; paternal: !! =0.001, P=0.97; maternal-by-paternal:

! !! =0.12, P=0.73), nor does a factor that distinguishes hybrid and parental seed types

! (!! =0.08, P=0.78). Accordingly, both barrier strength estimates were low (RIN->D=-0.003,

RID->N=0.006) and had confidence intervals that overlapped zero (Table 2.2, Fig. 2.5).

43

2.3.7 Hybrid pollen sterility

The pollen produced by hybrid plants was largely viable (93%) and comparable to the viability of pollen produced by parental types (dune=95%, non-dune=87%, Fig. 2.10).

There were no significant differences between pollen produced by plants crossed

! within and between types (!! =0.92, P=0.35). Again, these estimates resulted in weak barrier strengths that are not significantly different from zero (RIN->D=0.01, RID->N=-0.03,

Table 2.2, Fig. 2.5). le) b 1.4

1.2

1.0 tion of pollen that is via r 0.8

DxD NxN N*xN N*xN* NxD N*xD

Arcsine(propo Within Between ecotype ecotype

Figure 2.10 – The proportion of viable pollen grains produced by plants of different cross types. D = dune plants collected from the dunes at GSD, N = non-dune plants collected from the sand sheet at

GSD, N* = non-dune plants collected by the USDA.

44 2.3.8 Selection against hybrids

The emergence patterns of all seeds types (D1-3, F1D, BCD, BCN, F1N, N1-3) in the reciprocal transplant echoes those of the parental ecotypes; more seedlings emerged in

! the dune environment (!! =19.37, P<0.0001), the different seed types emerged at

! different rates (!! =337.51, P<0.0001), and there was no evidence of an environment-by-

! type interaction (!! =14.54, P=0.1497). In addition, seedling emergence was significantly

! associated with seed weight (!! =118.68, P<0.001), although weight did not explain as much variation as seed type (∆AIC=+43). This is illustrated in the emergence patterns of the four hybrid cross types (Fig. 2.3). The heavier F1D seedlings emerged better than

F1N seedlings, but BCD seedlings also emerged better than BCN seedlings even though they have similar weights. In general, the more dune-like hybrids (F1D and BCD) emerged well, though not as well as dune seeds, while the other hybrids emerged poorly (Fig. 2.3).

The fecundity of hybrids was similar in all cases to the less fecund parental type

(dune plants). Thus, while non-dune plants were taller, had more flowers, and produced more seeds on the sand sheet, the hybrids did not exhibit any advantage (Fig.

2.4).

In the full ASTER analysis, we found significant environment-by-type

! interactions (!! =137.6, P<0.001, Tables A.4) in which the three dune populations were the most fit in the dune environment and two of the three non-dune populations were the most fit in the non-dune environment, compared to any of the other hybrid or parental types. Essentially, local ecotypes tended to be more fit than the hybrids and the mismatched ecotype.

45 Selection against each hybrid type in the dune environment reduces gene flow coming from the sand sheet (RIN->D: F1D=0.44, BCD=0.77, BCN=0.94, F1N=0.99, Table

2.2, Fig. 2.4). The same might happen on the sand sheet (RID->N: F1D=0.43; BCD=0.74,

BCN=0.55, F1N=1), but small sample sizes resulted in uncertain estimates (Table 2.2,

Fig. 2.5). In fact, not one F1N individual survived on the sand sheet.

2.3.9 Total reproductive isolation

We found very strong total RI separating the ecotypes. The strength of RI inhibiting seed flow from the sand sheet onto the dunes was 0.999 and from the dunes to the sand sheet was 0.992, though the confidence intervals include 0 in the second case due to low survival of all seeds (Table 2.2). Similarly, total RI specific to pollen flow was 0.986 from sand sheet to dune and 1 from dune to sand sheet. These calculations were not particularly sensitive to the exclusion of any single barrier. The biggest drop in total barrier strength (1 to 0.809) was observed after excluding selection against F1Ns from the calculation of total RI for pollen flow from the dunes to the sand sheet. The exclusion of any other barrier resulted in a less than 0.1 drop in total RI.

2.4 Discussion

Despite their young age, surprisingly many reproductive barriers separate dune and non-dune populations of Helianthus petiolaris. We found several extrinsic barriers that act before and after zygote formation and include selection against immigrants, different pollinator assemblages, and selection against hybrids. In addition, we found evidence for an intrinsic barrier, post-pollination assortative mating, which is likely caused by differential fertilization, possibly acting alongside ovule abortion and/or an

46 early acting hybrid incompatibility. Together these barriers generate very strong total reproductive isolation between the ecotypes.

The number and size of seeds and their effects on seedling emergence and plant fecundity appear to be important to RI in this system. Dune plants produce few large seeds that emerge better in both environments, while non-dune plants produce many small seeds on the sand sheet. It is likely that differences in seed size were driven by divergent natural selection and that they contribute to selection against immigrants and hybrids with intermediate phenotypes (Rice and Hostert 1993, Hatfield and Schluter

1999). Large seeds are associated with dune adaptation in other systems (Maun 1994,

Cordazzo 2002, Donovan et al. 2010), and seed manipulations in the field suggest that seed size per se is under selection in the sand dunes (R. Andrew unpublished). Perhaps large seeds provide seedlings with enough resources to emerge after being buried by sand or send stabilizing taproots deep into the dunes. Alternatively or in addition, large seeds may stay closer to the surface of shifting sands, again facilitating seedling emergence.

An unexpected aspect of these results it that we have little evidence for flowering time differences. In several plant systems (e.g., Hurlbert 1970, Lowery et al.

2008b, Husband and Schemske 2000, Briscoe Runquist et al. 2014) flowering time is one of the strongest barriers, favoring pollinator specialization as well as forming a barrier in its own right. There is scope for flowering time differences to evolve in the dune sunflowers, owing to the variability of flowering time, both in the field and in the greenhouse (pers. obs. and unpublished data). Yet pollinator filtering by habitat is a more important barrier to hybridization. Reproductive isolation due to pollinator assemblages is probably a common consequence of moving into a new habitat but has not often been quantified (but see Kay et al 2006, Roccaforte et al. 2015). 47 The rapid evolution of post-pollination assortative mating is also surprising, as this has not been reported in other recently diverged ecotypes (see Husband et al. 2002 and Briscoe Runquist et al. 2014 for related examples). Differential pollen success is thus worth investigating further in other incipient species. H. petiolaris is self- incompatible and unlikely to be pollen limited in these populations (they grow in nutrient poor soils and are visited by many potential pollinators). These conditions suggest that could be strong (Willson and Burley 1983), and it is possible that biased siring is the result of acting through differential fertilization success or biased ovule abortion. Moreover, mate choice may have evolved in response to selection against hybrids. If so, this would represent an example of reinforcement acting early in speciation. Alternative explanations for the evolution of post-pollination assortative mating are that biased siring is genetically linked or pleiotropic with other traits under selection or that the underlying genes drifted apart during an unknown period of allopatry.

Overall, reproductive isolation was consistently stronger from the sand sheet into the dune environment. These asymmetrical barrier strengths match population genetic estimates of gene flow that infer more gene flow onto the sand sheet than the reverse (Andrew et al. 2012, 2013). Asymmetric RI is common in studies that measure barrier strengths (e.g., Bolnick and Near 2005, Rahme et al. 2009, Sánchez Guillén et al.

2012, Ishizaki et al. 2013), though most often reported for postmating barriers in plants

(Lowry et al. 2008a).

Also in line with previous studies (Lowry et al. 2008a), we found several strong prezygotic barriers (selection against immigrants, different pollinator assemblages and the differential fertilization component of post-pollination assortative mating).

However, we also found strong postzygotic barriers caused by selection against 48 hybrids. In fact, postzygotic barriers alone yield stronger values of total RI (seed flow:

RIN->D=0.970, RID->N=0.906; pollen flow: RIN->D= 0.905, RID->N=1) than prezygotic barriers alone (seed flow: RIN->D=0.934, RID->N=0.847; pollen flow: RIN->D= 0.747, RID->N=0.488).

Because the prezygotic barriers are not complete, the postzygotic barriers contribute meaningful reductions in gene flow between the ecotypes.

Extrinsic ecological barriers (selection against immigrants and hybrids, different pollinator assemblages) tended to be stronger than intrinsic barriers (flowering time, post-pollination assortative mating, intrinsic hybrid inviability and sterility) in our data. As such, total isolation caused by extrinsic barriers (seed flow: RIN->D=0.981, RID-

>N=0.987; pollen flow: RIN->D= 0.971, RID->N=1) was stronger than total isolation caused by intrinsic barriers (seed flow: RIN->D=0.255, RID->N=0.222; pollen flow: RIN->D= 0.441,

RID->N=0.175). This pattern has been found in other systems (e.g., Ramsey et al. 2003,

Kay 2006, Melo et al. 2014) and is not surprising in a case of ecological speciation.

Total barrier strength is strong in both directions and with respect to both pollen and seed flow. However, the true values of unidirectional total RI are probably closer to the pollen flow estimates because pollen flow can be an order of magnitude more common than seed flow (Petit et al. 2005) and the object that flows more will have a disproportionate effect on isolation. Another consideration is that our total RI calculations assume each barrier is independent of one another (Martin and Willis

2007). However, if barriers were perfectly positively correlated, some genotypes would be stopped by multiple barriers while other genotypes would be stopped by none, allowing as much gene flow as the strongest individual barrier. Regardless, gene flow must be strongly reduced given the strong single isolating barriers we observed (Table

2.2). The nearly complete isolation resulting from these barriers explains the

49 maintenance of separate ecotypes in the face of gene flow.

Additional caveats to our measure of total RI are that we may have missed reproductive barriers and that some of the barriers we measured required simplifying assumptions. For example, we did not account for RI mediated via pollinator behavior.

Although dune and non-dune flower heads look superficially similar, pollinators could discriminate between them based on unmeasured characteristics like ultraviolet reflectance, volatile compounds or flower head shape and size. Also, our analysis of potential pollinator assemblages makes a number of simplifying assumptions that could over- or underestimate its effect. These assumptions include that all visiting species are equally effective pollinators and that species found in both environments actually travel between the two environments. It would be useful to quantify the extent of pollinator movement between ecotypes and determine whether the primary pollinators discriminate between dune and non-dune plants growing in the same population. We also did not measure ecogeographic isolation, which is a barrier caused by spatial separation as a result of genetically based ecological differences between taxa

(Schemske 2000, Sobel 2014). This barrier is important in other systems (e.g., Ramsey et al. 2003, Dell’Olivo et al. 2011, Sobel 2014). However, dune H. petiolaris has a very small range adjacent to the much larger range of typical H. petiolaris, and most dune populations are within the range of potential pollen and seed dispersal from the sand sheet. Because ecogeographic isolation would not be very effective at stopping migration into the dune environment and reduced migration into the sand sheet is limited by the extent of dune environments, we focused on local processes.

One consideration when using ecotypes to study speciation is that we cannot be sure that they will complete the speciation process (Coyne and Orr 2004). It is possible that many ‘incipient species’ never complete speciation, and instead persist at an 50 intermediate stage of divergence indefinitely. Although this scenario is possible for dune H. petiolaris we believe it is unlikely because the divergence is so recent and yet characterized by a surprising number of reproductive barriers. Furthermore, H. neglectus, another sand dune specialist sister to H. petiolaris, has completed the speciation process and exhibits many similarities to the ecotypes studied here, including divergence in seed size.

2.5 Conclusion

We tested dune and non-dune ecotypes of H. petiolaris for reproductive barriers that reduce gene flow between them. This is one of very few studies investigating the evolution of reproductive isolation within populations known to be separated by short time scales (<~10,000 years). We found several strong extrinsic barriers in line with local adaptation, as well as the intrinsic barrier, post-pollination assortative mating.

Our results highlight the importance of postzygotic barriers in addition to prezygotic barriers, even within the earliest stages of speciation. Taken together, the most striking result of these experiments is that multiple diverse reproductive barriers separate the incipient species despite a very recent divergence and in the presence of gene flow.

51 Chapter 3: Rapid chromosomal evolution between closely related sunflower species

3.1 Introduction

Mapping chromosomal rearrangements between species is a window into the nature of reproductive isolation and ultimately speciation. Many species are differentiated by changes in karyotype. These divergent kayotypes often cause sterility and inviability when brought together in hybrids (“underdominance”) and therefore contribute to reproductive isolation between species (e.g., Stathos and Fishman 2014). Additionally, chromosomal rearrangements can facilitate speciation by extending genomic regions protected from introgression (Noor 2001, Rieseberg 2001), accumulating genetic incompatibilities (Navarro and Barton 2003), and simplifying reinforcement (Trickett &

Butlin 1994).

However, explaining how underdominant chromosomal rearrangements fix in populations is still an open question. In addition to the classic but often unlikely explanation that rearrangements drift to fixation in small populations, several other mechanisms have been proposed. These include neutral chromosome fissions and fusions (Baker and Bickham 1986), meiotic drive (Chmátal et al. 2014) and association with locally adapted genes (Kirkpatrick and Barton 2006, Lowry and Willis 2010). The importance of each of these mechanisms is likely to depend largely on the types of rearrangements fixed (e.g., chromosome fusions, pericentric inversions) and the genes associated with them. Therefore, mapping and characterizing chromosomal rearrangements is a critical first step towards understanding how and why they spread.

52 Genetic maps are also useful tools in themselves. Quantitative trait locus (QTL) mapping studies and genome scans rely on measures of genetic distance (i.e. recombination) between markers. Knowing the locations and sizes of rearrangements can also be used to make predictions about the extent of gene flow between species.

This information is especially valuable for crop species and their relatives because plant breeders use it to help introgress beneficial alleles from wild species into domesticated lineages (Chetelat and Meglic 2000, Foulongne et al. 2003, Dirlewanger et al. 2004).

Plant breeders also use genetic maps to identify genes in newly sequenced species based on collinearity with model species (e.g., Choi et al. 2004, Dilbirligi et al. 2006).

Helianthus is an economically important genus that includes the domesticated sunflower, H. annuus (ANN). Wild Helianthus species are ecologically diverse and likely harbor genes that can be used to improve domesticated sunflower linages (Seiler and

Rieseberg 1997). Helianthus also has one of the highest rates of chromosomal evolution across all plants and (Burke et al. 2004) and many rearrangements appear to be strongly underdominant (Lai et al. 2005) suggesting that chromosomal evolution is particularly important to speciation in the group. Recently, high-density genetic maps of H. niveus (NIV) and H. argophyllus (ARG) were created and compared to ANN (Barb et al. 2014). In this paper, we extend that work with the addition of high-density genetic maps of H. petiolaris (PET) and H. neglectus (NEG). Together, these closely related species (Fig. 3.1) deepen our understanding of chromosomal evolution across the annual species of Helianthus and allow us to infer ancestral karyotypes.

53 4 H. niveus ssp. tephrodes (NIV) 1 5 3 H. neglectus (NEG) 6 H. petiolaris (PET) 7 2 H. argophyllus (ARG) 8 H. annuus (ANN)

Figure 3.1 - The phylogenic relationships between the species referenced in this paper based on

Stephens et al. (2015). Branch lengths are not to scale and the branches are numbered for referencing.

PET is a widespread species that is found in sandy soils throughout the United

States. It is made up of two major subspecies: PET ssp. petiolaris, which is commonly found in the southern Great Plains, and PET ssp. fallax, which is limited to more arid regions in Colorado, Utah, New Mexico, and Arizona (Heiser 1961). Across most of the western United States, PET is sympatric with the similarly widespread ANN.

Throughout this region, gene flow occurs between the species despite strong reproductive barriers including strongly reduced pollen viability of early generation hybrids (Heiser 1947). Previous genetic maps suggest that several chromosomal rearrangements differentiate the species (Rieseberg et al. 1995b, Ungerer 1998). Most recently, Burke et al. (2004) found 8 translocations and 3 inversions between ANN and

PET. However, all the previously created maps had limited marker density (e.g., averaging 40 markers per chromosome, Burke et al. 2004).

NEG is sister to PET and has a range restricted to deep sand dunes in NW Texas and SE New Mexico. Crosses between PET and NEG are somewhat less fertile than intraspecific crosses and hybrids have pollen viabilities of 75‐80% (Chandler et al. 1986, 54 Heiser 1958). One or two reciprocal translocations between the species are thought to contribute to this intrinsic inviability (Chandler et al. 1986, Heiser 1958). Here we describe high-density genetic maps (averaging 91 markers per chromosome) for both

PET and NEG.

3.2 Methods

3.2.1 Study species and crosses

PET and NEG are self-incompatible annual species, while domesticated ANN lineages are self-compatible and annual. All three species are genetically diploid with a haploid chromosome number of 17, however, like other plant species, the sunflower genome shows evidence of several events in its evolutionary history (Barker et al. 2008). The most recent of these was about 20 million years ago. We used pollen from a wild PET ssp. petiolaris plant (PI435836) and a wild NEG plant (PI435768) to fertilize ANN plants from the highly inbred and cytoplasmic male sterile line,

HA89cms. The wild accessions were collected in central Colorado (latitude: 39.741, longitude: -105.342) and the SE corner of New Mexico (latitude: 32.3, longitude: -104.0), respectively, and maintained at fairly large population sizes by the USDA. Using this crossing design, we can effectively subtract the homozygous ANN contributions to the resulting F1 offspring. This allows us to infer the recombination events that took place in the highly heterozygous wild parents.

55 3.2.2 Genotyping

We extracted DNA from 116 ANN x PET and 132 ANN x NEG F1 seedlings using a modified CTAB protocol (Doyle and Doyle 1987). For each cross type, we prepared an individually barcoded genotyping-by-sequencing (GBS) library using the restriction enzymes PstI and MspI (Poland et al. 2012) and sequenced paired-ends of up to 192 samples per Illumina HiSeq lane. We modified the GBS protocol to exclude DNA fragments outside of 300-800 bp and deplete repetitive sequences (Appendix B). The

NEG genetic map also included a small amount of data from a library prepared using only PstI.

To avoid biases caused by aligning sequences to the ANN reference genome, we used the program UNEAK to identify SNPs (Lu et al. 2013). This reference-free method trims and compares 67 bp sequencing reads (tags) to find single base pair mismatches and then uses a strict network filter to select the tags found in networks with only one or two other tags. We used a custom shell script to genotype samples based on the frequency of each tag. We scored samples with the same number of each tag up to a 10- fold difference between the number of each tag as heterozygous, samples with 100 times more of one tag as homozygous for the frequent tag, and samples with intermediate ratios or fewer than 10 tags as unknown genotypes. Finally, we filtered the resulting SNP tables for markers with 40-60% heterozygosity. The vast majority of markers that pass through this filter will have two alleles in the wild species parent, with one of those alleles fixed in the ANN parent. Rare but informative markers in which the wild species has two alleles that are both different from the allele fixed in

ANN were excluded from the analysis because they required different filtering criteria.

56 3.2.3 Genetic mapping

After excluding duplicate markers, markers with less than 80% of individuals genotyped, and individuals with less than 50% of markers scored, we used R/qtl to make genetic maps (Broman et al 2003). Briefly, we used the function

‘formLinkageGroups’ with a maximum recombination fraction of 0.35 and a minimum

LOD score of 8 to form linkage groups (LGs). In both cases, this resulted in two sets of

LGs that were mirror images of each other. We then used the function ‘switchAlleles’ to reverse the genotype scores of one set of mirror image LGs. After that we reformed the linkage groups and used the functions ‘orderMarkers’ and ‘ripple’ (with a 7 marker window size) to determine the best marker order in each LG. We removed internal markers whose absence decreased chromosome length by more than 10 cM as they are likely markers with high genotyping error rates and genotype calls that had error LOD scores greater than 4. We repeated these steps until each genetic map stabilized.

Finally, we excluded several very small (1-5 markers) LGs and one slightly larger (24 markers covering 16.3 cM) LG, which were probably artifacts due to the chance clustering of markers with high genotyping error rates or markers that were polymorphic in the ANN individual used for crosses, from the final maps. Based on observations of 2-4% residual heterozygosity in sunflower inbred lines (Mandel et al.

2013), the latter explanation seems most plausible.

3.2.4 Synteny Analysis

To compare marker orders, we used bwa mem (Li and Durbin 2010) to align the marker sequences produced by UNEAK (including the duplicate markers) to the

‘HA412Bronze’ ANN genome assembly (http://www.sunflowergenome.org). We

57 filtered the resulting sam files for map quality scores greater than one to exclude markers that map to multiple locations in the genome. We used a more conservative version of methods outlined in Barb et al. (2014) to determine marker synteny between maps. Essentially, we looked for 4 or more markers that were shifted or reversed in a linkage group relative to the ANN reference. However, because small-scale ordering errors in our genetic maps and the reference genome are likely, we ignored inversions that were less than 2 cM long in either map. Finally, we measured the lengths of rearrangements in cM from the first marker and last marker bounding the rearrangement. This is a conservative estimate given that the rearrangements could extend as far as the closest non-rearranged marker.

3.3 Results

The final PET map is made up of 1923 markers at 1833 unique positions. The 17 LGs are between 37.2-97.8 cM long and together span a total of 1329.6 cM (Fig. 3.2). The majority of the genetic distance (96%) covered by the PET map consists of regions with markers at least every 10 cM and most of it (77%) has a marker every 5 cM (Fig. 3.3).

The longest distance between markers is 25 cM. Similarly, the NEG map is made up of

1164 markers at 1148 unique positions that cluster into 17 LGs spanning 1119.4 cM. The

LGs range from 10.4 to 112.2 cM long and the longest distance between markers is 26.7 cM. Again, the majority (85%) of the NEG map has a marker every 10 cM, and most

(68%) of it has a marker every 5 cM.

58 PET 1 2 3 4/7 5 6 7/4 8 9 10 11 12/16 13 14 15 16/12/17 17 cM 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 NEG cM 1 2 3 4 5 6/16 7 8 9 10 11 12/15 13 14 15 16/1217/16/12 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 M 4 6 7 105 ANN 12 15 16 17 110

Figure 3.2 - Helianthus petiolaris (PET) and H. neglectus (NEG) genetic maps. Each linkage group is colored based on its relationship to the H. annuus (ANN) reference. Regions in which markers map to multiple regions of the ANN references are shown in grey (M) and regions in which marker order is reversed are shown with hatching (I). Inverted translocations are shown with arrows pointing from the proximal to distal end of each translocated segment as it is ordered in the ANN reference.

59 1.0 ered v

o 0.8

0.6 0.4 PET

tion of map c PET & ANN r 0.2 NEG 0.0 NEG & ANN Propo 0 5 10 15 20 25 30 Distance between markers (cM)

Figure 3.3 - Cumulative frequency distributions of nearest-neighbor distances for all markers (solid lines) and markers with homologs in the H. annuus (ANN) reference (dashed lines) for the Helianthus petiolaris (PET) and the H. neglectus (NEG) genetic map.

We were able to find homologous regions in the ANN reference for a large proportion of the markers that make up each map (66% in both cases). In total, about

92% and 78% of the ANN reference genome is represented in the PET and NEG maps respectively and most of those regions are syntenic. Specifically, LGs 1-3, 5, 8-11 and 14 from the PET and NEG maps, as well as PET 6, PET 15, NEG 4 and NEG 7, are roughly collinear with the ANN reference (Fig. 3.2, Fig. 3.4). About 94% of the PET map and

78% of the NEG map has a homologous marker at least every 10 cM and 67% and 53% of the maps have a homologous marker every 5 cM (Fig. 3.3). Because of this high density of homologous markers, most large rearrangements should be detectable in our maps.

60 PET1 ANN1 NEG1 PET2 ANN2 NEG2 PET3 ANN3 NEG3 PET14 ANN14 NEG14 0 10 20 30 40 50 60 70 80 90 PET5 ANN5 NEG5 PET8 ANN8 NEG8 PET9 ANN9 NEG9 PET4/7 ANN4 NEG4 0 10 20 30 40 50 60 70 80 90 100 PET10 ANN10 NEG10 PET11 ANN11 NEG11 PET13 ANN13 NEG13 0 10 20 30 40 50 60 PET7/4 ANN7 NEG7 70 80 90

61 PET ANN NEG PET ANN NEG 12/16 12 12/15 15 15 15 0 10 20 30 40 50 PET ANN NEG 17 17 17/16/12 60 70 80 90 100 110 120 130 140 150 160 170 180 190

PET ANN NEG PET ANN NEG 16/12/17 16 16/12 6 6 6/16

Figure 3.4 - Helianthus petiolaris (PET) and H. neglectus (NEG) genetic maps compared to the H. annuus (ANN) reference genome. Lines connect homologous markers and regions in which marker order is reversed are shown with hatching (I). The scale on the left is in centimorgans (cM). See Fig.

C.1-C.17 for details.

Consequently, we found five clear chromosomal translocations between LGs on each map relative to the ANN reference (Table 3.1, Fig. 3.2, Fig. 3.4). However, the specific segments translocated were not always consistent between maps. The PET map has two reciprocal translocations, one between PET 4/7 (where PET 4/7 refers to a PET

62 LG carrying segments of both ANN chromosomes 4 and 7) and PET 7/4 and another between PET 12/16 and PET 17/16/12. In addition, the distal portion of ANN 17 is attached to the proximal portion of PET 17/16/12. In the NEG map, the same portions of ANN 12 and ANN 16 that were found in PET 17/16/12 are found in NEG 16/12.

However, the distal part of ANN 16, which was found in PET 12/16, is now found in

NEG 6/16 and is replaced by the distal portion of ANN 15 to make NEG 12/15. Finally, parts of ANN 12 and ANN 16 moved to NEG 17/16/12. In addition to these clear translocations, we found several regions in the PET and NEG maps where markers mapped to multiple ANN LGs (Table 3.1, Fig. 3.2). These could be multiple small translocations, clusters of poorly mapped markers, or regions with complicated paralogy. The latter explanation is reasonable given known whole genome duplication events at the base of the sunflower tribe and of the sunflower

(Barker et al. 2008). Excluding the mixed regions, a total of 8% and 6% of the PET and

NEG regions mapped are translocated.

Table 3.1 – The number of inversions (Inv), translocations (Trans, where “micro-translocations” are made up of 2-3 markers less than 2cM apart) and regions of mixed homology for each linkage group in Helianthus petiolaris (PET) and H. neglectus (NEG) genetic maps.

PET linkage groups NEG linkage groups Trans Mixed Trans Mixed Number of: Inv Number of: Inv (micro) regions (micro) regions Pet1 2 0(1) 0 Neg1 2 0(1) 0 Pet2 0 0(2) 0 Neg2 1 0(1) 0 Pet3 0 0 0 Neg3 1 0 0 Pet4/7 1 1(1) 0 Neg4 0 0 0 Pet5 0 0 0 Neg5 0 0 0 Pet6 3 0 0 Neg6/16 1 1 0

63 PET linkage groups NEG linkage groups Trans Mixed Trans Mixed Number of: Inv Number of: Inv (micro) regions (micro) regions Pet7/4 0 1(1) 1 Neg7 2 0(1) 1 Pet8 3 0 0 Neg8 2 0(1) 1 Pet9 2 0 0 Neg9 1 0 0 Pet10 3 0 0 Neg10 2 0 0 Pet11 0 0(4) 0 Neg11 0 0(1) 0 Pet12/16 0 1 0 Neg12/15 1 1 0 Pet13 1 0 0 Neg13 2 0 0 Pet14 2 0 0 Neg14 3 0 0 Pet15 2 0(3) 0 Neg15 2 0(1) 1 Pet16/17/12 1 2(2) 0 Neg16/12 0 1 0 Pet17 0 0(1) 0 Neg17/16/12 0 2 0 Totals 20 5(15) 1 Totals 20 5(6) 3

There is also evidence of 20 inversions in each map (Table 3.1, Fig. 3.2, Fig. 3.4), and these inversions make up 23% and 26% of the PET and NEG maps respectively.

There were two places (PET 6 and NEG 13) where several inversions or a combination of intra-chromosomal translocations and inversions could explain marker order.

However, after taking the inversions on homologous linkage groups into account, the multiple inversions scenarios seems more likely and are presented here.

Most of the inversions that we see between PET and ANN are roughly matched by inversions we see between NEG and ANN. Accordingly, when we reduce our data set to markers that align to the same place in the ANN reference (only 108 markers), most LGs are largely syntenic between PET and NEG. However, we do see translocations between NEG 12/15 and PET 4/7, and there are hints (2 reversed markers) of inversions on LG 9 and LG 14. When we compare the rearrangements seen

64 between each wild species and ANN it appears there could be up to 6 translocations that differentiate PET and NEG. However, the portion of ANN 7 found in PET

17/16/12 is not represented in the NEG map meaning that it is possible that NEG shares this rearrangement. Furthermore, there are two markers that map to ANN 4 in the mixed region of NEG 7 suggesting that the NEG map might also share the PET 7/4 rearrangement. With this in mind, a more conservative estimate of the number of translocations between PET and NEG is 4.

3.4 Discussion

When we compare the PET and NEG maps to the ARG and NIV maps presented in

Barb et al. (2014), a couple of patterns emerge. First, given the phylogenetic relationships of the species (Fig. 3.1), we can infer that some of the rearrangements are quite old. For example, the reciprocal translocation between LG 12/16 and LG 16/12 is consistent across the wild species maps and probably represents the ancestral condition. In this scenario, a reciprocal translocation between LG 12/16 and LG 16/12 along the ANN branch (Fig. 3.1, branch 8) would create the patterns observed here.

Also, the distal portion of ANN 17 that moved to LG17/16/12 may have occurred along the branch to NIV, PET and NEG (Fig. 3.1, branch 1). Second, there are some linkage groups that are involved in rearrangements more often than others. For example, ANN 6 is uniquely positioned in all four wild species maps while ANN 3 is almost always completely syntenic. It would be very interesting to know if there are any traits correlated with the LGs that tend to get rearranged. For example, perhaps these LGs tend to be acrocentric (where the centromeres are located near the edge of a chromosome) or made up of more repetitive DNA.

65 Overall, we found fewer translocations (5) and more inversions (20) between

ANN and PET than Burke et al. 2004 (8 translocations and 3 inversions). This is probably due to a combination of three factors. First, there appears to be considerable intraspecific karyotypic variation in Helianthus, especially H. petiolaris (Heiser 1961), and the PET and ANN genotypes employed for synteny analyses do not overlap between the two studies. Second and more importantly, higher marker densities were employed, which allowed us to identify many small inversions that were undetected in the previous comparison. Lastly, we used more conservative estimates of translocations in the present study. In our analysis, we required at least 4 markers to be found on another LG to call a translocation, while there is less support for some of the rearrangements suggested by Burke et al. (2004). Furthermore, in each of our inferred translocations, the synteny is correctly inferred within the translocation, which is highly unlikely if the inference were due solely to mapping errors. However, had we included small multi-marker translocations (“micro-translocations”, 2-3 markers less than 2 cM apart) in our counts, we would have found 20 and 11 translocations in the

PET and NEG maps respectively (Table 3.1). It is unclear whether such small rearrangements represent mapping errors, regions that are paralogous in PET and

NEG, or true translocations. Regardless, we are very likely underestimating the total number of the chromosomal rearrangements in PET and NEG.

ANN and PET diverged about 1.8 million years ago (Sambatti et al. 2012) while

NEG and PET are thought to be much younger. Therefore, it is surprising that NEG and PET are separated by almost as many translocations as ANN and PET (4 versus 5).

This extremely rapid chromosomal evolution might explain why NEG persists despite extremely low genetic divergence and presumably high gene flow from PET (Raduski et al. 2010). Evidence from comparisons between ANN and PET and ANN and ARG 66 suggests that rearranged portions of genomes experience less introgression than collinear ones in Helianthus (Rieseberg et al. 1999, Barb et al. 2014). In the case of NEG, perhaps enough of the genome is protected from introgression that the genetic integrity of the species is maintained. It is also possible that these translocations facilitated speciation between H. petiolaris and H. neglectus. In a case of ecological speciation within the plant genus Mimulus, chromosomal translocations associated with underdominant male sterility and quantitative trait loci for premating barriers are thought to have facilitated speciation (Fishman et al. 2013, Stathos and Fishman 2014).

In order to get the best estimate of chromosomal evolution specific to NEG, we need a genetic map for PET ssp. fallax. This is because an allozyme and restriction fragment analysis (Rieseberg et al. 1990) and a more recent study using GBS sequencing

(Baute 2015) suggests that NEG is more closely related to PET ssp. fallax than PET ssp. petiolaris. Also, there is some evidence that there is cytological variation within PET ssp. petiolaris and a chromosomal rearrangement between PET ssp. petiolaris and PET ssp. fallax (Heiser 1961). It is likely that the ANN 6 and ANN 15 fusion seen in the NIV and

ARG maps, and also reported by (Burke et al. 2004) in PET, is polymorphic within PET, given that we did not find in our map. If that were the case and the ancestral karyotype to NEG had the LG 6/15 fusion, then a single reciprocal translocation in the NEG lineage would result in the NEG LG 6/16 and NEG LG 12/15.

These genetic maps add to the growing picture of chromosomal evolution across

Helianthus, but there is a lot more work to be done. For example, it would also be interesting to see whether the rearrangements found here include centromeres more often than expected by chance. This is because changes in the position of centromeres might cause female meitoic drive to fix chromosomal rearrangements (Chmátal et al.

67 2014). We hope that these data lead to a more comprehensive understanding of chromosomal evolution and speciation in Helianthus.

68 Chapter 4: Parallel genetic changes underlie adaptation to sand dunes in two sunflower species

4.1 Introduction

An emerging question from studies of genotype-phenotype relationships is whether the genetic changes underling adaptive phenotypic evolution are predictable. This question is best explored in lineages that have undergone parallel phenotypic evolution. An association between parallel phenotypic evolution and similar environmental transitions represents strong evidence that the trait changes are adaptive (Endler 1986, Schluter et al. 2004). Moreover, observations that the same genes underlie such parallel phenotypic changes imply that genetic solutions to selection are constrained. This could be because the genomic target size of sites that lead to the phenotype is limited, because certain mutations arise more frequently (e.g., repeat number changes), because some alleles already are segregating within the population

(i.e., standing genetic variation), or because genetic constraints like pleiotropy and epistasis cause selection to favor some mutations over others (Streisfeld and Rausher

2009, Conte et al. 2012). In contrast, when different genes underlie repeated and adaptive trait changes we can infer that selection is relatively unconstrained and therefore less predictable (Conte et al. 2012).

Many systems with repeated habitat divergence exhibit parallel phenotypic evolution (e.g., Jones et al. 1992, Pidgeon et al. 1997, Rundle et al. 2000, Rajakaruna et al.

2003, Rosenblum 2006, Østbye et al. 2006, Foster et al. 2007). Across the genus

Helianthus, sister taxa typically occur in different habitats, while similar habitats may be independently colonized by different sunflower species or by different lineages within

69 the same species (Kantar et al. 2015). For example, the H. anomalus, H. niveus ssp. tephrodes and H. neglectus are found in deep sand dunes, while their closest relatives are not (Heiser et al. 1969). At the same time, an ecotype of the widespread non-dune species, H. petiolaris, has colonized deep sand dunes in Colorado (Andrew et al. 2012). It is likely that these dune sunflowers have experienced similar phenotypic changes in response to their shared habitat type.

One trait that is particularly likely to contribute to dune adaptation is seed size.

Large seeds and fast seedling growth are thought to be to avoid burial by shifting sands (Chen and Maun 1999, Schwarzbach et al. 2001, Seiler 2014, Chapter 2) and seed size is correlated with improved emergence under field conditions in H. annuus (Ahmad 2001). Furthermore, the homoploid hybrid, H. anomalus, has heavier seeds than both its parental species (Ludwig et al. 2008) and H. niveus ssp. tephrodes has seeds that approach the size of those in H. anomalus (Heiser et al. 1969). It has been hypothesized that large seeds are a result of parallel evolution in H. anomalus and H. niveus spp. tephrodes (Schwarzbach et al. 2001). In addition, we have already shown that the dune ecotype of H. petiolaris has seeds that are three times heavier than the non- dune ecotype (Chapter 2). However, whether similar shifts have taken place in dune H. neglectus has yet to be investigated.

Together these observations suggest that identifying the genetic basis of seed size in dune sunflowers represents an opportunity to examine genetic constraint during adaptive evolution. We chose to focus on dune and non-dune ecotypes of H. petiolaris and H. neglectus because H. anomalus is a homoploid hybrid species with two non-dune parental species and the non-dune counterpart to H. niveus ssp. tephrodes is rare.

Furthermore, we have shown that seed size differences in H. petiolaris likely contribute to strong reproductive barriers (i.e., extrinsic immigrant inviability and extrinsic 70 selection against hybrids, Chapter 2) that separate the ecotypes. Given that seed size has been implicated in divergence between the ecotypes, we expect that the genes underlying variation in seed size should map to the few large regions of increased genomic divergence that separate dune and non-dune H. petiolaris identified by

Andrew et al. (2013).

Therefore, in this chapter, we address the following questions: (1) Are large seeds associated with dune habitats in Helianthus neglectus? (2) Are the genes that underlie seed size in H. petiolaris found in the regions of elevated divergence that have been identified between dune and non-dune ecotypes of the species? (3) Is there evidence that the same genetic regions underlie seed size differences in both H. petiolaris and H. neglectus?

4.2 Methods

4.2.1 Seed size of natural populations

We collected seeds from natural populations of H. neglectus found on and off sand dunes near Monahans, TX. We collected seeds from 10 non-dune populations, 6 dune populations within Monahans Sandhills State Park, and 2 dune populations outside the park (Table D.1). We weighed groups of 5 seeds from 16-21 individuals per population.

To determine whether seed weight was associated with dune habitats, we fit a linear mixed effect model describing seed weight that included habitat as a fixed effect and population as a random effect using the R package nlme (Pinheiro et al 2015).

Throughout this paper, we used R version 3.1.3 for statistical analyses (R Core Team

2015).

71

4.2.2 Seed size common garden

Differences in seed weight associated with natural populations could be caused by phenotypic plasticity. To explore this possibility, we grew 1-4 plants from each of 2 dune and 3 non-dune populations of H. neglectus under standard greenhouse conditions (16 hours of light at 600W/m2 and 18-23°C). If phenotypic plasticity is responsible for the seed size differences we observed in the field, we expect those differences to disappear in the common garden. We collected the seeds produced by these plants and weighed individual seeds. In our statistical analysis, we fit a linear mixed effect model using the R package nlme to describe seed weight. The model included habitat as a fixed effect and maternal plant nested within population as a random effect.

4.2.3 Crossing and phenotyping

We made two mapping populations for each species by reciprocally crossing a single wild individual with large seeds obtained from a dune population with a single individual with small seeds obtained from a non-dune population. Plants from the following H. petiolaris populations were crossed (dune x non-dune): 1701x1791 (cross

Pet1) and 1547x1500 (cross Pet2, Table D.1, Fig. 4.1). Likewise, plants from the following H. neglectus populations were crossed (dune x non-dune): KING

153xPI468777 (cross Neg1) and KING153xPI468781 (cross Neg2, Table D.1). The H. neglectus crosses were made with two separate dune plants each grown from seeds collected from different plants in population KING153. To make the crosses, we covered immature flower heads with mesh to prevent natural pollination and hand

72 pollinated the flowering heads once they were mature. We grew up several F1 seeds produced by each of these crosses and crossed pairs of F1s that had the same parents

(i.e., parent A and parent B) but where one F1 had parent A as a mother while the other

F1 had parent B as a mother. This resulted in an F2 seed lot with dune cytoplasm and an

F2 seed lot with non-dune cytoplasm for each of the four cross types (Table 4.1).

Helianthus petiolaris Helianthus neglectus

0 5 10 kilometres 0 150 300 kilometres

Non-dune population Dune population Dune habitat

Great Sand Dunes National Park and Preserve New Mexico Texas 1791 1701 PI468781 PI468777 1547 1500 KING153 Monahans Sandhills State Park

Park boundry

Figure 4.1 – Map of populations used to make QTL mapping populations in H. petiolaris and H. neglectus.

73 Table 4.1 – The Helianthus petiolaris and H. neglectus F2 mapping populations generated in this study.

The maternal plants are listed first in each cross and grandparent plants from dune populations are

shown in bold. See table D.1 for information about the locations of the populations used.

Cross Crossing design Cytoplasm # F2s Pet1A (1547-8 x 1500-9) x (1500-9 x 1547-8) dune 300 Pet1B (1500-9 x 1547-8) x (1547-8 x 1500-9) non-dune 300 Pet2A (1701-1 x 1791-1) x (1791-1 x 1701-1) dune 175 Pet2B (1791-1 x 1701-1) x (1701-1 x 1791-1) non-dune 325 Neg1A (KING153-19 x PI468777-1) x (PI468777-1 x KING153-19) dune 255 Neg1B (PI468777-1 x KING153-19) x (KING153-19 x PI468777-1) non-dune 245 Neg2A (KING153-22 x PI468781-1) x (PI468781-1 x KING153-22) dune 317 Neg2B (PI468781-1 x KING153-22) x (KING153-22 x PI468781-1) non-dune 208

We grew up 175-325 F2 plants of each cross-by-cytoplasm type under the

standard greenhouse conditions described above. In addition, we grew a few plants

from wild collected seeds (half siblings of the grandparent plants) and F1 seeds (full

siblings of the parent plants) used to make the F2 seed lots. We allowed the F2 plants to

be open pollinated, however, when natural pollinators were scarce we crossed

neighboring F2 plants to ensure seed set. Finally, we allowed the plants to set seed and

weighed a group of 5 seeds from each plant.

4.2.4 Genotyping

To reduce costs, we used a selective genotyping approach in which we only genotyped

individuals with extreme seed size phenotypes (Lander and Botstein 1989). Specifically,

we genotyped the 12 plants that produced the heaviest seeds and the 12 plants that

produced the lightest seeds from each cross-by-cytoplasm type (Figure 4.2), for a total

of 192 plants. We extracted DNA from these individuals using Qiagen DNAesay kits

74 and prepared a GBS library for each species using the restriction enzymes MspI and

PstI and individual barcodes (see Appendix B for details, Poland et al. 2012). We sequenced the libraries using Illumina paired-end technology and found SNPs using the UNEAK method described in Chapter 3. We used bwa to align the marker sequences produced by UNEAK to the ‘HA412Bronze’ H. annuus genome assembly

(http://www.sunflowergenome.org) and then used the genetic maps described in

Chapter 3 to recalibrate the genetic distances between markers for each species.

H. petiolaris H. neglectus Cross Pet1 Cross Neg1

30 60 Dune cytoplasm Dune cytoplasm 40 20 20 10 0 0 Non-dune cytoplasm Non-dune cytoplasm 40 10 5 0 0 0 5 10 15 2 4 6 8 Cross Pet2 Cross Neg2

40 Dune cytoplasm Dune cytoplasm 20 10 0 0 30 Non-dune cytoplasm 30 Non-dune cytoplasm 20 20 10 10 0 0 0 5 10 15 2 4 6 8 Seed weight (mg) Seed weight (mg)

Non-dune F1 Dune F (genotyped) F (not genotyped) 2 2

Figure 4.2 - Distribution of seed weights produced by each F2 mapping population used in this study.

The plants selected for genotyping are shaded blue on each histogram. The mean and 95% confidence

intervals of the weight of seeds produced by relatives of the parents and grandparents of each F2 population are displayed above each pair of histograms. There was only one relative of the non-dune

grandparent of the F2 mapping population, cross Neg1, so there are no error bars for that point.

75 4.2.5 QTL mapping

We filtered markers for 50% coverage and input genotypes for between 453 and 627 markers from each cross type into R/qtl using the 4way cross notation (Broman et al.

2003). The 4way cross notation allows parental haplotypes to be tracked separately during QTL analysis. This is necessary when individuals from outbred populations are used as grandparents in crosses, as is the case here. To minimize bias associated with selective genotyping, we included phenotypes from every individual that we scored for seed weight in our analysis (Sen et al. 2005). The genotypes of individuals with typical seed weights (i.e., those that were not genotyped) were recorded as missing data. We performed multiple imputation interval mapping (Sen and Churchill 2001) using the

R/qtl function ‘scanone’. We used 100 imputations during interval mapping, and we included cytoplasm type as a covariate in each model. For each species, we analyzed each cross individually and in combination. To determine the LOD threshold for significant QTL in each model, we used 1000 stratified permutations of our phenotypic data in which individuals with genotypes were permuted independently from individuals without genotypes (Manichaikul et al. 2007). When two crosses were analyzed together, we fit models that included cross type as either an additive or interactive covariate and used the difference between these two models to determine whether there was evidence of genotype-by-cross interactions. Finally, we used 1.5-

LOD confidence intervals to estimate the position of each QTL, the function

‘find.marker’ to identify the marker nearest to each QTL peak, and the function ‘fitqtl’ to estimate the percentage of variance explained by each QTL.

76 4.3 Results

Seeds collected from H. neglectus plants found growing on sand dunes are about 3.96 mg or 73% (95% CI = 2.92-5.00) heavier than those collected from plants growing on nearby sandy soils (F1,16=55.9, P<0.0001, Fig. 4.3). This difference is maintained under greenhouse conditions where dune plants produced seeds that are 3.83 mg or 120% heavier (95% CI = 1.37-6.3) than non-dune plants, although it is reduced to marginal statistical significance (F1,3=9.3, P=0.055, Fig. 3) because of reduced sample size.

Natural populations Common garden dune 15 non-dune

10

5 Seed weight (mg) 0 1305 Dune Dune MON001 MON050 MON100 MON150 MON200 MON250 MON300 MON350 MON400 MON450 MON500 MON550 MON600 MON650 MON700 MON750 MON800 MON850 PI468777 PI468781 Non-dune Non-dune KING 153 KING 154

Figure 4.3 - The weight of seeds produced by H. neglectus plants in natural populations and grown in a common garden. Each point represents the average seed weight of an individual plant based on at least 5 seeds. Squares with error bars represent the mean and 95% confidence intervals of the seed weight produced by plants from each habitat type.

Each F2 mapping population produced a roughly normal distribution of seed sizes that spanned most of the variation seen between the grandparent populations

77 (Fig. 1.2). In all cases, F1 plants produced intermediate seed sizes, plants from dune populations produced larger seeds than plants from non-dune populations, and cytoplasm type did not have a detectable effect on the size of seeds produced by the F2s

(Fig. 1.2). Between 3 and 7 loci produce the seed size variation in these crosses according to Castle-Wright estimations (Pet1: 7.0, Pet2: 5.9, Neg1: 4.8, Neg2: 3.1, Castle

1921, Wright 1968). However, this equation is known to underestimate the actual number of loci when its assumptions (e.g., loci are additive, unlinked and have equal effects) are not met. Therefore, in this case, the estimated number of loci should be viewed as minimum values (Lynch and Walsh 1998).

Interval mapping in the first H. petiolaris F2 mapping population (cross Pet1) revealed three significant QTL regions on chromosomes 5, 11, and 14 (Table 2, Fig.

4.4A). When these three QTL are combined in a single model that included interactions, it accounts for 15.8 % of the variation in seed size in that cross and is highly significant

(LOD=20.15, p=0.009). In contrast, the second H. petiolaris F2 mapping population (cross

Pet2) did not contain any regions that exceeded our LOD threshold for significance.

When we analyze both crosses together, the QTL region on chromosomes 14 and most of the region on chromosome 11 remain significant and neither region shows evidence of an interaction between genotype and cross (Fig. 4.4B). This suggests that these regions may also contribute to seed size differences in the second H. petiolaris cross despite not being detectable on their own. The QTL on chromosome 5 and the distal portion of the QTL on chromosome 11 show significant cross-by-genotype interactions.

In fact, the direction of the effect at both these regions appears to be opposite in the two crosses.

78 Table 4.2 – Description of QTL regions identified in chapter 4.

Variance Interval H. annuus Crosses CHR Marker LOD explained (%) (cM) interval (cM) Pet1 5 TP24762 4.59 3.78 14.01-26.01 41.10-50.56 Pet1 11 TP14842 3.67 3.03 21.74-52.74 43.89-54.69 Pet1 14 TP25186 2.81 2.33 0-32 0-39.89 Pet1 & Pet2 11 TPB24556 3.27 1.71 11.16-44.16 42.42-46.91 Pet1 & Pet2 14 TP25186 3.54 1.92 0.01-12.55 0.02-34.97 Neg2 14 TP18407 5.07 8.89 4.48-40.01 0-40 Neg1 & Neg2 11 TP13073 3.36 3.44 2.57-51.29 14.28-51.29 Neg1 & Neg2 14 TPB42968 7.07 7.18 24.01-44.01 22.4-44.8

A H. petiolaris C H. neglectus Cross Pet1 5 Cross Neg1 4 Cross Pet2 Cross Neg2 4 3 3 2 2 LOD score 1 LOD score 1 0 0 B D 4 Cross Pet1+Pet2 7 Cross Neg1+Neg2 Pet1xPet2 Interaction 6 Neg1xNeg2 Interaction 3 5 4 2 3 1 2 LOD score LOD score 1 0 0 5 11 14 5 11 14 Chromosomes Chromosomes

Figure 4.4 - The locations of seed size QTL. The QTL maps show LOD scores along chromosomes 5,

11, and 14 from models of two F2 mapping populations in H. petiolaris (A), the combination of the H. petiolaris crosses and their interaction (B), two F2 mapping populations in H. neglectus (C), and the combination of the H. neglectus crosses and their interaction (D) from interval mapping. The dashed lines in each graph represents LOD thresholds for significant QTL determined from 1000 stratified permutations of the data.

79 Much like the H. petiolaris crosses, we were able to detect QTL in one of the H. neglectus crosses but not the other. Specifically, we detected a single QTL region on chromosome 14 that explains 8.9% of the variation in seed size in our second H. neglectus F2 population (cross Neg2, Fig. 4.4C). When we combine the two crosses together, we see that part of this region has opposite effects in the two crosses and a significant interaction, while the other part has similar effects in the two crosses that are statistically indistinguishable (Fig. 4.4D). The region on chromosome 14 with similar effects is significant in the combined analysis (Fig. 4.4D). In addition, a new region on chromosome 11, that was suggestive but not significant in either individual cross, is detectable when the two H. neglectus crosses are combined into one analysis.

After converting the QTL intervals found above back to the H. annuus genetic map, we can compare their positions to the previously published genome scan between several dune and non-dune populations (including the populations used here) of H. petiolaris (Andrew et al. 2013) and to one another (Fig. 4.5). As expected, one of the QTL found in H. petiolaris completely overlaps with a region of elevated divergence between dune and non-dune plants on chromosome 11. In addition, the QTL on chromosome 5 is within 1 cM of a second region of elevated divergence. It is possible that the regions do in fact overlap but that problems with small-scale marker order obscure the pattern.

On the other hand, the seed size QTL on chromosome 14 is not associated with increased divergence between the H. petiolaris ecotypes.

80 2 Neg1&2 QTL Neg2 QTL 5 Pet1&2 QTL Pet1 QTL 7 Elevated divergence between dune and 9 non-dune Pet

11

13

14

17 0 20 40 60 80 100 cM along chromosomes

Figure 4.5 – The locations of seed size QTL and elevated divergence between dune and non-dune H. petiolaris. Colored bars represent 1.5 LOD intervals of the seed size QTL identified in H. petiolaris and H. neglectus in this study and the regions of elevated divergence identified between dune and non-dune H. petiolaris (from Andrews et al. 2013). All regions here have been transformed to the H. annuus reference and only the chromosomes with regions of interest are represented.

When we compare the locations of seed size between the two species, the QTL regions on chromosomes 11 and 14 are shared between the species, while the QTL on chromosome 5 is unique to H. petiolaris (Fig. 4.5).

81 4.4 Discussion

We find substantial differences in seed size in dune versus non-dune populations of H. neglectus consistent with seed size differences in three other Helianthus dune species.

Although a reciprocal transplant or seed size manipulation would be more direct evidence that seed size is adaptive in H. neglectus, the repeated evolution of increased seed size concurrent with transitions to sand dunes across Helianthus suggests that natural selection is responsible for larger dune seeds here.

Overall, we found 3 QTL for seed size in H. petiolaris and 2 QTL for seed size in

H. neglectus. In all cases, the QTL were detected in only one of the two crosses made within each species, and we found several significant interactions between genotype and cross. This suggests that some of the alleles for seed size or epistatic modifiers of seed size are segregating within the ecotypes. This level of variation is unsurprising in

H. petiolaris where population genetic analyses of the dune and non-dune ecotypes show that overall divergence is low (Fst = 0.011) and where there is no evidence of fixed differences (maximum Fst at a single site = 0.58, Andrew et al. 2013).

The QTL we detected explain a maximum of 15% of variation we see in any one

F2 cross. This means that there are almost certainly small-effect QTL lying below our significance thresholds that account for some of the remaining variation. Accordingly, the minimum number of loci expected by Castle-Wright estimation exceeds the number of loci detected here. On one hand, it is not surprising that a quantitative trait like seed size is made up of several to many QTL of small to moderate effects. In H. annuus, where larger seeds were strongly selected during domestication, QTL for seed size have been found on at least 11 of the 17 chromosomes (Bert el al. 2003, Wills and Burke

82 2007, Mokrani et al. 2002, Tang et al. 2006, Baack et al. 2007, Chapman et al. 2008). On the other hand, one might have expected a few large-effect QTL to underlie a trait that maintains divergence in the face of substantial gene flow, which is a pattern frequently observed in other systems and simulations (e.g. Bradshaw et al. 1998, Griswold 2006,

Yeaman and Whitlock 2011).

The locations of QTL for seed size on chromosome 5 and 11 in H. petiolaris are consistent with the hypothesis that divergent selection on seed size is important to reducing gene flow between the two ecotypes in H. petiolaris. However, the QTL on chromosome 14 does not correspond to a region of elevated divergence between the ecotypes. The simplest explanation for the lack of significant divergence is that this

QTL may be too weakly selected to counterbalance on-going migration. A related possibility is that pleiotropic or epistatic effects of the large-seed size allele on chromosome 14 may counter its benefits, preventing it from rising to high frequency on the dunes. The major seed size QTL we found in H. neglectus is on chromosome 14 in a region that may be inverted relative to H. petiolaris and possibly non-dune H. neglectus

(Chapter 3). Although the evidence for the inversion is currently weak (only two markers are reversed), the pattern is worth following up. Chromosomal inversions are expected to be associated with genes under selection both when adaptive traits drive the fixation of otherwise underdominant inversions (Kirkpatrick and Barton 2006,

Lowry and Willis 2010) and when the reduced recombination inside inversions favors multiple adaptive loci to co-localize there (Navarro and Baron 2003, Gavrilets 2004,

Feder and Nosil 2009).

There is overlap between some of the QTL underlying seed size in H. petiolaris and H. neglectus, and the degree of overlap (2 of 3 genes are shared) is roughly consistent with patterns of QTL reuse during parallel phenotypic evolution across taxa 83 (Conte et al. 2012). However, because it unlikely that we are detecting all the QTL responsible for seed size differences, and there appears to be segregating variation for seed size within each comparison, we could be under- or overestimating of the extent of QTL reuse between the species. In addition, the QTL regions identified here are broad and it is possible that different but tightly linked genes underlie the phenotype in the two species. A goal for future work will be to identify the actual genes and mutations that underlie seed size phenotypes in dune sunflowers.

Standing genetic variation might be responsible for the parallelism observed here. In fact, adaptation from standing genetic variation seems quite likely because non-dune H. petiolaris is distributed widely and experiences gene flow from both dune sunflowers studied here (Raduski et al. 2010, Andrew et al. 2012). Moreover, small sand dunes scattered throughout the range of H. petiolaris could maintain large seed alleles.

Interestingly, small effect alleles are expected to contribute more to adaptation from standing genetic variation than de novo mutation (Barrett and Schluter 2008). To determine whether dune alleles originate from standing genetic variation it is again important to first identify the genes of interest.

Dune adaptation might be the ancestral condition for these sunflowers. If that is true, recent dune adaptation may represent independent reversals to the ancestral condition. It is also possible that this pattern is caused by a single selective event in which there was a change from a dune to non-dune habit in the widespread H. petiolaris plants; though, this is not likely given the geographic distance and lack of dune habitat between the systems studied here. However, if this were the case, the transitions between dune and non-dune types would not be independent (i.e. the phenotype evolved only once) and the QTL reuse observed here would not be indicative of genetic constraints or biases. More detailed genetic mapping and phylogenetic analyses within 84 the genomic regions associated with dune adaptation would help tease apart these explanations.

An interesting question about QTL reuse in general is whether it promotes similarities in subsequent ecological and evolutionary dynamics. It seems possible that when the same genes, which could have the same epistatic and pleiotropic effects, underlie a particular trait, the consequences of that phenotypic evolution will be comparable. For example, if the same genes underlie seed size changes associated with multiple transitions to sand dunes and those genes have pleiotropic effects on assortative mating, we might expect speciation to proceed in a more similar manner than if different genes were responsible for seed size changes in each case.

The results presented here add to the growing body of work that explores whether the genes that underlie adaptive parallel evolution are predictable. We find that seed size in H. neglectus has evolved in parallel to that in H. petiolaris and that some of the genetic regions that underlie these changes are shared between the species.

Future work will explore whether these repeated genetic underpinnings correspond to parallelism at smaller scales, such as genes or individual nucleotides.

85 Chapter 5: Conclusion

5.1 Discussion

Each of the preceding data chapters has contributed to our understanding of adaptation and speciation individually. I have shown that multiple and diverse reproductive barriers can separate very young species, that chromosomal evolution can be surprisingly fast and that the genetic basis of adaptation to sand dunes via larger seeds is repeatable and involves similar genetic regions in Helianthus. Here, I will compare and contrast progress towards ecological speciation between dune and non- dune H. petiolaris in Colorado and H. neglectus and H. petiolaris in Texas.

When I combine the data from previous studies with the data presented in this thesis, several similarities between these systems become apparent (Fig. 5.1). Both sand dune habitats are more than superficially similar with lower nutrients and qualitatively loose sand. There is parallel evolution of larger seeds in association with sand dunes, and there is some repeatability of the genetic regions that underlie that repeated phenotype. In both cases, reproductive barriers (e.g., poor F1 seed formation) separate the dune and non-dune types despite evidence of ongoing gene flow. There is an additional intrinsic reproductive barrier in both cases (pollen competition in Colorado and hybrid sterility in Texas). Furthermore, because selection against immigrants and hybrids in Colorado is probably due to differences in seed size, I expect this barrier to be present in Texas as well. Similarly, it is likely that dune specific pollinator communities in Texas would reduce gene flow between the environments in a manner similar to that in Colorado. Overall, these data are consistent with sand dune adaptation facilitating speciation between dune and non-dune types in both places.

86 Coloardo Texas dune Helianthus petiolaris Helianthus neglectus vs. non-dune H. petiolaris vs. Helianthus petiolaris

Low nutrients Low nutrients in sand dunes in sand dunes

Very young Unknown

Gene fow Gene fow

Big seeds 2 /3 QTL Big seeds

Strong selection against Unknown hybrids and immigrants

Diferent pollinators Unknown

Pollen competion Unknown

Poor F1 seed formation Poor F1 seed formation

No hybrid sterility Hybrid sterility

Chromosomal Unknown rearrangements

Similarities Unknown Diferences

Figure 5.1 – Summary of information known about divergence between dune and non-dune types within H. petiolaris in Colorado and between H. neglectus and H. petiolaris in Texas (Chapters 2-4, Chandler et al. 1986, Heiser 1958, Raduski et al. 2010, Andrew et al. 2012, Andrew et al. 2013).

87 On the other hand, there is a least one difference between these incipient speciation events. Hybrids between dune and non-dune H. petiolaris from Colorado are perfectly fertile while hybrids between H. petiolaris and H. neglectus in Texas are not.

This lack of hybrid sterility in Colorado suggests the absence of extensive chromosomal rearrangement between the dune and non-dune types because most major chromosomal rearrangements in Helianthus are underdominant (Chandler et al. 1986,

Lai et al. 2005). Therefore, the extent of chromosomal rearrangement between the pairs of sunflowers is probably also unequal. It is possible that H. neglectus and H. petiolaris started to diverge more than 10,000 years ago and that the increased divergence time was necessary for the extent of chromosomal rearrangement and hybrid sterility observed between the species.

Finally, it remains to be seen whether assortative mating based on pollen competition between types is a common feature of sand dune adaptation. However, if pollen competition is a pleiotropic effect of one or more loci that underlie seed size differences (or of the seed size phenotype itself) in the Colorado sunflowers, it could separate H. neglectus and H. petiolaris in Texas as well.

So far, I have presented the repeated divergence between dune and non-dune sunflowers in Colorado and Texas as a case of replicated ecological speciation.

However, it is important to consider whether this could be a case of parallel ecological speciation, in which the dune H. petiolaris in Colorado is equivalent to H. neglectus and the two sunflowers are reproductively compatible. If that were the case, it would suggest that divergent natural selection alone is responsible for the evolution of reproductive barriers between the types and that any differences observed are irrelevant to speciation. However, given that at least one reproductive barrier measured here is unique to a system, true parallel ecological speciation seems unlikely. 88

5.2 Strengths and limitations

This thesis explores several aspects of ecological speciation and is strongly rooted in the natural history of the systems studied. In an age where it is tempting to focus solely on genomic divergence, studies that link that divergence with reproductive barriers, phenotypes, genetic architectures and ecological contexts are rare and extremely informative. In addition, this thesis developed a new system useful to studying the repeatability of speciation. Indeed, there is enough information presented here to make testable predictions about speciation in each system.

The disadvantage of exploring the dune and non-dune types in Colorado and

Texas in depth is that I was unable to extend the work to additional cases of dune adaptation within the timeframe of a PhD. Clearly, any associations between the characteristics of a system and the evolution of reproductive isolation would be more convincing with more replicates. Also, my ability to compare and contrast the systems is limited by the extent to which the various experiments were replicated in each pair of species. Finally, there is the underlying limitation of studying incipient species in that we cannot be sure that either pair of divergent lineages will progress to “good” species, exhibiting complete reproductive isolation (Coyne and Orr 2004).

89 5.3 Future directions

There are several outstanding questions about divergence between H. petiolaris ecotypes in Colorado and between H. petiolaris and H. neglectus in Texas that can be relatively easily addressed. First, it is important to confirm that the two dune types are independently derived. Unfortunately, gene flow between dune and non-dune types in both systems makes empirically distinguishing multiple versus single origins difficult

(Quesada 2007). However, I can use the phylogenetic relationships among the genomic regions that are diverged between types (e.g., seed size QTL) to determine the number of origins. Second, I would like to test for reproductive compatibility between dune H. petiolaris in Colorado and H. neglectus to determine whether this is a case of parallel ecological speciation. Crossing the two dune types and assessing the viability and fertility of the resulting hybrids would be a good first step towards this, but a reciprocal transplant between sand dune habitats is also necessary to confirm that the dune types are equivalent.

Another obvious extension of this work is to measure additional reproductive barriers between H. neglectus and H. petiolaris at MSSP in Texas. It would be best to repeat all the barriers that I measured between the ecotypes at GSD (chapter 2).

However, testing for selection against immigrants and hybrids with a reciprocal transplant and testing for pollen competition between the ecotypes are a top priority. I would also like to more precisely measure the reduced seed set in crosses between types in both systems and compare the magnitude of those effects. To continue collecting the same information about both systems, I would also confirm that chromosomal rearrangements do not separate dune and non-dune H. petiolaris using the methods presented in chapter 3 of this thesis. 90 When studying parallelism it is useful to determine the actual genes and mutations responsible for phenotypes (Rausher and Delph 2015). In this case, fine mapping the genes that underlie seed size and the various aspects of pollen competition (i.e., pollen competitive ability and choosiness) would be particularly useful. Finding the mutations that underlie seed size would determine whether the parallel phenotypes are caused by parallel genetic pathways, genes or mutations and could rule out standing genetic variation as the cause of QTL reuse. Second, discovering the mutations that underlie pollen completion would determine whether that barrier arose due to pleiotropic effects of seed size.

The data on seed size QTL reuse presented here would be complemented by parallel population genomic scans that quantify the degree of similarity between loci associated with sand dune adaptation in the two systems (Conte et al. 2015). In fact, I already have whole genome sequence data from 10 dune and 10 non-dune populations near both sand dune systems in Colorado and Texas to be used for the purpose (data currently being analyzed). I will also use these data to perform the phylogenetic analysis aimed at confirming independence that was described above and to estimate the time at which H. neglectus started to diverge from H. petiolaris.

Finally, this work could be extended to more cases of sunflowers transitioning to sand dunes such as H. niveus ssp. tephrodes, H. anomalus, and additional populations of dune H. petiolaris found in Canada (L. Rieseberg personal communication). Ultimately, it would be interesting to move outside of Helianthus and compare these systems to other cases of ecological divergence associated with sand dunes (e.g., to dune and headland pairs in Sececio lautus, Roda et al. 2012, Melo et al. 2014), allowing us to discover the extent to which speciation is repeatable with increasing divergence time.

91 References

Ahmad, S. 2001. Environmental effects on seed characteristics of sunflower (Helianthus annuus L.). Journal of Agronomy and Crop Science 187:213–216.

Andrew, R. L., and L. H. Rieseberg. 2013. Divergence is focused on few genomic regions early in speciation: Incipient speciation of sunflower ecotypes. Evolution 67:2468–2482.

Andrew, R. L., K. L. Ostevik, D. P. Ebert, and L. H. Rieseberg. 2012. Adaptation with gene flow across the landscape in a dune sunflower. Molecular Ecology 21:2078– 2091.

Andrew, R. L., N. C. Kane, G. J. Baute, C. J. Grassa, and L. H. Rieseberg. 2013. Recent nonhybrid origin of sunflower ecotypes in a novel habitat. Molecular Ecology 22:799–813.

Baack, E. J., Y. Sapir, M. A. Chapman, J. M. Burke, and L. H. Rieseberg. 2007. Selection on domestication traits and quantitative trait loci in crop-wild sunflower hybrids. Molecular Ecology 17:666–677.

Baker, R. J., and J. W. Bickham. 1986. Speciation by monobrachial centric fusions. Proceedings of the National Academy of Sciences 83:8245–8248.

Barb, J. G., J. E. Bowers, S. Renaut, J. I. Rey, S. J. Knapp, L. H. Rieseberg, and J. M. Burke. 2014. Chromosomal evolution and patterns of introgression in Helianthus. Genetics 197:969–979.

Barker, M. S, N. C. Kane, M. Matvienko, A. Kozik, R. W. Michelmore, S. J. Knapp and L. H. Rieseberg. 2008. Multiple paleopolyploidizations during the evolution of the Compositae reveal parallel patterns of duplicate gene retention after millions of years. Molecular Biology and Evolution 25: 2445-2455.

Barrett, R., and D. Schluter. 2008. Adaptation from standing genetic variation. Trends in Ecology & Evolution 23:38–44.

Bates, D., M. Maechler, B. Bolker and S. Walker. 2015a. lme4: Linear mixed-effects models using Eigen and S4. R package version 1.1-8.

Bates, D., M. Maechler, B. Bolker and S. Walker. 2015b. Fitting linear mixed-effects models using lme4. Journal of Statistical Software: In press.

Baute, G. 2015. Genomics of sunflower improvement: from wild relatives to a global oil seed. University of British Columbia, PhD thesis.

Bert, P. F., I. Jouan, D. Tourvieille de Labrouhe, F. Serre, J. Philippon, P. Nicolas, and F. Vear. 2003. Comparative genetic analysis of quantitative traits in sunflower (Helianthus annuus L.). 2. Characterisation of QTL involved in developmental and agronomic traits. Theoretical and Applied Genetics 107:181–189.

92 Bikard, D., D. Patel, C. Le Metté, V. Giorgi, C. Camilleri, M. J. Bennett, and O. Loudet. 2009. Divergent evolution of duplicate genes leads to genetic incompatibilities within A. thaliana. Science 323: 623-626.

Bolnick, D. I., and T. J. Near. 2005. Tempo of hybrid inviability in centrarchid fishes (Teleostei: Centrarchidae). Evolution 59:1754–1767.

Boughman, J., H. Rundle, and D. Schluter. 2005. Parallel evolution of sexual isolation in sticklebacks. Evolution 59:361–373.

Bradshaw, H. D., K. G. Otto, B. E. Frewen, J. K. McKay, and D. W. Schemske. 1998. Quantitative trait loci affecting differences in floral morphology between two species of monkeyflower (Mimulus). Genetics 149:367–382.

Bray, J. R., and J. T. Curtis. 1957. An ordination of the upland forest communities of southern Wisconsin. Ecological monographs 27:325–349.

Briscoe Runquist, R. D., E. Chu, J. L. Iverson, J. C. Kopp, and D. A. Moeller. 2014. Rapid evolution of reproductive isolation between incipient outcrossing and selfing Clarkia species. Evolution 68:2885–2900.

Broman, K. W., H. Wu, S. Sen, and G. A. Churchill. 2003. R/qtl: QTL mapping in experimental crosses. Bioinformatics 19:889–890.

Burke, J. M., Z. Lai, M. Salmaso, T. Nakazato, S. Tang, A. Heesacker, S. J. Knapp, and L. H. Rieseberg. 2004. Comparative mapping and rapid karyotypic evolution in the genus Helianthus. Genetics 167:449–457.

Castle, W. E. 1921. An improved method of estimating the number of genetic factors concerned in cases of blending inheritance. Science 54:223.

Chandler, J. M., and C. C. Jan. 2007. Comparison of germination techniques for wild Helianthus seeds. Crop Science 25:356–358.

Chandler, J. M., C. C. Jan, and B. H. Beard. 1986. Chromosomal differentiation among the annual Helianthus species. Systematic Botany: 354-371.

Chapman, M. A., C. H. Pashley, J. Wenzler, J. Hvala, S. Tang, S. J. Knapp, and J. M. Burke. 2008. A genomic scan for selection reveals candidates for genes involved in the evolution of cultivated sunflower (Helianthus annuus). Plant Cell 20:2931–2945.

Chen, H., and M. A. Maun. 1999. Effects of sand burial depth on seed germination and seedling emergence of Cirsium pitcheri. Plant Ecology 140:53–60.

Chetelat, R. T., and V. Meglic. 2000. Molecular mapping of chromosome segments introgressed from Solanum lycopersicoides into cultivated tomato (Lycopersicon esculentum). Theoretical and Applied Genetetics 100:232–241.

93 Chmátal, L., S. I. Gabriel, G. P. Mitsainas, J. Martínez-Vargas, J. Ventura, J. B. Searle, R. M. Schultz, and M. A. Lampson. 2014. Centromere strength provides the cell biological basis for meiotic drive and karyotype evolution in mice. Current Biology 24:2295–2300.

Choi, H. K., D. Kim, T. Uhm, E. Limpens, H. Lim, J. H. Mun, P. Kalo, R. V. Penmetsa, A. Seres, O. Kulikova, and B. A. Roe. A sequence-based genetic map of Medicago truncatula and a comparison of marker colinearity with M. sativa. Genetics 166:1463-1502.

Conte, G. L., M. E. Arnegard, C. L. Peichel, and D. Schluter. 2012. The probability of genetic parallelism and convergence in natural populations. Proceedings of the Royal Society B: Biological Sciences 279:5039–5047.

Conte, G. L., M. E. Arnegard, J. Best, Y. F. Chan, F. C. Jones, D. M. Kingsley, D. Schluter, and C. L. Peichel. 2015. Extent of QTL reuse during repeated phenotypic divergence of sympatric threespine stickleback. Genetics 201:1189–1200.

Cordazzo, C. V. 2002. Effect of seed mass on germination and growth in three dominant species in southern Brazilian coastal dunes. Brazilian Journal of Biology 62:427–435.

Coyne, J., and H. Orr. 2004. Speciation. Sinauer associates Sunderland, Sunderland, MA.

Dell'Olivo, A., M. E. Hoballah, T. Gübitz, and C. Kuhlemeier. 2011. Isolation barriers between Petunia axillaris and Petunia integrifolia (Solanaceae). Evolution 65:1979– 1991.

Dilbirligi, M., M. Erayman, B. T. Campbell, H. S. Randhawa, P. S. Baenziger, I. Dweikat, and K. S. Gill. 2006. High-density mapping and comparative analysis of agronomically important traits on wheat chromosome 3A. Genomics 88:74–87.

Dirlewanger, E., E. Graziano, T. Joobeur, F. Garriga-Calderé, P. Cosson, W. Howad, and P. Arús. 2004. Comparative mapping and marker-assisted selection in Rosaceae fruit crops. Proceedings of the National Academy of Sciences 101:9891– 9896.

Donovan, L. A., D. M. Rosenthal, M. Sanchez-Velenosi, L. H. Rieseberg, and F. Ludwig. 2010. Are hybrid species more fit than ancestral parent species in the current hybrid species habitats? J. Evol. Biol. 23:805–816.

Doyle, J., and J. Doyle. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochem Bull 19:11–15.

Efron, B. 1987. Better bootstrap confidence intervals. Journal of the American Statistical Association 82:171–185.

Endler, J. A. 1986. Natural selection in the wild. Princeton University Press.

94 Feder, J. L., and P. Nosil. 2009. Chromosomal inversions and species differences: When are genes affecting adaptive divergence and reproductive isolation expected to reside within inversions? Evolution 63:3061–3075.

Felsenstein, J. 1981. Skepticism towards Santa Rosalia, or why are there so few kinds of animals? Evolution 124–138.

Fishman, L., A. Stathos, P. M. Beardsley, C. F. Williams and J. P. Hill. 2013. Chromosomal rearrangements and the genetics of reproductive barriers in Mimulus (monkey flowers). Evolution 67:2547–2560.

Foster, S. A., G. E. McKinnon, D. A. Steane, B. M. Potts, and R. E. Vaillancourt. 2007. Parallel evolution of dwarf ecotypes in the forest tree Eucalyptus globulus. New Phytol. 175:370–380.

Foulongne, M., T. Pascal, P. Ar s, and J. Kervella. 2003. The potential of Prunus davidiana for introgression into peach [Prunus persica (L.) Batsch] assessed by comparative mapping. Theor Appl Genet 107:227–238.

Fournier, D. A., H. J. Skaug, J. Ancheta, J. Ianelli, A. Magnusson, M. N. Maunder, A. Nielsen, and J. Sibert. 2012. AD Model Builder: Using automatic differentiation for statistical inference of highly parameterized complex nonlinear models. Optimization Methods and Software 27:233–249.

Gavrilets, S. 2004. Fitness landscapes and the origin of species. Princeton University Press, Princton, NJ.

Geyer, C. J. 2015. aster: Aster Models. R package version 0.8-31.

Griswold, C. K. 2006. Gene flow's effect on the genetic architecture of a local adaptation and its consequences for QTL analyses. Heredity 96:445–453.

Hatfield, T., and D. Schluter. 1999. Ecological speciation in sticklebacks: Environment- dependent hybrid fitness. Evolution 53:866–873.

Heiser, C. B., 1947. Hybridization between the sunflower species Helianthus annuus and H. petiolaris. Evolution 1:249–262.

Heiser, C. B., 1961. Morphological and cytological variation in Helianthus petiolaris with notes on related species. Evolution 15:247–258.

Heiser, C. B., 1958. Three new annual sunflowers (Helianthus) from the southwestern United States. Rhodora 272–283.

Heiser, C. B., D. M. Smith, S. B. Clevenger, and W. C. Martin. 1969. The North American sunflowers (Helianthus). Seeman Printery, Durham, NC.

Hurlbert, S. H. 1970. Flower number, flowering time, and reproductive isolation among ten species of Solidago (Compositae). Bulletin of the Torrey Botanical Club 189– 195.

95 Husband, B. C., and D. W. Schemske. 2000. Ecological mechanisms of reproductive isolation between diploid and tetraploid Chamerion angustifolium. Journal of Ecology 88:689–701.

Husband, B. C., D. W. Schemske, T. L. Burton and C. Goodwillie. 2002. Pollen competition as a unilateral reproductive barrier between sympatric diploid and tetraploid Chamerion angustifolium. Proceedings of the Royal Society of London B: Biological Sciences 269:2565-2571.

Ishizaki, S., T. Abe, and M. Ohara. 2013. Mechanisms of reproductive isolation of interspecific hybridization between Trillium camschatcense and T. tschonoskii (Melanthiaceae). Plant Species Biology 28:204–214.

Johnson, R. B. 1968. The great sand dunes of southern Colorado. The Mountain Geologist.

Jones, R., D. C. Culver, and T. C. Kane. 1992. Are parallel morphologies of cave organisms the result of similar selection pressures? Evolution 46:353–365.

Kantar, M. B., C. C. Sosa, C. K. Khoury, N. P. Castañeda-Álvarez, H. A. Achicanoy, V. Bernau, N. C. Kane, L. Marek, G. Seiler, and L. H. Rieseberg. 2015. Ecogeography and utility to plant breeding of the crop wild relatives of sunflower (Helianthus annuus L.). Frontiers in Plant Science 6.

Kao, J. Y., S. Lymer, S. H. Hwang, A. Sung, and S. V. Nuzhdin. 2015. Postmating reproductive barriers contribute to the incipient sexual isolation of the United States and Caribbean Drosophila melanogaster. Ecology and Evolution 5:3171–3182.

Kay, K. 2006. Reproductive isolation between two closely related hummingbird pollinated neotropical gingers. Evolution 60:538–552.

Kim, S. C., and L. H. Rieseberg. 1999. Genetic architecture of species differences in annual sunflowers: Implications for adaptive trait introgression. Genetics 153:965– 977.

Kirkpatrick, M., and N. Barton. 2006. Chromosome inversions, local adaptation and speciation. Genetics 173:419–434.

Kozak, G. M., A. B. Rudolph, B. L. Colon, and R. C. Fuller. 2012. Postzygotic isolation evolves before prezygotic isolation between fresh and saltwater populations of the rainwater killifish, Lucania parva. International Journal of Evolutionary Biology 2012:1–11.

Lai, Z., T. Nakazato, M. Salmaso, J. M. Burke, S. Tang, S. J. Knapp, and L. H. Rieseberg. 2005. Extensive chromosomal repatterning and the evolution of sterility barriers in hybrid sunflower species. Genetics, 171(1), 291-303.

Lander, E. S., and D. Botstein. 1989. Mapping mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121:185–199.

96 Li, H., and R. Durbin. 2010. Fast and accurate long-read alignment with Burrows- Wheeler transform. Bioinformatics 26:589–595.

Li, N., and Y. Li. 2015. Maternal control of seed size in plants. Journal of Experimental Botany 66:1087–1097.

Lowry, D. B., and J. H. Willis. 2010. A widespread chromosomal inversion polymorphism contributes to a major life-history transition, local adaptation, and reproductive isolation. PLoS Biol 8:2227–2241.

Lowry, D. B., J. L. Modliszewski, K. M. Wright, C. A. Wu, and J. H. Willis. 2008a. The strength and genetic basis of reproductive isolating barriers in flowering plants. Philosophical Transactions of the Royal Society B: Biological Sciences 363:3009– 3021.

Lowry, D. B., R. C. Rockwood, and J. H. Willis. 2008b. Ecological reproductive isolation of coast and inland races of Mimulus guttatus. Evolution 62:2196–2214.

Lu , F., A. E. Lipka, R. J. Elshire, J. C. Glaubitz, J. H. Cherney, M. D. Casler , E. S. Buckler and D. E. 2013. Switchgrass genomic diversity, ploidy and evolution: novel insights from a network-based SNP discovery protocol. PLoS Geentics 9:e1003215.

Lynch, M., and B. Walsh. 1998. Genetics and analysis of quantitative traits. Sinauer associates Sunderland.

Madole, R.F., J. H. Romig, J. N. Aleinikoff, D. P. VanSistine, and E. Y. Yacob. 2008. On the origin and age of the Great Sand Dunes, Colorado, Geomorphology 99:99-119.

Mandel, J. R., S. Nambeesan, J. E. Bowers, L. F. Marek, D. Ebert, L. H. Rieseberg, S. J. Knapp, and J. M. Burke. 2013. Association Mapping and the Genomic Consequences of Selection in Sunflower. PLoS Genetics 9:e1003378.

Manichaikul, A., A. A. Palmer, S. Sen, and K. W. Broman. 2007. Significance thresholds for quantitative trait locus mapping under selective genotyping. Genetics 177:1963–1966.

Martin, N. H., and J. H. Willis. 2007. Ecological divergence associated with causes nearly complete reproductive isolation between sympatric Mimulus species. Evolution 61:68–82.

Matsubayashi, K. W., and H. Katakura. 2009. Contribution of multiple isolating barriers to reproductive isolation between a pair of phytophagous ladybird beetles. Evolution 63:2563–2580.

Maun, M. A. 1994. Adaptations enhancing survival and establishment of seedlings on coastal dune systems. Vegetatio 111:59–70.

McMillan, W. O., C. Jiggins, and J. Mallet. 1997. What initiates speciation in passion- vine butterflies? Proceedings of the National Academy of Sciences 94:8628–8633.

97 Melo, M. C., A. Grealy, B. Brittain, G. M. Walter, and D. Ortiz-Barrientos. 2014. Strong extrinsic reproductive isolation between parapatric populations of an Australian groundsel. New Phytologist 203:323–334.

Mizuta, Y., Y. Harushima, and N. Kurata. 2010. Rice pollen hybrid incompatibility caused by reciprocal gene loss of duplicated genes. Proceedings of the National Academy of Sciences 107:20417–20422.

Mokrani, L., L. Gentzbittel, F. Azanza, L. Fitamant, G. Al-Chaarani, and A. Sarrafi. 2002. Mapping and analysis of quantitative trait loci for grain oil content and agronomic traits using AFLP and SSR in sunflower (Helianthus annuus L.). Theoretical and Applied Genetics 106:149–156.

Morenno-Casasola, P. 1986. Sand movement as a factor in the distribution of plant communities in a coastal dune system. Vegetatio 65:67–76.

Mountford, M. D. 1962. An index of similarity and its application to classification problems. Pp. 43–50 in Progress in Soil Zoology. Springer Netherlands.

Navarro, A., and N. H. Barton. 2003. Accumulating postzygotic isolation genes in parapatry: A new twist on chromosomal speciation. Evolution 57:447–459.

Nepi, M., and G. G. Franchi. 2000. Cytochemistry of mature angiosperm pollen. in A. Dafni, M. Hesse, and E. Pacini, eds. Pollen and Pollination. Springer-Verlag Wien. 45–62

Noor, M. A., K. L. Grams, L. A. Bertucci, and J. Reiland. 2001. Chromosomal inversions and the reproductive isolation of species. Proceedings of the National Academy of Sciences 98:12084–12088.

Nosil, P. 2007. Divergent host plant adaptation and reproductive isolation between ecotypes of Timema cristinae walking sticks. The American Naturalist 169:151–162.

Nosil, P. 2012. Ecological Speciation. Oxford University Press.

Nosil, P., B. Crespi, and C. Sandoval. 2002. Host-plant adaptation drives the parallel evolution of reproductive isolation. Nature 417:440–443.

Oksanen, J., F. G. Blanchet, R. Kindt, P. Legendre, P. R. Minchin, R. B. O'Hara, G. L. Simpson, P. Solymos, M. H. H. Stevens and H. Wagner. 2015. vegan: Community Ecology Package. R package version 2.3-0.

Østbye, K., P. A. Amundsen, L. Bernatchez, A. Klemetsen, R. Knudsen, R. Kristoffersen, T. F. Naesje, and K. Hindar. 2006. Parallel evolution of ecomorphological traits in the european whitefish Coregonus lavaretus (l.) during postglacial times. Molecular Ecology 15:3983–4001.

Ostevik, K. L., B. T. Moyers, G. L. Owens and L. H. Rieseberg. 2012. Parallel ecological speciation in plants? International Journal of Ecology 2012.

98 Peters, G. 2015. userfriendlyscience: Quantitative analysis made accessible. R package version 0.3-0.

Petit, R. J., J. Duminil, S. Fineschi, A. Hampe, D. Salvini, and G. G. Vendramin. 2005. Comparative organization of chloroplast, mitochondrial and nuclear diversity in plant populations. Molecular Ecology 14:689–701.

Pinheiro, J., D. Bates, S. DebRoy, and D. Sarkar. 2015. R Development Core Team. 2013. nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1–109.

Pigeon, D., A. Chouinard, and L. Bernatchez. 1997. Multiple modes of speciation involved in the parallel evolution of sympatric morphotypes of lake whitefish (Coregonus clupeaformis, Salmonidae). Evolution 51:196–205.

Poland, J. A., P. J. Brown, M. E. Sorrells, and J. L. Jannink. 2012. Development of high- density genetic maps for barley and wheat using a novel two-enzyme genotyping- by-sequencing approach. PLoS ONE 7:e32253.

Quesada, H., D. Posada, A. Caballero, P. Morán, and E. Rolán-Alvarez. 2007. Phylogenetic evidence for multiple sympatric ecological diversification in a marine snail. Evolution 61:1600–1612.

R Core Team. 2015. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

Raduski, A. R., L. Rieseberg, and J. Strasburg. 2010. Effective population size, gene flow, and species status in a narrow endemic sunflower, Helianthus neglectus, compared to its widespread sister species, H. petiolaris. International Journal of Molecular Sciences 11:492–506.

Rahme, J., A. Widmer, and S. Karrenberg. 2009. Pollen competition as an asymmetric reproductive barrier between two closely related Silene species. Journal of Evolutionary Biology 22:1937–1943.

Rajakaruna, N., M. Siddiqi, J. Whitton, B. Bohm, and A. Glass. 2003. Differential responses to Na+/K+ and Ca2+/Mg2+ in two edaphic races of the Lasthenia californica (Asteraceae) complex: A case for parallel evolution of physiological traits. New Phytologist 157:93–103.

Ramsey, J., H. Bradshaw Jr, and D. Schemske. 2003. Components of reproductive isolation between the monkeyflowers Mimulus lewisii and M. cardinalis (Phrymaceae). Evolution 57:1520–1534.

Rausher, M. D., and L. F. Delph. 2015. Commentary: When does understanding phenotypic evolution require identification of the underlying genes? Evolution 69:1655–1664.

99 Renaut, S., C. J. Grassa, S. Yeaman, B. T. Moyers, Z. Lai, N. C. Kane, J. E. Bowers, J. M. Burke, and L. H. Rieseberg. 2013. Genomic islands of divergence are not affected by geography of speciation in sunflowers. Nature Communications 4:1–8. Nature Publishing Group.

Rice, W. R., and E. E. Hostert. 1993. Laboratory experiments on speciation: What have we learned in 40 years? Evolution 47:1637–1653.

Rieseberg, L. H. 2001. Chromosomal rearrangements and speciation. Trends in Ecology & Evolution 16:351–358.

Rieseberg, L. H., R. Carter, and S. Zona. 1990. Molecular tests of the hypothesized hybrid origin of two diploid Helianthus species (Asteraceae). Evolution 1498–1511.

Rieseberg, L. H., A. Desrochers, and S. Youn. 1995a. Interspecific pollen competition as a reproductive barrier between sympatric species of Helianthus (Asteraceae). American Journal of Botany 82:515–519.

Rieseberg, L. H., C. R. Linder, and G. J. Seiler. 1995b. Chromosomal and genic barriers to introgression in Helianthus. Genetics 141:1163–1171.

Rieseberg, L. H., J. Whitton, and K. Gardner. 1999. Hybrid zones and the genetic architecture of a barrier to gene flow between two sunflower species. Genetics 152:713–727.

Roccaforte, K., S. E. Russo, and D. Pilson. 2015. Hybridization and reproductive isolation between diploid Erythronium mesochoreum and its tetraploid congener E. albidum (Liliaceae). Evolution 69:1375–1389.

Roda, F., L. Ambrose, G. M. Walter, H. L. Liu, A. Schaul, A. Lowe, P. B. Pelser, P. Prentis, L. H. Rieseberg, and D. Ortiz-Barrientos. 2013. Genomic evidence for the parallel evolution of coastal forms in the Senecio lautus complex. Molecular Ecology 22:2941–2952.

Rogers, C. E., T. E. Thompson, and G. J. Seiler. 1982. Sunflowers species of the United States. National Sunflower Association.

Rosenblum, E. B. 2006. Convergent evolution and divergent selection: Lizards at the white sands ecotone. The American Naturalist 167:1–15.

Rosenblum, E. B., and L. J. Harmon. 2011. “Same Same but Different”: Replicated ecological speciation at white sands. Evolution 65: 946-960.

Rundle, H. D., L. Nagel, J. W. Boughman, and D. Schluter. 2000. Natural selection and parallel speciation in sympatric sticklebacks. Science 287:306–308.

Sambatti, J. B. M., and K. J. Rice. 2006. Local adaptation, patterns of selection, and gene flow in the Californian serpentine sunflower (Helianthus exilis). Evolution 60:696– 710.

100 Sambatti, J. B. M., J. L. Strasburg, D. Ortiz-Barrientos, E. J. Baack, and L. H. Rieseberg. 2012. Reconciling extremely strong barriers with high levels of gene exchange in annual sunflowers. Evolution 66:1459–1473.

Sánchez-Guillén, R. A., M. Wellenreuther, and A. Cordero Rivera. 2012. Strong asymmetry in the relative strengths of prezygotic and postzygotic barriers between two damselfly sister species. Evolution 66:690–707.

Schemske, D. W. 2000. Understanding the origin of species. Evolution 54:1069–1073.

Schluter, D. 2009. Evidence for ecological speciation and its alternative. Science 323:737–741.

Schluter, D., and L. M. Nagel. 1995. Parallel speciation by natural selection. American naturalist 292–301.

Schluter, D., E. Clifford, M. Nemethy and J. McKinnon. 2004. Parallel evolution and inheritance of quantitative traits. The American Naturalist 163:809–822.

Schneider, C. A., W. S. Rasband, and K. W. Eliceiri. 2012. NIH Image to ImageJ: 25 years of image analysis. Nature Methods 9:671–675.

Schwarzbach, A. E., and L. Donovan. 2001. Transgressive character expression in a hybrid sunflower species. American Journal of Botany 88:270–277.

Scopece, G., A. Croce, C. Lexer, and S. Cozzolino. 2013. Components of reproductive isolation between Orchis mascula and Orchis pauciflora. Evolution 67:2083–2093.

Seiler, G. 2014. Wild Helianthus anomalus and H. deserticola from the desert southwest usa: A potential source of stress genes for cultivated sunflower.

Seiler, G. J., and L. H. Rieseberg. 1997. Systematics, origin, and germplasm resources of the wild and domesticated sunflower. Agronomy 35:21–66.

Sen, S., and G. A. Churchill. 2001. A statistical framework for quantitative trait mapping. Genetics 159:371–387.

Sen, S., J. M. Satagopan, and G. A. Churchill. 2005. Quantitative trait locus study design from an information perspective. Genetics 170:447–464.

Sobel, J. M. 2014. Ecogeographic isolation and speciation in the genus Mimulus. The American Naturalist 184:565–579.

Sobel, J. M., and G. F. Chen. 2014. Unification of methods for estimating the strength of reproductive isolation. Evolution 68:1511–1522.

Sobel, J. M., and M. A. Streisfeld. 2015. Strong premating reproductive isolation drives incipient speciation in Mimulus aurantiacus. Evolution 69:447–461.

101 Stathos, A and L. Fishman. 2014. Chromosomal rearrangements directly cause underdominant F1 pollen sterility in Mimulus lewisii-Mimulus cardinalis hybrids. Evolution 68:3109–3119.

Stephens, J. D., W. L. Rogers, C. M. Mason, L. A. Donovan, and R. L. Malmberg. 2015. Species tree estimation of diploid Helianthus (Asteraceae) using target enrichment. American Journal of Botany 102:910–920.

Strasburg, J., and L. Rieseberg. 2008. Molecular demographic history of the annual sunflowers Helianthus annuus and H. petiolaris—Large effective population sizes and rates of long-term gene flow. Evolution 62:1936–1950.

Strecker, U., B. Hausdorf, and H. Wilkens. 2012. Parallel speciation in Astyanax cave fish (Teleostei) in Northern Mexico. Molecular Phylogenetics and Evolution 62:62– 70.

Streisfeld, M. A, and M. D. Rausher. 2009. Genetic changes contributing to the parallel evolution of red floral pigmentation among Ipomoea species. New Phytologist 183:751–763.

Tang, S., A. Leon, W. C. Bridges, and S. J. Knapp. 2006. Quantitative trait loci for genetically correlated seed traits are tightly linked to branching and pericarp pigment loci in sunflower. Crop Science 46:721.

Timme, R. E., B. B. Simpson and C. R. Linder. 2007. High-resolution phylogeny for Helianthus (Asteraceae) using the 18S-26S ribosomal DNA external transcribed spacer. American Journal of Botany, 94:1837-1852.

Therneau, T. 2015. A Package for Survival Analysis in S. R package version 2.38.

Therneau, T. M. and P. M. Grambsch. 2000. Modeling Survival Data: Extending the Cox Model. Springer, New York. ISBN 0-387-98784-3.

Tobler, M., M. Palacios, L. J. Chapman, I. Mitrofanov, D. Bierbach, M. Plath, L. Arias- Rodriguez, F. J. García de León, and M. Mateos. 2011. Evolution in extreme environments: Replicated phenotypic differentiation in livebearing fish inhabiting sulfidic springs. Evolution 65:2213–2228.

Trickett, A. J., and R. K. Butlin. 1994. Recombination suppressors and the evolution of new species. Heredity 73:339–345.

Ungerer, M. C., S. J. Baird, J. Pan, and L. H. Rieseberg. 1998. Rapid hybrid speciation in wild sunflowers. Proceedings of the National Academy of Sciences 95:11757– 11762.

Via, S., and D. J. Hawthorne. 2002. The genetic architecture of ecological specialization: Correlated gene effects on host use and habitat choice in pea aphids. The American Naturalist 159:S76–88.

102 Wickham, H. 2011. The split-apply-combine strategy for data analysis. Journal of Statistical Software 40:1-29.

Wills, D. M., and J. M. Burke. 2007. Quantitative trait locus analysis of the early domestication of sunflower. Genetics 176:2589–2599.

Willson, M. F., and N. Burley. 1983. Mate choice in plants: tactics, mechanisms and consequences. Vol 19. Princeton University Press 1983.

Wright, S. 1968. Evolution and genetics of populations. University of Chicago Press.

Yatabe, Y., N. C. Kane, C. Scotti-Saintagne, and L. H. Rieseberg. 2007. Rampant gene exchange across a strong reproductive barrier between the annual sunflowers, Helianthus annuus and H. petiolaris. Genetics 175:1883–1893.

Yeaman, S., and M. Whitlock. 2011. The genetic architecture of adaptation under migration-selection balance. Evolution 65:1897–1911.

103 Appendices

Appendix A - Chapter 2 supplementary materials

A.1 Seed size common garden methods and results

We grew 1-6 plants from each of 11 dune and 7 non-dune populations under standard greenhouse conditions (16 hours of light at 600W/m2, temperature 18-23°C). We covered immature flower heads with mesh to prevent natural pollination and pollinated flowering heads with mixtures of either dune or non-dune pollen (see the methods in chapter 2 for details about the pollen mixtures). In most cases, we pollinated one flower head with a dune pollen mixture and one flower head with a non-dune pollen mixture per plant. We collected the seeds produced by each pollination and weighed groups of 5 seeds. In our statistical analysis, we fit a linear mixed effect model describing seed weight that included maternal ecotype as a fixed effect and maternal plant, maternal population and the ecotype of the pollen mixture as random effects. We found that seeds produced by dune plants were significantly

! heavier than those produced by non-dune plants (Table 2.1, !! =32.16, P<0.001).

A.2 2010 reciprocal transplant methods and results

We used seeds from 20 H. petiolaris populations (10 dune, 6 non-dune, and 4 intermediate where intermediate populations are those collected at the margins of dunes. However, intermediate populations are not relevant to this analysis and are not considered further) in a reciprocal transplant similar to the one described in the main text. Each population seed lot was made up of equal numbers of seeds from 10

104 maternal families collected at Great Sand Dune National Park and Preserve in 2008. We refrigerated half of each seed lot on wet filter paper at 4°C for 30 days and scarified the other half by cutting away a small portion of the distil end of each seed. We established

20 plots in the dune and non-dune reciprocal transplant sites described in the main text and subdivided each plot into 6 subplots. We assigned a refrigerated and a scarified seed lot from each of two populations to each plot (one dune and one non-dune or intermediate). We planted 25 seeds for these 4 seed types into 4 of the subplots in each plot and left the other 2 subplots to measure the rate of natural sunflower recruitment

(volunteers). Finally, we monitored the subplots for seedling emergence and survival and counted the number of mature flower heads produced by each surviving plant.

In this year, many fewer seedlings emerged in the dune environment (8) than in the non-dune environment (139), in contrast to the main experiment carried out in 2012.

Also, no seedling emerged in the control plots so we do not consider them further.

Using zero-inflated generalized mixed effect models, we explored the effects of treatment (refrigeration versus scarification), environment, ecotype, and their interactions on the number of seedling that emerged in each subplot. We used a

Poisson distribution to model the errors and included plot and population as random effects.

First, we found that seeds treated with refrigeration emerged significantly better

! than those that were scarified (!! =62.71, P<0.001) but that treatment did not interact

! ! with either environment (!! =2.49, P=0.11) or ecotype (!! =0.11, P=0.74). Therefore, we pooled across treatment in subsequent analyses. We found that more seedlings

! emerged in the non-dune environment (!! =20.47, P<0.001), that more dune seedlings

! emerged than non-dune seedlings (!! =16.86, P<0.001), and that there was no

105 ! interaction between environment and ecotype (!! =2.32, P=0.13). These results are very similar to those found in 2012 except that the environment with more seedling emergence switched.

Very few plants survived to maturity in the dune environment so we restricted our analysis of flower heads to the plants on the sand sheet. In this case, we wanted to test the effect of ecotype on the number of flower heads produced using linear mixed effect models that included plot and population random effects. Again there was no

! evidence of a treatment-by-ecotype interaction (!! =0.02, P=0.899) so we pooled the data across treatments. We found that, on the sand sheet, non-dune plants produce an

! average of 7 more flower heads than dune plants (!! =6.35, P=0.012, Fig. A.1), with the significance of this result confirmed by a randomization test (P=0.0027).

In general, the dune seeds emerged better in both environments while non-dune plants had higher fecundity in the non-dune environment. This is consistent with the results described in the main text.

106 25 20 er heads w 15 o 10 5 Number of fl 0 dune non−dune plants plants

Figure A.1 - The number of flower heads produced by dune (red) and non-dune (blue) plants transplanted into the non-dune environment. Note that neither of the 2 dune plants that survived in the dune environment produced flower heads.

107 A.3 Supplementary tables

Table A.1 List of the populations and sites used in different experiments. DRT = dune reciprocal transplant site, NRT = non-dune reciprocal transplant site, RT = pure populations used in reciprocal transplant, RTH = populations used in the generation of the hybrid seeds (F1D, BCD, BCN, F1N) used in the reciprocal transplant, FT = flowering time measured by timelapse cameras, PC = pollen competition experiment, SS = seed set experiment, PT = pollination timing experiment, G = seed germination experiment Included in experiments Name Population Environment Latitude Longitude RT RTH FT PC SS PT G D1 1300 Dune 37.745N 105.546W Y Y Y* Y Y Y D2 1701 Dune 37.803N 105.524W Y Y Y Y Y Y D3 1547 Dune 37.761N 105.570W Y Y Y Y Y Y D4 1270 Dune 37.764N 105.524W Y Y Y Y D5 1240 Dune 37.786N 105.531W Y Y Y Y D6 1181 Dune 37.773N 105.553W Y Y Y Y D7 1003 Dune 37.774N 105.576W Y Y Y D8 1033 Dune 37.787N 105.569W Y Y Y Y D9 1731 Dune 37.807N 105.554W Y Y Y D10 1117 Dune 37.818N 105.597W Y Y Y Y D11 2061 Dune 37.758N 105.515W Y Y Y Y D12 Updune Dune 37.762N 105.516W Y D13 170 Dune 37.768N 105.516W Y D14 417 Dune 37.761N 105.514W Y

108 Included in experiments Name Population Environment Latitude Longitude RT RTH FT PC SS PT G D15 2121 Dune 37.764N 105.522W Y D16 2141 Dune 37.763N 105.519W Y D17 2161 Dune 37.763N 105.515W Y DRT DRT Dune 37.767N 105.515W Y N1 2001 Sand sheet 37.757N 105.507W Y Y Y Y Y Y N2 1791 Sand sheet 37.813N 105.515W Y Y Y Y Y Y N3 1363 Sand sheet 37.716N 105.533W Y Y Y Y Y Y Y N4 970 Sand sheet 37.663N 105.625W Y N5 2250 Sand sheet 37.672N 105.592W Y Y Y Y N6 1147 Sand sheet 37.836N 105.603W Y Y Y N7 1063 Sand sheet 37.774N 105.598W Y Y Y Y N8 1500 Sand sheet 37.765N 105.615W Y Y Y N9 221 Sand sheet 37.765N 105.499W Y N10 356 Sand sheet 37.762N 105.501W Y NRT NRT Sand sheet 37.679N 105.591W Y * The camera at this site broke halfway through the experiment. The data were excluded from analysis but were consistent with the other dune populations

109

Table A.2 Reproductive barriers included in each total RI calculation.

Seed flow Pollen flow

Barriers: RIN->D RID->N RIN->D RID->N Co-occurrence barriers: Flowering time NA NA 0.093 0.093 Pollinator assemblages NA NA 0.55 0.36 Prezygotic barriers: Selection against immigrants 0.89 0.76 NA NA Post-pollination assortative mating 0.25* 0.25* 0.38 0.12 Postzygotic barriers: F1 germination -0.0028 0.0065 -0.0028 0.0065

Selection against F1s 0.67 0.65 0.44 (F1D) 1 (F1N) F1 pollen sterility 0.012 -0.032 0.012 -0.032 Selection against BCs 0.85 0.64 0.77 (BCD) 0.55 (BCN)

* This is the average of RIN->D and RID->N for post-pollination assortative mating

110 Table A.3 Parameter estimates from an ASTER analysis that includes the fitness components emergence, survival and number of seeds produced from the dune and non-dune plants used ! in the reciprocal transplant. There is a significant environment-by-ecotype effect (!!=1951.2, P<0.001). Fixed effects: Estimate SE z-value P (Intercept) -1.85 0.299 -6.18 < 0.001 Seed variable 6.22 0.095 65.64 < 0.001 Survival variable -136.5 1.648 -82.83 < 0.001 Non-dune environment -1.85 0.227 -8.15 < 0.001 Non-dune ecotype -2.21 0.347 -6.36 < 0.001 Non-dune environment * Non-dune ecotype 2.94 0.072 41.06 < 0.001

Random effects: Variance SE Populations 0.421 0.130 Plots 0.829 0.090 Subplots 0.810 0.082

111 Table A.4 Parameter estimates from an ASTER analysis that includes the fitness components emergence, survival and number of seeds produced from all plant types used in the reciprocal ! transplant. There is a significant environment-by-type effect (!!=137.6, P<0.001). Fixed effects: Estimate SE z-value P (Intercept) -2.470 0.083 -29.79 < 0.001 Seed variable 7.270 0.077 95.02 < 0.001 Survival variable -160.930 1.374 -117.14 < 0.001 Non-dune environment -0.489 0.097 -5.03 < 0.001 Type D3 0.275 0.044 6.29 < 0.001 Type D2 0.020 0.036 0.57 0.568 Type F1D -0.327 0.030 -10.80 < 0.001 Type BCD -0.125 0.040 -3.15 0.003 Type BCN -0.207 0.121 -1.72 0.086 Type F1N -0.443 0.127 -3.49 < 0.001 Type N3 -0.166 0.035 -4.80 < 0.001 Type N2 -0.537 0.024 -22.68 < 0.001 Type N1 -0.430 0.056 -7.63 < 0.001 Non-dune environment * Type D3 -0.191 0.068 -2.84 0.005 Non-dune environment * Type D2 0.098 0.050 1.99 0.047 Non-dune environment * Type F1D 0.135 0.041 3.34 < 0.001 Non-dune environment * Type BCD 0.391 0.092 4.26 < 0.001 Non-dune environment * Type BCN 0.365 0.123 2.97 0.003 Non-dune environment * Type F1N -0.256 0.438 -0.58 0.560 Non-dune environment * Type N3 0.581 0.071 8.16 < 0.001 Non-dune environment * Type N2 1.290 0.085 15.18 < 0.001 Non-dune environment * Type N1 -0.660 0.514 -1.28 0.199

Random effects: Variance SE Plots 0.241 0.033 Subplots 0.242 0.030

112 Table A.5 Parameter estimates from an ASTER analysis that includes the fitness components emergence, survival, and number of seeds and flowers produced from the dune and non-dune plants used in the reciprocal transplant. There is a significant environment-by-ecotype effect

! (!!=1911.1, P<0.001).

Fixed effects: Estimate SE z-value P (Intercept) -1.850 0.287 -6.45 < 0.001 Flower variable 2.796 0.102 27.43 < 0.001 Seed variable 6.271 0.095 66.32 < 0.001 Survival variable -142.419 1.688 -84.37 < 0.001 Non-dune environment -1.792 0.221 -8.10 < 0.001 Non-dune ecotype -2.128 0.330 -6.45 < 0.001 Non-dune environment * Non- dune ecotype 2.828 0.070 40.32 < 0.001

Random effects: Variance SE Populations 0.400 0.124 Plots 0.802 0.088 Subplots 0.781 0.080

Table A.6 Parameter estimates from an ASTER analysis that includes the fitness components emergence, survival and number of seeds and flowers produced from all plant types used in

! the reciprocal transplant. There is a significant environment-by-type effect (!!=1187.6,

P<0.001).

Fixed effects: Estimate SE z-value P (Intercept) -2.485 0.079 -31.58 < 0.001 Flower variable 3.897 0.080 48.52 < 0.001 Seed variable 7.327 0.073 100.65 < 0.001 Survival variable -167.219 1.392 -120.16 < 0.001 Non-dune environment -0.428 0.090 -4.74 < 0.001

113 Fixed effects: Estimate SE z-value P Type D3 0.249 0.035 7.11 < 0.001 Type D2 0.031 0.036 0.87 0.387 Type F1D -0.293 0.024 -12.03 < 0.001 Type BCD -0.107 0.040 -2.66 0.008 Type BCN -0.180 0.048 -3.77 < 0.001 Type F1N -0.459 0.116 -3.96 < 0.001 Type N3 -0.147 0.024 -6.26 < 0.001 Type N2 -0.497 0.023 -21.30 < 0.001 Type N1 -0.384 0.032 -11.92 < 0.001 Non-dune environment * Type D3 -0.175 0.060 -2.91 0.004 Non-dune environment * Type D2 0.088 0.046 1.89 0.059 Non-dune environment * Type F1D 0.117 0.041 2.83 0.005 Non-dune environment * Type BCD 0.326 0.088 3.70 < 0.001 Non-dune environment * Type BCN 0.328 0.055 6.01 < 0.001 Non-dune environment * Type F1N -0.272 0.437 -0.62 0.534 Non-dune environment * Type N3 0.514 0.054 9.46 < 0.001 Non-dune environment * Type N2 1.176 0.087 13.51 < 0.001 Non-dune environment * Type N1 -0.743 0.511 -1.46 0.146

Random effects: Variance SE Plots 0.216 0.028 Subplots 0.218 0.026

114 Table A.7 Insect collection details and locations.

Environment Date Type Time Latitude Longitude Name Dune 8 Aug 2012 Net 11:00-13:00 37.767N 105.515W DRT Non-dune 9 Aug 2012 Net 11:00-13:00 37.679N 105.591W NRT Dune 15 Aug 2012 Net 13:00-15:00 37.767N 105.515W DRT Non-dune 16 Aug 2012 Net 13:00-15:00 37.679N 105.591W NRT Dune 28 Aug 2012 Net 9:00-11:00 37.767N 105.515W DRT Non-dune 29 Aug 2012 Net 9:00-11:00 37.679N 105.591W NRT Dune 6 Sep 2012 Net 7:00-9:00 37.767N 105.515W DRT Non-dune 7 Sep 2012 Net 7:00-9:00 37.679N 105.591W NRT Dune 9 Aug 2012 Malaise 24 hours 37.762N 105.516W D12 Dune 9 Aug 2012 Malaise 24 hours 37.767N 105.516W DRT Non-dune 10 Aug 2012 Malaise 24 hours 37.670N 105.593W N5 Non-dune 15 Aug 2012 Malaise 24 hours 37.757N 105.507W N1 Non-dune 16 Aug 2012 Malaise 24 hours 37.674N 105.593W Non-dune 28 Aug 2012 Malaise 24 hours 37.715N 105.534W N3 Dune 29 Aug 2012 Malaise 24 hours 37.771N 105.516W Non-dune 29 Aug 2012 Pan 9:00-15:00 37.679N 105.590W NRT Dune 9 Aug 2012 Pan 9:00-15:00 37.764N 105.515W Non-dune 15 Aug 2012 Pan 9:00-15:00 37.757N 105.507W N1 Non-dune 16 Aug 2012 Pan 9:00-15:00 37.664N 105.593W

115 Environment Date Type Time Latitude Longitude Name Dune 28 Aug 2012 Pan 9:00-15:00 37.767N 105.516W DRT

116 Table A.8 Parameter estimates from a generalized mixed effect model of seedling emergence that includes all the seed types. In this case the intercept represent emergence in the unseeded

! control subplots. This model has significant effects of environment (!!=19.37, P<0.0001), type

! ! (!!=337.51, P<0.0001) but no environment-by-type interaction (!!= 14.54, P=0.1497).

Fixed effects: Estimate SE t P Intercept -3.343 0.320 -10.46 <0.001 Non-dune environment -0.938 0.213 -4.40 <0.001 Type N3 1.312 0.458 2.86 0.004 Type N2 1.445 0.444 3.25 0.001 Type N1 1.866 0.394 4.73 <0.001 Type BCN 1.170 0.477 2.45 0.014 Type F1N 2.133 0.375 5.70 <0.001 Type F1D 3.030 0.320 9.47 <0.001 Type BCD 2.877 0.330 8.73 <0.001 Type D1 3.794 0.301 12.59 <0.001 Type D2 3.698 0.303 12.20 <0.001 Type D3 3.465 0.310 11.19 <0.001

Zero inflation: 0.245 0.060

Random effects: Variance SD Plot 0.568 0.754

117 Table A.9 Parameter estimates from a generalized mixed effect model of seedling emergence that included dune and non-dune subplots only. This model has significant effects of

! ! environment (!!=19.85, P<0.001), ecotype (!!=20.82, P<0.001) but no environment-by-ecotype

! interaction (!!=0.74, P=0.39).

Fixed effects: Estimate SE t P Intercept 0.316 0.182 1.73 0.083 Non-dune environment -1.094 0.249 -4.39 <0.001 Non-dune ecotype -2.095 0.206 -10.16 <0.001

Zero inflation: 0.234 0.071

Random effects: Variance SD Plot 0.669 0.818 Population 0.003 0.051

118 Table A.10 Parameter estimates from LMMs for three difference proxies of fecundity. Only surviving plants were included in the analysis, and those plants were pooled into three types (dune, hybrid and non-dune).

Plant height Flower number Log(seed number) Fixed effects: Estimate SE t Estimate SE t Estimate SE t Intercept 100.46 13.06 7.689 54.589 6.85 7.967 7.718 1.107 6.972 Hybrid plant -47.35 13.97 -3.390 -38.794 5.26 -7.381 -2.995 1.205 -2.486 Dune plant -66.49 15.50 -4.290 -54.223 7.26 -7.470 -3.889 1.312 -2.964 Dune environment -69.22 15.27 -4.534 -47.270 7.81 -6.050 -3.633 1.299 -2.797 Hybrid plant * dune environment 47.74 16.53 2.889 38.284 5.86 6.531 3.264 1.422 2.295 Dune plant * dune environment 64.89 17.39 3.731 55.016 7.66 7.181 3.710 1.482 2.503

Random effects: Variance SD Variance SD Variance SD Plot 135.2 11.63 203.40 14.26 0.677 0.82

119

Table A.11 Aculeate insects caught visiting sunflowers. The USDA-ARS Bee Biology & Systematics

Laboratory provided identifications.

Percent/number of specimens collected Total number on of specimens Dune Non-dune Dunes Species collected flowers flowers Agapostemon femoratus 3 0% 0 2 Agapostemon texanus 2 100% 1 0 Andrena haynesi 2 50% 1 1 Andrena helianthi 2 0% 0 2 Andrena pecosana 2 100% 2 0 Apis mellifera 40 100% 40 0 Bembix pallidipicta 14 7% 1 3 Bembix sayi 1 100% 1 0 Bicyrtes ventralis 1 0% 0 1 Bombus huntii 5 100% 5 0 Bombus morrisoni 3 67% 2 1 Colletes gypsicolens 2 0% 0 2 Colletes lutzi 10 10% 0 5 Colletes phaceliae 4 0% 0 1 Dieunomia triangulifera 11 100% 10 0 Euodynerus sp. 1 100% 1 0 Gorytes sp. 1 0% 0 1 Hesperapis carinata 1 0% 0 1 Lasioglossum sp. 66 2% 0 23 parallela 2 100% 2 0 Megachile onobrychidis 1 0% 0 1 Megachile n. sp. 1 0% 0 1 Megachile agustini 12 100% 12 0 Melissodes agilis 4 75% 2 1

120 Total number Percent/number of specimens collected of specimens on Species collected Dunes Dunes Dunes Microbembex monodonta 61 97% 21 0 Oxybelus sp. 16 0% 0 3 Perdita albipennis 24 8% 2 16 Perdita dolichocephala 38 0% 0 21 Perdita electa 7 29% 2 1 Perdita tridentata 37 8% 2 28 Pterocheilus sp. 1 0% 0 1 Svastra obliqua 1 100% 1 0 Triepeolus grindeliae 1 0% 0 1

Table A.12 Parameter estimates from a GLMM analysis of pollen competition data. There is a

! significant effect of maternal ecotype (!!=24.4, P<0.0001) on the proportion of offspring that are sired by dune plants after mixed pollinations.

Fixed effects: Estimate SE z P Intercept 0.944 0.160 5.907 <0.001 Non-dune ecotype -1.255 0.192 -6.525 <0.001

Random effects: Variance SD Maternal plant 0.451 0.672 Maternal population 1.26x10-9 3.55x10-5 Pollen mix 0.337 0.580

121 Appendix B - GBS protocol

This protocol is largely based on a protocol by Poland et al. (2012). It had been modified by several members of the Rieseberg lab including, Greg Baute, Dan Ebert, Kristin

Nururski, Greg Owens, Brook Moyers, and Marco Todesco. The section used to deplete repetitive DNA sequences was modified by Marco Todesco from protocols by Shagina et al. (2010) and Matvienko et al. (2013).

B.1 Protocol summary

The basic steps that make up this GBS protocol are: digestion (DIG) of DNA by restriction enzymes, ligation (LIG) of barcoded and common adapters to the ends of the digested fragments, amplification (PCR) of the adapter-ligated fragments, depletion

(DSN) of high-copy fragments (e.g., cpDNA) from the library (which is followed by additional amplification), and size selection (SS) of fragments to a specific range. In addition, there are several places in the protocol where the DNA is cleaned (i.e., primer/adapter sequence is removed) and concentrated (CC) using SPRI beads.

DIG à LIG à CC (weak SS) à PCR à CC à DSN à PCR à Gel SSà CC à QC

B.2 Adaptor and primer sequences

Barcoded adaptor (***=barcode & ^^^=reverse compliment of barcode):

ACACTCTTTCCCTACACGACGCTCTTCCGATCT***TGCA

^^^AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT

122 Common adapter (from Poland et al. 2012):

CGAGATCGGAAGAGCGGGGACTTTAAGC

GATCGGTCTCGGCATTCCTGCTGAACCGCTCTTCCGATCT

PCR primers (from Poland et al. 2012):

AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCG

ATCT

CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAA

B.3 Protocol

1. Digestion

• Aliquot 10 µL of normalized DNA (10 ng/µL eluted in 10 mM Tris-HCl ) into a 96 well plate. • Make a digestion master mix:

DOUBLE DIGESTION Catalogue 1 sample 1 plate / 100 MASTER MIX Number (µl) samples (µl)

10mM Tris-HCl, pH 8 7.2 720

NEB 10X Cutsmart buffer B7204 2.0 200

NEB Pst I–HF R3140 8 units: 0.4 40

NEB Msp I R0106 8 units: 0.4 40

Total 10.0 1000

123 • Add 10 µL of double digestion master mix to each sample, bringing the total volume in each well to 20 µl. • Digest for 5 hrs at 37 ˚C, followed by 20 min at 65 ˚C, then hold at 4 ˚C. • Proceed to ligation as soon as possible.

2. Ligation

• Aliquot 4.5 µl of barcoded adapters and 18 µl of digested samples into a new plate (a unique barcode adaptor for each samples). • Make a master ligation mix:

LIGATION MASTER MIX Catalogue Number 1 sample 1 plate / 100 (µl) samples (µl)

10mM Tris-HCl, pH 8 8.0 800

NEB 10X Cutsmart buffer B7204 2.0 200

10 mM ATP P0756 4.0 400

Common adaptor 1.0 100

NEB T4 DNA Ligase M0202 ~200 U: 0.5 50

Total 15.5 1550

• Aliquot 15.5 µl of the ligation mater mix to each well brining the total volume to 38 µl. • Ligate for 3 hrs at 22 ˚C, followed by 20 min at 65 ˚C, then hold at 4 ˚C.

3. Clean and concentrate with weak size selection

• Pool 36 µl from each well across each row into a 1.5 mL tube, for a total of 432 µl in each of eight tubes. • Add 367.2 µl (0.85X) of well-mixed room temperature beads to each tube and let stand (5 mins). • Place on magnet, allow pellet to form (2 mins) and then discard supernatant. • Wash pellet 2X with 1 mL of 80% EtOH and then 1X with 1 ml 100% EtOH. • Let tubes dry for 15 mins with caps open.

124 • Remove tubes from magnet, add 50 µl 10 mM Tris-HCl to each tube, mix by pipetting and let stand for 5 minutes. • Place tubes on magnet and allow pellet to form (2 min). • Transfer 48 µl of supernatant to a new tubes • Add 28.8 µl (0.6X) of well-mixed room temperature beads to each tube and let stand (5 mins). • Place on plate magnet and allow pellet to form (2 mins). • Transfer 72 µl of supernatant to a new tube. • Add 72 µl (1X) of well-mixed room temperature beads to each tube and let stand (5 mins). • Place on magnet, allow pellet to form (2 mins) and then discard supernatant. • Wash pellet 2X with 1 mL of 80% EtOH and then 1X with 1 ml 100% EtOH. • Let tubes dry for 15 mins with caps open. • Remove tubes from magnet, add 32 µl 10 mM Tris-HCl to each tube, mix by pipetting and let stand (5 mins). • Place on magnet and allow pellet to form (2 min). • Carefully transfer supernatant to fresh tubes.

4. Amplification

• Aliquot 2 µl of the concentrated DNA into 8 tubes. • Make a master PCR mix:

AMPLIFICATION MIX 1 rxn 8 rxns

Kapa HIFI HotStart 2X 12.5 106.25

Primer F 1.0 8.5

Primer R 1.0 8.5

10mM Tris-HCl, pH 8 8.5 72.25

Total 23

• Aliquot 23 µl of the PCR mater mix to each well brining the total volume to 25µl. • Amplify DNA using the following cycles: 30 sec at 98°C, repeat 12 times, 30 sec at 98°C, 20 sec at 62°C and 30 sec and 72°C, then 5 mins at 72°C and 4°C forever. • Pool the pcr products.

125 • Add 1.6X volumes of well-mixed room temperature beads to the pooled pcr product and let stand (5 mins). • Place on magnet, allow pellet to form (2 mins) and then discard supernatant. • Wash pellet 2X with 1 mL of 80% EtOH and then 1X with 1 ml 100% EtOH. • Let tubes dry for 15 mins with caps open. • Remove tubes from magnet, add 22 µl 10 mM Tris-HCl to each tube, mix by pipetting and let stand (5 mins). • Place on magnet and allow pellet to form (2 min). • Carefully transfer supernatant to fresh tubes.

5. Depletion of high copy fragments

• Aliquot 3uL of very concentrated DNA (80 ng/uL minimum) into a 0.2mL pcr tube • Add 1 uL of 4X hybridization buffer (200 mM HEPES pH 7.5, 2 M NaCl, 0.8 mM EDTA) and mix. • Overlay with 10 uL of mineral oil and spin down. • Heat the sample to 98°C for 2 min and then 78°C for 5 hours. • Add 5 uL of DSN buffer (0.1 M Tris pH 8, 10 mM MgCl2, 2 mM DTT, pre- warmed to 70°C) to sample, mix, spin, and move to 70°C for 5 mins (it’s important to do this as quickly as possible). • Add 0.2 uL of a 1:2 dilution of DSN enzyme, mix, spin, and return to 70°C for 15 mins. • Add 10 uL of 10 mM EDTA, mix, spin and put on ice. • Repeat the amplification and concentration steps in section 4.

6. Gel size selection

• Pour a 1.5% gel using high molecular weight agarose that includes 1 uL of ethidium bromide for every 100mL of gel. • Load a 25 uL of sample (with a maximum of 10 µg of DNA) and 5 uL of dye into alternating wells on the gel. • Load 20–25 µl water mixed with 5 µl dye in the empty wells. • Add 5 µl EtBr to the bottom reservoir of the gel rig. • Run for ~10 minutes at 100 V. • Increase to 150 V and run for 60 mins. • Excise bands between 300 and 600bp using new, sterile blades. • Proceed to follow the QIAquick gel extraction kit protocol. • Concentrate your extracted library using the concentration steps in section 4.

126 Appendix C - Chapter 3 supplementary figures

PET1 ANN1 NEG1

TP64546 TP61739 TP7795 TP34376 TP61638 TP45484 TP57675 TP40673 TP54604 TP13122 TP16790 TP43625 TP22349 TP59424 TP16809 TP3487 TP54869 TP34126 1ENTP4853 TP40656 TP34161 TP6432 2ENTP55593 cM TP18786 TP39245 TP9407 2ENTP7034 TP25594 0 2ENTP30683 2ENTP1623 TP18340 TP25040 2ENTP53959 2ENTP80018 2ENTP90340 TP26821 5 2ENTP48882 2ENTP28028 2ENTP67859 TP46526 TP537 TP31452 2ENTP53830 2ENTP5501 2ENTP57843 TP3641 TP42196 TP17263 10 2ENTP98735 2ENTP61015 2ENTP53609 TP33023 2ENTP95796 2ENTP76350 TP6867 TP50612 TP33769 1ENTP19861 2ENTP71557 2ENTP38346 15 TP22596 2ENTP2891 TP32497 TP46771 TP7264 2ENTP95984 2ENTP88048 20 TP2610 TP20964 TP31344 2ENTP83873 TP20035 TP30714 2ENTP3792 2ENTP10017 25 TP13032 TP17930 TP37689 TP32866 TP61855 TP50300 2ENTP15850 2ENTP48341 2ENTP34862 30 TP3210 TP23533 1ENTP20797 TP7083 TP51902 TP32795 2ENTP67914 2ENTP16013 2ENTP48481 TP22647 1ENTP15857 1ENTP849 35 TP12059 TP2302 TP31728 2ENTP4248 TP33455 2ENTP39331 2ENTP96668 2ENTP73479 40 TP30958 2ENTP5818 2ENTP88566 2ENTP57946 TP36538 TP55709 TP55689 2ENTP92949 45 TP20173 TP14663 2ENTP96788 2ENTP48350 2ENTP32614 TP34912 TP61617 TP7980 2ENTP100108 1ENTP12314 2ENTP3808 50 TP20902 TP5685 2ENTP45917 TP54240 TP65540 2ENTP20796 2ENTP53709 2ENTP81812 55 TP46557 TP15363 TP37918 2ENTP11964 2ENTP81797 TP62605 TP30812 **TP43234** 2ENTP80020 60 **TP54216** 1ENTP25412 2ENTP566 2ENTP4676 TP5937 TP35483 2ENTP35590 2ENTP59823 65 TP38230 TP47897 1ENTP20728 TP3297 TP50259 TP29167 2ENTP1015 2ENTP91732 TP37611 TP42890 TP66072 2ENTP78915 70 TP12061 TP27208 TP30750 2ENTP30684 2ENTP31870 2ENTP94829 TP2249 2ENTP52246 75 TP46773 2ENTP68547 TP10787 2ENTP34897 2ENTP35377 2ENTP60743 80 TP56778TP44482 1ENTP12486 TP12208 2ENTP90380 2ENTP11395 85 TP36449 TP23468 TP1590 2ENTP15204 TP19697 TP35033 2ENTP27523 2ENTP93796 2ENTP13077 90 TP59978 2ENTP17149 2ENTP81881 **2ENTP2119** TP20615 2ENTP77703 2ENTP11527 **2ENTP86878** 95 TP13270 TP56207 TP3805 1ENTP22223 2ENTP39486 TP45982 TP24311 2ENTP4177 TP10288 TP38966 TP44643 100 TP16622 TP2647 TP44904

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.1 – PET1 and NEG1 compared to ANN1. Lines connect homologous markers on the same page. Marker names are colored based on their homology with the ANN reference (black denotes that the marker could not be reliably mapped in ANN reference). Markers that part of a “micro- translocation” (2-3 markers less than 2 cM apart that map to another ANN LG) are identified with the symbol **.

127 PET2 ANN2 NEG2 cM 2ENTP84898 0 TP31818 TP5671 2ENTP15140 **2ENTP95185**2ENTP15581** 2ENTP16340 5 TP15497 2ENTP95081 TP31925 TP36732 2ENTP68467 10 TP59970 2ENTP69064 TP29938 1ENTP22047 2ENTP38356 15 TP22579 TP30860 TP44924 2ENTP85539 2ENTP57289 2ENTP15550 TP12578 1ENTP20621 20 1ENTP6908 TP43553 TP7598 TP39558 TP30529 25 TP40282 TP24173 TP1766 TP36347 TP59579 30 TP19340 TP57528 TP36678 TP62714 35 TP36153 **TP26334**TP28825 TP37786 TP57483 **TP1543** 40 TP53154 TP53616 TP53153 45 TP28949 TP39285 TP35048 50 TP60844 TP55230 TP4855 TP38081 55 TP46922 TP39122 TP3978 TP14644 TP15739 TP3748 TP35908 TP34012 TP8101 60 TP50861 TP7128 TP64441 65 TP65838TP48779TP29668 TP13917 70 TP56945 TP47943 TP37462 TP48049 75 TP752 TP44669 TP45389 80 TP28000 TP57978 TP32244 TP65272 85 TP1408 TP5509

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.2 – PET2 and NEG2 compared to ANN2. See figure C.1 caption for details.

128 PET3 ANN3 NEG3

TP10361 TP55268 TP62162 TP65253 TP54012 TP16123 TP56853 TP61976 TP66095 TP45440 TP24931 TP54874 TP21443 TP44590 TP46973 TP45276 TP56493 TP8210 TP3105 TP35879 TP9285 TP10364 TP34939 TP40892 TP34216 TP2910 TP9692 TP41625 2ENTP11307 TP62439 TP24593 TP2401 1ENTP673 cM TP46141 TP50510 TP60482 2ENTP34858 TP5190 TP2232 TP39546 2ENTP96559 2ENTP100343 0 TP29825 TP2038 2ENTP24730 2ENTP90429 2ENTP48150 TP65833 TP14725 TP25935 2ENTP94258 2ENTP6679 5 TP64141 TP45153 TP16382 2ENTP73240 TP57791 TP25989 TP61135 2ENTP35249 2ENTP967311ENTP10012 10 TP48876 TP24202 TP40963 1ENTP3901 1ENTP14689 TP61103 TP7518 2ENTP32081 2ENTP738 15 TP9104 TP52595 TP51919 2ENTP13592 2ENTP87602 2ENTP93120 TP35445 TP32120 TP19153 1ENTP665 2ENTP77384 20 TP17321 2ENTP24505 2ENTP75757 TP56518 TP20578 TP54382 1ENTP5598 25 TP56955 2ENTP26625 2ENTP80008 2ENTP61318 TP27846 2ENTP81238 2ENTP58578 2ENTP53723 30 TP46515 TP27924 TP52590 2ENTP18567 2ENTP69248 TP28031 TP35613 TP38319 2ENTP75034 2ENTP92881 TP61399 2ENTP3163 35 TP11570 2ENTP98862 2ENTP16658 TP42930 TP59486 TP22944 2ENTP68584 40 2ENTP13949 1ENTP24326 TP50439 TP36905 TP47572 1ENTP18634 2ENTP88061 2ENTP94981 TP24629 TP46246 1ENTP4424 1ENTP21463 2ENTP5374 45 TP10849 TP44822 2ENTP29694 1ENTP2039 TP61118 TP34065 2ENTP22031 2ENTP72897 2ENTP81965 50 TP30821 TP38473 2ENTP92561 2ENTP16515 2ENTP53643 TP27177 TP34592 TP6116 2ENTP4132 2ENTP93297 55 TP4463 TP37616 2ENTP99838 TP58808 2ENTP21752 2ENTP100232 2ENTP96718 60 TP43409 TP54489 1ENTP5498 TP6171 TP11870 TP14454 2ENTP79318 TP2123 65 2ENTP1578 TP32564 2ENTP77447 2ENTP81711 TP24239 1ENTP22288 70 TP52575 TP34681 2ENTP17684 2ENTP83703 TP43930 2ENTP82252 2ENTP67674 2ENTP68377 75 TP27048 TP59287 TP4301 2ENTP87965 TP62091 2ENTP48316 1ENTP21593 1ENTP24479 80 TP50845 TP18582 2ENTP93464 TP51074 1ENTP732 85 TP65730 TP59520 TP61266 TP43800 TP65011 90 TP43769 TP47174 TP10390 TP3179 95 TP13584 TP39063 TP15105 TP66133 TP37974 100 TP11063 TP61255 TP9765 TP59478 TP65635 TP25275 TP2323 TP53473

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.3 – PET3 and NEG3 compared to ANN3. See figure C.1 caption for details.

129 PET4/7 ANN4 NEG4

TP48848 TP15276 cM TP10549 TP19969 1ENTP23702 0 **TP11747**TP35148**TP40939 TP42579 TP64559 TP48986 1ENTP21660 TP10071 2ENTP14202 2ENTP61234 1ENTP21907 5 TP1495 TP7953 1ENTP22178 2ENTP98502 TP7211 TP61095 TP24025 2ENTP59652 1ENTP25465 2ENTP23425 10 TP20146 TP39332 TP51720 2ENTP11932 2ENTP95747 TP49139 TP36256 2ENTP16681 15 TP63394 TP17182 TP34434 TP63672 20 TP14958 TP23210 TP15002 TP22323 25 TP506 TP31706 TP37275 TP11909 TP2085 TP12829 30 TP59026 TP7007 TP47715 TP21322 TP42689 TP37908 TP41409 35 TP58592 TP37477 40 TP45886 TP17228 45 TP12454 TP17453 TP52006 TP52256 50 TP17374 TP754 TP62270 55 TP775 TP32651 TP36298 60 TP24690 TP60015 TP17128 65 TP7522 TP61333 TP11023 TP61769 TP21055 TP37461 TP10554 TP62056 70 TP60995 TP37025 75 TP61239 TP40313 80 TP26699 TP50330 TP63479 85

90

95

100

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.4 – PET4/7 and NEG4 compared to ANN4. See figure C.1 caption for details.

130 PET5 ANN5 NEG5

TP45193 TP19375 TP57678 TP45546 TP34526 2ENTP97404 2ENTP81122 2ENTP14082 cM TP24179 TP4530 2ENTP91690 TP12358 0 2ENTP1232 2ENTP53167 TP44037 TP40168 TP16884 1ENTP25893 2ENTP48312 2ENTP59571 TP61282 TP13824 TP20417 5 2ENTP4110 2ENTP18215 2ENTP23636 TP38732 TP30015 1ENTP25876 2ENTP32471 TP64679 TP61347 TP10115 10 2ENTP61715 2ENTP93208 TP55102 TP219 TP55302 2ENTP34600 2ENTP35183 TP62646 2ENTP58600 1ENTP5280 15 TP58561 TP60997 TP1575 2ENTP99203 TP1366 2ENTP38018 20 TP19649 TP36111 TP48645 1ENTP25885 2ENTP771 TP66217 TP25208 TP40586 2ENTP77150 25 TP26408 TP6292 TP60193 2ENTP6793 2ENTP10461 2ENTP78535 TP40791 TP32569 TP65316 1ENTP13806 30 TP43531 TP8232 2ENTP56711 2ENTP16592 TP7458 TP11441 2ENTP94700 2ENTP1219 TP35826 TP19337 35 2ENTP51303 2ENTP55487 2ENTP77917 TP46190 TP39669 TP24943 2ENTP91296 TP17563 40 2ENTP32254 TP54570 TP36260 2ENTP95319 2ENTP55335 2ENTP2718 TP28301 2ENTP99468 2ENTP91703 1ENTP3567 45 TP17560 TP23600 2ENTP13086 2ENTP53306 2ENTP88665 TP20604 TP4251 TP43988 2ENTP53592 2ENTP76162 2ENTP2935 50 TP25145 2ENTP24598 2ENTP51080 2ENTP73701 TP11953 TP46580 TP38417 2ENTP78514 2ENTP15213 2ENTP48924 55 TP4565 TP19312 TP18093 1ENTP1866 TP12047 TP60926 TP10670 2ENTP98329 2ENTP94851 60 TP12365 TP39756 TP57706 2ENTP34666 2ENTP33535 2ENTP14528 TP22836 2ENTP22605 2ENTP30078 2ENTP42824 TP45681 65 2ENTP49237 2ENTP91702 2ENTP115 TP18924 2ENTP74090 2ENTP49288 2ENTP99697 TP32638 TP45639 TP47397 70 2ENTP6138 2ENTP14357 TP50672 TP9818 2ENTP12312 2ENTP8717 2ENTP5929 TP33991 1ENTP2186 75 TP39417 TP28522 TP35456 TP50502 TP19623 80

85

90

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.5 – PET5 and NEG5 compared to ANN5. See figure C.1 caption for details.

131 PET6 ANN6 NEG6/16

TP1952 TP30164 TP18496 TP63392 TP843 TP25019 TP11431 TP24062 TP35633 TP1281 TP19889 TP25497 TP53773 TP50470 TP37496 cM TP914 0 TP16992 1ENTP8365 TP13115 2ENTP80094 2ENTP56286 2ENTP84594 TP62362 TP7055 TP27739 5 2ENTP1858 TP44379 TP47012 TP19395 TP55611 TP21111 TP62618 10 2ENTP55435 TP28630 TP21547 TP15215 2ENTP37523 TP18763 TP19068 TP44884 2ENTP24877 15 TP42161 TP12316 TP30181 TP28013 TP22937 TP62633 2ENTP56382 2ENTP5177 2ENTP89007 20 TP6105 TP14408 TP48748 2ENTP16537 TP11878 TP54287 TP6234 2ENTP84691 2ENTP24235 25 TP30748 TP19485 TP3294 2ENTP6332 2ENTP25048 TP26617 TP45871 TP16104 2ENTP15093 30 TP4479 2ENTP96813 2ENTP1326 2ENTP68843 TP12564 TP39891 TP64960 2ENTP17238 2ENTP78250 35 TP8129 TP46118 2ENTP14487 2ENTP83266 2ENTP96971 TP22695 2ENTP39131 TP15150 2ENTP9518 40 TP6656 TP43286 TP40629 2ENTP5646 2ENTP14150 2ENTP16502 TP46910 TP4063 2ENTP85055 2ENTP39168 45 TP21301 TP22611 1ENTP25596 TP35335 1ENTP20025 2ENTP55485 1ENTP12516 50 TP15305 TP27169 TP41542 2ENTP87217 TP57386 2ENTP44459 1ENTP13875 2ENTP22484 55 TP16994 2ENTP89277 2ENTP18979 TP58271 TP30018 TP40801 1ENTP13876 60 TP18743 TP59762 TP27482 TP5576 TP34145 TP20676 65 TP34374 TP41830 TP7292 TP20852 TP54959 TP44976 70 TP56404 TP6414 TP40498 TP38389 TP35084 TP63629 TP63891 2ENTP48544 75 TP64850 TP16911 2ENTP72444 2ENTP38363 TP11105 TP48226 TP56471 2ENTP12082 80 TP64026 TP43424 2ENTP72881 2ENTP48731 TP10761 TP43181 2ENTP48917 2ENTP4282 2ENTP32026 85 TP13759 2ENTP14957 2ENTP72994 90 2ENTP95371 1ENTP24490 95 1ENTP6563 2ENTP100122 1ENTP1177 100 1ENTP1982 2ENTP62467 105 2ENTP1643 1ENTP9838 2ENTP32107 2ENTP95757 2ENTP34592 110 2ENTP34360 2ENTP8200 2ENTP28946 2ENTP99940 2ENTP84042 1ENTP13846 2ENTP59342 2ENTP10702 2ENTP88877 1ENTP5006

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.6 – PET6 and NEG6/16 compared to ANN6. See figure C.1 caption for details.

132 Pet4/7 Ann7 Pet7/4 Ann4

TP16102 TP36733 cM TP48848 TP15276 TP16536 TP10549 TP19969 TP26697TP65054 0 **TP11747**TP35148**TP40939 TP5148 TP42579 TP64559 TP48986 TP16538 TP10071 5 TP46189 TP21469 TP19801 TP1495 TP7953 TP26939 TP48767 TP41627 TP7211 TP61095 TP24025 10 TP40893 TP41238 TP48478 TP20146 TP39332 TP51720 TP27976 TP49139 TP36256 15 TP52960 TP63394 TP24677 TP20779 TP51353 TP17182 TP34434 TP63672 TP6184 TP63154 TP64820 20 TP14958 TP34839 TP16140 TP13969 TP23210 TP15002 TP22323 TP6981 TP20089 25 TP506 TP31706 TP37275 TP13369 TP44147 TP11909 TP2085 TP12829 TP2329 TP42790 TP59948 30 TP59026 TP12150 TP7007 TP47715 TP21322 TP53976 TP502 TP26741 35 TP42689 TP37908 TP41409 TP54503 TP58592 TP34188 TP37477 TP28264 TP43662 40 TP43790 TP45886 TP34745 TP50652 TP17228 45 TP13412 TP12078 TP12454 TP35001 TP17453 TP52006 TP52256 TP440 50 TP17374 TP57747 TP63616 TP754 TP62270 TP11954 55 TP775 TP23804 TP32651 TP36298 TP8604 60 TP24690 TP48325 TP37060 TP54673 TP60015 TP17128 TP12377 TP41306 TP2305 65 TP7522 TP61333 TP33202 TP22853 TP11023 TP61769 TP21055 TP9872 TP13959 TP37461 TP10554 TP62056 70 TP5782 TP60995 TP50842 TP6816 TP37025 75 TP39442 TP61239 TP17576 TP40313 80 TP52824 TP9937 TP26699 TP60639 TP50330 TP63479 TP2382 85 TP23343 TP63325 90 TP29147 TP3103 TP38145 TP53026 TP6294 TP53277 95 TP61953 TP65812 TP50435 TP27530 100

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.7 – PET4/7 compared to ANN7 and PET7/4 compared to ANN4. See figure C.1 caption for details.

133 PET7/4 ANN7 NEG7

TP16102 TP36733 TP16536 cM TP26697TP65054 0 TP5148 TP16538 5 TP46189 TP21469 TP19801 TP26939 TP48767 TP41627 10 TP40893 TP41238 TP48478 TP27976 2ENTP51298 2ENTP56334 15 TP52960 TP24677 TP20779 TP51353 2ENTP84636 2ENTP36644 20 TP6184 TP63154 TP64820 TP34839 TP16140 TP13969 2ENTP82023 TP6981 TP20089 2ENTP27970 1ENTP8172 25 TP13369 TP44147 2ENTP55779 TP2329 TP42790 TP59948 1ENTP3753 1ENTP23317 30 TP12150 2ENTP22458 2ENTP87500 2ENTP14110 TP53976 TP502 TP26741 **1ENTP4011**1ENTP21705** 35 TP54503 2ENTP84112 2ENTP46807 2ENTP30118 TP28264 TP43662 TP34188 2ENTP56767 1ENTP4681 2ENTP88496 40 TP43790 2ENTP88384 2ENTP62384 TP34745 TP50652 1ENTP18170 45 TP13412 TP12078 1ENTP3461 TP35001 2ENTP14160 2ENTP37605 2ENTP82941 2ENTP91803 50 TP440 TP57747 TP63616 2ENTP91751 2ENTP98767 2ENTP73848 TP11954 2ENTP24238 55 TP23804 2ENTP15216 2ENTP88982 2ENTP49028 TP8604 2ENTP316 1ENTP17766 60 TP48325 TP37060 TP54673 2ENTP34843 2ENTP22944 TP12377 TP41306 TP2305 1ENTP3507 65 TP33202 TP22853 2ENTP1764 2ENTP68936 2ENTP30094 TP9872 TP13959 2ENTP5522 2ENTP2086 2ENTP45674 70 TP5782 2ENTP99220 TP50842 TP6816 2ENTP98742 2ENTP35958 75 TP39442 2ENTP24407 2ENTP61402 2ENTP96082 TP17576 2ENTP14083 80 TP52824 TP9937 2ENTP83403 2ENTP33461 2ENTP27668 TP60639 1ENTP8935 2ENTP36928 2ENTP11504 2ENTP85202 85 TP2382 TP23343 2ENTP19096 2ENTP1114 TP63325 2ENTP53298 2ENTP22071 90 TP29147 TP3103 **TP38145** 1ENTP13739 1ENTP13730 TP53026 TP6294 TP53277 2ENTP16588 2ENTP33318 95 TP61953 **TP65812**TP50435** 2ENTP51134 2ENTP2205 TP27530 2ENTP6898 2ENTP16842 2ENTP42295 100 2ENTP85719 2ENTP2920

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.8 – PET7/4 and NEG7 compared to ANN7. See figure C.1 caption for details.

134 PET8 ANN8 NEG8

TP52361 TP20053 TP6243 TP8434 TP15997 TP51212 TP48501 TP10354 TP1568 TP59723 TP28484 TP15352 TP33649 TP60591 TP63836 TP42867 TP45688 TP13003 TP30962 TP24155 TP51601 TP34965 cM TP18349 TP23130 TP38334 TP63982 0 TP5607 2ENTP74897 TP42691 2ENTP32560 2ENTP58685 2ENTP69572 5 TP17428 2ENTP15171 TP20248 TP15734 TP7434 1ENTP15111 2ENTP30554 10 TP61768 TP65556 TP16528 2ENTP19327 TP12538 2ENTP98180 15 TP58425 2ENTP72035 2ENTP84350 2ENTP5884 TP10136 TP65499 20 TP54939 TP39869 TP30390 TP15240 TP46896 25 TP16693 TP54992 TP7242 TP49221 TP40985 TP45215 TP60705 TP51769 30 TP45658 TP5854 TP43526 TP43690 2ENTP58445 35 TP24386 1ENTP15201 TP47549 TP38182 TP32658 2ENTP54827 40 TP1372 TP24098 TP55501 2ENTP15611 TP8268 TP6551 TP57336 2ENTP93750 45 TP23991 TP59911 TP1435 1ENTP14247 1ENTP9751 TP50512 TP42888 TP35468 1ENTP6107 2ENTP38345 50 TP25368 TP52449 2ENTP1466 TP47096 2ENTP83774 2ENTP712 **1ENTP12363** 55 TP48227 TP66206 **1ENTP22218** 2ENTP91300 TP27664 TP2684 TP35791 1ENTP3494 1ENTP5194 60 TP23668 TP14602 TP25794 2ENTP51778 TP30527 TP10090 TP45549 2ENTP20936 2ENTP55030 TP50379 2ENTP95821 2ENTP50960 2ENTP28515 65 TP698 1ENTP24354 TP45378 TP46029 2ENTP87643 70 TP51842 TP29176 2ENTP98569 2ENTP87088 2ENTP58757 TP39265 TP46192 TP31784 75 TP23268 TP4241 TP19638 2ENTP11473 TP23212 TP39057 TP66060 2ENTP1005 2ENTP18776 80 TP9989 TP63222 TP1131 85 TP40403 TP6402 TP18036 TP20210 90 TP47612 TP26349 TP49909 TP39585 95 TP7244 TP1782 TP54637 TP58944 TP25393 TP64140 100 TP28475 TP60342 TP1729 TP23961 TP18887 TP39072 TP34294 TP16497 TP32801 TP57805 TP58190 TP19909 TP65409 TP25248 TP44329 TP10335 TP43090 TP36724 TP3322 TP30956 TP16970

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.9 – PET8 and NEG8 compared to ANN8. See figure C.1 caption for details.

135 PET9 ANN9 NEG9

TP42520 TP33879 TP4309 TP40693 TP576 TP36309 2ENTP23043 2ENTP15176 2ENTP93155 TP43382 TP41211 2ENTP83683 2ENTP8191 TP21890 TP46491 TP2451 2ENTP56273 TP44330 TP40527 TP11339 1ENTP19989 1ENTP24698 TP44836 TP39227 TP41807 2ENTP58273 TP23512 2ENTP18660 2ENTP34733 2ENTP31674 TP18523 TP49951 2ENTP93334 cM TP48947 TP59565 TP7302 1ENTP11873 1ENTP5571 TP30135 TP58690 TP52551 2ENTP29640 0 TP53692 1ENTP9189 2ENTP3233 1ENTP10085 TP27911 TP18698 2ENTP9326 2ENTP57018 5 TP27038 TP22902 TP30562 2ENTP80702 TP57192 TP43576 1ENTP24127 2ENTP73683 10 TP44770 TP50084 TP63579 2ENTP11523 2ENTP30916 TP23022 TP39312 TP65353 2ENTP82 15 TP31247 2ENTP80279 TP36919 TP15489 TP56025 2ENTP19767 2ENTP62478 TP52089 TP1259 20 2ENTP96982 1ENTP5688 TP23121 2ENTP78361 TP46858 TP33028 TP34547 25 2ENTP68783 TP17358 TP31978 TP47814 2ENTP91418 2ENTP32965 2ENTP33320 TP60724 2ENTP61591 2ENTP74216 30 TP14912 TP7717 TP33179 2ENTP89970 TP4713 2ENTP83207 1ENTP12948 35 TP36123 TP27469 TP39125 2ENTP84634 TP59310 TP33176 TP11719 2ENTP57226 40 TP53406 TP48185 TP42876 2ENTP91105 2ENTP96248 1ENTP3594 TP23838 TP50741 TP9687 2ENTP70184 2ENTP55759 1ENTP23544 45 TP9980 TP28919 TP10606 2ENTP60478 2ENTP14022 TP40688 TP4488 1ENTP5803 50 TP58068 TP38783 2ENTP56938 1ENTP24410 2ENTP6677 TP54497 TP38243 TP59042 1ENTP25374 TP27618 TP63042 TP50898 55 1ENTP5009 TP7293 TP38656 TP25876 2ENTP28720 2ENTP77321 2ENTP15549 TP44128 TP42648 TP11864 60 2ENTP14508 1ENTP7539 TP4371 TP24609 1ENTP23216 2ENTP32922 2ENTP23718 TP56464 TP21847 TP56117 1ENTP14446 2ENTP57680 65 TP26955 TP37898 TP6470 2ENTP90753 2ENTP92427 2ENTP98063 TP47028 TP7476 TP10402 2ENTP38084 70 TP11943 TP44557 TP11477 2ENTP16310 TP6783 2ENTP73201 75 TP7779 TP39338 2ENTP3929 2ENTP6376 TP11335 1ENTP25142 80 TP1409 TP50574 TP31691 2ENTP2375 TP64602 TP39358 TP55093 2ENTP72087 1ENTP24260 2ENTP8174 85 TP21726 TP22056 2ENTP32630 TP62033 2ENTP7078 TP12450 90 2ENTP93762 TP26313 2ENTP91938 2ENTP73539 TP14473 95 TP37160 TP35050 TP16491 100 TP61413 TP9218 TP20269 TP5887 TP1124 TP23799 TP41897 TP27436 TP2453 TP64645 TP30200 TP54504 TP56507 TP56969

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.10 – PET9 and NEG9 compared to ANN9. See figure C.1 caption for details.

136 PET10 ANN10 NEG10

TP60940 TP16416 TP7705 TP14784 TP22325 TP33756 TP5323 TP14398 TP13088 TP1490 TP6231 TP24467 TP18107 TP44641 TP28194 2ENTP19123 2ENTP94747 TP16662 2ENTP88631 cM TP7194 TP18548 TP11837 2ENTP39775 2ENTP41559 1ENTP22571 1ENTP10098 0 TP8833 TP1411 TP28100 TP51103 2ENTP49810 2ENTP4558 2ENTP73021 TP24169 2ENTP75745 5 TP1562 2ENTP99572 2ENTP6814 2ENTP91611 TP50923 TP3668 TP62153 1ENTP9843 1ENTP25438 10 TP4974 TP40376 2ENTP83718 2ENTP7698 TP55299 TP15746 2ENTP19159 2ENTP1224 2ENTP93874 15 TP47916 TP60485 2ENTP1372 1ENTP5750 1ENTP22711 TP51381 TP56803 TP42716 2ENTP20685 20 TP47776 TP61791 TP59165 2ENTP18500 TP47355 TP38925 TP56792 2ENTP18610 25 TP8990 TP62347 TP59716 2ENTP24600 2ENTP18420 2ENTP32528 TP21977 TP33085 TP26481 1ENTP18505 2ENTP20910 2ENTP12191 1ENTP23335 1ENTP4657 2ENTP90652 30 TP39486 TP28968 TP36128 TP31638 1ENTP21997 2ENTP8446 35 TP38407 TP5750 TP11664 TP40997 2ENTP68096 TP10171 TP10682 2ENTP35569 2ENTP16855 40 TP34905 TP4726 2ENTP79960 2ENTP96015 TP62155 1ENTP1662 2ENTP75985 2ENTP57758 45 TP55745 TP16363 TP40792 2ENTP92691 1ENTP8390 2ENTP13711 TP40790 2ENTP99938 1ENTP14620 50 TP6620 TP35977 1ENTP22290 2ENTP5445 TP56030 TP35756 TP53131 2ENTP10940 1ENTP22401 55 TP26618 TP54915 TP21144 2ENTP100007 2ENTP54230 2ENTP59696 TP10632 TP2406 TP4869 2ENTP1390 2ENTP58857 60 TP56922 TP54913 TP54162 2ENTP53633 2ENTP48642 2ENTP98393 TP12424 TP5706 TP26337 1ENTP166 2ENTP98219 65 TP59471 TP28266 TP1783 TP40384 2ENTP48135 2ENTP25768 2ENTP21428 70 TP44544 TP5189 TP31042 TP13654 TP18703 TP21195 1ENTP12682 TP25631 TP32847 TP54423 2ENTP22079 75 TP44984 1ENTP8148 2ENTP48925 TP21169 TP40045 TP13681 2ENTP90515 2ENTP4189 2ENTP1779 80 TP62506 2ENTP72041 TP16258 2ENTP29244 85 TP49395 TP53351 TP1250 TP34805 90 TP15100 TP18685 TP9187 TP56916 TP56878 95 TP35333 TP6269 TP50152 TP22855 TP49019 TP25135 100 TP27403 TP4012 TP43003 TP6199 TP57232 TP51251 TP22664 TP39406

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.11 - PET10 and NEG10 compared to ANN10. See figure C.1 caption for details.

137 PET11 ANN11 NEG11

TP31358 TP10040 TP42265 TP43117 TP53681 TP57949 TP63822 **TP46589**TP15919** TP22312 TP30674 TP18123 cM TP23553 TP51537 0 TP34182 TP3771 2ENTP77162 2ENTP2006 TP51281 TP50825 2ENTP88582 2ENTP59789 2ENTP6375 5 TP7820 TP51699 TP57457 2ENTP73460 2ENTP71462 2ENTP59268 TP64380 TP32633 2ENTP81609 10 TP39779 TP4577 TP43099 TP45504 15 TP50029 TP33666 TP3658 TP24044 1ENTP15892 1ENTP23923 TP38567 TP56849 TP55205 20 1ENTP18501 TP41820 2ENTP2191 TP4115 TP45038 25 2ENTP87219 TP26213 TP305 TP5426 2ENTP96496 TP36804 TP23775 TP6793 2ENTP54423 30 TP1943 TP6999 2ENTP14223 2ENTP72951 2ENTP50703 TP11619 2ENTP14157 1ENTP24306 2ENTP7201 35 TP22341 2ENTP76228 TP25979 **TP55453** TP20564 2ENTP87821 2ENTP32282 1ENTP6213 40 **TP27098** **TP14351** TP8215 2ENTP72806 TP19469 TP29892 TP54922 2ENTP81691 2ENTP37555 1ENTP5789 45 **TP38559**TP197 TP35403 2ENTP99134 2ENTP35295 TP19696 TP43519 TP14862 2ENTP96526 2ENTP60835 2ENTP23659 50 TP11729 TP41629 TP35368 2ENTP7504 1ENTP4764 2ENTP2755 TP4385 2ENTP77431 2ENTP1405 TP64975 TP63801 TP46889 55 2ENTP83375 1ENTP25194 1ENTP21321 TP1603 TP45725 2ENTP60997 TP15790 60 1ENTP347 TP25430 **2ENTP74945 **2ENTP98684** TP12014 TP48590 TP37582 2ENTP26707 2ENTP38518 65 TP4853 2ENTP7356 TP21159 TP9448 TP3402 2ENTP11230 2ENTP132 70 TP24733 TP36619 TP60476 2ENTP67063 2ENTP17694 TP52102 TP64431 2ENTP30642 1ENTP9791 75 **TP54406**TP16073 **TP21736** 2ENTP63191 2ENTP31869 TP51945 TP65359 TP57885 2ENTP23919 80 TP31091 TP30972 TP8233 TP47228 TP3653 85 TP19731 TP22484 TP65923 TP60933 TP16058 90 TP8805 TP43869 TP47659 95 TP42296 TP54973 TP56590 TP50720 100 TP33925 TP41225 TP5085 TP187 TP22607 TP7186 TP58960

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.12 - PET11 and NEG11 compared to ANN11. See figure C.1 caption for details.

138 PET12/16 ANN12 NEG12/15

TP20908 TP34235 TP39588 TP2019 TP3483 TP18119 TP32679 TP31036 TP4878 TP15026 TP37476 2ENTP4135 TP20867 TP29665 2ENTP90301 TP726 TP1290 TP7595 2ENTP58627 2ENTP31140 TP17807 TP1577 TP60454 2ENTP77037 cM TP7477 2ENTP17572 TP50790 TP65803 TP36139 2ENTP53014 2ENTP87086 2ENTP74147 0 TP31780 TP20303 TP33762 2ENTP51129 2ENTP6879 2ENTP97580 TP58999 TP35726 TP41557 2ENTP20693 2ENTP94228 2ENTP18988 2ENTP99117 2ENTP97010 2ENTP35632 5 TP27713 TP28790 TP26728 TP43250 TP66108 2ENTP52048 2ENTP48834 2ENTP80145 10 TP51731 TP50156 TP21196 TP34871 TP49955 1ENTP1875 2ENTP72125 2ENTP90653 15 TP1637 TP36505 TP25768 2ENTP38410 2ENTP36611 20 TP34243 TP55636 TP59597 2ENTP82190 2ENTP1448 1ENTP5661 TP40120 2ENTP99860 2ENTP669 25 TP52829 TP22510 2ENTP78685 TP21774 TP53945 2ENTP40441 30 TP38790 TP10636 TP50796 2ENTP18938 TP44294 2ENTP22416 2ENTP59265 35 TP20450 TP36894 TP34699 2ENTP100065 2ENTP839 2ENTP32559 TP10121 TP27044 2ENTP85319 2ENTP81632 1ENTP21571 1ENTP21610 40 TP21027 TP15802 TP16524 1ENTP21359 2ENTP12697 2ENTP80984 45 TP18007 TP36259 TP30602 TP24247 1ENTP22020 2ENTP39293 2ENTP81956 2ENTP27276 50 TP30219 TP58699 TP19314 TP27981 TP48309 TP42463 2ENTP78090 1ENTP21811 TP27804 TP62123 TP20262 2ENTP26958 2ENTP83836 2ENTP31470 55 TP32716 TP50786 TP60458 2ENTP32782 1ENTP23846 2ENTP57958 TP49877 TP25152 TP51367 2ENTP87425 60 TP36658 TP11138 2ENTP55554 2ENTP9864 TP22624 TP12312 TP65179 2ENTP9706 65 TP12816 TP26490 TP40994 2ENTP5949 2ENTP82756 2ENTP2107 TP45610 2ENTP5825 70 TP62627 2ENTP67538 TP11465 TP66284 TP33045 2ENTP93966 2ENTP15472 1ENTP18100 75 TP4664 TP31496 TP34642 TP44144 TP48681 TP59510 2ENTP81619 2ENTP18653 80 TP50662 TP37739 TP2964 TP23643 TP63026 1ENTP1337 2ENTP27199 2ENTP24739 1ENTP3979 85 TP3701 TP22586 2ENTP58920 2ENTP28615 2ENTP72809 TP35484 TP54732 TP41205 2ENTP4313 90 TP13629 2ENTP59786 TP62501 2ENTP83990 95 TP53639 TP34112 TP9910 TP9611 100 TP33259 TP47266 TP62083 TP24872 TP8649 TP45812 TP21552 TP20092 TP55688 TP32279 TP14836 TP61596 TP10180

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.13 - PET12/16 and NEG12/15 compared to ANN12. See figure C.1 caption for details.

139 Pet12/16/17 Ann12 Neg16/12

TP49561 TP5182 TP49487 TP48522 TP20392 TP40842 TP38555 TP47959 TP2198 TP58345 TP50628 TP27701 TP25892 TP36560 2ENTP2016 cM TP7116 TP14042 2ENTP77935 TP2786 0 2ENTP82316 TP811 TP50250 2ENTP5918 2ENTP8848 TP43775 TP1938 TP8137 2ENTP3208 5 TP783 2ENTP94959 2ENTP75849 TP47534 TP64057 TP39924 2ENTP25585 10 TP3488 TP21370 TP25986 1ENTP21435 TP104 TP33187 2ENTP85596 2ENTP49061 15 TP45680 TP25354 TP53754 2ENTP32651 1ENTP24340 TP61796 2ENTP18759 20 TP25931 TP23253 2ENTP60007 TP33379 TP21895 TP57897 2ENTP1068 2ENTP36316 25 TP35891 TP30775 TP58638 2ENTP98555 2ENTP83954 1ENTP16235 TP46533 TP56884 2ENTP74031 2ENTP57503 2ENTP21161 30 TP19495 TP44758 TP56228 2ENTP92015 1ENTP1814 TP50447 2ENTP60444 2ENTP97437 2ENTP67648 TP6003 TP49765 TP5814 35 2ENTP13845 2ENTP19834 TP11806 TP59267 TP32904 2ENTP87466 TP30279 TP63916 2ENTP74376 2ENTP14113 40 TP19453 TP39005 TP56251 2ENTP4612 TP56521 2ENTP51866 45 TP34460 2ENTP20592 2ENTP27951 2ENTP4788 TP43430 TP34056 2ENTP95741 50 TP20616 2ENTP5896 2ENTP74184 TP26717 1ENTP6146 55 TP56345 TP50117 TP35840 TP60899 60 TP51752 TP32995 TP10233 TP36158 TP2152 TP60157 TP38202 TP46214 65 TP26615 TP24783 TP48065 TP35495 70 TP18939 TP53995 TP34634 TP57342 75 TP18630 TP28328 TP44817 80 TP31045 TP13193 85 TP9636 TP50573 TP59057 TP60991 90 TP21938 TP25450 TP34383 TP62134 95 TP61965 TP23473 TP43240 100 TP64388 TP36348 TP39608 105 TP37297 TP28217 110 TP11794 TP21236 TP20723 TP54474 TP35987 TP17401 TP4476 TP3906 TP17680 TP45837 TP61580 TP56156

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.14 - PET12/16/17 and NEG16/12 compared to ANN12. See figure C.1 caption for details.

140 PET13 ANN13 NEG13

TP13366 TP27449 TP25511 TP47251 TP46840 TP18657 TP38127 TP10825 TP28149 TP21415 TP54308 TP21367 TP24400 TP25304 TP58233 TP38932 2ENTP12208 TP26755 2ENTP78700 cM TP17183 2ENTP11821 TP3251 TP43475 TP16963 2ENTP21096 0 TP60318 2ENTP83329 TP55148 TP45308 TP59584 2ENTP11358 5 TP12602 TP31324 TP63530 2ENTP24889 TP45011 TP26386 TP54395 2ENTP91560 2ENTP52775 2ENTP12987 10 TP32748 TP58503 TP51886 2ENTP475 1ENTP22980 1ENTP971 TP9947 TP59318 TP15548 2ENTP8108 2ENTP11703 TP51778 2ENTP17952 15 TP15326 TP25113 2ENTP88945 2ENTP81650 2ENTP32080 TP59000 TP30916 TP23399 2ENTP82974 20 TP46281 1ENTP22317 2ENTP88010 2ENTP82373 TP49456 2ENTP1719 2ENTP18510 2ENTP55417 25 TP2731 TP58439 TP12388 2ENTP18659 2ENTP59749 2ENTP2843 TP50677 TP54744 TP3320 1ENTP21983 2ENTP1429 2ENTP78632 30 TP55287 2ENTP33120 2ENTP22791 2ENTP94178 TP3158 TP4141 TP44143 2ENTP92475 35 TP48082 TP5518 TP33532 1ENTP20033 2ENTP77892 2ENTP32948 TP53077 2ENTP16764 2ENTP7039 2ENTP41082 40 TP37550 TP46315 TP31404 2ENTP92871 TP10236 TP34470 TP9619 2ENTP99498 2ENTP55185 2ENTP48373 45 TP61133 TP6349 TP38476 1ENTP5600 2ENTP13390 2ENTP96391 TP18262 2ENTP71982 2ENTP94536 TP24754 2ENTP62426 50 TP13384 2ENTP13242 TP41925 2ENTP90189 55 TP61351 TP2447 2ENTP24566 TP33718 TP9909 TP6437 2ENTP23344 2ENTP87232 60 TP26474 TP39907 TP32876 1ENTP23752 2ENTP16914 1ENTP4092 TP2088 TP24602 TP42199 2ENTP32348 65 TP5205 TP13915 TP34984 2ENTP56705 TP32097 TP20849 TP15977 1ENTP21643 1ENTP22319 2ENTP19775 70 TP46342 TP1359 TP5068 1ENTP22380 TP1393 TP42684 TP20387 1ENTP5209 1ENTP25192 2ENTP90312 75 TP4531 TP34458 TP41805 1ENTP12368 2ENTP27901 1ENTP9927 TP41420 1ENTP3710 2ENTP20562 2ENTP99547 2ENTP55963 2ENTP20607 80 TP29970 TP54781 TP27017 TP12120 TP4306 2ENTP94614 TP2170 TP33167 2ENTP21085 2ENTP3759 2ENTP4592 85 TP65290 2ENTP77371 TP60103 2ENTP26525 90 TP5381 TP33953 TP30262 2ENTP35535 TP49177 95 TP20302 TP23479 TP3975 TP26862 TP56707 100 TP27805 TP5107 TP5839 TP26676 TP50679 TP42806 TP47220 TP40500 TP18826 TP36349 TP44187 TP3949 TP46878 TP57860 TP10759 TP45125 TP50832 TP30059

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.15 - PET13 and NEG13 compared to ANN13. See figure C.1 caption for details.

141 PET14 ANN14 NEG14

TP27593 TP43182 TP47030 TP56164 TP45053 TP50436 TP65749 TP5144 TP26198 TP50631 TP16666 TP17690 TP60953 TP44317 TP15572 1ENTP25403 TP24405 TP16779 TP26462 1ENTP7718 TP10919 TP60651 TP65592 2ENTP77633 TP7411 TP19741 TP18927 1ENTP5568 TP60980 TP24104 TP12340 1ENTP8092 TP12949 2ENTP30844 2ENTP82911 2ENTP7948 TP35006 2ENTP83460 2ENTP7792ENTP88505 TP1097 TP30639 2ENTP32280 2ENTP74468 2ENTP22260 TP40260 TP21343 2ENTP3944 TP41262 TP35232 TP4378 2ENTP92198 2ENTP19137 2ENTP22804 TP44400 2ENTP19861 1ENTP23840 2ENTP95866 TP48303 TP16176 TP9897 2ENTP9032 2ENTP26771 2ENTP36061 TP53524 TP11144 TP61558 2ENTP17546 TP55456 TP11434 TP24233 2ENTP12698 cM TP49091 TP8071 TP50998 2ENTP72202 2ENTP98464 TP18137 TP55015 TP20563 2ENTP84772 0 TP8538 2ENTP7947 TP31729 2ENTP4869 2ENTP52114 5 TP10557 2ENTP15077 TP40128 TP27440 2ENTP19037 1ENTP13566 10 TP56317 TP13853 2ENTP83610 2ENTP80477 2ENTP56442 TP23384 TP31705 TP13219 2ENTP3077 2ENTP96744 1ENTP22512 15 TP26513 TP38255 1ENTP25576 TP3925 TP35803 2ENTP9704 TP63344 TP60821 TP19561 20 1ENTP1401 1ENTP13831 1ENTP16316 TP42892 TP39238 1ENTP25677 1ENTP21033 1ENTP8744 TP12318 TP25387 TP54364 25 2ENTP82963 2ENTP604 2ENTP83864 TP5723 2ENTP48391 TP15729 TP10981 30 2ENTP57859 2ENTP34468 2ENTP89248 TP33715 TP12503 TP33298 1ENTP1838 1ENTP2011 TP12625 TP53989 TP16355 1ENTP14550 35 TP31360 2ENTP94587 TP32565 1ENTP13786 2ENTP99968 2ENTP18589 40 TP355 TP13300 2ENTP74638 1ENTP142 TP56871 TP36761 1ENTP718 2ENTP71124 45 TP4661 TP38831 2ENTP13687 2ENTP87275 TP5206 TP52302 2ENTP17223 2ENTP12757 50 TP30913 2ENTP91514 2ENTP5807 1ENTP20875 TP32360 TP57132 2ENTP12038 2ENTP19618 2ENTP74240 TP10192 TP10076 55 2ENTP83649 TP44760 TP64149 2ENTP96786 TP48507 TP10638 60 1ENTP10006 1ENTP2151 2ENTP49337 TP62602 TP41615 TP18502 2ENTP84488 TP12793 2ENTP27971 2ENTP61635 2ENTP92701 65 TP52184 TP25185 TP33773 2ENTP56455 2ENTP75807 2ENTP95033 TP13528 TP22291 TP37603 2ENTP51373 70 TP23595 TP7504 TP58617 2ENTP14373 TP17252 TP50580 2ENTP25598 2ENTP11483 2ENTP14370 75 TP17029 TP44869 TP57843 2ENTP81478 2ENTP47873 TP35785 TP32834 2ENTP16816 2ENTP13276 2ENTP38942 80 TP12997 TP20673 TP49317 1ENTP9951 TP9637 TP18405 2ENTP82544 2ENTP77277 85 TP21316 TP23870 TP64932 2ENTP59527 2ENTP8240 TP52840 2ENTP34645 1ENTP14843 2ENTP77354 TP35422 TP65057 90 2ENTP2584 2ENTP73802 TP36982 2ENTP2957 TP5459 TP62440 95 2ENTP51152 2ENTP30618 TP22713 TP49523 TP8087 2ENTP10062 2ENTP97136 2ENTP21636 TP16231 TP25422 TP4338 2ENTP88337 100 TP37939 TP42804 TP32697 2ENTP20763 2ENTP78822 1ENTP5885 TP36305 1ENTP5885 2ENTP23548 2ENTP46525 TP53116 TP65537 2ENTP75107 2ENTP56967 TP55314 TP46231 1ENTP13868 TP23858 TP51538 TP49827 TP46727 TP51181 TP32600 TP66007 TP23825 TP22693 TP19733

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.16 - PET14 and NEG14 compared to ANN14. See figure C.1 caption for details.

142 PET15 ANN15 NEG15

TP13327 TP55221 TP63226 TP33571 TP61857 **TP18273** TP15127 **TP8586** TP22234 TP60461 TP2980 TP6007 TP55756 TP48409 TP29656 TP28529 TP48024 TP35383 TP16808 TP16220 TP40349 TP3651 TP29805 TP56824 cM TP4698 TP64545 TP47328 TP31048 2ENTP50568 0 TP61298 TP54007 TP39040 2ENTP61343 TP32315 2ENTP55138 2ENTP16802 2ENTP55137 TP57660 TP42399 1ENTP23250 5 TP13748 TP31707 TP10790 TP51683 TP27589 2ENTP91285 2ENTP19105 10 TP19793 2ENTP57155 2ENTP73064 2ENTP97855 TP28757 TP32704 2ENTP2915 2ENTP39161 15 TP39284 2ENTP80713 TP379 TP39289 2ENTP49332 2ENTP96111 2ENTP40678 20 TP35250 TP26745 TP11442 2ENTP71687 2ENTP87102 TP58370 2ENTP73 25 TP39748 TP12322 TP11697 1ENTP24380 2ENTP83720 2ENTP55113 TP46767 2ENTP16702 2ENTP100047 2ENTP68847 30 TP39974 1ENTP23570 TP44186 2ENTP35231 35 TP27706 2ENTP33112 TP17990 2ENTP1131 2ENTP66280 TP12801 **TP53862**TP30915** 2ENTP86788 2ENTP2017 40 TP21287 TP487 TP10446 2ENTP55908 2ENTP445 TP41092 TP17913 TP63531 2ENTP57598 45 TP38194 TP8263 2ENTP31685 1ENTP25541 2ENTP71994 TP54009 2ENTP77502 50 TP36269 1ENTP21128 1ENTP21861 TP44558 TP60051 2ENTP89900 55 TP33004 2ENTP59614 TP37553 TP31154 TP48114 1ENTP22531 60 TP33002 TP65423 TP20948 1ENTP15882 **2ENTP71979** 1ENTP2119 TP20162 2ENTP60917 2ENTP58433 **2ENTP23468** 65 TP61026 1ENTP2888 2ENTP37952 **TP63052**TP21142** 1ENTP25605 70 TP55238 TP30188 1ENTP15548 1ENTP13252 TP47645 TP33845 TP54414 TP51940 75 TP24466 TP60070 TP8713 TP14947 80 TP29522 TP40443 TP48870 85 TP2195 TP2208 TP49777 1ENTP22535 TP500 TP31308 TP62626 90 TP62597 TP25776 95 TP54117 TP65374 TP46486 100 TP19245 TP18203 TP22099 TP60751 TP18403 TP36062 TP32784 TP15784 TP12751 TP38395 TP62665 TP10610 TP15256 TP10282 TP22457 TP22539 TP37063 TP43101 TP47492 TP48939 TP21138

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.17 - PET15 and NEG15 compared to ANN15. See figure C.1 caption for details.

143 Pet12/16 Ann16 Neg6/16

TP20908 TP34235 TP39588 TP2019 TP3483 TP18119 TP32679 TP31036 TP4878 TP15026 TP37476 TP20867 TP29665 TP726 TP1290 TP7595 TP17807 TP1577 TP60454 cM TP7477 0 TP50790 TP65803 TP36139 TP31780 TP20303 TP33762 1ENTP8365 5 TP58999 TP35726 TP41557 2ENTP80094 2ENTP56286 2ENTP84594 TP27713 TP28790 2ENTP1858 TP26728 TP43250 TP66108 10 TP51731 TP50156 2ENTP55435 TP21196 TP34871 TP49955 2ENTP37523 15 TP1637 TP36505 2ENTP24877 TP25768 20 2ENTP56382 2ENTP5177 TP34243 2ENTP16537 TP55636 TP59597 2ENTP84691 2ENTP24235 25 TP40120 2ENTP6332 2ENTP25048 TP52829 TP22510 2ENTP15093 30 TP21774 TP53945 2ENTP96813 2ENTP1326 TP38790 TP10636 TP50796 2ENTP17238 2ENTP78250 35 TP44294 2ENTP14487 2ENTP83266 TP20450 TP36894 TP34699 2ENTP39131 TP10121 TP27044 40 2ENTP9518 TP21027 2ENTP5646 2ENTP14150 2ENTP16502 TP15802 TP16524 45 2ENTP85055 2ENTP39168 TP18007 TP36259 TP30602 1ENTP25596 TP24247 50 1ENTP20025 2ENTP55485 1ENTP12516 TP30219 TP58699 TP19314 2ENTP87217 TP27981 TP48309 TP42463 55 2ENTP44459 1ENTP13875 2ENTP22484 TP27804 TP62123 TP20262 2ENTP89277 2ENTP18979 TP32716 TP50786 TP60458 1ENTP13876 60 TP49877 TP25152 TP51367 TP36658 TP11138 65 TP22624 TP12312 TP65179 TP12816 TP26490 TP40994 70 TP45610 TP62627 TP11465 TP66284 TP33045 75 2ENTP48544 TP4664 TP31496 TP34642 2ENTP72444 2ENTP38363 TP44144 TP48681 TP59510 80 2ENTP12082 TP50662 TP37739 TP2964 2ENTP72881 2ENTP48731 TP23643 TP63026 85 2ENTP48917 2ENTP4282 2ENTP32026 TP3701 2ENTP14957 TP22586 2ENTP72994 90 TP35484 TP54732 TP41205 2ENTP95371 TP13629 1ENTP24490 95 TP62501 1ENTP6563 TP53639 TP34112 TP9910 2ENTP100122 100 TP9611 1ENTP1177 TP33259 1ENTP1982 105 TP47266 2ENTP62467 TP62083 2ENTP1643 1ENTP9838 TP24872 TP8649 TP45812 110 2ENTP32107 2ENTP95757 2ENTP34592 TP21552 TP20092 2ENTP34360 2ENTP8200 2ENTP28946 TP55688 2ENTP99940 2ENTP84042 1ENTP13846 TP32279 TP14836 2ENTP59342 2ENTP10702 TP61596 2ENTP88877 1ENTP5006 TP10180

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.18 - PET12/16 and NEG6/16 compared to ANN16. See figure C.1 caption for details.

144 PET16/12/17 ANN16 NEG16/12

TP49561 TP5182 TP49487 TP48522 TP20392 TP40842 TP38555 TP47959 TP2198 TP58345 TP50628 TP27701 TP25892 TP36560 2ENTP2016 cM TP7116 TP14042 2ENTP77935 TP2786 0 2ENTP82316 TP811 TP50250 2ENTP5918 2ENTP8848 TP43775 TP1938 TP8137 2ENTP3208 5 TP783 2ENTP94959 2ENTP75849 TP47534 TP64057 TP39924 2ENTP25585 10 **TP3488**TP21370**TP25986 1ENTP21435 TP104 TP33187 2ENTP85596 2ENTP49061 15 TP45680 TP25354 TP53754 2ENTP32651 1ENTP24340 TP61796 2ENTP18759 20 TP25931 TP23253 2ENTP60007 TP33379 TP21895 TP57897 2ENTP1068 2ENTP36316 25 TP35891 TP30775 TP58638 2ENTP98555 2ENTP83954 1ENTP16235 TP49856 TP46533 TP56884 2ENTP74031 2ENTP57503 2ENTP21161 30 TP19495 TP44758 TP56228 2ENTP92015 1ENTP1814 TP50447 2ENTP60444 2ENTP97437 2ENTP67648 TP6003 TP49765 TP5814 35 2ENTP13845 2ENTP19834 TP11806 TP59267 TP32904 2ENTP87466 **TP30279**TP63916** 2ENTP74376 2ENTP14113 40 TP19453 TP39005 TP56251 2ENTP4612 TP56521 2ENTP51866 45 TP34460 2ENTP20592 2ENTP27951 2ENTP4788 TP43430 TP34056 2ENTP95741 50 TP20616 2ENTP5896 2ENTP74184 TP26717 1ENTP6146 55 TP56345 TP50117 TP35840 TP60899 60 TP51752 TP32995 TP10233 TP36158 TP2152 TP60157 65 TP38202 TP46214 TP26615 TP24783 TP48065 TP35495 70 TP18939 TP53995 TP34634 TP57342 75 TP18630 TP3512 TP28328 TP44817 80 TP31045 TP13193 85 TP9636 TP50573 TP59057 TP60991 90 TP21938 TP25450 TP34383 TP62134 95 TP61965 TP23473 TP43240 100 TP64388 TP36348 TP39608 TP37297 TP28217 TP11794 TP21236 TP20723 TP54474 TP35987 TP17401 TP4476 TP3906 TP17680 TP45837 TP61580 TP56156

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.19 - PET16/12/17 and NEG16/12 compared to ANN16. See figure C.1 caption for details.

145 PET17 ANN17 NEG17/16/12

2ENTP19376 2ENTP74608 2ENTP73287 2ENTP10948 2ENTP70110 TP8710 2ENTP25156 2ENTP88170 TP65911 TP9169 2ENTP51170 TP64842 TP47796 1ENTP4170 2ENTP1271 2ENTP4295 **TP31007**TP7703** 2ENTP82960 2ENTP48362 TP50667 2ENTP19337 TP14795 TP55904 TP46434 2ENTP99012 2ENTP16909 2ENTP46293 TP10079 cM 1ENTP24251 TP19053 2ENTP91098 2ENTP12366 TP1470 0 2ENTP58946 TP21927 TP20952 TP28055 2ENTP23414 TP30987 TP4192 TP63319 5 2ENTP55080 TP58778 TP26180 TP19994 2ENTP100035 2ENTP78107 2ENTP9760 TP61394 10 2ENTP88859 TP32938 TP20567 TP10161 2ENTP15334 2ENTP93015 2ENTP15106 TP11238 TP13198 TP26176 2ENTP57361 15 TP26907 TP28259 TP37645 1ENTP22963 TP44026 TP45197 TP46222 2ENTP80969 2ENTP82878 20 TP46326 TP50608 TP55644 2ENTP23140 TP56078 TP57677 TP59127 1ENTP25718 2ENTP77151 25 TP62777 TP64370 TP65780 1ENTP11116 1ENTP22616 2ENTP53395 TP20687 TP31250 TP50480 1ENTP25769 30 TP54525 TP64579 2ENTP19545 2ENTP79604 TP8533 2ENTP94605 1ENTP792 2ENTP93056 TP44138 TP40967 35 2ENTP15776 TP58043 2ENTP15918 2ENTP12149 TP5738 40 2ENTP98266 2ENTP10730 2ENTP32574 TP13712 2ENTP91519 TP13860 TP10323 45 2ENTP9278 1ENTP7046 2ENTP58766 TP16110 1ENTP8145 2ENTP97054 1ENTP9552 TP59387 2ENTP12427 50 TP52994 2ENTP77623 TP31323 2ENTP10820 2ENTP89182 2ENTP40670 55 TP3983 TP43527 2ENTP38312 2ENTP17470 2ENTP93341 TP10616 TP19682 2ENTP90545 60 1ENTP5257 2ENTP69507 1ENTP4979 2ENTP51694 2ENTP91505 1ENTP23328 65 2ENTP14829 2ENTP59356 2ENTP44801 2ENTP35724 70 2ENTP98519 2ENTP88732 1ENTP23327 75 2ENTP82727

80

85

90

95

100

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.20 - PET17 and NEG17/16/12 compared to ANN17. See figure C.1 caption for details.

146 Neg12/15 Ann15 Pet12/16/17 Ann17

TP49561 TP5182 TP49487 TP48522 TP20392 TP40842 2ENTP4135 TP38555 2ENTP90301 TP47959 2ENTP58627 2ENTP31140 TP2198 TP58345 2ENTP77037 TP50628 2ENTP17572 TP27701 TP25892 TP36560 cM 2ENTP53014 2ENTP87086 2ENTP74147 TP7116 TP14042 0 2ENTP51129 2ENTP6879 2ENTP97580 TP2786 2ENTP20693 2ENTP94228 2ENTP18988 TP811 TP50250 2ENTP99117 2ENTP97010 2ENTP35632 TP43775 TP1938 TP8137 5 2ENTP52048 TP783 2ENTP48834 2ENTP80145 TP47534 TP64057 TP39924 10 1ENTP1875 TP3488 TP21370 TP25986 2ENTP72125 2ENTP90653 TP104 TP33187 15 2ENTP38410 TP45680 TP25354 TP53754 2ENTP36611 TP61796 20 2ENTP82190 2ENTP1448 1ENTP5661 TP25931 TP23253 2ENTP99860 2ENTP669 TP33379 TP21895 TP57897 25 2ENTP78685 TP35891 TP30775 TP58638 2ENTP40441 TP49856 TP46533 TP56884 30 2ENTP18938 TP19495 TP44758 TP56228 2ENTP22416 2ENTP59265 TP50447 35 2ENTP100065 2ENTP839 2ENTP32559 TP6003 TP49765 TP5814 2ENTP85319 2ENTP81632 1ENTP21571 TP11806 TP59267 TP32904 1ENTP21610 TP30279 TP63916 40 1ENTP21359 2ENTP12697 TP19453 TP39005 TP56251 2ENTP80984 TP56521 45 1ENTP22020 TP34460 2ENTP39293 2ENTP81956 TP43430 TP34056 50 2ENTP78090 1ENTP21811 TP20616 2ENTP26958 2ENTP83836 2ENTP31470 TP26717 55 2ENTP32782 1ENTP23846 2ENTP57958 TP56345 TP50117 2ENTP87425 TP35840 TP60899 60 2ENTP55554 2ENTP9864 TP51752 TP32995 TP10233 2ENTP9706 TP36158 TP2152 TP60157 65 2ENTP5949 2ENTP82756 2ENTP2107 TP38202 TP46214 2ENTP5825 TP26615 TP24783 TP48065 TP35495 70 2ENTP67538 2ENTP93966 2ENTP15472 TP18939 TP53995 TP34634 1ENTP18100 TP57342 75 2ENTP81619 TP18630 2ENTP18653 TP3512 TP28328 TP44817 80 1ENTP1337 2ENTP27199 2ENTP24739 TP31045 1ENTP3979 TP13193 85 2ENTP58920 2ENTP28615 2ENTP72809 TP9636 TP50573 2ENTP4313 TP59057 TP60991 90 2ENTP59786 TP21938 TP25450 2ENTP83990 TP34383 TP62134 95 TP61965 TP23473 100 TP43240 TP64388 TP36348 TP39608 105 TP37297 TP28217 110 TP11794 TP21236 TP20723 TP54474 TP35987 TP17401 TP4476 TP3906 TP17680 TP45837 TP61580 TP56156

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.21 - NEG12/15 compared to ANN15 and PET12/16/17 compared to ANN17. See figure C.1 caption for details.

147 Neg17/16/12 Ann16 Ann12

2ENTP19376 2ENTP74608 2ENTP73287 2ENTP10948 2ENTP70110 2ENTP25156 2ENTP88170 2ENTP51170 1ENTP4170 2ENTP1271 2ENTP4295 2ENTP82960 2ENTP48362 2ENTP19337 2ENTP99012 2ENTP16909 2ENTP46293 1ENTP24251 cM 2ENTP91098 2ENTP12366 0 2ENTP58946 2ENTP23414 5 2ENTP55080 2ENTP100035 2ENTP78107 2ENTP9760 10 2ENTP88859 2ENTP15334 2ENTP93015 2ENTP15106 2ENTP57361 15 1ENTP22963 2ENTP80969 2ENTP82878 20 2ENTP23140 1ENTP25718 2ENTP77151 25 1ENTP11116 1ENTP22616 2ENTP53395 1ENTP25769 30 2ENTP19545 2ENTP79604 2ENTP94605 1ENTP792 2ENTP93056 35 2ENTP15776 2ENTP15918 2ENTP12149 40 2ENTP98266 2ENTP10730 2ENTP32574 2ENTP91519 45 2ENTP9278 1ENTP7046 2ENTP58766 1ENTP8145 2ENTP97054 1ENTP9552 2ENTP12427 50 2ENTP77623 2ENTP10820 2ENTP89182 2ENTP40670 55 2ENTP38312 2ENTP17470 2ENTP93341 2ENTP90545 60 1ENTP5257 2ENTP69507 1ENTP4979 2ENTP51694 2ENTP91505 1ENTP23328 65 2ENTP14829 2ENTP59356 2ENTP44801 2ENTP35724 70 2ENTP98519 2ENTP88732 1ENTP23327 75 2ENTP82727

80

85

90

95

100

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Figure C.22 – Neg17/16/12 compared to ANN16 and ANN12. See figure C.1 caption for details.

148 Appendix D - Chapter 4 supplementary tables

Table D.1 Description of the populations used in chapter 4.

Experiment Species Population Ecotype Latitude Longitude State Wild seed H. neglectus MON001 dune 31.63151 -102.80999 TX Wild seed H. neglectus MON050 dune 31.63679 -102.81361 TX Wild seed H. neglectus MON100 dune 31.64052 -102.81797 TX Wild seed H. neglectus MON150 dune 31.64075 -102.81345 TX Wild seed H. neglectus MON200 dune 31.62984 -102.81567 TX Wild seed H. neglectus MON250 non-dune 31.61642 -102.81197 TX Wild seed H. neglectus MON300 non-dune 32.3204 -103.82318 TX Wild seed H. neglectus MON350 non-dune 32.21017 -103.58173 TX Wild seed H. neglectus MON400 non-dune 32.17832 -103.38889 TX Wild seed H. neglectus MON450 non-dune 32.0889 -103.17894 TX Wild seed H. neglectus MON500 dune 32.03659 -103.15085 TX Wild seed H. neglectus MON550 dune 32.03527 -103.15001 TX Wild seed H. neglectus MON600 non-dune 31.99352 -103.13029 TX Wild seed H. neglectus MON650 non-dune 31.82584 -103.07813 TX Wild seed H. neglectus MON700 non-dune 31.64043 -102.98539 TX Wild seed H. neglectus MON750 non-dune 31.58296 -102.87074 TX Wild seed H. neglectus MON800 non-dune 31.61138 -102.83128 TX Wild seed H. neglectus MON850 dune 31.62488 -102.81174 TX CG H. neglectus 1305 non-dune 31.595 -102.89167 TX CG & QTL H. neglectus KING 153 dune 31.6475 -102.82306 TX CG H. neglectus KING 154 dune 31.6475 -102.82306 TX CG & QTL H. neglectus PI468777 non-dune 32.285 -104.09167 NM CG & QTL H. neglectus PI468781 non-dune 32.773333 -108.28 NM QTL H. petiolaris 1701 dune 37.803 -105.524 CO QTL H. petiolaris 1547 dune 37.761 -105.570 CO QTL H. petiolaris 1791 non-dune 37.813 -105.515 CO QTL H. petiolaris 1500 non-dune 37.765 -105.615 CO

149