Detecting Hybridization in Hawaiian Cyrtandra Using

DETECTING HYBRIDIZATION IN HAWAIIAN CYRTANDRA USING

GENOME-WIDE DATA

JOSEPH A. KLEINKOPF

A thesis submitted in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE IN PLANT BIOLOGY

WASHINGTON STATE UNIVERSITY School of Biological Sciences

MAY 2018

To the Faculty of Washington State University:

The members of the Committee appointed to examine thesis of JOSEPH A.

KLEINKOPF find it satisfactory and recommend that it be accepted.

Eric H. Roalson, Ph.D., Chair

Joanna L. Kelley, Ph.D.

Larry D. Hufford, Ph.D.

Warren L. Wagner, Ph.D.

ACKNOWLEDGMENTS

I would like to first thank my advisor, Eric Roalson. He introduced me to Cyrtandra, a fascinating group of flowering plants with questions aligned to my interests, while encouraging me to identify and pursue my goals beyond graduate school. Secondly, I would like to express my sincerest gratitude to my committee members, Joanna Kelley, Larry Hufford, and Warren

Wagner for their support, feedback, and time despite their busy schedules.

Additionally, I would like to thank others who helped me along the way. Thank you to

Corey Quackenbush for showing me best practices for library preparation and members and associates of the Roalson lab - Lucy Allison, Wade Roberts, Kimberly Hansen, Josh Brindley,

Luciano Soares, and Nan Jiang - who helped make every day of graduate school interesting and enjoyable. I am especially grateful to Wade for his guidance and input throughout the research process.

Finally, I would like to thank my friends and family, whose enduring support encouraged me to complete this study.

iii

DETECTING HYBRIDIZATION IN HAWAIIAN CYRTANDRA USING

GENOME-WIDE DATA

Abstract

by Joseph A. Kleinkopf, M.S. Washington State University May 2018

Chair: Eric H. Roalson

Cyrtandra (Gesneriaceae) is a genus of flowering plants with over 800 species distributed throughout Southeast Asia and the Pacific Islands. On the Hawaiian Islands, 60 named species and over 62 putative hybrids exist, most of which are identified on the basis of morphology.

Despite many previous studies on the Hawaiian lineage of Cyrtandra, questions regarding the reconciliation of morphology and genetics remain, many of which can be attributed to the relatively young age and evidence of hybridization between species. We utilized targeted enrichment, high-throughput sequencing, and modern phylogenomics tools to test 33 Hawaiian

Cyrtandra samples (including 21 species, five putative hybrids, and two outgroups) for species relationships and hybridization in the presence of incomplete lineage sorting (ILS). Both concatenated and species-tree methods were used to reconstruct species relationships, and network analyses were conducted to test for hybridization. Based on prior studies of Cyrtandra and other Hawaiian lineages, we expected to see both a classic stepping-stone model, where species relationships group by island from oldest to youngest, and species relationships grouping by morphology. Additionally, we expected to see high levels of ILS and putative hybrids intermediate to their parent species. Phylogenies reconstructed from the concatenated and

species-tree methods were highly incongruent, most likely due to high levels of incomplete lineage sorting. Network analyses inferred gene flow within this lineage, but not always between taxa that we expected. Of the five putative hybrids tested, only one was inferred to incorporate genetic material from both putative parents. The genomic methods utilized here have allowed us to reconstruct better resolved hypotheses and to test for hybridization in the presence of incomplete lineage sorting. Wider sampling within this lineage will allow us to better understand species relationships and gene flow within Hawaiian Cyrtandra in future studies.

TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS ...... iii

ABSTRACT ...... iv

LIST OF APPENDICES ...... viii

LIST OF TABLES ...... ix

LIST OF FIGURES ...... x

CHAPTER

ISLAND SYSTEMS AS NATURAL LABORATORIES FOR INFERENCES OF ECOLOGICAL AND EVOLUTIONARY PROCESSES...... 1

Introduction ...... 1

Formation of Island Biota ...... 2

The Progression Rule and Adaptive Radiation ...... 3

Biodiversity and Conservation of Plants...... 5

The Hawaiian Islands...... 6

Literature Cited ...... 8

DETECTING HYBRIDIZATION IN HAWAIIAN CYRTANDRA USING GENOME- WIDE DATA ...... 17

Introduction ...... 17

Materials and Methods ...... 23

Results ...... 27

Discussion ...... 33

Conclusions ...... 44

Literature Cited ...... 46

SUMMARY OF RESULTS AND FUTURE DIRECTIONS ...... 91

Literature Cited ...... 94

vii

APPENDICES

Page

Appendix 1: Sampled specimens ...... 58

viii

TABLES

Page

Table 1: Summary of targeted enrichment sequencing and mapped reads ...... 60

FIGURES

Page

Figure 1: Concatenated maximum likelihood tree ...... 61

Figure 2: ASTRAL tree for all taxa ...... 62

Figure 3: ASTRAL tree for all taxa with quartet scores ...... 63

Figure 4: ASTRID tree for all taxa ...... 64

Figure 5: ASTRAL tree for the core dataset ...... 65

Figure 6: ASTRAL tree for the core dataset with C. calpidicarpa x hawaiensis ...... 66

Figure 7: ASTRAL tree for the core dataset with C. grayi-1 ...... 67

Figure 8: ASTRAL tree for the core dataset with C. hawaiensis-3 ...... 68

Figure 9: ASTRAL tree for the core dataset with C. hawaiensis x grayana ...... 69

Figure 10: ASTRAL tree for the core dataset with C. longifolia x kauaiensis ...... 70

Figure 11: ASTRAL tree for the core dataset with C. munroi x grayi ...... 71

Figure 12: ASTRAL tree for the core dataset with C. sandwicensis x grandiflora ...... 72

Figure 13: ASTRID tree for the core dataset...... 73

Figure 14: ASTRID tree for the core dataset with C. calpidicarpa x hawaiensis ...... 74

Figure 15: ASTRID tree for the core dataset with C. grayi-1 ...... 75

Figure 16: ASTRID tree for the core dataset with C. hawaiensis-3 ...... 76

Figure 17: ASTRID tree for the core dataset with C. hawaiensis x grayana ...... 77

Figure 18: ASTRID tree for the core dataset with C. longifolia x kauaiensis...... 78

Figure 19: ASTRID tree for the core dataset with C. munroi x grayi...... 79

Figure 20: ASTRID tree for the core dataset with C. sandwicensis x grandiflora ...... 80

Figure 21: Model selection using negative pseudolikelihood scores...... 81

Figure 22: Introgression model for all taxa ...... 82

Figure 23: Introgression model for the core dataset ...... 83

Figure 24: Introgression model for the core dataset with C. calpidicarpa x hawaiensis ...... 84

Figure 25: Introgression model for the core dataset with C. grayi-1 ...... 85

Figure 26: Introgression model for the core dataset with C. hawaiensis-3 ...... 86

Figure 27: Introgression model for the core dataset with C. hawaiensis x grayana...... 87

Figure 28: Introgression model for the core dataset with C. longifolia x kauaiensis ...... 88

Figure 29: Introgression model for the core dataset with C. munroi x grayi ...... 89

Figure 30: Introgression model for the core dataset with C. sandwicensis x grandiflora ...... 90

CHAPTER ONE

ISLAND SYSTEMS AS NATURAL LABORATORIES FOR INFERENCES OF

ECOLOGICAL AND EVOLUTIONARY PROCESSES

Introduction – Ever since Charles Darwin’s (1859) first observations of species diversity on the Galapagos Islands, oceanic island systems have played a key role in the study of evolutionary and ecological processes. Throughout the 20th century, biologists such as Eugene

Gordon Munroe (1948), Edward O. Wilson (1959), and Frank W. Preston (1962a, 1962b) studied and contributed to the small but growing body of work on island biogeography, but it was not until Robert H. MacArthur and E. O. Wilson published their seminal works on the island biogeography theory (1963, 1967) that island lineages gained acknowledgement as powerful models to understand the processes of ecology and evolution. MacArthur and Wilson developed a general theory of biogeography to explain community assembly dynamics through fundamental processes such as immigration and extinction, and how geographical context can predict the rates of those processes. Ever since, ecologists and evolutionary biologists have been attracted to the study of island lineages for their endemic species and their value as “natural laboratories,” where processes of community assembly and evolution can be observed and studied in comparable and replicated conditions across the globe (Kueffer et al. 2014; Warren et al. 2015). More recently, biologists have started studying ecology and evolution in tandem to understand the evolution of community assembly, where the large majority of species in an ecological community of an island system arise from in situ speciation (Heaney 2007; Rominger et al. 2016). Today, phylogeographic tools are used to study the processes of community assembly and evolution that are thought to be important to creating the diversity of life we see today.

Formation of Island Biota – Prior to Darwin and Wallace’s work on the theory of evolution, differences in community assembly between different island systems were explained by historical patterns of species distribution from where they originated. Darwin’s The Origin of

Species (Darwin 1859) gave examples of island lineages throughout the world that conflicted with these ideas. Unsurprisingly, the community assembly of islands is reflected in the way island systems arise. The isolation of a section of land from an existing mainland results in fragment islands, which begin with nearly the same biota as the landmass from which it was derived, but begin to change soon after formation as extinction rates are expected to be greater than colonization rates for these islands (Watson 2002). De novo island systems, on the other hand, are created from the movement of tectonic plates over a volcanic hotspot, often resulting in an island chain that progresses by age (Wilson 1963; Carson and Clague 1995; Clouard and

Bonneville 2004). These islands tend to have smaller community assemblies as they are extremely isolated and start without life, and species number only rises as a result of the rate of colonization and diversification (Gillespie et al. 2012). While there are similarities between the two island types, there are fundamental differences in how their biotas were formed. Because lineages on de novo island systems have been studied extensively and lineages are introduced over time, these islands are ideal locations to study how in situ speciation gives rise to ecological communities (MacArthur and Wilson 1967; Warren et al. 2015).

Compared to continental land masses, de novo island systems are relatively simple settings often characterized by small total areas, relatively young ages, and isolation beyond the range of dispersal for most organisms (Fleischer et al. 1998; Price and Clague 2002; Price and

Wagner 2004; Rosindell and Phillimore 2011). Because islands (especially remote islands) are small specks of land in vast oceans, colonization is improbable. Additionally, Simpson (1940)

hypothesized that with increasing distance from a continental mainland, the probability of island colonization should decline even further. However, the probability of colonization is dependent on additional factors such as mode of dispersal and wind or ocean currents. The simplicity of island systems allows for the use of phylogenetic methods to infer the effects that distance and direction of the source pool have on the community assembly of islands (Ronquist and Sanmartín

2011). Studies conforming to predictions show how Hawaiian taxa with aerial dispersal abilities predominantly colonize islands from the east in the direction of large storms, while organisms capable of rafting colonize the islands from the southwest (Gillespie et al. 2012). Other studies, however, have found patterns of less appreciated dispersal methods that contradict expectations, such as how extreme, long-distance dispersal events are more probable than consecutive shorter dispersal events for some taxa (Higgins et al. 2003; Gillespie et al. 2008; Nathan et al. 2008).

The Progression Rule and Adaptive Radiation – De novo oceanic hot-spot island systems tend to be archipelagos of age-sorted islands, with lineages often exhibiting a pattern of progression from oldest to youngest. The hypothesis that older lineages tend to map to older islands while younger lineages map to younger islands in an archipelago (Funk and Wagner

1995) has been described as the progression rule (Hennig 1966). Older lineages are inferred to have colonized the oldest islands after they were created but prior to the emergence of the younger islands, and younger lineages are a consequence of subsequent colonizations occurring as new islands form. Progression is thus apparent in the phylogenies for island lineages, where lineages from younger islands form clades derived from the lineages of older islands. Because the ages of volcanic islands can be accurately estimated (Carracedo et al. 2011), divergence times of these lineages can also be estimated with accuracy, making island systems useful for understanding the timing of colonization and ecological and evolutionary processes. While the

simple model of progression cannot explain the diversity of all organisms found on all island systems, the general model is still apparent to varying degrees on many age-sorted archipelagos including the Galapagos Islands (Parent and Crespi 2006), Canary Islands (Juan et al. 2000),

Azores Islands (Díaz-Pérez et al. 2008; Rumeu et al. 2011), and Hawaiian Islands (Baldwin and

Sanderson 1998), among others.

Island systems may be most famous for their displays of adaptive radiation, where lineages multiply rapidly and evolve a multitude of ecological and phenotypic characters

(Schluter 2000; Losos and Ricklefs 2009). Schluter (2000) defines an adaptive radiation as a lineage consisting of organisms that 1) share a single common ancestor, 2) exhibit phenotype- environment correlation, 3) acquire adaptive traits that have performance or fitness advantages, and 4) display rapid speciation. Although the underlying mechanisms of diversification within species radiations remains poorly understood (Yoder et al. 2010), it is thought that the functional diversity of all life today may have arisen in adaptive radiation events (Seehausen 2004). Though not limited to island systems (Hodges and Arnold 1994), adaptive radiation is most often seen on islands thanks to their insular nature and variety of open habitats. In many cases, the diversity of an island lineage is often the result of a single colonist ancestor which differentiated into an array of species occupying different niches, and which exhibit various morphological character traits used to exploit their environments. A hybrid origin for many radiations is highly plausible, as hybrids may have elevated genetic variability and phenotypic evolution (Seehausen 2004; Yang and Berry 2011). Famous examples of adaptive radiation we can see today include Darwin’s finches (Schluter et al. 1997; Petren et al. 1999), where finch species on the Galapagos Islands exhibit differences in beak morphology that correspond to different diet types, and the Hawaiian

silversword alliance (Robichaux et al. 1990; Baldwin and Sanderson 1998), where leaf morphology and physiology correlate with habitat types.

Because many island lineages diversify quickly over short periods of time, hybridization is rampant within many young island lineages confounding species relationships (Francisco-

Ortega et al., 2000; Schluter, 2000; Kim et al., 2007). It is estimated that up to 10% of all animal species and 25% of all plant species hybridize to some extent (Mallet 2005), and many of these species inhabit island systems. Inferring highly resolved species relationships using Sanger sequencing has proven to be difficult in these lineages, as many species hybridize, leading to gene regions with conflicting histories, and because some species have not had sufficient time to evolve distinct molecular signals (Grant and Grant 1992; Sang 2002; Cronk et al. 2005; Shaw and Gillespie 2016). Until recently, many putative hybrid taxa were identified primarily by morphological traits, where exhibited traits were intermediate to the putative parent species.

However, recent advances in sequencing technology, computational efficiency, and lab methods have made phylogeographic and hybridization studies using molecular data from across the genome a possibility, and have allowed us to infer evidence of hybridization, progression, or other complex processes such as back-migrations within a lineage (Sang 2002; Howarth and

Baum 2005; Pillon et al. 2013a; Seehausen et al. 2014; Weitemier et al. 2014; Solís-Lemus and

Ané 2016; Shaw and Gillespie 2016). Further advancements in phylogeographic methods are sure to yield fascinating insights into the processes of community assembly and evolution on island systems.

Biodiversity and Conservation of Plants – In addition to allowing us to study ecological and evolutionary processes, island systems are important to biologists because of the biodiversity they harbor. Though ocean archipelagos represent only about 5% of the terrestrial surface of the

Earth, they harbor nearly 25% of all described species of vascular plants (Kreft et al. 2008) including approximately 70,000 endemic plant species, of which up to 6,800 may be highly threatened and up to 2,800 may be in critical danger of extinction (Caujapé-Castells et al. 2010).

Humans in particular have heavily affected island ecosystems through urbanization, crop and pasture development, and the introduction of invasive species (Francisco-Ortega et al. 2000; Kier et al. 2009; Kingsford et al. 2009; Kueffer et al. 2010; Graham et al. 2017). Though it is difficult to tell how continued human activity on island systems will influence the trajectories of insular lineages, the continued study of these lineages remains necessary for understanding and preventing loss of biodiversity, especially in the face of climate change and development of land as the human population continues to grow (Kueffer et al. 2010, 2014; Graham et al. 2017).

Perhaps most importantly for the local peoples of island systems, conservation of island lineages remains important to island economies, as many island states and nations depend on their natural attractions, particularly through tourism and subsistence farming (Connell 2013).

The Hawaiian Islands – Among the most well studied island systems in the world, the

Hawaiian archipelago is an exemplary site for studying ecological and evolutionary processes.

Hawaii has a unique geologic history, small total area, and is highly isolated nearly 4000 km from the nearest landmass, and the accessibility and plethora of environmental data available for the archipelago makes Hawaii one of the best settings to understand biological processes over time and in relation to other environmental variables. The oldest island (Kaua’i) was formed around 4.7 million years ago (Mya) and is situated to the northwest, while the youngest

(Hawai’i) was formed approximately 0.5 Mya and is situated to the southeast (Price and Clague

2002). Northeast of the main high islands of Hawaii lie the Emperor Seamounts, former high islands that are now eroded atolls which formed between 7 and 28 Mya (Price and Clague 2002).

Most of the Hawaiian flora and fauna are likely to have evolved on the main high islands within the last 5 My due to a hiatus of high islands being formed, though there are some lineages that are thought to have colonized the now eroded atolls when they were higher islands prior to colonizing the current main Hawaiian Islands (Price and Clague 2002; Givnish et al. 2009).

While most Hawaiian groups contain a single species, most of the diversity in Hawaii is a result of adaptive radiation in just few lineages (Price 2004). Hawaiian lineages are inferred to have origins from all over the world, including North America (Baldwin et al. 1991, Baldwin and

Wagner 2010), the Pacific Islands (Johnson et al. 2017), and Africa (Kim et al. 1998). Over the years, a number of phylogeographic studies on both Hawaiian plants and animals have been conducted and it is clear that evolution on the Hawaiian Islands has occurred recently and natively (Funk and Wagner 1995). Among angiosperms, approximately 270 natural introductions are inferred to have given rise to over 1000 species, of which nearly 90% are endemic to the

Hawaiian Islands and over 50% are endemic to a single island (Wagner et al. 1990; Price 2004).

In Hawaiian plant lineages, morphological, physiological, and ecological diversification is apparent across many endemic lineages, but genetic change is not. As evidence of this, many

Hawaiian lineages hybridize readily, which may be because species are relatively young, or because many species within a lineage are not sympatric (Carlquist 1995), and therefore have not undergone selection for prezygotic isolation mechanisms. Furthermore, the lack of genetic change in some Hawaiian lineages has made phylogenetic studies using Sanger sequencing data difficult, as gene regions may have conflicting histories (Pillon et al. 2013b). High-throughput sequencing and modern analytical approaches are clearly the best way to further our understanding the underlying processes of ecological assembly and evolution, not only on the

Hawaiian Islands, but on other island systems as well.

LITERATURE CITED

Baldwin B.G., Kyhos D.W., Dvorak J., Carr G.D. 1991. Chloroplast DNA evidence for a North

American origin of the Hawaiian silversword alliance (Asteraceae). Proceedings of the

National Academy of Sciences. 88:1840–1843.

Baldwin B.G., Sanderson M.J. 1998. Age and rate of diversification of the Hawaiian silversword

alliance (Compositae). Proceedings of the National Academy of Sciences. 95:9402–9406.

Baldwin B.G., Wagner W.L. 2010. Hawaiian angiosperm radiations of North American origin.

Annals of Botany. 105:849-879.

Carlquist S.J. 1995. Introduction. In: Wagner W.L., Funk V.A., editors. Hawaiian Biogeography:

Evolution on a Hot Spot Archipelago. Washington DC: Smithsonian Institution Press.

Carracedo J.C., Guillou H., Nomade S., Rodríguez-Badiola E., Pérez-Torrado F.J., Rodríguez-

González A., Paris R., Troll V.R., Wiesmaier S., Delcamp A., Fernández-Turiel J.L.

2011. Evolution of ocean-island rifts: The northeast rift zone of Tenerife, Canary Islands.

Geological Society of America Bulletin. 123:562–584.

Carson H.L., Clague D.A. 1995. Geology and biogeography of the Hawaiian Islands. In: Wagner

W.L., Funk V.A., editors. Hawaiian Biogeography: Evolution on a Hot Spot Archipelago.

Washington DC: Smithsonian Institution Press.

Caujapé-Castells J., Tye A., Crawford D.J., Santos-Guerra A., Sakai A., Beaver K., Lobin W.,

Vincent Florens F.B., Moura M., Jardim R., Gómes I., Kueffer C. 2010. Conservation of

oceanic island floras: Present and future global challenges. Perspectives in Plant Ecology,

Evolution and Systematics. 12:107–129.

Clouard V., Bonneville A. 2004. Ages of seamounts, islands, and plateaus on the Pacific Plate.

Available from http://www.mantleplumes.org/WebDocuments/PacificSeamountAges.pdf.

Connell J. 2013. Islands at risk? Environments, economies and contemporary change.

Cheltenham, UK ; Northampton, MA: Edward Elgar.

Cronk Q.C., Kiehn M., Wagner W.L., Smith J.F. 2005. Evolution of Cyrtandra (Gesneriaceae) in

the Pacific Ocean: The origin of a supertramp clade. American Journal of Botany.

92:1017–1024.

Darwin C. 1859. The origin of species. New York, NY: Oxford University Press.

Díaz-Pérez A., Sequeira M., Santos-Guerra A., Catalán P. 2008. Multiple colonizations, in situ

speciation, and volcanism-associated stepping-stone dispersals shaped the

phylogeography of the Macaronesian Red Fescues (Festuca L., Gramineae). Systematic

Biology. 57:732–749.

Fleischer R.C., Mcintosh C.E., Tarr C.L. 1998. Evolution on a volcanic conveyor belt: using

phylogeographic reconstructions and K–Ar-based ages of the Hawaiian Islands to

estimate molecular evolutionary rates. Molecular Ecology. 7:533–545.

Francisco-Ortega J., Santos-Guerra A., Kim S.-C., Crawford D.J. 2000. Plant genetic diversity in

the Canary Islands: a conservation perspective. American Journal of Botany 87:909–919.

Funk V.A., Wagner W.L. 1995. Biogeography of seven ancient Hawaiian plant lineages.

Hawaiian Biogeography: Evolution on a Hotspot Archipelago. Washington DC:

Smithsonian Institution Press.

Gillespie R.G., Baldwin B.G., Waters J.M., Fraser C.I., Nikula R., Roderick G.K. 2012. Long-

distance dispersal: a framework for hypothesis testing. Trends in Ecology & Evolution.

27:47–56.

Gillespie R.G., Claridge E.M., Goodacre S.L. 2008. Biogeography of the fauna of French

Polynesia: diversification within and between a series of hot spot archipelagos.

Philosophical Transactions of the Royal Society B: Biological Sciences. 363:3335–3346.

Givnish T.J., Millam K.C., Mast A.R., Paterson T.B., Theim T.J., Hipp A.L., Henss J.M., Smith

J.F., Wood K.R., Sytsma K.J. 2009. Origin, adaptive radiation and diversification of the

Hawaiian lobeliads (Asterales: Campanulaceae). Proceedings of the Royal Society B:

Biological Sciences. 276:407–416.

Graham N.R., Gruner D.S., Lim J.Y., Gillespie R.G. 2017. Island ecology and evolution:

challenges in the Anthropocene. Environmental Conservation. 44:323–335.

Grant P.R., Grant B.R. 1992. Hybridization of bird species. Science. 256:193–197.

Heaney L.R. 2007. Is a new paradigm emerging for oceanic island biogeography? Journal of

Biogeography. 34:753–757.

Hennig W. 1966. Phylogenetic Systematics. Urbana, IL: University of Illinois Press.

Higgins S.I., Nathan R., Cain M.L. 2003. Are long-distance dispersal events in plants usually

caused by nonstandard means of dispersal? Ecology. 84:1945–1956.

Hodges S.A., Arnold M.L. 1994. Columbines: a geographically widespread species flock.

Proceedings of the National Academy of Sciences. 91:5129–5132.

Howarth D.G., Baum D.A. 2005. Genealogical evidence of homoploid hybrid speciation in an

adaptive radiation of Scaevola (Goodeniaceae) in the Hawaiian Islands. Evolution.

59:948.

Johnson M.A., Clark J.R., Wagner W.L., McDade L.A. 2017. A molecular phylogeny of the

Pacific clade of Cyrtandra (Gesneriaceae) reveals a Fijian origin, recent diversification,

and the importance of founder events. Molecular Phylogenetics and Evolution. 116:30–

48.

Juan C., Emerson B.C., Oromı́ P., Hewitt G.M. 2000. Colonization and diversification: towards a

phylogeographic synthesis for the Canary Islands. Trends in Ecology & Evolution.

15:104–109.

Kier G., Kreft H., Lee T.M., Jetz W., Ibisch P.L., Nowicki C., Mutke J., Barthlott W. 2009. A

global assessment of endemism and species richness across island and mainland regions.

Proceedings of the National Academy of Sciences. 106:9322–9327.

Kim H.-G., Keeley S.C., Vroom P.S., Jansen R.K. 1998. Molecular evidence for an African

origin of the Hawaiian endemic Hesperomannia (Asteraceae). Proceedings of the

National Academy of Sciences. 95:15440–15445.

Kim S.-C., Chunghee L., Mejías J.A. 2007. Phylogenetic analysis of chloroplast DNA matK

gene and ITS of nrDNA sequences reveals polyphyly of the genus Sonchus and new

relationships among the subtribe Sonchinae (Asteraceae: Cichorieae). Molecular

Phylogenetics and Evolution. 44:578–597.

Kingsford R.T., Watson J.E.M., Lundquist C.J., Venter O., Hughes L., Johnston E.L., Atherton

J., Gawel M., Keith D.A., Mackey B.G., Morley C., Possingham H.P., Raynor B., Recher

H.F., Wilson K.A. 2009. Major conservation policy issues for biodiversity in Oceania.

Conservation Biology. 23:834–840.

Kreft H., Jetz W., Mutke J., Kier G., Barthlott W. 2008. Global diversity of island floras from a

macroecological perspective. Ecology Letters. 11:116–127.

Kueffer C., Daehler C.C., Torres-Santana C.W., Lavergne C., Meyer J.-Y., Otto R., Silva L.

2010. A global comparison of plant invasions on oceanic islands. Perspectives in Plant

Ecology, Evolution and Systematics. 12:145–161.

Kueffer C., Drake D.R., Fernandez-Palacios J.M. 2014. Island biology: looking towards the

future. Biology Letters. 10:20140719.

Losos J.B., Ricklefs R.E. 2009. The theory of island biogeography revisited. Princeton, NJ:

Princeton University Press.

MacArthur R.H., Wilson E.O. 1963. An equilibrium theory of insular zoogeography. Evolution.

17:373–387.

MacArthur R.H., Wilson E.O. 1967. The Theory of Island Biogeography. Princeton, NJ:

Princeton University Press.

Mallet J. 2005. Hybridization as an invasion of the genome. Trends in Ecology & Evolution.

20:229–237.

Munroe E.G. 1948. The geographical distribution of butterflies in the West Indies. Ph.D.

dissertation. Ithaca, NY: Cornell University.

Nathan R., Schurr F.M., Spiegel O., Steinitz O., Trakhtenbrot A., Tsoar A. 2008. Mechanisms of

long-distance seed dispersal. Trends in Ecology & Evolution. 23:638–647.

Parent C.E., Crespi B.J. 2006. Sequential colonization and diversification of Galápagos endemic

land snail genus Bulimulus (Gastropoda, Stylommatophora). Evolution. 60:2311.

Petren K., Grant B.R., Grant P.R. 1999. A phylogeny of Darwin’s finches based on microsatellite

DNA length variation. Proceedings of the Royal Society B: Biological Sciences. 266:321.

Pillon Y., Johansen J., Sakishima T., Chamala S., Barbazuk W.B., Roalson E.H., Price D.K.,

Stacy E.A. 2013a. Potential use of low-copy nuclear genes in DNA barcoding: a

comparison with plastid genes in two Hawaiian plant radiations. BMC evolutionary

biology. 13:1.

Pillon Y., Johansen J.B., Sakishima T., Roalson E.H., Price D.K., Stacy E.A. 2013b. Gene

discordance in phylogenomics of recent plant radiations, an example from Hawaiian

Cyrtandra (Gesneriaceae). Molecular Phylogenetics and Evolution. 69:293–298.

Preston F.W. 1962a. The canonical distribution of commonness and rarity: Part II. Ecology.

43:410–432.

Preston F.W. 1962b. The canonical distribution of commonness and rarity: Part I. Ecology.

43:185–215.

Price J.P. 2004. Floristic biogeography of the Hawaiian Islands: Influences of area, environment

and paleogeography. Journal of Biogeography. 31:487–500.

Price J.P., Clague D.A. 2002. How old is the Hawaiian biota? Geology and phylogeny suggest

recent divergence. Proceedings of the Royal Society B: Biological Sciences. 269:2429–

2435.

Price J.P., Wagner W.L. 2004. Speciation in Hawaiian angiosperm lineages: Cause,

consequence, and mode. Evolution. 58:2185.

Robichaux R.H., Carr G.D., Liebman M., Pearcy R.W. 1990. Adaptive radiation of the Hawaiian

silversword alliance (Compositae-Madiinae): Ecological, morphological, and

physiological diversity. Annals of the Missouri Botanical Garden. 77:64–72.

Rominger A.J., Goodman K.R., Lim J.Y., Armstrong E.E., Becking L.E., Bennett G.M., Brewer

M.S., Cotoras D.D., Ewing C.P., Harte J., Martinez N.D., O’Grady P.M., Percy D.M.,

Price D.K., Roderick G.K., Shaw K.L., Valdovinos F.S., Gruner D.S., Gillespie R.G.

2016. Community assembly on isolated islands: macroecology meets evolution:

Community assembly on isolated islands. Global Ecology and Biogeography. 25:769–

780.

Ronquist F., Sanmartín I. 2011. Phylogenetic methods in biogeography. Annual Review of

Ecology, Evolution, and Systematics. 42:441–464.

Rosindell J., Phillimore A.B. 2011. A unified model of island biogeography sheds light on the

zone of radiation. Ecology Letters. 14:552–560.

Rumeu B., Caujapé-Castells J., Blanco-Pastor J.L., Jaén-Molina R., Nogales M., Elias R.B.,

Vargas P. 2011. The colonization history of Juniperus brevifolia (Cupressaceae) in the

Azores Islands. PLOS ONE. 6:e27697.

Sang T. 2002. Utility of low-copy nuclear gene sequences in plant phylogenetics. Critical

Reviews in Biochemistry and Molecular Biology. 37:121–147.

Schluter D. 2000. The ecology of adaptive radiation. Oxford: Oxford University Press.

Schluter D., Price T., Mooers A.Ø., Ludwig D. 1997. Likelihood of ancestor states in sdaptive

radiation. Evolution. 51:1699–1711.

Seehausen O. 2004. Hybridization and adaptive radiation. Trends in Ecology & Evolution.

19:198–207.

Seehausen O., Butlin R.K., Keller I., Wagner C.E., Boughman J.W., Hohenlohe P.A., Peichel

C.L., Saetre G.-P., Bank C., Brännström Å., Brelsford A., Clarkson C.S., Eroukhmanoff

F., Feder J.L., Fischer M.C., Foote A.D., Franchini P., Jiggins C.D., Jones F.C.,

Lindholm A.K. 2014. Genomics and the origin of species. Nature Reviews Genetics.

15:176–192.

Shaw K.L., Gillespie R.G. 2016. Comparative phylogeography of oceanic archipelagos: Hotspots

for inferences of evolutionary process. Proceedings of the National Academy of Sciences.

113:7986–7993.

Simpson G.G. 1940. Mammals and land bridges. Journal of the Washington Academy of

Sciences. 30:137–163.

Solís-Lemus C., Ané C. 2016. Inferring phylogenetic networks with maximum pseudolikelihood

under incomplete lineage sorting. PLOS genetics. 12:e1005896.

Wagner W., Herbst D., Sohmer S.H. 1990. Manual of Flowering Plants of Hawai’i. Honolulu:

University of Hawaii Press: Bishop Museum Press.

Warren B.H., Simberloff D., Ricklefs R.E., Aguilée R., Condamine F.L., Gravel D., Morlon H.,

Mouquet N., Rosindell J., Casquet J., Conti E., Cornuault J., Fernández-Palacios J.M.,

Hengl T., Norder S.J., Rijsdijk K.F., Sanmartín I., Strasberg D., Triantis K.A., Valente

L.M., Whittaker R.J., Gillespie R.G., Emerson B.C., Thébaud C. 2015. Islands as model

systems in ecology and evolution: Prospects fifty years after MacArthur-Wilson. Ecology

Letters. 18:200–217.

Watson D.M. 2002. A conceptual framework for studying species composition in fragments,

islands and other patchy ecosystems. Journal of Biogeography. 29:823–834.

Weitemier K., Straub S.C.K., Cronn R.C., Fishbein M., Schmickl R., McDonnell A., Liston A.

2014. Hyb-Seq: combining target enrichment and genome skimming for plant

phylogenomics. Applications in Plant Sciences. 2:1400042.

Wilson E.O. 1959. Adaptive shift and dispersal in a tropical ant fauna. Evolution. 13:122–144.

Wilson J.T. 1963. A possible origin of the Hawaiian Islands. Canadian Journal of Physics.

41:863–870.

Yang Y., Berry P.E. 2011. Phylogenetics of the Chamaesyce clade (Euphorbia, Euphorbiaceae):

Reticulate evolution and long-distance dispersal in a prominent C4 lineage. American

Journal of Botany. 98:1486–1503.

Yoder J.B., Clancey E., Des Roches S., Eastman J.M., Gentry L., Godsoe W., Hagey T.J.,

Jochimsen D., Oswald B.P., Robertson J., Sarver B.A.J., Schenk J.J., Spear S.F., Harmon

L.J. 2010. Ecological opportunity and the origin of adaptive radiations: Ecological

opportunity and origin of adaptive radiations. Journal of Evolutionary Biology. 23:1581–

1596.

CHAPTER TWO

DETECTING HYBRIDIZATION IN HAWAIIAN CYRTANDRA USING

GENOME-WIDE DATA

INTRODUCTION

The Hawaiian Islands – Since Darwin’s (1859) first observation of species diversity on the Galapagos Islands, oceanic island systems have played a key role in the study of evolutionary processes such as speciation and adaptive radiation (Carlquist 1974; Schluter 2000; Emerson

2002). Although the underlying processes of diversification within species radiations remain poorly understood (Yoder et al. 2010), the natural isolation, young ages, and small areas of island systems make them ideal settings for studying the mechanisms of evolution (Funk and

Wagner 1995; Fleischer et al. 1998; Price and Clague 2002; Price 2004; Price and Wagner

2004). The Hawaiian Islands in particular is an excellent site for biogeographic and phylogenetic studies due to its unique geological history, small total area, and highly isolated location nearly

4000 km from the nearest land mass. The Hawaiian Islands are a hotspot archipelago, a chain of age-sorted islands formed by the northwest movement of the Pacific plate over an area of active volcanism. Kaua’i, the oldest of the six main high islands, was formed approximately 4.7 million years ago (Mya) and sits in the northwest, while Hawai’i, the youngest and largest of the high islands, was formed approximately 0.5 Mya and sits in the southeast (Price and Clague 2002).

The three high islands of Moloka’i, Lana’i, and Maui are products of submergence and erosion of a larger prehistoric island called Maui Nui which formed around 1.0 Mya (Carson and Clague

1995). Evidence has shown that most of the Hawaiian flora and fauna likely diversified on the current high islands over the most recent 5 My (Price and Clague 2002). However, there are

lineages, such as the Hawaiian lobeliads (Givnish et al. 2009), which are thought to have colonized the now eroded atolls when they were higher islands situated northwest of the main high islands, which are estimated to have formed between 7 and 28 Mya, prior to dispersing to the current high Hawaiian Islands (Price and Clague 2002).

Due to its extreme isolation, the Hawaiian Islands have one of the highest rates of floral endemism, with 89% of its flowering plant species restricted to these islands (Wagner et al.

1990; Carson and Clague 1995; Price 2004). Among angiosperms, approximately 270 natural introductions are inferred to have given rise to over 1000 species, of which nearly 90% are endemic to the Hawaiian Islands and over 50% are endemic to a single island (Wagner et al.

1990; Price 2004). Much of the diversity within each of the Hawaiian angiosperm lineages is likely to have resulted from a single colonization event to the archipelago followed by dispersal throughout the islands and includes lineages such as the Hawaiian silversword alliance (Baldwin and Sanderson 1998), Hawaiian Psychotria (Nepokroeff et al. 2003), Hawaiian lobeliads

(Givnish et al. 2009), Schiedea (Willyard et al. 2011), Hawaiian Stachydae (Roy et al. 2013),

Hawaiian Silene (Eggens et al. 2007) and Hesperomannia (Kim et al. 1998; Morden and Harbin

2013), though there are angiosperm groups such as the Hawaiian Scaevola (Howarth et al. 2003) and Hawaiian Coprosma (Cantley et al. 2014) which are inferred to have colonized the islands more than once. The rapid radiation of the Hawaiian flora has resulted in an array of morphologically diverse lineages exhibiting low levels of genetic variation (Baldwin and

Sanderson 1998; Lindqvist and Albert 2002; Clark et al. 2009; Givnish et al. 2009; Baldwin and

Wagner 2010), and many are thought to hybridize leading to further difficulty in understanding their evolutionary histories (Smith et al. 1996; Wagner et al. 1999; Pillon et al. 2013b; Roy et al.

2013; Cantley et al. 2014; Stacy et al. 2014). One such hybridization-prone lineage with high

levels of phenotypic variation and low levels of genetic diversity is Hawaiian Cyrtandra

(Gesneriaceae).

Advantages of High-Throughput Sequencing in Phylogenetic Studies – Hybridization has long been known to play a role in the diversity and evolution of plants as illustrated by the body of literature available on the topic (Zirkle 1934; Anderson 1953; Anderson and Stebbins

1954; Stebbins 1959; Mallet 2005), and evidence for its importance continues to grow (Mallet

2008; Soltis and Soltis 2009; Abbott et al. 2013). It is estimated that approximately 25% of all plant species, especially younger lineages, are involved in hybridization and introgression with other plant species (Mallet 2005). Hybridization studies prior to molecular sequence data relied on phenotypic observations and comparisons, which were largely insufficient for understanding the importance of introgression in plant evolution as the extent of gene flow could not be accurately measured (Anderson 1948; Goulet et al. 2017). The advent of Sanger sequencing has made molecular phylogenetics possible, allowing for greater insight into the evolutionary histories of plant lineages and mechanisms of hybridization (Savolainen and Chase 2003).

Phylogenetic evidence suggests that introgression in the ancestry of many lineages is common, and that hybridization is a major factor in species evolution, especially among plants (McDade

2000). However, testing for hybridization in the presence of incomplete lineage sorting (ILS) has been difficult without the use of genome or transcriptome data. Modern phylogenomic methods and current high-throughput sequencing (HTS) techniques of whole genomes and transcriptomes has furthered our understanding of evolutionary processes and has given us greater insight into hybridization in plants by allowing us to test for underlying genetic structure and relationships within and between lineages (Soltis and Soltis 2009; Jackson et al. 2016; Solís-

Lemus and Ané 2016).

A multitude of different HTS techniques have been used to resolve difficult groups and understand how their evolutionary histories have been affected by hybridization and/or ILS.

Restriction-site associated DNA sequencing (RAD-seq; Hipp et al. 2014; Vargas et al. 2017), transcriptomics (Yang and Smith 2014), genome skimming (Straub et al. 2012; Bock et al. 2014;

Weitemier et al. 2014), and targeted enrichment (Cronn et al. 2012; Stull et al. 2013; Weitemier et al. 2014; Folk et al. 2015, 2016), among other methods, have been used in phylogenomic studies. Targeted enrichment is among the leading strategies for obtaining large amounts of data for relatively low cost, and has yielded highly informative data for resolving difficult groups.

The influence of hybridization and/or ILS also can be studied using data yielded from targeted enrichment strategies, and has been used in studies on Asclepias (Weitemier et al. 2014) and

Heuchera (Folk et al. 2016) with promising results. Furthermore, modern phylogenomic methods running on accessible and powerful computing clusters has allowed researchers to test for hybridization in the presence of ILS using large datasets in relatively short amounts of time

(Than et al. 2008; Solís-Lemus and Ané 2016; Blischak et al. 2017).

Cyrtandra – Cyrtandra J.R. & G. Forster is the most species rich genus in the family

Gesneriaceae, with over 800 species representing 15-20% of all species within the family

(Wagner et al., 1999; Cronk et al., 2005). Found in the understory of many Old World tropical forests, the morphologically diverse species within this genus are often long-lived shrubs and small trees, though they are also sometimes herbs, lianas, and epiphytes (Wagner et al. 1990).

This genus is likely to have evolved in the Indo-Malaysia region of Southeast Asia and later dispersed throughout the Pacific to become one of the most widely distributed plant genera in the region (Burtt 2001; Cronk et al. 2005; Clark et al. 2008, 2009). Based on the highly similar habitats across the broad geographic region occupied by the genus, Cyrtandra is likely to have

diverged under a dispersal-mediated allopatry model rather than ecological pressures (Cronk et al. 2005; Clark et al. 2008). To date, all examined species of Cyrtandra are diploid (Kiehn 2005) and there is no evidence of polyploidy in the group. Prior phylogenetic studies support monophyly of the Pacific clade, suggesting that the ~300 species within this clade diversified within the last 22 million years (Clark et al. 2009). While species of Cyrtandra are morphologically quite diverse across their range, species of Pacific clade are much more homogeneous as they are nearly exclusively understory shrubs or small trees with white flowers and fleshy berries (Cronk et al. 2005). However, the Fijian and Hawaiian lineages of Cyrtandra exhibit marked morphological diversity in other characters (Gillett 1967).

Hawaiian Cyrtandra are most likely of Fijian origin and colonized Kaua’i after its formation approximately 4.7 Mya (Clark et al. 2008, 2009; Johnson et al. 2017). To date, the most comprehensive study on morphological diversity was conducted on the Hawaiian species of

Cyrtandra (Wagner et al. 1990), but many questions regarding their evolutionary histories still remain. Currently 60 named species of Hawaiian Cyrtandra exist based on morphological

(Wagner et al. 1990) and genetic (Clark et al. 2008, 2009; Pillon et al. 2013a, 2013b) characters.

Interestingly, 57 of the 60 species are single-island endemics and the majority of species have a narrow distribution within islands, with up to eight species observed to inhabit a single community (Wagner et al. 1990). Eighty-seven putative Hawaiian hybrid combinations have been listed to date (Wagner et al. 1990; Wagner et al., unpubl.), primarily based on morphological intermediacy and sympatry of their parent species, though molecular evidence does exist for a few putative hybrids (Smith et al. 1996).

Species-diagnostic haplotypes have been found for species on Kaua’i but not for species on the younger islands of O'ahu or Hawai’i (Pillon et al. 2013a) and phylogenetic analyses using

nuclear and chloroplast loci show that species of Cyrtandra from Kaua’i form a well-supported lineage in the Hawaiian clade while Hawai’i taxa do not (Clark et al. 2009; Johnson et al. 2015).

This evidence suggests that the diversity of Cyrtandra across the Hawaiian Islands is the result of a classic stepping-stone model from the oldest to youngest as has been hypothesized for a number of other Hawaiian lineages such as Metrosideros (Percy et al. 2008) and Psychotria

(Nepokroeff et al. 2003), though Johnson et al. (2017) present evidence that Cyrtandra is likely to have dispersed in both directions across the Hawaiian Islands.

Outstanding Questions in Cyrtandra – Hawaiian Cyrtandra poses a complex problem, where morphology and genetics have not been reconciled despite numerous attempts to do so

(Smith et al. 1996; Cronk et al. 2005; Clark et al. 2008, 2009, Pillon et al. 2013a, 2013b; Johnson et al. 2017). The confusing array of morphological diversity has been attributed to hybridization

(Smith et al. 1996; Pillon et al. 2013a, 2013b) and convergence (Clark et al. 2008), though understanding the patterns of introgression or convergence has been difficult with the molecular data that has been generated for these organisms. Here, we use targeted enrichment to sequence hundreds of loci from 33 taxa including putative hybrids. We hypothesize that both incomplete lineage and hybridization will be evident in the genetic data recovered for these taxa.

Additionally, we hypothesize that genetic data will reflect the morphology of these species.

While the stepping-stone model of dispersal is well accepted, we expect Cyrtandra to also exhibit evidence of back-dispersal on the basis on morphology and the results from Johnson et al.

(2017). The advent of HTS and modern phylogenomic techniques has allowed us to further our understanding of Hawaiian Cyrtandra, and we hope the results from this research positively influence how other Hawaiian and island-dwelling flora are studied.

MATERIALS AND METHODS

Sampling – We sampled 33 specimens of Cyrtandra from across the Hawaiian Islands

(Appendix 1). We chose to sample across the islands rather than on a single island in order to look for biogeographical patterns that may inform the direction of future studies on this and other

Hawaiian lineages. Many species of Cyrtandra are found only on a single island, but for those found across multiple islands, we attempted to sample multiple specimens. We sampled 22 species and six putative hybrids, and included parent species in the sampling. These individuals were hypothesized to be hybrids on the basis of morphology in the field, and include C. calpidicarpa x hawaiensis (O’ahu, det. J. Clark), C. hawaiensis x grayana (Moloka’i, det. K.

Wood), C. longifolia x kauaiensis (Kaua’i, det. J. Clark), C. munroi x grayi (Maui, det. J. Clark),

C. sandwicensis x grandiflora (O’ahu, det. J. Clark), and C. procera x grayana (Moloka’i, det.

H. Oppenheimer; Appendix 1). We included two species of Cyrtandra from Fiji to use as outgroups based on prior studies of this genus (Cronk et al. 2005; Clark et al. 2008; Johnson et al. 2017) and additionally included a New World gesneriad, Diastema vexans, to test the utility of our probes across the family. Genomic DNA was extracted from silica-dried leaf material housed in the Roalson lab or supplied by the National Tropical Botanic Garden (NTBG) using a

CTAB protocol (Doyle and Doyle 1987).

Library Preparation and Sequencing – DNAs were checked for quality using a

NanoDrop 2000 (Thermo Fisher Scientific, Waltham, MA) and poor-quality samples were cleaned using an alcohol precipitation step. DNAs were then quantified with a Qubit (hsDNA assays, Thermo Fisher Scientific). We prepared 36 genomic DNA libraries using a NEBNext

Ultra II kit (New England Biolabs, Ipswich, MA) starting with between 70 ng-1ug of input DNA sheared for 350bp fragments with a Covaris sonicator using the manufacturer recommended

settings (model M220; Covaris, Woburn, MA). Ampure XP beads (Beckman Coulter, Brea, CA) were used in bead-based size selection steps, using 0.25x volumes for the first step and 0.1x volumes for the second. PCR enrichment used between five and ten cycles depending on the amount of input DNA. Libraries were barcoded using NEBNext multiplex oligos (New England

Biolabs) and quantified using qPCR (NEBNext quant kit, New England Biolabs). Probes for target enrichment of our libraries were designed with the Hyb-Seq protocol (Weitemier et al.

2014) using a Boea hygrometrica genome (Xiao et al. 2015) and Primulina pteropoda transcriptome (Ai et al. 2015) – two old-world species of Gesneriaceae. The Hyb-Seq protocol was meant for use with a transcriptome and genome from the same species, but because there are no transcriptomes or complete genomes available for Cyrtandra, we opted to use two available and related Old World gesneriad taxa. An unfiltered biotinylated baitset was synthesized for us using 3x tiling and a bait length of 120 bp totaling 3,420 unique baits by MYcroarray (Ann

Arbor, MI; now Arbor Biosciences). We followed the MyBaits v.3 protocol but using pools of six libraries and between 98-126 ng of input DNA (14-18ng DNA per library) in 7 uL per enrichment reactions (six total reactions). Hybridization was performed at 65°C for 25.5 hours and NEB 2x HiFi MasterMix (New England Biolabs) was used in the final amplification step after removing the beads from the enriched library pools. The thermocycler profile used 15 ul of template, an annealing temperature of 60°C for 30 seconds, an extension temperature of 72°C for

35 seconds, and 18 total cycles. Enriched library pools were cleaned using a Monarch PCR &

DNA cleanup kit (New England Biolabs), quantified using a Qubit, and checked for quality and size distribution using a Fragment Analyzer (Advanced Analytical, Ankeny, IA). Enriched pools were then combined in equimolar amounts to create a final pool of 36 samples, followed by a final clean-up step using 0.7x volumes of AxyPrep purification beads. The final library was

sequenced on an Illumina MiSeq (Clinical Genomics Center, Oklahoma Medical Research

Foundation, Oklahoma City, OK) using paired-end 250 bp reads. Reads were demultiplexed through the MiSeq controller software.

Marker Assembly and Dataset Creation – Raw reads were first assessed for quality using the tools implemented in FastQC (Simon 2010). Trimmomatic v. 0.36 (Bolger et al. 2014) was then used to remove contaminating adapter sequences, trim the first 13 bases from each read, and filter low-quality paired-end reads with a sliding window of 5bp and only accepting quality scores of at least Q20 (options HEADCROP:13 LEADING:3 TRAILING:3

SLIDINGWINDOW5:20 MINLEN:36). Two samples (C. procera and C. procera x grayana) were discarded because of an unexpectedly low number of mapped reads for these samples. We used the default settings in Geneious 8 to merge paired-end reads and then map reads to the probe-set file as the reference. Consensus sequences were extracted for each locus for each sample. “Splash-zone” regions, reads from the intron sequences flanking the exon sequences from the probe-set, were also extracted. Default settings allowed for ambiguity calls in cases of multiple alleles, and regions where coverage was not at least 5x were called as “?” and treated as missing data. Diastema vexans was not included in subsequent analyses. Individual loci were aligned using MAFFT v. 7.271 (Katoh et al. 2009) using the automatic setting (option --auto).

Following alignment, putative hybrids and two samples that exhibited unexpected placement in preliminary analyses (C. grayi-1 and C. hawaiensis-3) were removed from the dataset to create a

“core” dataset of 26 taxa. Putative hybrid taxa were added back to the core dataset, one putative hybrid at a time, to create a new dataset for each hybrid taxon. Following creation of these datasets, loci were trimmed using Phyutility v 2.7.1 (Smith and Dunn 2008) to a minimum column occupancy of 0.5, and sequences for samples with less than 10 characters were removed

following column filtering. Because one locus had only sequences from three samples, it was removed for subsequent analyses.

Concatenated and Species Tree Estimations – Because assumptions behind any one phylogenetic analysis are unlikely to be met for our dataset, we used both concatenated and coalescent methods in a variety of analyses. A supermatrix was created for our dataset including all of our taxa by concatenating all loci and a maximum likelihood analysis was conducted in

RAxML v. 8.2.11 (Stamatakis 2014) using an unpartitioned GTRGAMMA model for computational efficiency. 2500 rapid bootstrap replicates (option -f a) and an ML tree search from every fifth replicate were used for the concatenated alignment.

Under scenarios of incomplete lineage sorting (ILS), it has been shown that concatenation has lower power to reconstruct phylogenetic relationships (Mirarab et al. 2014;

Chou et al. 2015). Therefore, we utilized two methods of coalescent phylogenetic inference.

ASTRAL-III (Mirarab and Warnow 2015; Zhang et al. 2017) and ASTRID (Vachaspati and

Warnow 2015) are coalescent methods that both use unrooted gene trees to infer species trees in the presence of ILS. While ASTRAL utilizes quartets from the gene trees to infer the species tree, ASTRID is a distance-based and ILS-aware method that uses the BIONJ algorithm (Gascuel

1997). RAxML was used to conduct an ML analysis using 100 rapid bootstrap replicates (option

-f a) and an ML tree search from every fifth replicate using an unpartitioned GTRGAMMA for each of the loci for each dataset. These gene trees were used as input for both ASTRAL and

ASTRID. 100 bootstrap replicates were also conducted to assess branch support in both species tree methods, and quartet scores were calculated in ASTRAL to measure the percentage of gene trees that agree with each branch.

Analyses of Introgression – The gene trees and ASTRAL species trees from each dataset were used in the Julia package PHYLONETWORKS (Solís-Lemus and Ané 2016) to build phylogenetic networks in the presence of incomplete lineage sorting. We used SNaQ (Species

Network applying Quartets; Solís-Lemus and Ané 2016) to evaluate the most likely network and to calculate γ, the vector of inheritance probabilities describing the proportion of loci inherited by a hybrid node from each of its parents. We performed nested analyses that allowed for up to eight hybridization (h) events and compared the negative log psuedolikelihood scores. We stopped SNaQ runs when the pseudolikelihood scores reached their optimum. Optimization in each nested analysis was performed for 5 independent runs. The run with the lowest negative log pseudolikelihood score was kept as the best estimate. A sharp improvement in negative log pseudolikelihood scores is expected for each nested analysis, with a slow, linear improvement after h reaches the best value. Finally, we selected the best network with h hybridization events when the pseudolikelihood values hit a plateau. Because the output for SNaQ is a semi-directed, unrooted network, we rooted our networks using the outgroups even if the direction of hybrid edges constrained rooting.

RESULTS

Hyb-Seq and Read Mapping – Our probe-set design from the Hyb-Seq protocol

(Weitemier et al. 2014) contained 570 sequences for enrichment totaling 180,784 bp, with an average length of 317.2 bp per probe. The largest and shortest probe sequences were 2053 bp and

120 bp, respectively. Preliminary analyses on datasets with and without the intron regions flanking the targeted exons were conducted. Here we focus on the dataset including the intron sequences flanking the exons as the data was more informative. After removing the two samples

(C. procera and C. procera x grayana) with low percentages of mapped reads, the average

sequencing coverage across samples was 226.3X (Table 1). The least-covered sample had 58.0X average coverage and the most-covered sample had 593.3X average coverage. 546 loci were extracted for D. vexans with an average of 218.7X coverage across loci. Only three samples had sequences for one of the loci, which was removed prior to analyses. Following individual alignment and filtering of 569 loci, each locus had an average length of 819.3 bp. The largest aligned locus was 2590 bp and the shortest was 233 bp. The aligned length of the concatenated

33-taxon, 569 locus supermatrix was 466,201 bp. 58,683 characters (12.6%) were parsimony informative and 8.31% of the matrix was missing data. Considering only the ingroup taxa,

52,029 characters (11.2%) were parsimony informative and 8.37% of the matrix was missing data.

Concatenated Phylogenetic Inference – An unpartitioned maximum likelihood analyses was conducted on the concatenated dataset using 2500 bootstrap replicates. The concatenated analysis yielded a phylogeny with bootstrap support (BS) ranging from BS = 9 to BS = 100 (Fig.

1). Most Kaua'i taxa formed a grade paraphyletic to the rest of the taxa. Cyrtandra wainihaensis,

C. longifolia, C. wawrae, and C. aff. wawrae formed a paraphyletic group in relation to the rest of the taxa with moderately strong support (BS > 89). Two Kaua'i samples, C. kauaiensis and C. longifolia x kauaiensis were sister to each other (BS = 100) and were in a clade with O'ahu and

Maui Nui taxa. The two C. wawrae and C. aff. wawrae were closely related, but C. wawrae-1 formed a paraphyletic relationship with C. wawrae-2, C. aff. wawrae, and C. longifolia. O'ahu,

Maui Nui, and Hawai’i taxa were interspersed throughout the concatenated phylogeny. Bootstrap support for the clade containing all but the five Kaua'i taxa was moderate (BS = 75), but support for the two nested clades were very low (BS = 9 and BS = 18). The three sampled C. hawaiensis were in a clade with low bootstrap support (BS = 9). Cyrtandra calpidicarpa x hawaiensis, C.

grandiflora, C. hawaiensis-3, and C. kaulantha formed a clade (BS = 57) sister to a clade with C. hawaiensis-1, C. hawaiensis-2, C. calpidicarpa, C. munroi, C. paludosa, C. longifolia x kauaiensis, and C. kauaiensis (BS = 50). Cyrtandra longifolia x kauaiensis was sister to C. kauaiensis (BS = 100). The clade containing the three C. hawaiensis samples was sister to the clade formed by the rest of the non-Kaua’i taxa (BS = 18). Cyrtandra grayi-2 was sister to C. hawaiensis x grayana (BS = 28), and this group was sister to the rest of the taxa (BS = 17).

Cyrtandra giffardii and C. tintinnabula (BS = 65) was sister to the clade formed by C. grayi-1,

C. oxybapha, and C. platyphylla (BS = 97). This clade (BS = 94) was sister to the clade formed by the rest of the taxa (BS = 40). Cyrtandra lysiosepala, C. sandwicensis, C. sandwicensis x grandiflora, and C. cordifolia formed a clade (BS = 40) sister to the clade formed by C. wagneri,

C. munroi x grayi, C. macrocalyx, and C. grayana (BS = 38).

Coalescent Phylogenetic Inference – We generated maximum likelihood trees with 100 bootstrap replicates for each gene, and used these gene trees to create species trees in ASTRAL and ASTRID. The ASTRAL analysis on our dataset for all taxa yielded a phylogeny with a different topology than that found for the concatenated dataset (Fig. 2). The ASTRAL analysis reconstructed a species tree where the Kaua'i taxa formed a paraphyletic group in relation to the rest of our sampled taxa. Cyrtandra wainihaensis, C. kauaiensis, C. longifolia x kauaiensis, and

C. longifolia formed a clade (BS = 99) sister to all Hawaiian taxa (BS = 27), while the two sampled C. wawrae and C. aff. wawrae formed a clade (BS = 62) sister to all non-Kaua'i taxa

(BS = 70). Cyrtandra grandiflora, C. kaulantha, C. calpidicarpa x hawaiensis, C. hawaiensis-3, and C. calpidicarpa formed a clade (BS = 79) sister to the clade formed by C. hawaiensis-1 and

C. hawaiensis-2 (BS = 100). These taxa formed a clade (BS=25) sister to the rest of the taxa (BS

= 25). Cyrtandra hawaiensis x grayana was sister to the rest of the taxa (BS = 92). Cyrtandra

munroi and C. paludosa formed a group (BS = 88) sister to the remaining taxa (BS = 94).

Cyrtandra oxybapha, C. munroi x grayi, C. cordifolia, C. sandwicensis, and C. sandwicensis x grandiflora formed a clade (BS = 62) sister to a clade formed by the remaining Moloka’i, O’ahu, and Hawai’i taxa (BS = 65). Cyrtandra grayi-2, C. grayana, and C. macrocalyx formed a clade

(BS = 100) sister to the clade formed by C. grayi-1, C. platyphylla, and the taxa from Hawai’i

(BS = 82). Cyrtandra lysiosepala, C. tintinnabula, and C. giffardii formed a clade (BS = 100) sister to a clade formed by C. grayi-1, C. wagneri, and C. platyphylla. Quartet scores for ingroup branches of ASTRAL species trees found very low scores (< 50), with most branches showing nearly equal proportions of quartets (Fig. 3). A higher quartet score indicates that a larger proportion of the gene trees share the same topology as the species tree.

The ASTRID analysis (Fig. 4) gave similar results, but with a few notable differences.

The Kaua'i taxa formed a paraphyletic group in relation to the other Hawaiian taxa. The two C. wawrae and C. aff wawrae formed a clade (BS = 3) sister to all taxa (BS = 3), while C. longifolia, C. wainihaensis, C. kauaiensis, and C. longifolia x kauaiensis formed a clade (BS =

98) sister to all non-Kaua'i taxa (BS=3). Cyrtandra grandiflora, C. kaulantha, C. calpidicarpa,

C. calpidicarpa x hawaiensis, and C. hawaiensis-3 formed a clade (BS = 4) sister to the remaining taxa. Cyrtandra hawaiensis x grayana formed a clade with the remaining sampled C. hawaiensis (BS = 75) sister to the remaining taxa (BS = 41). Cyrtandra munroi and C. paludosa

(BS = 33) were sister to the remaining taxa, which were divided into two clades, one comprised of C. munroi x grayi, C. oxybapha, C. cordifolia, C. sandwicensis and C. sandwicensis x grandiflora (BS = 43), and the other comprised of a Moloka’i clade (C. grayana, C. grayi-2, and

C. macrocalyx; BS = 99) and a clade formed by the Hawai’i taxa C. platyphylla, and C. grayi-1

(BS = 6). The bootstrap support values for the ASTRID analysis in general were very low in

comparison to both the ASTRAL and concatenated analyses. ASTRAL and ASTRID analyses for the other datasets were generally very similar (Figs. 5-20), though the placement of some taxa and bootstrap support values changed depending on the dataset.

Evidence of Gene Flow in Hawaiian Cyrtandra – The networks (Figs. 21-30) that best fit our data included vectors of inheritance probabilities (γ) for between two and five hybridization events depending on the dataset. Following rooting, major-tree topologies for our networks differed slightly from the topologies of the species trees inferred in ASTRAL. Four of the networks inferred hybridization events directed in the incorrect direction, in that the edge with the smaller inheritance probability led from a derived lineage to an ancestral lineage (i.e., internal branch). The network analysis for our dataset with all samples inferred four hybridization events. One hybrid edge led from C. platyphylla to an internal branch in the incorrect direction. Of the other three hybrid edges, one led from the branch leading from C. sandwicensis and C. sandwicensis x grandiflora to C. wawrae and C. aff wawrae (γ = 0.212), another led from C. kauaiensis to C. longifolia x kauaiensis (γ = 0.438), and another led from the branch leading from C. grandiflora and C. kaulantha to C. hawaiensis-3 (γ = 0.475). Network analyses also inferred four hybridization events for the dataset without any hybrids (the core dataset). In this network (Fig. 23), one hybrid edge led from C. sandwicensis to the two C. wawrae and C. aff. wawrae (γ = 0.273), another led from C. longifolia to C. kauaiensis (γ =

0.142), a third led from C. grayana to the taxa from Hawai’i (γ = 0.127), and a fourth led from C. hawaiensis-1 to an internal branch at the base of a clade formed by C. hawaiensis-1, C. hawaiensis-2, C. calpidicarpa, C. grandiflora, and C. kaulantha (γ = 0.217).

Network analysis for the core dataset plus C. calpidicarpa x hawaiensis inferred three hybrid edges (Fig. 24). One led from C. munroi to C. wawrae and C. aff. wawrae (γ = 0.298),

another led from C. hawaiensis-2 to C. calpidicarpa (γ = 0.263), and a third led from C. grayana to a clade formed by all Hawai’i taxa except C. wagneri (γ = 0.08). No hybrid branches were inferred to or from C. calpidicarpa x hawaiensis. Similarly, network analysis of the core dataset plus C. grayi-1 also inferred three hybrid edges (Fig. 25). The first led from C. munroi to C. wawrae and C. aff. wawrae (γ = 0.295), a second led from C. sandwicensis to C. cordifolia (γ =

0.465), and a third led from C. grayana and C. macrocalyx to C. grayi-1 (γ = 0.358). Network analysis for the core dataset plus C. hawaiensis-3 inferred three hybrid edges as well (Fig. 26).

Two of the hybrid edges were directed in the incorrect direction. One led from C. wawrae-1 to an internal branch of both the sampled C. wawrae and C. aff. wawrae (γ = 0.21), and the other led from C. calpidicarpa to an internal branch leading from all taxa but the outgroup and C. wainihaensis, C. kauaiensis, and C. longifolia (γ = 0.362). The third hybrid edge led from C. paludosa to C. platyphylla (γ = 0.308).

Network analysis for the core dataset plus C. hawaiensis x grayana inferred five hybridization events (Fig. 27). One hybrid edge led in an incorrect direction, from C. grayi-2 to an internal branch leading from C. grayana, C. macrocalyx, C. grayi-2, C. platyphylla, and the taxa from Hawai’i (γ = 0.217). The other four hybrid edges led from C. calpidicarpa to C. hawaiensis-2 (γ = 0.499), from C. sandwicensis to C. cordifolia (γ = 0.463), from C. wawrae-2 to C. aff. wawrae (γ = 0.185), and from C. munroi to C. wawrae and C. aff wawrae (γ = 0.338).

For the core dataset with C. longifolia x kauaiensis, network analysis inferred two hybridization events (Fig. 28). One hybrid edge led from C. kauaiensis to C. longifolia x kauaiensis (γ =

0.417), and the other led from the clade made up of C. macrocalyx, C. grayana, and C. grayi-2 to the clade containing C. hawaiensis-1, C. hawaiensis-2, C. kaulantha, and C. grandiflora (γ =

0.253).

For the core dataset with C. munroi x grayi, network analysis inferred four hybridization events (Fig. 29). One hybrid branch led from C. sandwicensis to C. wawrae and C. aff. wawrae

(γ = 0.272), another led from C. hawaiensis-2 to C. calpidicarpa (γ = 0.148), a third led from C. grayana to the clade formed by C. lysiosepala, C. tintinnabula and C. giffardii (γ = 0.087), and a fourth led from C. cordifolia to C. munroi x grayi (γ = 0.464). Finally, network analysis inferred three hybridization events for the core dataset plus C. sandwicensis x grandiflora (Fig. 30). One hybrid edge led from C. oxybapha to C. wawrae and C. aff. wawrae (γ = 0.246), another led from

C. hawaiensis-1 to C. hawaiensis-2 (γ = 0.316), and a third led from C. giffardii to C. grayana (γ

= 0.113).

DISCUSSION

On island systems, the evolutionary forces of hybridization and gene flow have been thought to play important roles in the diversification of lineages, though the underlying processes are not well understood (Yoder et al. 2010). High-throughput sequencing methods and modern genomic tools provide new avenues to test for the effects of hybridization in diversification

(Gompert and Buerkle 2016; Payseur and Rieseberg 2016; Vallejo-Marín and Hiscock 2016).

Here we use targeted enrichment strategies implemented through Hyb-Seq (Weitemier et al.

2014) to demonstrate its utility in understanding phylogenetic relationships of Hawaiian

Cyrtandra, and to test morphologically-based hypotheses of gene flow in this lineage. Through species tree methods and analyses of gene discordance, we provide evidence of ILS and gene flow in Hawaiian Cyrtandra.

Concatenated vs ASTRAL vs ASTRID Analyses – The concatenated and coalescent analyses conducted here yielded phylogenies with incongruent topologies (Figs 1, 2, and 12).

Because concatenated analyses assume that all loci share the same evolutionary history, the species tree estimated is identical to estimating a single gene tree. Long loci provide greater signal than short loci in concatenation, which can result in a resolved tree with an incorrect topology. Furthermore, concatenation has been shown to output incorrect trees when incomplete lineage sorting (ILS) is widespread (Kubatko and Degnan 2007). Coalescent methods on the other hand are able to produce accurate species trees in comparison to concatenation methods when ILS is present (Liu et al. 2008; Degnan and Rosenberg 2009; Edwards 2009). A number of recent phylogenomic studies using both concatenated and coalescent methods (Song et al. 2012;

Stephens et al. 2015; Folk et al. 2016) resulted in phylogenies that were largely congruent, but it is clear in Hawaiian Cyrtandra that the high level of ILS has influenced the topology of the concatenated phylogeny. ILS is expected in Hawaiian Cyrtandra due to its young age and is evident through the quartet values generated in ASTRAL (Fig. 11). In lineages with low levels of

ILS, the major quartet score (q1) at nodes is often high (> 60), while low scores (< 50) indicate high levels of ILS (Mirarab et al. 2014). In our ASTRAL species trees, we see that most nodes have almost equal proportions of quartet scores (q1, q2, q3 = ~33), indicating very high levels of

ILS. The ASTRID species trees reconstructed here have topologies more similar to the ASTRAL trees than the concatenated tree (Figs. 12-20), though there are still many obvious differences in topology. Perhaps most notably, the node support values are much lower for the ASTRID analyses. These differences can most likely be attributed to the methods used. Both are statistically consistent under the multispecies coalescent, but ASTRAL uses quartets from gene trees while ASTRID utilizes the BIONJ algorithm, a distance-based method to build species trees, which is much faster than ASTRAL especially when using very large numbers of loci

(Mirarab et al. 2014; Vachaspati and Warnow 2015). We used the ASTRAL species trees in tests

of gene flow and to compare our findings with those of prior studies, as ASTRAL has been shown to be more accurate than distance-based methods in reconstructing phylogenetic relationships when ILS levels are very high (Sayyari and Mirarab 2016).

Phylogenetic Relationships among Species of Cyrtandra – The phylogenetic hypotheses presented here largely agree with the previous phylogenetic studies on this group (Cronk et al.

2005; Clark et al. 2008, 2009; Johnson et al. 2017). Prior studies relied on relatively few loci to reconstruct hypotheses of species relationships in Hawaiian Cyrtandra. Cronk et al. (2005) utilized the nuclear ITS spacer, and Clark et al. (2008, 2009) utilized nuclear ITS, nuclear ETS, and plastid psbA-trnH spacer regions to estimate species relationships. Pillon et al. (2013b) showed how discordance between gene histories due to hybridization may skew hypotheses of species relationships, especially when few loci are used, and demonstrated the use of low-copy nuclear loci in testing for species relationships in this and other recent plant radiations (Pillon et al. 2013a). Johnson et al. (2017) utilized the nuclear low-copy Cyrt1 and plastid rpl32-trnL regions in addition to loci from prior studies, and was able to reconstruct a more resolved picture of species relationships in Hawaiian Cyrtandra, further demonstrating the use of larger datasets including low-copy exon regions in young plant lineages. Throughout our species tree, nodes are generally more strongly supported than the previous phylogenetic hypotheses, most likely due to the large number of loci used in comparison to prior studies (569 to ≤ five) and species-tree methods which account for ILS. However, there are still unsupported nodes (BS < 50) indicating that: (1) gene flow and ILS are obscuring the relationships in Hawaiian Cyrtandra at these nodes, or (2) despite having a large amount of data, our datasets are not informative at these nodes given our sampling, or (3) these very short branches are real and represent very rapid divergence of

those lineages. Additionally, the discordance between ASTRAL species trees from our datasets indicate that taxon sampling is important in testing for species relationships in this lineage. In this study, we used relatively few taxa (21 (or 22) species and 5 putative hybrids), and only for three species did we sample more than one individual. Wider sampling across species and hybrids of Hawaiian Cyrtandra, and including more individuals per species, is likely to yield a clearer picture of species relationships in this lineage.

From our ASTRAL species tree (Fig. 2), we infer an initial colonization of Kaua'i (the oldest island) with subsequent dispersal to O'ahu, Maui Nui, and then Hawai’i (progression rule) which has been hypothesized in prior studies (Clark et al. 2008, 2009; Johnson et al. 2017).

Furthermore, our phylogeny supports the back-dispersal hypotheses of Johnson et al. (2017). We infer through species relationships that a back-dispersal event from Maui Nui to O'ahu (C. cordifolia, C. sandwicensis, and C. sandwicensis x grandiflora) and a second back-dispersal event from Hawai’i to Maui Nui (C. grayi-1 and C. platyphylla) are probable due to the placement of three O'ahu taxa among Maui Nui taxa, and the placement of two Maui Nui taxa among the Hawai’i taxa.

Species and putative hybrids in Hawaiian Cyrtandra are identified on the basis of morphology and/or geographical location. In our sampling, we included multiple samples of C. wawrae, C. hawaiensis, and C. grayi in order to see if the morphology of these species is reflected in a molecular phylogeny. We also included C. aff. wawrae to test the hypothesis that this individual belongs to a new and currently unnamed species, and C. kaulantha to test the hypothesis that this species is actually the hybrid progeny of C. calpidicarpa and C. hawaiensis.

Both of the sampled C. wawrae and C. aff. wawrae were from Kaua'i and formed a clade, separate from the other Kaua’i taxa implying that C. aff. wawrae is at least closely related to C.

wawrae, but more work on that possibly undescribed species needs to be completed (Fig. 2).

Morphologically, C. wawrae is thought to be a lineage distinct from C. longifolia, C. kauaiensis, and C. wainihaensis, a hypothesis that is supported through the paraphyletic relationship formed by C. wawrae and the other Kaua’i taxa in our phylogeny. Additionally, C. longifolia is morphologically distinct from C. kauaiensis and C. wainihaensis, and in the phylogeny is sister to these taxa.

The three C. hawaiensis grouped closely in our species tree. The two from Maui Nui (C. hawaiensis-1, Maui; C. hawaiensis-2, Moloka'i) formed a strongly supported clade sister to the clade formed by C. hawaiensis-3, C. calpidicarpa, C. calpidicarpa x hawaiensis, C. grandiflora, and C. kaulantha (all from O’ahu). Because Maui Nui eroded into the three islands of Moloka’i,

Lana’i, and Maui relatively recently (0.5 Mya), it is likely that gene flow among populations of

C. hawaiensis on these islands were more recent than gene flow between populations of C. hawaiensis on O’ahu and Moloka’i or Maui. Alternatively, our phylogeny shows that C. hawaiensis-3 shares more genetic similarity with the other specimens from O’ahu. Though populations of C. hawaiensis from different islands can certainly be classified as the same species despite genetic dissimilarity, our results indicate cryptic diversity may be present in widespread Hawaiian Cyrtandra species.

Unlike our results for C. hawaiensis, C. grayi-1 (Maui) and C. grayi-2 (Moloka'i; Fig. 2) did not group together despite both being from Maui Nui. Instead, the individual from Moloka'i grouped closely with two other Moloka'i taxa (C. grayana and C. macrocalyx), while the individual from Maui grouped more closely with C. platyphylla (also from Maui) and the taxa from Hawai’i. While C. platyphylla was thought to be closely related to C. oxybapha (Wagner et al. 1990), it is not unexpected that this taxa grouped with the other taxa from Hawai’i in our

analyses, as it has a distribution across both Maui and Hawai’i. However, the placement of C. grayi-1 with C. platyphylla and the taxa of Hawai’i is unexpected, as this species is only found on Maui and Moloka'i.

Impacts of Gene Flow – Species tree methods such as ASTRAL are efficient for reconstructing species relationships in the presence of ILS, but do not take into account the effects of hybridization or other forms of gene flow. Therefore, network analyses such as those we conducted in SNaQ (Solís-Lemus and Ané 2016) are important in understanding the species relationships for lineages whose species are thought to hybridize in nature. The phylogenetic network analyses conducted here indicated some level of gene flow in the diversification of

Hawaiian Cyrtandra (Figs. 22-30). Specimens from across the Hawaiian Islands have been hypothesized to be hybrids on the basis of morphology and sympatry, and therefore have not been identified as their own species. Our networks show that nearly all of the morphologically identified hybrids are not likely to be hybrids based on their genetics and that hybridization may be occurring between other lineages that are currently identified as species. We expected to see hybrid edges extending from parent species to the putative hybrids in each analysis, but we saw this only for C. longifolia x kauaiensis. Additionally, different numbers of hybridization events inferred for each dataset (Fig. 21; all taxa, all putative hybrids including C. hawaiensis-3 and C. grayi-1 removed, and one for each hybrid included) demonstrates the importance of sampling for this lineage in phylogenetic studies.

Across the nine network analyses conducted, seven inferred a hybrid edge to C. wawrae from either O’ahu (C. sandwicensis) or Maui (C. munroi and C. oxybapha) taxa. This hybrid edge is fairly consistent across network analyses and was often the first hybrid edge inferred

(h=1), indicating that C. wawrae is most likely an introgressed lineage. However, gamma values

for these hybrid edges were relatively low (0.20 < γ < 0.30), indicating that introgression in C. wawrae was likely not recent or occurred with a species not sampled here. It is not clear which species are contributing to C. wawrae as our analyses inferred parents to be C. munroi, C. oxybapha, or C. sandwicensis, but it is likely that a species from Maui or O'ahu back-dispersed to Kaua'i and hybridized with C. wawrae. Additionally, in none of the network analyses did we see a hybrid edge leading to or from C. kaulantha, which has been hypothesized on the basis of morphology to be a hybrid species between C. calpidicarpa and C. hawaiensis, indicating that other processes are likely to have shaped the diversification of C. kaulantha. For C. kaulantha and many other species, their status as endangered or threatened hinges on whether or not they are considered to be species. Cyrtandra kaulantha does not seem to exhibit signs of being a hybrid taxon, and therefore may be a rare species that will require protection and status as a threatened or endangered species.

Across the analyses, only one putative hybrid showed evidence of having genes from both putative parents. The putative hybrids C. calpidicarpa x hawaiensis, C. hawaiensis x grayana, C. munroi x grayi, and C. sandwicensis x grandiflora are not inferred to be hybrids based on the network analyses conducted here, while C. longifolia x kauaiensis is most likely a first or second generation hybrid between C. longifolia and C. kauaiensis. In both the analysis of the dataset using all taxa and the core dataset plus C. calpidicarpa x hawaiensis, no hybrid edges lead to or from C. calpidicarpa x hawaiensis, indicating that this putative hybrid is unlikely to be a hybrid of C. calpidicarpa and C. hawaiensis given our sampling, and there another hypothesis is needed. It may be that some putative hybrids are just part of morphologically variable species.

When this putative hybrid is analyzed with the core dataset, there is a hybrid edge leading from

C. hawaiensis-2 to C. calpidicarpa (γ = 0.263), indicating that the C. calpidicarpa sample is

more likely to be introgressed than C. calpidicarpa x hawaiensis. Similarly, network analyses of the dataset including all taxa, the core dataset plus C. hawaiensis x grayana, and the core dataset plus C. sandwicensis x grandiflora did not infer hybrid edges to or from C. hawaiensis x grayana or C. sandwicensis x grandiflora, indicating that while these individuals may exhibit morphological intermediacy to their putative parent species, these putative hybrids are more likely to be something else. However, it should be noted that without more detailed sampling these results are far from conclusive.

Two network analyses inferred hybrid edges to putative hybrids. The networks from the dataset of all taxa and the core dataset with C. longifolia x kauaiensis showed that C. longifolia x kauaiensis is likely to be a hybrid of C. longifolia and C. kauaiensis. In these networks, the proportion of genes inferred to come from each parent agreed with what we would expect for an

F1 or F2 hybrid (γ = ~0.50). Of all the putative hybrids, this was the only one from Kaua'i and the only to be correctly identified on the basis of morphology. Because species on Kaua’i have had more time to establish themselves, putative hybrids may be easier to identify especially on the basis of morphology than those on the younger islands, which could still be undergoing processes of adaptive radiation and may not have had time to evolve prezygotic barriers to gene flow (Carlquist 1974; Johnson et al. 2015).

Network analysis of the dataset with hybrids removed except for C. munroi x grayi inferred a hybrid edge to C. munroi x grayi from C. cordifolia (γ = 0.464), indicating that this individual also may be a hybrid. However, this hybrid edge implies that gene flow between C. cordifolia and C. oxybapha, neither of which are the species hypothesized to have given rise to this individual, are the parent species. Similar to the hybrid edge to C. wawrae from O’ahu or

Maui taxa, C. cordifolia (O’ahu) does not exist where C. munroi x grayi was collected (Maui).

The parent species therefore is likely to be closely an unsampled close relative of C. cordifolia that dispersed to Maui and hybridized C. oxybapha. Additionally, the lack of hybrid edge from either putative parent means this hybrid was simply mis-identified on the basis of morphology, or that hybridization resulted in morphological features not intermediate to their parental species.

Throughout our analyses, many taxa were inferred to be hybrids despite being identified as an individual from a previously named and identified species. Though many of the vectors of inheritance had low probabilities (around γ = 0.0 to 0.3) and may be artifacts of the effects of sampling or ILS, hybrid edges with relatively high inheritance probabilities (γ = 0.3 to 0.5) were also inferred. For example, network analysis of the dataset with all taxa inferred a hybrid edge from the branch leading from C. grandiflora and C. kaulantha to C. hawaiensis-3 (γ = 0.475). In the network for the core dataset plus C. grayi-1, a hybrid edge led from the branch leading from

C. grayana and C. macrocalyx to C. grayi-1 (γ = 0.358) and another led from C. sandwicensis to

C. cordifolia (γ = 0.465). These inferences of hybrid edges between taxa that are hypothesized to belong to species, and the lack of hybrid edges to putative hybrids indicate that while gene flow may be prevalent for Hawaiian Cyrtandra, the ways by which species and hybrids are identified may need to be reconsidered. Wider sampling of this lineage will certainly give us greater insight into whether or not the methods used to identify taxa in this lineage should be revised going forward.

SNaQ (Solís-Lemus and Ané 2016) infers unrooted species networks with directed hybrid edges. An unexpected result of this is that, following rooting with the outgroup, hybridization events were inferred in directions that were implausible for some of our networks.

These edges imply that more recent Hawaiian taxa contributed genes to more ancient taxa (i.e., internal branches) including the Fijian outgroups, which is highly unlikely for this lineage. These

hybrid edges indicated to us that, though we do not know exactly between which taxa, ancient gene flow is likely to have played a role in shaping the morphological and genetic landscape of

Hawaiian Cyrtandra. Alternatively, this could be an artifact of hybrid edges with unsampled lineages. Greater sampling of species (including multiple sampling with species and more species lineages) should allow us to understand the direction and timing of gene flow in this lineage and allow us to make better sense of these hybrid edges. Along with evidence of gene flow contributing to C. wawrae, this evidence of ancient gene flow implies that hybridization in

Hawaiian Cyrtandra plays a natural role in the process of diversification and is not just a consequence of human activity in Hawaii.

Herbaria and High-Throughput Sequencing – The importance of herbaria in the phylogenetic study of rare and extinct plants cannot be overstated. Though ocean archipelagos represent only around 5% of the terrestrial surface of the Earth, they are home to nearly 25% of all described vascular plant species (Kreft et al. 2008). Nearly 10,000 of these species are highly threatened or in critical danger of extinction (Caujapé-Castells et al. 2010), making herbaria more important than ever for understanding species relationships and processes of evolution in island lineages. Extraction of DNA from herbarium specimens results in about five times less

DNA than extraction from fresh samples (Staats et al. 2011), much of which is highly degraded and difficult to sequence with the Sanger method which works best with longer contiguous segments of DNA (> 500bp). In high-throughput methods, very little input DNA is needed for library preparation and high-throughput sequencing can make use of smaller fragments of DNA

(100 – 400bp) making it very effective for obtaining genetic data from degraded herbarium specimens (Ribeiro and Lovato 2007; Staats et al. 2013; Bakker et al. 2016). In the present study, we extracted and sequenced DNA from silica-dried plant material over a decade old with high

coverage and strong results. High-throughput sequencing of DNA extracted from silica dried or herbarium materials can be an effective way to test for species relationships in rare and difficult plant radiations such as Hawaiian Cyrtandra.

With recent advances and falling costs in sequencing and computing technology, genomic methods in systematics research has gained traction as a cost-effective way of understanding evolutionary relationships in difficult lineages. Methods such as RNA-seq (Roda et al. 2013; Yang and Smith 2014), RAD-seq (Eaton and Ree 2013; Cavender-Bares et al. 2015), and targeted enrichment (Weitemier et al. 2014; Folk et al. 2015, 2016) have allowed systematists to obtain and use large datasets to construct highly resolved phylogenies and test for gene flow between lineages. Though RNA-seq is effective in gene expression studies, a limitation of this method is the requirement of fresh material, which is often unavailable for many rare and extinct taxa. The most cost effective ways of obtaining data for phylogenomic studies are RAD-seq and targeted enrichment. RAD-seq takes advantage of restriction-sites and sequences DNA near those sites, resulting in large datasets of genomic data from across the genome. While simulation and empirical studies show that RAD-seq data is adequate for systematics studies including those on recent radiations (Cariou et al. 2013; Wagner et al. 2013),

RAD-seq data often has a large percentage of missing data which may affect tests of gene flow.

For our system, targeted enrichment via Hyb-Seq (Weitemier et al. 2014) was a cost- effective method of obtaining large amounts of data from multiple samples for phylogenetic analyses. A limitation of targeted enrichment is that a genome and transcriptome from the genus of interest are often required to identify low-copy regions for sequencing. However, as

Weitemier et al. (2014) and we demonstrate, a genome and transcriptome from closely related

taxa may be sufficient for identifying highly-conserved regions for sequencing. In our study, we utilized the genome of Boea hygrometrica (Xiao et al. 2015) and the transcriptome of Primulina pteropoda (Ai et al. 2015) as input for the Hyb-Seq protocol (Weitemier et al. 2014) because a complete genome and transcriptome were not available for Cyrtandra. Our probe-set thus targets less data than those of other studies that utilize a genome and transcriptome from the genus of interest (Straub et al. 2012; Weitemier et al. 2014; Folk et al. 2016). However, we obtained and aligned over 400,000 bp of data for each sample, which is significantly more than prior studies

(Cronk et al. 2005; Clark et al. 2008, 2009; Pillon et al. 2013b; Johnson et al. 2017) and has allowed us to better hypothesize species relationships and instances of gene flow within this lineage. Moreover, this method has allowed us to see that putative hybrids may actually be unique species, which has great implications for how species are circumscribed and conserved going forward. Additionally, our probe-set captured 546 loci from Diastema vexans, a New

World gesneriad (Table 1), indicating that this probe-set may be useful for understanding the evolutionary relationships within other Gesneriad lineages.

CONCLUSIONS

Many hypotheses of species relationships in Hawaiian Cyrtandra have been proposed over the years on the basis of morphology and Sanger sequences (Gillett 1967; Wagner et al.

1990; Smith et al. 1996; Cronk et al. 2005; Clark et al. 2008; Johnson et al. 2017). However, these hypotheses struggled to overcome the influences of incomplete lineage sorting and hybridization evident in this lineage. Here we demonstrated the utility of high-throughput sequencing and a phylogenomic approach to understanding species relationships and gene flow in the presence of ILS. The limitations of concatenated analyses in lineages exhibiting high

levels of ILS is demonstrated here through the incongruence of the concatenated phylogeny with the ASTRAL and ASTRID phylogenies. Evident in our ASTRAL species trees is progression of

Cyrtandra from the oldest island of Kaua'i to the youngest of Hawai’i, with possible back- dispersals from Maui Nui to O'ahu and Hawai’i to Maui Nui. Network analyses show that C. wawrae is most likely an introgressed lineage, and putative hybrids with the exception of C. longifolia x kauaiensis and C. munroi x grayi are not likely to be hybrids.

These results additionally have implications for the conservation of species. While some taxa are currently considered to be hybrids based on their morphological traits, some of these might not be of hybrid origin based on the results here and may require a more careful evaluation of identification by an experienced individual, as many of these were collected and identified by different individuals, and additionally may require a reevaluation of species circumscriptions and descriptions to determine what they might represent. We acknowledge, however, that wider sampling of both sympatric and allopatric species and populations across their ranges will allow us to better understand species relationships and the timing, direction, and magnitude of gene flow in this group, in addition to how many species of Hawaiian Cyrtandra may actually exist.

Future studies incorporating wider sampling are certain to refine and solidify the hypotheses we have proposed here. We hope this study provides a strong basis for further phylogenomic and phylogeographic studies into evolution and diversification of Cyrtandra, as well as a framework for studying evolutionary patterns of other recent plant radiations across the planet.

LITERATURE CITED

Abbott R., Albach D., Ansell S., Arntzen J.W., Baird S.J.E., Bierne N., Boughman J., Brelsford

A., Buerkle C.A., Buggs R., Butlin R.K., Dieckmann U., Eroukhmanoff F., Grill A.,

Cahan S.H., Hermansen J.S., Hewitt G., Hudson A.G., Jiggins C., Jones J., Keller B.,

Marczewski T., Mallet J., Martinez-Rodriguez P., Möst M., Mullen S., Nichols R., Nolte

A.W., Parisod C., Pfennig K., Rice A.M., Ritchie M.G., Seifert B., Smadja C.M.,

Stelkens R., Szymura J.M., Väinölä R., Wolf J.B.W., Zinner D. 2013. Hybridization and

speciation. Journal of Evolutionary Biology. 26:229–246.

Ai B., Gao Y., Zhang X., Tao J., Kang M., Huang H. 2015. Comparative transcriptome resources

of eleven Primulina species, a group of ‘stone plants’ from a biodiversity hot spot.

Molecular Ecology Resources. 15:619–632.

Anderson E. 1948. Hybridization of the Habitat. Evolution. 2:1–9.

Anderson E. 1953. Introgressive Hybridization. Biological Reviews. 28:280–307.

Anderson E., Stebbins G.L. 1954. Hybridization as an Evolutionary Stimulus. Evolution. 8:378–

388.

Bakker F.T., Lei D., Yu J., Mohammadin S., Wei Z., van de Kerke S., Gravendeel B.,

Nieuwenhuis M., Staats M., Alquezar-Planas D.E., Holmer R. 2016. Herbarium

genomics: plastome sequence assembly from a range of herbarium specimens using an

Iterative Organelle Genome Assembly pipeline. Biological Journal of the Linnean

Society. 117:33–43.

Baldwin B.G., Sanderson M.J. 1998. Age and rate of diversification of the Hawaiian silversword

alliance (Compositae). Proceedings of the National Academy of Sciences. 95:9402–9406.

Baldwin B.G., Wagner W.L. 2010. Hawaiian angiosperm radiations of North American origin.

Annals of Botany. 105:849–879.

Blischak P.D., Chifman J., Wolfe A.D., Kubatko L.S. 2018. HyDe: a Python package for

genome-scale hybridization detection. Systematic Biology.:syy023.

Bock D.G., Kane N.C., Ebert D.P., Rieseberg L.H. 2014. Genome skimming reveals the origin of

the Jerusalem Artichoke tuber crop species: neither from Jerusalem nor an artichoke.

New Phytologist. 201:1021–1030.

Bolger A.M., Lohse M., Usadel B. 2014. Trimmomatic: a flexible trimmer for Illumina sequence

data. Bioinformatics. 30:2114–2120.

Burtt B.L. 2001. A survey of the genus Cyrtandra (Gesneriaceae). Phytomorphology. 51:393–

404.

Cantley J.T., Swenson N.G., Markey A., Keeley S.C. 2014. Biogeographic insights on Pacific

Coprosma (Rubiaceae) indicate two colonizations to the Hawaiian Islands. Botanical

Journal of the Linnean Society. 174:412–424.

Cariou M., Duret L., Charlat S. 2013. Is RAD-seq suitable for phylogenetic inference? An in

silico assessment and optimization. Ecology and Evolution. 3:846–852.

Carlquist S. 1974. Island Biology. New York: Columbia University Press.

Carson H.L., Clague D.A. 1995. Geology and biogeography of the Hawaiian Islands. In: Wagner

W.L., Funk V.A., editors. Hawaiian biogeography: Evolution on a hot spot archipelago.

Washington: Smithsonian Institution Press.

Caujapé-Castells J., Tye A., Crawford D.J., Santos-Guerra A., Sakai A., Beaver K., Lobin W.,

Vincent Florens F.B., Moura M., Jardim R., Gómes I., Kueffer C. 2010. Conservation of

oceanic island floras: Present and future global challenges. Perspectives in Plant Ecology,

Evolution and Systematics. 12:107–129.

Cavender-Bares J., González-Rodríguez A., Eaton D.A.R., Hipp A.A.L., Beulke A., Manos P.S.

2015. Phylogeny and biogeography of the American live oaks (Quercus subsection

Virentes): a genomic and population genetics approach. Molecular Ecology. 24:3668–

3687.

Chou J., Gupta A., Yaduvanshi S., Davidson R., Nute M., Mirarab S., Warnow T. 2015. A

comparative study of SVDquartets and other coalescent-based species tree estimation

methods. BMC genomics. 16:S2.

Clark J., Ree R., Alfaro M., King M., Wagner W., Roalson E. 2008. A comparative study in

ancestral range reconstruction methods: Retracing the uncertain histories of insular

lineages. Systematic Biology. 57:693–707.

Clark J.R., Wagner W.L., Roalson E.H. 2009. Patterns of diversification and ancestral range

reconstruction in the southeast Asian–Pacific angiosperm lineage Cyrtandra

(Gesneriaceae). Molecular Phylogenetics and Evolution. 53:982–994.

Cronk Q.C., Kiehn M., Wagner W.L., Smith J.F. 2005. Evolution of Cyrtandra (Gesneriaceae) in

the Pacific Ocean: the origin of a supertramp clade. American Journal of Botany.

92:1017–1024.

Cronn R., Knaus B.J., Liston A., Maughan P.J., Parks M., Syring J.V., Udall J. 2012. Targeted

enrichment strategies for next-generation plant biology. American Journal of Botany.

99:291–311.

Darwin C. 1859. The origin of species. New York: Oxford University Press.

Degnan J.H., Rosenberg N.A. 2009. Gene tree discordance, phylogenetic inference and the

multispecies coalescent. Trends in Ecology & Evolution. 24:332–340.

Eaton D.A.R., Ree R.H. 2013. Inferring Phylogeny and Introgression using RADseq Data: An

Example from Flowering Plants (Pedicularis: Orobanchaceae). Systematic Biology.

62:689–706.

Edwards S.V. 2009. Is a new and general theory of molecular systematics emerging? Evolution.

63:1–19.

Eggens F., Popp M., Nepokroeff M., Wagner W.L., Oxelman B. 2007. The origin and number of

introductions of the Hawaiian endemic Silene species (Caryophyllaceae). American

Journal of Botany. 94:210–218.

Emerson B.C. 2002. Evolution on oceanic islands: molecular phylogenetic approaches to

understanding pattern and process. Molecular ecology. 11:951–966.

Fleischer R.C., Mcintosh C.E., Tarr C.L. 1998. Evolution on a volcanic conveyor belt: using

phylogeographic reconstructions and K–Ar-based ages of the Hawaiian Islands to

estimate molecular evolutionary rates. Molecular Ecology. 7:533–545.

Folk R.A., Mandel J.R., Freudenstein J.V. 2015. A protocol for targeted enrichment of intron-

containing sequence markers for recent radiations: A phylogenomic example from

Heuchera (Saxifragaceae). Applications in Plant Sciences. 3:1500039.

Folk R.A., Mandel J.R., Freudenstein J.V. 2016. Ancestral gene flow and parallel organellar

genome capture result in extreme phylogenomic discord in a lineage of angiosperms.

Systematic Biology.:syw083.

Funk V.A., Wagner W.L. 1995. Biogeography of seven ancient Hawaiian plant lineages.

Hawaiian Biogeography: Evolution on a Hotspot Archipelago. Washington DC:

Smithsonian Institution Press.

Gascuel O. 1997. BIONJ: an improved version of the NJ algorithm based on a simple model of

sequence data. Molecular Biology and Evolution. 14:685–695.

Gillett G.W. 1967. The genus Cyrtandra in Fiji. Contributions of the U.S. National Herbarium.

37:107–159.

Givnish T.J., Millam K.C., Mast A.R., Paterson T.B., Theim T.J., Hipp A.L., Henss J.M., Smith

J.F., Wood K.R., Sytsma K.J. 2009. Origin, adaptive radiation and diversification of the

Hawaiian lobeliads (Asterales: Campanulaceae). Proceedings of the Royal Society B:

Biological Sciences. 276:407–416.

Gompert Z., Buerkle C.A. 2016. What, if anything, are hybrids: enduring truths and challenges

associated with population structure and gene flow. Evolutionary Applications. 9:909–

923.

Goulet B.E., Roda F., Hopkins R. 2017. Hybridization in Plants: Old Ideas, New Techniques.

Plant Physiology. 173:65–78.

Hipp A.L., Eaton D.A.R., Cavender-Bares J., Fitzek E., Nipper R., Manos P.S. 2014. A

framework phylogeny of the American Oak clade based on sequenced RAD data. PLOS

ONE. 9:e93975.

Howarth D.G., Gustafsson M.H., Baum D.A., Motley T.J. 2003. Phylogenetics of the genus

Scaevola (Goodeniaceae): implication for dispersal patterns across the Pacific Basin and

colonization of the Hawaiian Islands. American Journal of Botany. 90:915–923.

Jackson N.D., Carstens B.C., Morales A.E., O’Meara B.C. 2016. Species delimitation with gene

flow. Systematic Biology.:syw117.

Johnson M.A., Clark J.R., Wagner W.L., McDade L.A. 2017. A molecular phylogeny of the

Pacific clade of Cyrtandra (Gesneriaceae) reveals a Fijian origin, recent diversification,

and the importance of founder events. Molecular Phylogenetics and Evolution. 116:30–

48.

Johnson M.A., Price D.K., Price J.P., Stacy E.A. 2015. Postzygotic barriers isolate sympatric

species of Cyrtandra (Gesneriaceae) in Hawaiian montane forest understories. American

Journal of Botany. 102:1870–1882.

Katoh K., Asimenos G., Toh H. 2009. Multiple alignment of DNA sequences with MAFFT. In:

Posada D., editor. Bioinformatics for DNA Sequence Analysis. Totowa, NJ: Humana

Press. p. 39–64.

Kiehn M. 2005. Chromosome numbers of Hawaiian angiosperms: New records and comments.

Pacific Science. 59:363–377.

Kim H.-G., Keeley S.C., Vroom P.S., Jansen R.K. 1998. Molecular evidence for an African

origin of the Hawaiian endemic Hesperomannia (Asteraceae). Proceedings of the

National Academy of Sciences. 95:15440–15445.

Kreft H., Jetz W., Mutke J., Kier G., Barthlott W. 2008. Global diversity of island floras from a

macroecological perspective. Ecology Letters. 11:116–127.

Kubatko L.S., Degnan J.H. 2007. Inconsistency of phylogenetic estimates from concatenated

data under coalescence. Systematic Biology. 56:17–24.

Lindqvist C., Albert V.A. 2002. Origin of the Hawaiian endemic mints within North American

Stachys (Lamiaceae). American Journal of Botany. 89:1709–1724.

Liu L., Pearl D.K., Brumfield R.T., Edwards S.V. 2008. Estimating species trees using multiple-

allele DNA sequence data. Evolution. 62:2080–2091.

Mallet J. 2005. Hybridization as an invasion of the genome. Trends in Ecology & Evolution.

20:229–237.

Mallet J. 2008. Hybridization, ecological races and the nature of species: empirical evidence for

the ease of speciation. Philosophical Transactions of the Royal Society B: Biological

Sciences. 363:2971–2986.

McDade L.A. 2000. Hybridization and phylogenetics. In: Wiens J.J., editor. Phylogenetic

Analysis of Morphological Data. Washington DC: Smithsonian Institution Press. p. 146–

164.

Mirarab S., Reaz R., Bayzid M.S., Zimmermann T., Swenson M.S., Warnow T. 2014. ASTRAL:

genome-scale coalescent-based species tree estimation. Bioinformatics. 30:i541–i548.

Mirarab S., Warnow T. 2015. ASTRAL-II: coalescent-based species tree estimation with many

hundreds of taxa and thousands of genes. Bioinformatics. 31:i44–i52.

Morden, C.W. and Harbin, S.C., 2013. Evolution and Biogeographic Origins of the Endemic

Hawaiian Genus Hesperomannia (Asteraceae). Pacific science, 67: 219-235.

Nepokroeff M., Sytsma K.J., Wagner W.L., Zimmer E.A. 2003. Reconstructing ancestral

patterns of colonization and dispersal in the Hawaiian understory tree genus Psychotria

(Rubiaceae): A comparison of parsimony and likelihood approaches. Systematic Biology.

52:820–838.

Payseur B.A., Rieseberg L.H. 2016. A genomic perspective on hybridization and speciation.

Molecular Ecology. 25:2337–2360.

Percy D.M., Garver A.M., Wagner W.L., James H.F., Cunningham C.W., Miller S.E., Fleischer

R.C. 2008. Progressive island colonization and ancient origin of Hawaiian Metrosideros

(Myrtaceae). Proceedings of the Royal Society B: Biological Sciences. 275:1479–1490.

Pillon Y., Johansen J., Sakishima T., Chamala S., Barbazuk W.B., Roalson E.H., Price D.K.,

Stacy E.A. 2013a. Potential use of low-copy nuclear genes in DNA barcoding: a

comparison with plastid genes in two Hawaiian plant radiations. BMC evolutionary

biology. 13:1.

Pillon Y., Johansen J.B., Sakishima T., Roalson E.H., Price D.K., Stacy E.A. 2013b. Gene

discordance in phylogenomics of recent plant radiations, an example from Hawaiian

Cyrtandra (Gesneriaceae). Molecular Phylogenetics and Evolution. 69:293–298.

Price J.P. 2004. Floristic Biogeography of the Hawaiian Islands: Influences of Area,

Environment and Paleogeography. Journal of Biogeography. 31:487–500.

Price J.P., Clague D.A. 2002. How old is the Hawaiian biota? Geology and phylogeny suggest

recent divergence. Proceedings of the Royal Society B: Biological Sciences. 269:2429–

2435.

Price J.P., Wagner W.L. 2004. Speciation in Hawaiian angiosperm lineages: Cause,

consequence, and mode. Evolution. 58:2185.

Ribeiro R.A., Lovato M.B. 2007. Comparative analysis of different DNA extraction protocols in

fresh and herbarium specimens of the genus Dalbergia. Genetics and Molecular

Research. 6:173–187.

Roda F., Ambrose L., Walter G.M., Liu H.L., Schaul A., Lowe A., Pelser P.B., Prentis P.,

Rieseberg L.H., Ortiz-Barrientos D. 2013. Genomic evidence for the parallel evolution of

coastal forms in the Senecio lautus complex. Molecular Ecology. 22.

Roy T., Chang T.-H., Lan T., Lindqvist C. 2013. Phylogeny and biogeography of New World

Stachydeae (Lamiaceae) with emphasis on the origin and diversification of Hawaiian and

South American taxa. Molecular Phylogenetics and Evolution. 69:218–238.

Savolainen V., Chase M.W. 2003. A decade of progress in plant molecular phylogenetics.

Trends in Genetics. 19:717–724.

Schluter D. 2000. The ecology of adaptive radiation. Oxford: Oxford University Press.

Simon A. 2010. FastQC: A quality control tool for high throughput sequence data. Available

from http://www.bioinformatics.babraham.ac.uk/projects/fastqc/.

Smith J.F., Burke C.C., Wagner W.L. 1996. Interspecific hybridization in natural populations of

Cyrtandra (Gesneriaceae) on the Hawaiian Islands: Evidence from RAPD markers. Plant

Systematics and Evolution. 200:61–77.

Smith S.A., Dunn C.W. 2008. Phyutility: a phyloinformatics tool for trees, alignments and

molecular data. Bioinformatics. 24:715–716.

Solís-Lemus C., Ané C. 2016. Inferring phylogenetic networks with maximum pseudolikelihood

under incomplete lineage sorting. PLOS genetics. 12:e1005896.

Soltis P.S., Soltis D.E. 2009. The Role of hybridization in plant speciation. Annual Review of

Plant Biology. 60:561–588.

Song S., Liu L., Edwards S.V., Wu S. 2012. Resolving conflict in eutherian mammal phylogeny

using phylogenomics and the multispecies coalescent model. Proceedings of the National

Academy of Sciences. 109:14942–14947.

Staats M., Cuenca A., Richardson J.E., Ginkel R.V., Petersen G., Seberg O., Bakker F.T. 2011.

DNA damage in plant herbarium tissue. PLOS ONE. 6:e28448.

Staats M., Erkens R.H.J., Vossenberg B. van de, Wieringa J.J., Kraaijeveld K., Stielow B., Geml

J., Richardson J.E., Bakker F.T. 2013. Genomic treasure troves: Complete genome

sequencing of herbarium and insect museum specimens. PLOS ONE. 8:e69189.

Stacy E.A., Johansen J.B., Sakishima T., Price D.K., Pillon Y. 2014. Incipient radiation within

the dominant Hawaiian tree Metrosideros polymorpha. Heredity. 113:334–342.

Stamatakis A. 2014. RAxML version 8: a tool for phylogenetic analysis and post-analysis of

large phylogenies. Bioinformatics. 30:1312–1313.

Stebbins G.L. 1959. The Role of hybridization in evolution. Proceedings of the American

Philosophical Society. 103:231–251.

Stephens J.D., Rogers W.L., Heyduk K., Cruse-Sanders J.M., Determann R.O., Glenn T.C.,

Malmberg R.L. 2015. Resolving phylogenetic relationships of the recently radiated

carnivorous plant genus Sarracenia using target enrichment. Molecular Phylogenetics and

Evolution. 85:76–87.

Straub S.C.K., Parks M., Weitemier K., Fishbein M., Cronn R.C., Liston A. 2012. Navigating the

tip of the genomic iceberg: Next-generation sequencing for plant systematics. American

Journal of Botany. 99:349–364.

Stull G.W., Moore M.J., Mandala V.S., Douglas N.A., Kates H.-R., Qi X., Brockington S.F.,

Soltis P.S., Soltis D.E., Gitzendanner M.A. 2013. A targeted enrichment strategy for

massively parallel sequencing of angiosperm plastid genomes. Applications in Plant

Sciences. 1:1200497.

Than C., Ruths D., Nakhleh L. 2008. PhyloNet: a software package for analyzing and

reconstructing reticulate evolutionary relationships. BMC Bioinformatics. 9:322.

Vachaspati P., Warnow T. 2015. ASTRID: accurate species trees from internode distances. BMC

Genomics. 16:S3.

Vallejo-Marín M., Hiscock S.J. 2016. Hybridization and hybrid speciation under global change.

New Phytologist. 211:1170–1187.

Vargas O.M., Ortiz E.M., Simpson B.B. 2017. Conflicting phylogenomic signals reveal a pattern

of reticulate evolution in a recent high-Andean diversification (Asteraceae: Astereae:

Diplostephium). New Phytologist. 214:1736-1750.

Wagner C.E., Keller I., Wittwer S., Selz O.M., Mwaiko S., Greuter L., Sivasundar A., Seehausen

O. 2013. Genome-wide RAD sequence data provide unprecedented resolution of species

boundaries and relationships in the Lake Victoria cichlid adaptive radiation. Molecular

Ecology. 22:787–798.

Wagner W., Herbst D., Sohmer S.H. 1990. Manual of Flowering Plants of Hawaii. Honolulu:

University of Hawaii Press: Bishop Museum Press.

Wagner W., Herbst D., Sohmer S.H. 1999. Manual of Flowering Plants of Hawaii. Honolulu:

University of Hawaii Press: Bishop Museum Press.

Weitemier K., Straub S.C.K., Cronn R.C., Fishbein M., Schmickl R., McDonnell A., Liston A.

2014. Hyb-Seq: Combining target enrichment and genome skimming for plant

phylogenomics. Applications in Plant Sciences. 2:1400042.

Willyard, A., Wallace L. E., Wagner W. L., Weller S. G., Sakai A. K., Nepokroeff M. 2011.

Estimating the species tree for Hawaiian Schiedea (Caryophyllaceae) from multiple loci

in the presence of reticulate evolution. Molecular Phylogenetics and Evolution 60: 29-48.

Xiao L., Yang G., Zhang L., Yang X., Zhao S., Ji Z., Zhou Q., Hu M., Wang Y., Chen M., Xu

Y., Jin H., Xiao X., Hu G., Bao F., Hu Y., Wan P., Li L., Deng X., Kuang T., Xiang C.,

Zhu J.-K., Oliver M.J., He Y. 2015. The resurrection genome of Boea hygrometrica: A

blueprint for survival of dehydration. Proceedings of the National Academy of Sciences.

112:5833–5837.

Yang Y., Smith S.A. 2014. Orthology inference in nonmodel organisms using transcriptomes

and low-coverage genomes: Improving accuracy and matrix occupancy for

phylogenomics. Molecular Biology and Evolution. 31:3081–3092.

Yoder J.B., Clancey E., Des Roches S., Eastman J.M., Gentry L., Godsoe W., Hagey T.J.,

Jochimsen D., Oswald B.P., Robertson J., Sarver B.A.J., Schenk J.J., Spear S.F., Harmon

L.J. 2010. Ecological opportunity and the origin of adaptive radiations: Ecological

opportunity and origin of adaptive radiations. Journal of Evolutionary Biology. 23:1581–

1596.

Zhang C., Sayyari E., Mirarab S. 2017. ASTRAL-III: Increased scalability and impacts of

contracting low support branches. Comparative Genomics. RECOMB-CG:53–75.

Zirkle C. 1934. More records of plant hybridization before Koelreuter. Journal of Heredity.

25:3–18.

Appendix 1. Sampled specimens, collectors, herbarium, and location. All specimens unless otherwise indicated are from Hawaii so only island name is given.

Cyrtandra calpidicarpa (Rock) H. St. John & Storey; J. Clark 571; SEL; O’ahu.

Cyrtandra calpidicarpa x hawaiensis; J. Clark 567; SEL; O’ahu. Cyrtandra cordifolia Gaudich.;

J. Clark 579; SEL; O’ahu. Cyrtandra giffardii Rock; K. Wood 12258; PTBG; Hawai’i

Cyrtandra grandiflora Gaudich.; J. Clark 577; SEL; O’ahu. Cyrtandra grayana Hillebr.; K.

Wood 11401; PTBG; Moloka’i. Cyrtandra grayi-1 C.B. Clarke; J. Clark 667; SEL; Maui.

Cyrtandra grayi-2 C.B. Clarke; H. Oppenheimer H110624; BISH; Moloka’i. Cyrtandra hawaiensis-1 C.B. Clarke; J. Clark 662; SEL; Maui. Cyrtandra hawaiensis-2 C.B. Clarke; K.

Wood 11385; PTBG; Moloka’i. Cyrtandra hawaiensis-3 C.B. Clarke; J. Clark 569; SEL; O’ahu.

Cyrtandra hawaiensis x grayana; K. Wood 11392; PTBG; Moloka’i. Cyrtandra kauaiensis

Wawra; J. Clark 556A; SEL; Kaua’i. Cyrtandra kaulantha Wawra; J. Clark 572; SEL; O’ahu.

Cyrtandra longifolia (Wawra) Hillebr.; J. Clark 551; SEL; Kaua’i. Cyrtandra longifolia x kauaiensis; J. Clark 552; SEL; Kaua’i. Cyrtandra lysiosepala (A. Gray) C.B. Clarke; K. Wood

12418; PTBG; Hawai’i. Cyrtandra macrocalyx Hillebr.; H. Oppenheimer H110622; BISH;

Moloka’i. Cyrtandra munroi C. N. Forbes; H. Oppenheimer H120638; BISH; Lana’i. Cyrtandra munroi x grayi; J. Clark 678; SEL; Maui. Cyrtandra occulta A.C.Sm.; J. Clark 694; PTBG; Fiji.

Cyrtandra oxybapha W.L. Wagner & D. R. Herbst.; K. Wood 6249; PTBG; Maui. Cyrtandra paludosa Gaudich.; J. Clark 680; SEL; Maui. Cyrtandra platyphylla C.B. Clarke; H.

Oppenheimer H100512; BISH; Maui. Cyrtandra procera Hillebr.; H. Oppenheimer H110621;

BISH; Moloka’i. Cyrtandra procera x grayana; H. Oppenheimer H110623; BISH; Moloka’i.

Cyrtandra sandwicensis (H. Lév) H. St. John & Storey; J. Clark 576; PTBG; O’ahu. Cyrtandra sandwicensis x grandiflora; J. Clark 578; SEL; O’ahu. Cyrtandra tintinnabula Rock; S.

Perlman 17676; PTBG; Hawai’i. Cyrtandra victoriae Gillespie; M. Johnson MJ049; SUVA; Fiji.

Cyrtandra wagneri Lorence & Perlman; S. Perlman 17675; PTBG; Hawai’i. Cyrtandra wainihaensis H. Lév; J. Clark 549; SEL; Kaua’i. Cyrtandra wawrae-1 C.B. Clarke; J. Clark 550;

PTBG; Kaua’i. Cyrtandra wawrae-2 C.B. Clarke; K. Wood 17317; PTBG; Kaua’i. Cyrtandra aff. wawrae; K. Wood 12775; PTBG; Kaua’i. Diastema vexans H.E.Moore; Skog 8260; US;

Colombia.

Table 1. Summary of targeted enrichment sequencing and mapped reads.

Sample Raw Assembled Unassembled Extracted Mean Reads Reads Reads Loci Coverage C. aff. wawrae 481400 307485 143211 569 152.69x C. calpidicarpa 699974 562663 92721 569 286.30x C. cordifolia 843912 657681 122291 570 344.06x C. calpidicarpa x hawaiensis 636918 485865 112401 569 265.38x C. giffardii 375466 324892 27954 569 188.37x C. grayana 736298 605696 74834 569 328.00x C. grandiflora 766312 601847 101423 569 311.88x C. grayi-1 288814 247223 26679 569 141.35x C. grayi-2 411964 343433 40143 569 194.30x C. hawaiensis-1 281798 233635 33675 569 132.45x C. hawaiensis-2 660106 545274 60644 569 284.31x C. hawaiensis-3 186366 115875 51919 567 57.98x C. hawaiensis x grayana 609880 508345 63963 569 294.62x C. kauaiensis 711810 618927 40471 569 324.02x C. kaulantha 1018210 792993 147151 570 423.28x C. longifolia 582082 505417 42693 569 288.90x C. longifolia x kauaiensis 623864 530140 46032 569 273.26x C. lysiosepala 1397914 1086373 224379 566 593.31x C. macrocalyx 734202 615422 67566 569 346.65x C. munroi 248414 181760 43932 568 83.82x C. munroi x grayi 409700 349810 35630 569 197.42x C. occulta 560408 339712 188270 569 169.97x C. oxybapha 423958 338505 63267 569 200.74x C. paludosa 179246 151344 15606 569 82.85x C. platyphylla 117110 98654 12554 568 59.12x C. sandwicensis 650000 412039 199341 569 200.59x C. sandwicensis x 639090 477252 108226 569 254.25x grandiflora C. tintinnabula 409766 352425 27621 569 201.66x C. victoriae 528964 322053 170657 569 166.63x C. wagneri 384676 342816 19580 569 199.89x C. wainihaensis 283526 235793 27991 569 136.71x C. wawrae-1 420752 278095 106721 568 140.35x C. wawrae-2 439702 287115 126679 569 141.40x D. vexans 629818 385805 198255 546 218.67x

Figure 1. Concatenated maximum likelihood tree of all taxa inferred in RaxML with 2500 bootstrap replicates. Bootstrap

values are plotted on branches. Scale represents branch lengths in per-site substitution rate.

Figure 2. ASTRAL tree for all taxa from 569 gene trees and 100 bootstrap replicates. Bootstrap values are plotted on branches.

Scale represents branch lengths in coalescent units.

Figure 3. ASTRAL tree for all taxa with quartet scores given for each internal branch. The three scores on each branch

correspond to the species tree, first alternate, and second alternate.

Figure 4. ASTRID tree for all taxa from 569 gene trees and 100 bootstrap replicates. Bootstrap values are plotted on branches.

Figure 5. ASTRAL tree for the core dataset from 569 gene trees and 100 bootstrap replicates. Bootstrap values are plotted on

branches. Scale represents branch lengths in coalescent units.

Figure 6. ASTRAL tree for the core dataset with C. calpidicarpa x hawaiensis from 569 gene trees and 100 bootstrap

replicates. Bootstrap values are plotted on branches. Scale represents branch lengths in coalescent units.

Figure 7. ASTRAL tree for the core dataset with C. grayi-1 from 569 gene trees and 100 bootstrap replicates. Bootstrap values

are plotted on branches. Scale represents branch lengths in coalescent units.

Figure 8. ASTRAL tree for the core dataset with C. hawaiensis-3 from 569 gene trees and 100 bootstrap replicates. Bootstrap

values are plotted on branches. Scale represents branch lengths in coalescent units.

Figure 9. ASTRAL tree for the core dataset with C. hawaiensis x grayana from 569 gene trees and 100 bootstrap replicates.

Bootstrap values are plotted on branches. Scale represents branch lengths in coalescent units.

Figure 10. ASTRAL tree for the core dataset with C. longifolia x kauaiensis from 569 gene trees and 100 bootstrap replicates.

Bootstrap values are plotted on branches. Scale represents branch lengths in coalescent units.

Figure 11. ASTRAL tree for the core dataset with C. munroi x grayi from 569 gene trees and 100 bootstrap replicates.

Bootstrap values are plotted on branches. Scale represents branch lengths in coalescent units.

Figure 12. ASTRAL tree for the core dataset with C. sandwicensis x grandiflora from 569 gene trees and 100 bootstrap

replicates. Bootstrap values are plotted on branches. Scale represents branch lengths in coalescent units.

Figure 13. ASTRID tree for the core dataset from 569 gene trees and 100 bootstrap replicates. Bootstrap values are plotted on

branches.

Figure 14. ASTRID tree for the core dataset with C. calpidicarpa x hawaiensis from 569 gene trees and 100 bootstrap

replicates. Bootstrap values are plotted on branches.

Figure 15. ASTRID tree for the core dataset with C. grayi-1 from 569 gene trees and 100 bootstrap replicates. Bootstrap values

are plotted on branches.

Figure 16. ASTRID tree for the core dataset with C. hawaiensis-3 from 569 gene trees and 100 bootstrap replicates. Bootstrap

values are plotted on branches.

Figure 17. ASTRID tree for the core dataset with C. hawaiensis x grayana from 569 gene trees and 100 bootstrap replicates.

Bootstrap values are plotted on branches.

Figure 18. ASTRID tree for the core dataset with C. longifolia x kauaiensis from 569 gene trees and 100 bootstrap replicates.

Bootstrap values are plotted on branches.

Figure 19. ASTRID tree for the core dataset with C. munroi x grayi from 569 gene trees and 100 bootstrap replicates.

Bootstrap values are plotted on branches.

Figure 20. ASTRID tree for the core dataset with C. sandwicensis x grandiflora from 569 gene trees and 100 bootstrap

replicates. Bootstrap values are plotted on branches.

Figure 21. Model selection using negative pseudolikelihood scores for A) all taxa, B) core dataset, C) core dataset with C. calpidicarpa x hawaiensis, D) core dataset with C. grayi-1, E) core dataset with C. hawaiensis-3, F) core dataset with C. hawaiensis x grayana, G) core dataset with C. longifolia x kauaiensis, and H) core dataset with C. munroi x grayi, I) core dataset with

C. sandwicensis x grandiflora. Negative pseudolikelihood values are plotted on the Y-axis and hybridization events are plotted on the X-axis. A sharp improvement in negative pseudolikelihood until h reaches the optimal value, and a slower, linear improvement thereafter, is expected.

Figure 22. Introgression model for all taxa. Hybrid edges (h=4) are indicated by red (minor hybrid edges) or green (major hybrid edges). Arrows indicate the direction of gene flow and numbers indicate percentage of genes inferred to contribute to a lineage.

Figure 23. Introgression model for the core dataset. Hybrid edges (h=4) are indicated by red

(minor hybrid edges) or green (major hybrid edges). Arrows indicate the direction of gene flow and numbers indicate percentage of genes inferred to contribute to a lineage.

Figure 24. Introgression model for the core dataset with C. calpidicarpa x hawaiensis.

Hybrid edges (h=3) are indicated by red (minor hybrid edges) or green (major hybrid edges).

Arrows indicate the direction of gene flow and numbers indicate percentage of genes inferred to contribute to a lineage.

Figure 25. Introgression model for the core dataset with C. grayi-1. Hybrid edges (h=3) are indicated by red (minor hybrid edges) or green (major hybrid edges). Arrows indicate the direction of gene flow and numbers indicate percentage of genes inferred to contribute to a lineage.

Figure 26. Introgression model for the core dataset with C. hawaiensis-3. Hybrid edges (h=3) are indicated by red (minor hybrid edges) or green (major hybrid edges). Arrows indicate the direction of gene flow and numbers indicate percentage of genes inferred to contribute to a lineage.

Figure 27. Introgression model for the core dataset with C. hawaiensis x grayana. Hybrid edges (h=5) are indicated by red (minor hybrid edges) or green (major hybrid edges). Arrows indicate the direction of gene flow and numbers indicate percentage of genes inferred to contribute to a lineage.

Figure 28. Introgression model for the core dataset with C. longifolia x kauaiensis. Hybrid edges (h=2) are indicated by red (minor hybrid edges) or green (major hybrid edges). Arrows indicate the direction of gene flow and numbers indicate percentage of genes inferred to contribute to a lineage.

Figure 29. Introgression model for the core dataset with C. munroi x grayi. Hybrid edges

(h=4) are indicated by red (minor hybrid edges) or green (major hybrid edges). Arrows indicate the direction of gene flow and numbers indicate percentage of genes inferred to contribute to a lineage.

Figure 30. Introgression model for the core dataset with C. sandwicensis x grandiflora.

Hybrid edges (h=3) are indicated by red (minor hybrid edges) or green (major hybrid edges).

Arrows indicate the direction of gene flow and numbers indicate percentage of genes inferred to contribute to a lineage.

CHAPTER THREE

SUMMARY OF RESULTS AND FUTURE DIRECTIONS

The disparity between the morphology and genetics of Hawaiian Cyrtandra has confounded scientists for decades. Only recently has the cost and capability of technologies allowed us to investigate young lineages and those that are thought to hybridize readily in nature.

Here, we utilized targeted sequencing via the Hyb-Seq protocol (Weitemier et al. 2014) to sequence over 550 loci. We subsequently implemented traditional maximum likelihood methods in RAxML and modern phylogenomic methods in ASTRAL (Mirarab and Warnow 2015; Zhang et al. 2017), ASTRID (Vachaspati and Warnow 2015), and SNaQ (Solís-Lemus and Ané 2016) to recover robust hypotheses of species relationships and gene flow among species and putative hybrids.

In many cases, concatenated and coalescent analyses yield similar, if not congruent, hypotheses of species relationships. However, quartet scores (q1, q2, and q3) in our ASTRAL analysis were nearly equal in proportion at each node, indicating that this lineage exhibits very high levels of incomplete lineage sorting (ILS). In lineages where ILS is high, concatenated analyses have been shown to perform poorly, as they assume that all genes have the same evolutionary history (Kubatko and Degnan 2007). On the other hand, coalescent analyses perform much better when ILS is present (Liu et al. 2008; Degnan and Rosenberg 2009; Edwards

2009). Our ASTRAL and ASTRID analyses gave similar phylogenetic hypotheses (with some incongruence), and better support previous hypotheses of progression and back dispersal in this lineage (Clark et al. 2008, 2009; Johnson et al. 2017). Because simulation studies have shown that ASTRAL is likely to perform better in lineages with high levels of ILS, we used the species trees recovered in ASTRAL in comparisons to prior hypotheses of species relationships and in

subsequent analyses (Sayyari and Mirarab 2016). Our ASTRAL trees supported the hypotheses of progression of Cyrtandra from the oldest island (Kaua’i) to the youngest (Hawai’i; Clark et al.

2008, 2009; Johnson et al. 2017), as well as the back-dispersal of Cyrtandra from Maui Nui to

O’ahu (Johnson et al. 2017). However, in all of our analyses, there are poorly supported nodes

(BS < 50) which may be due to small sampling or insufficient data despite our use of high- throughput sequencing data.

Network analyses conducted in SNaQ (Solís-Lemus and Ané 2016) inferred hybridization in our dataset, though not always between taxa we expected. We conducted network analyses on nine datasets, including one with all taxa, one with the core dataset, and one with the core dataset with one putative hybrid included. Across these analyses, the most consistent hybrid-edge led from a Maui Nui or back-dispersed O’ahu taxon to the branch leading to C. wawrae and C. aff. wawrae, indicating that Cyrtandra may have back-dispersed from

O’ahu or Maui Nui to Kaua’i and hybridized with C. wawrae. Additionally, only one of the five putative hybrids included in this study (C. longifolia x kauaiensis) was inferred to be a hybrid of its putative parents. Network analyses inferred either no evidence of gene flow, or gene flow from unexpected parents, in the other hybrids.

Our results supported the prior hypotheses of progression and back-dispersal of

Cyrtandra in the Hawaiian Islands, as well as inferred some level of hybridization in this lineage.

The methods utilized in this study are useful for understanding the species relationships and magnitude, direction, and timing of gene flow in Hawaiian Cyrtandra. Additionally, our results call into question if the putative hybrids sampled here are simply mis-identified hybrids of species not sampled here, are morphologically diverse members of currently described species, or should be described as new species. Although species and putative hybrids of this lineage are

often identified in the field on the basis of morphology, our results indicate that how these species and putative hybrids are identified may need to be revised and may have implications for conservation of this group in this future.

The use of targeted enrichment via Hyb-Seq (Weitemier et al. 2014), high-throughput sequencing, coalescent methods, and network analyses can provide us with insight into the evolutionary history of Cyrtandra as well as the mechanisms of evolution in young lineages.

However, we acknowledge that wider sampling of sympatric and allopatric populations of species and putative hybrids from across Hawaii will yield greater insights than we can provide here. Future studies will likely focus on sampling at the population level and applying high- throughput sequencing and modern phylogenomic tools to these specimens to make sense of species relationships in Hawaiian Cyrtandra. Wider sampling is likely to give us better resolved phylogenetic hypotheses in this lineage, and is also likely to affect the inference of hybrid edges in network analyses. Furthermore, better resolved phylogenies and networks will allow us to investigate the evolution of character traits and conduct biogeographical analyses in this lineage.

Continued study of this group using cutting-edge techniques and methods is likely to yield fascinating insight into species relationships of this group and processes of ecology and evolution.

LITERATURE CITED

Clark, J., Ree, R., Alfaro, M., King, M., Wagner, W., Roalson, E.H. 2008. A Comparative Study

in Ancestral Range Reconstruction Methods: Retracing the Uncertain Histories of Insular

Lineages. Systematic Biology. 57:693–707.

Clark, J.R., Wagner, W.L., Roalson, E.H. 2009. Patterns of diversification and ancestral range

reconstruction in the southeast Asian–Pacific angiosperm lineage Cyrtandra

(Gesneriaceae). Molecular Phylogenetics and Evolution 53:982–994.

Degnan J.H., Rosenberg N.A. 2009. Gene tree discordance, phylogenetic inference and the

multispecies coalescent. Trends in Ecology & Evolution. 24:332–340.

Edwards S.V. 2009. Is a New and General Theory of Molecular Systematics Emerging?

Evolution. 63:1–19.

Johnson, M.A., Clark, J.R., Wagner, W.L., McDade, L.A. 2017. A molecular phylogeny of the

Pacific clade of Cyrtandra (Gesneriaceae) reveals a Fijian origin, recent diversification,

and the importance of founder events. Molecular Phylogenetics and Evolution 116:30–

48.

Kubatko L.S., Degnan J.H. 2007. Inconsistency of Phylogenetic Estimates from Concatenated

Data under Coalescence. Systematic Biology. 56:17–24.

Liu L., Pearl D.K., Brumfield R.T., Edwards S.V. 2008. Estimating Species Trees Using

Multiple-Allele DNA Sequence Data. Evolution. 62:2080–2091.

Weitemier, K., Straub, S.C.K., Cronn, R.C., Fishbein, M., Schmickl, R., McDonnell, A., Liston,

A. 2014. Hyb-Seq: Combining Target Enrichment and Genome Skimming for Plant

Phylogenomics. Applications in Plant Sciences 2:1400042.