And Yellow Fever Mosquito Aedes Aegypti (Linnaeus, 1762) in the Indo-Pacific

INVASION HISTORY AND EVOLUTION OF THE ASIAN TIGER MOSQUITO AEDES ALBOPICTUS (SKUSE, 1894) AND YELLOW FEVER MOSQUITO AEDES AEGYPTI (LINNAEUS, 1762) IN THE INDO-PACIFIC

ANDREW JAMES MAYNARD

Bachelor of Science (Honours Class I)

A thesis submitted for the degree of Doctor of Philosophy at

The University of Queensland in 2020

School of Biological Sciences

ABSTRACT

Mosquitoes cause more human deaths than any other animal in the world. Diseases such as dengue, chikungunya and Zika are transmitted to humans by two chief mosquito vectors: the yellow fever mosquito Aedes aegypti (Linneaus, 1762) and the Asian tiger mosquito Aedes albopictus (Skuse, 1894). Both species are highly invasive due to their close association with humans and ecological plasticity. They originate from different parts of the world, and have discrete behaviours, evolution, and ecology, but share key similarities in their global invasions, making them ideal candidates for investigating population genetic processes in invasive species. My dissertation explores population genetic patterns and attempts to reconstruct the invasion histories of these two species using microsatellites, gene sequencing, and whole-genome sequencing, specifically focusing on Australasia and Southeast Asia.

Within the last century, increases in human movement, globalisation, and trade have facilitated the establishment of several highly invasive mosquito species in new geographic locations causing major environmental, economic and health consequences. The Asian tiger mosquito, Ae. albopictus, has expanded globally in the last century, from its native range in Asia, chiefly due to increases of human movement. In Chapter II (https://doi.org/10.1371/journal.pntd.0005546), I used 13 nuclear microsatellite loci (911 individuals) and mitochondrial COI sequences to gain insight into the historical and contemporary movements of Ae. albopictus in the Indo-Pacific. Approximate Bayesian computation (ABC) was employed to test competing historical invasion routes within Southeast Asia and Australasia. I uncovered clear genetic clusters throughout the Indo-Pacific, but some geographically distant populations appear closely-related, likely due to human-associated movements. I also found that Ae. albopictus likely colonised New Guinea from mainland Southeast Asia, before spreading to the Solomon Islands via either Papua New Guinea (PNG) or Southeast Asia. In contrast, the recent (mid-2000s) incursion into northern Australia’s Torres Strait Islands likely stemmed from an Indonesian genetic source. These recently colonised populations displayed high spatio-temporal structure which could be due to genetic drift or represent a secondary invasion from an unknown source.

For Chapter III, I used similar tools to elucidate the invasion history and population genetics of Ae. aegypti across several populations in Southeast Asia and Australasia. This African native spread globally several centuries (~500 years ago) prior to Ae. albopictus mediated by global shipping. I used 11 nuclear microsatellites (366 individuals) and COI sequences to explore Ae. aegypti’s population structure and invasion history using more detailed analyses and ABC simulations than

i previous investigations. My results highlight that Ae. aegypti established in the region via multiple, independent invasions between the late-1700s and early 1900s. My research also revealed further genetic divisions between populations that had not previously been found. These results help to resolve the invasive origins of Ae. aegypti and act as a starting place for future studies.

In Chapter IV, I reanalysed some of the samples of Ae. albopictus used in Chapter II using a whole- genome sequencing approach. Until now, few studies have explored the genetics of this species using a genome-wide approach. To further investigate population structure and determine the sources of recently invaded regions, the genomes of 158 individuals from nine populations (one with temporal sampling) from Australasia and Southeast Asia were sequenced. Population patterns were mostly consistent with those obtained in Chapter II, but revealed less admixture between genetic clusters and clearer differentiation between populations than was shown with microsatellites. I found that certain genomic regions may be under selection in some native and introduced populations using sliding windows of FST across the genome. I estimated the demographic histories of populations using this genomic dataset to better understand how effective population size has varied over time to elucidate the biogeography of Ae. albopictus across Sundaland (exposed Southeast Asian landmass). I found strong support for post-glacial population isolation following the last glacial maximum (18-21 million years ago) when sea levels rose rapidly. Additionally, by analysing the mitochondrial genomes of my samples I was able to compare my dataset to a more global one, further detailing the relationships between various native and invasive populations of Ae. albopictus.

Overall, this thesis represents the most comprehensive analysis of the population structure and invasion history of Ae. albopictus and Ae. aegypti in Australasia and Southeast Asia using genetic approaches to date. It also represents the first investigation of the biogeographic history of Ae. albopictus in an unstudied region of its native and introduced range using genomic techniques. This knowledge is crucial to understand the invasion histories and current genetic population structure of these two medically significant species. I have been able to compare and critique the different population genetic approaches used here to inform others. My dissertation results could be useful for the successful deployment of control strategies and for identifying invasion pathways for biosecurity as it enhances our understanding of mosquito movements, population relatedness and invasion history.

ii DECLARATION BY AUTHOR

This thesis is composed of my original work, and contains no material previously published or written by another person except where due reference has been made in the text. I have clearly stated the contribution by others to jointly-authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance, survey design, data analysis, significant technical procedures, professional editorial advice, financial support and any other original research work used or reported in my thesis. The content of my thesis is the result of work I have carried out since the commencement of my higher degree by research candidature and does not include a substantial part of work that has been submitted to qualify for the award of any other degree or diploma in any university or other tertiary institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and, subject to the policy and procedures of The University of Queensland, the thesis be made available for research and study in accordance with the Copyright Act 1968 unless a period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright holder to reproduce material in this thesis and have sought permission from co-authors for any jointly authored works included in the thesis.

iii PUBLICATIONS DURING CANDIDATURE

Maynard, A.J., Ambrose, L., Cooper, R.D. Chow, W.K., Davis, J.B., Muzari, M.O., van den Hurk, A.F., Hall-Mendelin, S., Hasty, J.M., Burkot, T.R., Bangs, M.J., Reimer, L.J., Butafa, C., Lobo, N.F., Syafruddin, D., Maung, Y.N.M., Ahmad, R. & Beebe, N.W. (2017). Tiger on the Prowl: Spatio-temporal genetic structure of the Asian Tiger Mosquito Aedes albopictus (Skuse, 1894) in the Indo-Pacific. PLoS Neglected Tropical Diseases. 11(4): e0005546.

PUBLICATIONS INCLUDED IN THIS THESIS

SUBMITTED MANUSCRIPTS INCLUDED IN THIS

THESIS

No manuscripts submitted for publication.

iv CONTRIBUTIONS BY OTHERS TO THE THESIS

Luke Ambrose processed selected samples used in Chapters II and III. James Hereward provided major assistance and guidance for the bioinformatics part of the project used in Chapter IV. James Hereward, Caitlin Curtis and Luke Ambrose provided discussions regarding Chapter IV study conception, design and aims. Jacob Crawford provided sequencing in a collaborative effort that was used in Chapter IV; raw .bam files for 158 mosquitoes of Ae. albopictus were provided as an outcome. Nigel Beebe played a major role for formulating project aims for Chapter II and provided preliminary discussion for the development of Chapters III and IV. Samples were provided by several sources, most of which were co-ordinated by Nigel Beebe. Tom Burkot and Michael Bangs provided feedback and comments on the paper published from Chapter II and Michael Bangs additionally for Chapter III.

STATEMENT OF PARTS OF THE THESIS SUBMITTED

TO QUALIFY FOR THE AWARD OF ANOTHER

DEGREE

None.

RESEARCH INVOLVING HUMAN OR ANIMAL

SUBJECTS

Mosquito collections involving human-landing captures (HLC) from the Solomon Islands were approved by the Medical Research Ethics Committee in compliance with Australia’s National Statement on Ethical Conduct in Human Research (project no. 2011000603; Appendix). Collectors involved in HLC were taking anti-malarial medication and wore long-sleeved, protective clothing.

v ACKNOWLEDGEMENTS

First off, I would like to thank my supervisor Nigel Beebe for taking me on in his lab and for his continued support and approachability throughout my PhD. Thank you for networking me into the mosquito world and for keeping my PhD interesting and diverse with side gigs, field trips and travel.

To my co-supervisors Caitlin Curtis and James Hereward, I really would’ve struggled over the finish line and a few milestones if it weren’t for you both. Caitlin, thank you for continuously motivating me and for being so optimistic and uplifting. James, I wouldn’t have been able to get through the last data chapter without your bioinformatics wizardry and overall guidance, expertise and willingness to help me. I can’t thank you both enough.

Maddie James, James Wisdom and Luke Ambrose in particular, thank you for sharing all the ups and downs during my PhD and for providing useful advice or comments on drafts. Maddie James, my first true Brisbane/science friend. I’m honestly not sure I would’ve gotten far in my degree if I hadn’t met you early on. Boy am I glad I sat with you in that tutorial and that we just so happened to get along well, do the same major and take a similar path onto Honours and a PhD. Thank you for patiently listening to me whinge and for all the coffee walks over the last few years. I look forward to the day that you’re free of the PhD burden as well. James Wisdom, thank you for your good humour and for helping me live out my 20s in a proper, reckless fashion in those early care- free days. Luke Ambrose (bik brata Luke), I have very fond memories from field work together and I’m grateful you roped me into some of your trips. Hopefully we’ll get to go back to the Solomon Islands in the future. I couldn’t have had a better lab mate.

I’m extremely grateful for all the friends I’ve made along the way and how much they’ve all enriched this experience. Thank you to Carmen, Josh, Iva, Julian, Cara, Hugh, Brogan, Essie, Piet, Bec, Mel, Emily M., Andrew M., Penny and all the others I’ve hung with over the years.

All of those at UQ’s School of Biological Sciences – what an amazing school and connected group of people. All of the teaching staff that taught me and that I got to teach alongside have been a pleasure to work with. And the higher powers that keep ‘the roof’ going. I would like to also thank David Merritt and Lyn Cook for initially sparking my interest in entomology and evolutionary biology. John Hall and all of the UC Davis 2018 students – tutoring the field courses came at a time

vi when I was particularly unmotivated and going on those incredible trips and John’s style of teaching definitely reinspired me.

Thank you to everyone in Greg Devine’s group at QIMR and the Verily/CSIRO team, it was great working with you all and learning from experienced technicians and being part of a large and innovative project. Steve Whyard’s lab in Canada at the University of Manitoba – I had an unforgettable experience over there and thank you for being so warm in that cold, cold place. My colleagues at the Department of Agriculture for their support and encouragement throughout 2019 as I tried to finish the last of my PhD whilst working full-time. I appreciate the helpful information and collections that Michael Bangs has provided me with over the years. I would also like to thank all of the other collectors that have provided us with mosquitoes and without which these studies would not have been possible.

Of course, I have to thank the mosquitoes themselves, Aedes albopictus and Aedes aegypti, I hope my blood sacrifices and brief career as a mosquito care-taker have made up for the collections and swattings over the last several years.

A MASSIVE thank you to my family: Mum, Dad, Amy, Lauren and my Grandparents. Thank you for always being understanding, supportive and for nurturing my passion for nature and animals from a young age.

Lastly, thank you to Jessa for putting up with me during these trying times and for continuing to inspire me. Thanks for getting me away from the computer and back to my ento roots.

May my soul rest in peace.

vii FINANCIAL SUPPORT

This research was funded by an Australian Postgraduate Award funded by the Australian Federal Government. Research was additionally supported by the Commonwealth Scientific and Industrial Research Organisation (CSIRO) Cluster Collaboration Fund ‘Urbanism, Climate Change and Health’ as well as Western Australia Department of Health 'Funding initiatives for mosquito management in Western Australia' (FIMMWA, MBDC004). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript (Chapter II). I was awarded a Graduate School International Travel Award (2015, $3,282.00 AUD) from The University of Queensland, which I used to fund travel to The University of Manitoba in Winnipeg, Canada, between Jan-Feb 2015 where I met with my research collaborator Dr Steve Whyard and his lab to learn laboratory techniques. I was also awarded a Travel Award (2016, $1,600.00 AUD) from the School of Biological Sciences, The University of Queensland, which I used to fund travel to present my PhD research on Asian tiger mosquito population genetics at ICE 2016 XXV - International Congress of Entomology in Orlando, Florida.

KEYWORDS

Mosquito, population genetics, Aedes albopictus, Aedes aegypti, evolution, invasion, pest, entomology, genomics, biogeography.

AUSTRALIAN AND NEW ZEALAND STANDARD RESEARCH CLASSIFICATIONS (ANZSRC)

ANZSRC code: 060411, Population, Ecological and Evolutionary Genetics, 60% ANZSRC code: 060808, Invertebrate Biology, 20% ANZSRC code: 060207, Population Ecology, 20%

FIELDS OF RESEARCH (FoR) CLASSIFICATION

FoR code: 0604,Genetics, 60% FoR code: 0608, Zoology,20% FoR code: 0602, Ecology, 20%

viii TABLE OF CONTENTS

ABSTRACT ...... i

DECLARATION BY AUTHOR...... iii

PUBLICATIONS DURING CANDIDATURE ...... iv

PUBLICATIONS INCLUDED IN THIS THESIS ...... iv

SUBMITTED MANUSCRIPTS INCLUDED IN THIS THESIS ...... iv

CONTRIBUTIONS BY OTHERS TO THE THESIS ...... v

STATEMENT OF PARTS OF THE THESIS SUBMITTED TO QUALIFY FOR THE AWARD OF ANOTHER DEGREE ...... v

RESEARCH INVOLVING HUMAN OR ANIMAL SUBJECTS ...... v

ACKNOWLEDGEMENTS...... vi

FINANCIAL SUPPORT ...... viii

KEYWORDS ...... viii

AUSTRALIAN AND NEW ZEALAND STANDARD RESEARCH CLASSIFICATIONS (ANZSRC)...... viii

FIELDS OF RESEARCH (FoR) CLASSIFICATION ...... viii

CHAPTER I – GENERAL INTRODUCTION ...... 1 BIOLOGICAL INVASIONS ...... 1 A genetic approach ...... 1 Mosquitoes as invasive pests ...... 2 Aedes albopictus – A NEW THREAT ...... 3 A new threat ...... 3 Australia and surrounding areas ...... 3 Biology and vector status ...... 4 Invasion history and population genetics of Aedes albopictus ...... 5 Aedes aegypti – OUT OF AFRICA ...... 7 Complicated taxonomic and invasion histories ...... 7 Invasion history and population genetics of Aedes aegypti ...... 8 RESEARCH AIMS ...... 9 THESIS STRUCTURE ...... 10 REFERENCES ...... 11

CHAPTER II – Tiger on the Prowl: Invasion history and spatio-temporal genetic structure of the Asian tiger mosquito Aedes albopictus (Skuse, 1894) in the Indo-Pacific ...... 20 ABSTRACT ...... 21 INTRODUCTION ...... 22 MATERIALS & METHODS ...... 23 Ethics Statement ...... 23 Mosquito Samples...... 23 Microsatellite analysis ...... 27 Invasion History – ABC ...... 29 COI analysis ...... 32 RESULTS ...... 32 Microsatellite genetic diversity and population structure ...... 32 Invasion history – ABC ...... 35 COI haplotype networks and diversity ...... 36 Bottleneck tests ...... 40 DISCUSSION ...... 40 Population structure in the Indo-Pacific ...... 40 Invasion into Australasia ...... 43 Conclusion and Future Implications ...... 46 ACKNOWLEDGEMENTS ...... 47 REFERENCES ...... 47 SUPPLEMTARY FIGURES, TABLES & FILES ...... 56

CHAPTER III – Population structure and invasion origins of Aedes aegypti (Linnaeus, 1762) throughout Southeast Asia and Australasia ...... 75 ABSTRACT ...... 75 INTRODUCTION ...... 76 MATERIALS AND METHODS ...... 80 Collection sites and methods ...... 80 Microsatellite processing ...... 80 United States of America (USA) ...... 81 Microsatellite characteristics and genetic distance ...... 82 STRUCTURE analysis ...... 82 DAPC/K-means clustering ...... 83 COI analysis ...... 83 Invasion history...... 84 RESULTS ...... 86

Genetic diversity and gene flow (FST, Jost’s D & G”st) ...... 86 STRUCTURE ...... 89

DAPC ...... 91 COI ...... 93 Invasion history...... 95 DISCUSSION ...... 95 Invasion history in Australasia and Southeast Asia ...... 99 Future work and conclusion ...... 101 ACKNOWLEDGEMENTS ...... 102 REFERENCES ...... 102 SUPPLEMENTARY FIGURES & TABLES ...... 109

CHAPTER IV – Genome-wide SNPs reveal the demographic histories of native and invasive populations of Aedes albopictus (Skuse, 1894) in Australasia and Southeast Asia ...... 120 ABSTRACT ...... 120 INTRODUCTION ...... 121 MATERIALS AND METHODS ...... 123 Samples and sequencing ...... 123 Nuclear genome ...... 125

Sliding-window FST and population histories ...... 127 Mitochondrial genome ...... 127 RESULTS ...... 128 Mitochondrial genome ...... 128 Nuclear genome ...... 130

Sliding-window FST and population histories ...... 132 DISCUSSION ...... 134 Demographic history of Aedes albopictus in Australasia and Malaysia ...... 136 Future work and conclusion ...... 138 ACKNOWLEDGEMENTS ...... 139 REFERENCES ...... 139 SUPPLEMENTARY FIGURES & TABLES ...... 148

CHAPTER V – GENERAL DISCUSSION ...... 158 Problems with sampling ...... 159 Uncovering invasion histories ...... 160 The postgenomic era ...... 161 CONCLUSIONS AND FUTURE RESEARCH ...... 163 REFERENCES ...... 164

APPENDIX ...... 168

LIST OF FIGURES

Fig 2.1 ...... 26 Fig 2.2 ...... 30 Fig 2.3 ...... 38 Fig 3.1 ...... 79 Fig 3.2 ...... 86 Fig 3.3 ...... 89 Fig 3.4 ...... 90 Fig 3.5 ...... 92 Fig 3.6 ...... 94 Fig 4.1 ...... 124 Fig 4.2 ...... 129 Fig 4.3 ...... 133

LIST OF TABLES

Table 1.1...... 2 Table 2.1...... 24 Table 2.2...... 38 Table 2.3...... 39 Table 3.1...... 81 Table 3.2...... 88 Table 4.1...... 125 Table 4.2...... 131

CHAPTER I – GENERAL INTRODUCTION

BIOLOGICAL INVASIONS

A genetic approach

Our understanding of the biological characteristics which constitute successful invasive species is still developing, although it remains one of the most pressing questions in invasion biology (Kolar & Lodge, 2001; Lockwood et al., 2009). Recent research has attempted to capture how the genetic attributes of invasive species might influence their success, with some studies exploring how genetic variation in founding populations influences invasion success (Bock et al., 2015). Some invaders apparently arrive pre-adapted (including phenotypic plasticity) to their new environment, but the success of others is thought to rely on local adaptation, post-invasion. There are two main sources of genetic variation that facilitate adaptation: standing genetic variation and the gaining of new beneficial mutations. Because standing variation is already present in a founding population encountering novel environments, the speed of adaptation is more rapid than that resulting from new beneficial mutations, which often have an associated post-invasion lag phase (Barrett & Schluter, 2008; Bock et al., 2015). In contrast, new mutations are predicted to be much more beneficial to populations outside of their adaptive optimum (Fisher’s geometric model of adaptation; Fisher (1930)), which is often the case for many invasions.

The genetic makeup of an invading population can change over a short time frame (in tens of generations or fewer) due to the dramatic changes in environment and population demography that invaders experience (Chu et al., 2014; Fonseca et al., 2010; C. E. Lee et al. 2002; Prentis et al., 2008; Reznick & Ghalambor, 2001; Zhan et al., 2012). Alternatively, it can remain stable over time, despite low genetic diversity (Echodu et al., 2011; Guillemaud et al., 2011). In reality, invasions are complex and often difficult to detect and analyse at both temporal and spatial scales, especially when multiple colonisation events have occurred and when the strength of selection acting on the invasive lineage is not well known (Estoup & Guillemaud, 2010; Hermisson & Pennings, 2005). If the source of invasions can be identified and temporal monitoring of the population is carried out, it may be possible to gain insights into some of the evolutionary processes that underlie successful colonisations and to characterise spatio-temporal changes in genetic composition. Not only does this assist with the identification of invasion sources and pathways, but it may allow us to predict future colonisation events and design effective control strategies (Simberloff et al., 2013).

1 Mosquitoes as invasive pests

Many species of mosquitoes are amongst the most invasive pests in the world. They have a long history of human-mediated introductions (Bonizzoni et al., 2013; Powell & Tabachnick, 2013). These have resulted in the spread of major epidemics (see Table 1.1 for examples of major diseases spread by mosquitoes) and the establishment of invading mosquitoes as a biting nuisance. Species belonging to the Aedes genus are notable for their role in global disease transmission (Table 1.1). Two species: the yellow fever mosquito Aedes aegypti (Linneaus 1762) and the Asian tiger mosquito Aedes albopictus (Skuse, 1894) are mainly responsible for this. Both species are important vectors of these diseases because of their close association with humans. Due to increases in temporal and geographic spread of vectors, dengue fever has been identified by the World Health Organisation (WHO) as an issue that will demand close attention (WHO, 2019). The lengthening of wet seasons in some countries (such as India and Bangladesh) are exposing humans to mosquito- borne diseases for longer periods (WHO, 2019).

Population genetics can be extremely useful for inferring the source population of mosquito invasions and assessing the spread of insecticidal resistance (with important implications for mosquito control). This can allow state and federal authorities to more specifically identify invasion pathways or use more targeted control techniques (Brown et al., 2011; Endersby-Harshman et al., 2019; Maynard et al., 2017; Schmidt et al., 2019; Sherpa et al., 2019a; Sherpa et al., 2019b).

Table 1.1. Major diseases spread by different genera of mosquito.

Primary vertebrate Disease Chief vector Geographic distribution hosts Sub-Saharan Africa, Southeast Asia, Eastern Malaria Anopheles Humans Mediterranean, Western Pacific and the Americas Dengue fever Aedes Humans, primates Worldwide in the tropics Chikungunya Aedes Humans, primates Asia, Africa, Europe and the Americas Zika virus Aedes Humans, primates Africa, Southeast Asia, Pacific Islands, Americas Yellow fever Aedes Humans, primates Africa, South America West Nile virus Culex Birds Africa, Asia, Europe, USA Africa, Europe, the Middle East, North America Japanese encephalitis Culex Birds, pigs and West Asia Aedes, Anopheles, Sub-Saharan Africa, Southeast Asia, Pacific Lymphatic filariasis Humans Culex Islands, South America

2 AEDES ALBOPICTUS – A NEW THREAT

A new threat

The Asian tiger mosquito, Ae. albopictus, is a highly invasive and aggressive daytime-biting mosquito. It is emerging as a major public health threat throughout its invasive range (Bonizzoni et al., 2013). It is a competent vector for at least 26 arboviruses and it plays a significant role in the transmission and maintenance of dengue (DENV) and chikungunya (CHIKV) viruses (Paupy et al., 2009). While the species’ role in transmission cycles is often secondary to other mosquito vectors, chiefly the yellow fever mosquito Ae. aegypti, it has been the epidemic vector of recent DENV outbreaks in China, Central Africa, Hawaii and multiple Indian Oceans island (Rezza, 2012). Aedes albopictus was also responsible for the first autochthonous transmission of DENV in Europe since 1928 (Schaffner et al., 2013). More recently, the tiger mosquito was linked to the large dengue epidemic in the Solomon Islands that started in early 2013 (Nogareda et al., 2013; Shortus et al., 2016). In 2005 the Asian tiger mosquito was implicated during the global resurgence of CHIKV, an alphavirus clinically similar to DENV (Burt et al., 2012), where a mutation in the outer coat protein of the virus led to its more efficient uptake by Ae. albopictus. This resulted in outbreaks of chikungunya in Italy, Central Africa and several Indian Ocean islands (de Lamballerie et al., 2008; Paupy et al., 2010; Rezza et al., 2007). In 2013, this virus was moving through Papua New Guinea (PNG) transmitted by Ae. albopictus (Horwood et al., 2013a; Horwood et al., 2013b).

Australia and surrounding areas

Aedes albopictus was first detected in Australasia’s northern New Guinea region in the 1960s and was found in southern Papua New Guinea (PNG) 20 years later, just 150km from mainland Australia (Cooper et al., 1994; Elliott, 1980; Kay et al., 1990). In 2004, it was discovered for the first time in the Torres Strait just off northern Queensland and a year later was found to be widespread throughout the Torres Strait Islands (Ritchie et al., 2006). Should Ae. albopictus successfully invade the Australian mainland it has the potential to become established in much of Australia’s coastal and southern regions, extending as far as Tasmania (Hill et al., 2014; Nicholson et al., 2014a; Nicholson et al., 2014b). Because Ae. aegypti’s tropical biology restricts it from these cooler areas, an incursion of Ae. albopictus onto the mainland and into urban city landscapes would make them susceptible to the transmission of CHIKV and DENV during summer (van den Hurk, 2009). Additionally, Ae. albopictus’s savage biting behaviour would have a significant impact on Australia’s quality of life as it has in other invaded regions such as Italy, Spain and America (Curco et al., 2008; Darbro et al., 2017; Halasa et al., 2014; Romi, 2001; Worobey et al., 2013).

3 Based on the proximity of the Torres Strait Islands and high frequency of air and sea traffic between this area and mainland Australia, there is a high likelihood of a mainland invasion (Hill et al., 2014). However, entry via major international trading ports provides several opportunities for the species to enter Australia via a non-Torres Strait origin. To date, quarantine has successfully intercepted and eradicated the species in cities including Cairns, Darwin, Perth and Melbourne (Ritchie et al., 2006), although both pre- and post- border detections are common in Australia. Once in Australia the species’ plasticity in larval habitat selection and its zoophilic feeding behaviour will likely facilitate its rapid expansion in the country (Hill et al., 2014; Ritchie et al., 2006; Williams, 2012). Moreover, once established in urban and natural areas the species will be extremely difficult to control, particularly in the latter (Hawley, 1988; Unlu et al., 2011).

Biology and vector status

Aedes albopictus is endemic to the tropical and subtropical regions of Southeast Asia as well as Western Pacific and Indian Ocean islands. Its natural habitat consists mostly of forests and woodlands. The species also thrives in rural, suburban and urban areas and it has become abundant in many populous Asian cities such as Kuala Lumpur, Singapore and Tokyo (Hawley, 1988). When laying eggs, females seek water reservoirs (both indoors and outdoors) with still, clear to brackish water. This includes natural reservoirs such as tree holes, rock/stone pools, bamboo stumps, coconut shells/husks, leaf axils and pitcher plants as well as artificial sites such as used tyres, cans/bottles, barrels and water tanks.

Human habitation provides a range of artificial and more permanent breeding sites, but it also supplies an abundance of blood sources for this zoophilic and opportunistic feeder. Unlike Anopheles malaria vectors (which predominately feed at night), Ae. albopictus is a daytime-biting mosquito that is most active during the morning and evening (Bonizzoni et al., 2013; Hawley, 1988). In the urban landscape, females of Ae. albopictus will preferably obtain the bloodmeal required to develop eggs by feeding on humans, but also on domesticated animals such as dogs, cats, cows and goats (Kamgang et al., 2012). In more natural and rural areas, the mosquito feeds on a wider range of hosts including mammals, birds, reptiles and amphibians (Richards et al., 2006; Valerio et al., 2010).

While tiger mosquitos generally prefer well-vegetated and shaded outdoor sites in urban areas, the species has become more closely associated with humans and indoor activity in a domestication process similar to that of Ae. aegypti in Africa (Paupy et al., 2009; Powell & Tabachnick, 2013; Tabachnick, 1991). Unlike Ae. aegypti, which is more peri-domestic than Ae. albopictus, the tiger

4 mosquito may play an important role in the transfer of new strains of enzootic viruses from natural to human inhabited areas (Paupy et al., 2001). Additionally, Ae. aegypti is restricted to mostly tropical and subtropical habitats whereas Ae. albopictus can tolerate cooler temperate and subtemperate conditions due to its ability to diapause (Hawley et al., 1987; Urbanski et al., 2010a; Urbanski et al., 2010b; Urbanski et al., 2012). With its broader range of breeding environments (Ae. aegypti prefers indoor breeding sites whereas Ae. albopictus exploits both indoor and outdoor sites) this makes Ae. albopictus a strong coloniser that can use broader habitats than Ae. aegypti

The recent global expansion of Ae. albopictus has provided an ideal opportunity for mosquito-borne diseases to spread to new areas and adapt to this new host (Rezza, 2012). Convergent mutations in CHIKV improved the virus’s replication and transmission in Ae. albopictus, causing large outbreaks in several Indian Ocean islands and more northern regions such as Italy (de Lamballerie et al., 2008; Schuffenecker et al., 2006; Tsetsarkin et al., 2009). Grard et al. (2014) highlighted a similar scenario with recent outbreaks of the less known ZIKV, which is clinically similar to DENV and CHIKV, in Libreville, Gabon where Ae. albopictus was the only species positive for ZIKV and whose numbers largely predominated Ae. aegypti in Zika-affected suburbs. Under current and future climate models, Ae. albopictus is expected to become widespread in Australia and increase the range of these viruses (Hill et al., 2014; Kamal et al., 2018), thus it presents a significant threat to Australia’s public health (Hanna & McIver, 2018; Ryan et al., 2019).

Invasion history and population genetics of Aedes albopictus

The increase in human migration likely drove the early spread of Ae. albopictus from Southeast Asia toward the Indo-Malaysian Peninsula and Indian Ocean islands, whereas international trade likely spread Ae. albopictus more globally in the 20th century. In Europe, the species was first detected in 1979 in Albania and has since colonised numerous European countries including Italy, France, Greece, Spain, Switzerland, the Netherlands and Bosnia (to name a few) (Paupy et al., 2009). The species was found in South Africa in 1989 (Cornel & Hunt, 1991) and since has established in Nigeria, Equatorial Guinea, Gabon and Cameroon (Fontenille & Toto, 2001; Krueger & Hagen, 2007; Savage et al., 1992; Toto et al., 2003). Aedes albopictus has also invaded the Americas, ranging from the USA to Argentina, as well as numerous Pacific Islands (such as Hawaii, Fiji and the Solomon Islands) (Paupy et al., 2009). In Australasia, Ae. albopictus was found in Jayapura in the West Papua Province of Indonesia in 1963 and later detected in northern PNG near Madang in the 1970s (Cooper et al., 1994). By the 1980s, it had established in Port Moresby in southern PNG where it spread east to Bougainville Province and the Solomon Islands (Elliott,

5 1980). The tiger mosquito was detected in the coastal Fly Region, southern Western Province PNG, in 1988 and on the Torres Strait’s Daru Island.

Multiple studies have investigated the potential sources of some of these invasions. An allozyme study on North and South American populations of Ae. albopictus concluded that the USA and Brazilian introduced populations likely originated from Japan (Kambhampati et al., 1991). This was questioned more recently using mitochondrial markers because Brazilian populations were more closely related to Southeast Asian populations (Mousson et al., 2005). This is further supported by the lack of winter diapause in Brazilian Ae. albopictus (Hawley, 1988). Urbanelli et al. (2000) found that the temperate populations of Ae. albopictus from Italy, the USA and Japan form a distinct genetic cluster separate from Southeast Asian Ae. albopictus. Microsatellite and mitochondrial analyses revealed that multiple tropical sources led to the Ae. albopictus invasion of Cameroon, Africa (Kamgang et al., 2011). The results of many of these studies hint that genetic and physiological traits of an invading lineage play an important role in its colonising capacity, for instance tropical Ae. albopictus lineages appear to more efficiently establish in new tropical regions.

The Australian Torres Strait population of Ae. albopictus was initially hypothesised as originating from Southern PNG’s Fly Region as part of a range expansion (Beebe et al., 2013). However, the population genetic data from this study suggested that the incursion into the Torres Strait islands was more likely driven by a genetically distinct, non-PNG population, probably originating from the west (Indonesian region) (Beebe et al., 2013). Human-mediated movements, most likely in the form of fishing vessels from the west, drove this incursion. In support, haplotype networks using the cytochrome oxidase I (COI) gene region showed similarities between Indonesian populations of Ae. albopictus and those from the Torres Strait and Fly Region. There was also a high diversity of haplotypes in the Torres Strait suggesting a high degree of movement of females between islands.

In addition, the Fly Region showed considerable intermixing with the Torres Strait, supporting Ae. albopictus’s ability to exploit human movements as there is frequent traffic between the regions (Horwood et al., 2018). Multiple introductions of Ae. albopictus from the Indonesian region into the Torres Strait and Fly Region likely explain the high amount of genetic diversity in these recently colonised regions (Beebe et al., 2013). However, the exact origin/s of the Torres Strait incursion of Ae. albopictus remains unclear and needs to be addressed to better inform control programs. The use of genetic techniques to infer the routes and sources of invasive species provides us with critical information about the demographic history and genetic composition of founding populations (Beebe et al., 1999; Guillemaud et al., 2010; Schmidt et al., 2019; Sherpa et al., 2019a; Sherpa et al.,

6 2019b). However, the high dispersibility of Ae. albopictus mediated by human activities can make it challenging to detect genetic variation between populations (Mousson et al., 2005). Because of this, many (or highly variable) molecular markers are needed to explore Ae. albopictus population patterns at a fine scale.

AEDES AEGYPTI – OUT OF AFRICA

Complicated taxonomic and invasion histories

Historically, Ae. aegypti has had various synonyms (such as Stegomyia fasciatus, Culex bancrofti, Culex aegypti (Lee et al., 1980), but three major forms are typically recognised today based on variation in scaling colour/patterning and behavioural differences. These include the subspecies Ae. aegypti formosus (Walker), type form Ae. aegypti aegypti and varietal form Ae. aegypti var. queenslandensis (Theobald) (Mattingly, 1957). The ancestral, darkly coloured subspecies, Ae. aegypti formosus is found strictly in sub-Saharan Africa and prefers natural habitats and nonhuman blood. The globally-distributed, domesticated type forms are, Ae. aegypti aegypti and Ae. aegypti var. queenslandensis, which are both brown in colour, but with var. queenslandensis being the palest and more golden-brown. Both latter forms are more associated with humans than Ae. aegypti formosus, but appear to freely interbreed and are (currently) genetically inseparable (at least in Singapore and Australia) based on mitochondrial sequences and >16,000 nuclear single nucleotide polymorphisms (SNPs) (Rašić et al., 2016). Such genetic results support earlier conclusions by Mattingly (1957) and McClelland (1974) that Ae. aegypti queenslandesis should not be considered a valid subspecies.

The global spread of yellow fever mosquito, Ae. aegypti, appears to have commenced roughly 400- 550 years ago as a result of global trade, chiefly through shipping, which presented an ideal opportunity for the species to establish in non-native regions (Powell et al., 2018). Individual incursions of the species would have varied greatly and were likely seeded by multiple invasion waves (given the extraordinary number of historical ship movements from ~1500 CE to present). However, the general characteristics of Ae. aegypti’s global invasion based on historical records tend to involve the introduction and establishment of the species into major shipping ports, with conditions suitable for the mosquitoes’ container-breeding habit and indoor-feeding behaviour, before being spread coastally as a result of local human movements and further inland often by train lines or human boating movement upriver (Causey 1937; Daniels 1908; Hamlyn-Harris 1927; Smith 1956). Multiple studies have suggested that Africa was the likely native range and source of

7 Ae. aegypti that colonised the Americas between the 16th and 19th centuries (Failloux et al., 2002; Gloria‐Soria et al., 2016; Tabachnick, 1991), possibly during the early period of the Atlantic slave trade from West Africa (between 1500‐1650) (Kotsakiozi et al., 2018a; Powell et al., 2018). It was in the Americas the species gained its fierce reputation as the yellow fever mosquito (although the vector was not known for some time), leading to extensive outbreaks of the disease after its introduction from Africa where the virus is endemic (Gould et al., 2003; McNeill, 2010). Following the species’ arrival in the Americas, it is thought to have subsequently spread to islands of the Pacific Ocean, Australia and Asia between the 17-19th century.

Invasion history and population genetics of Aedes aegypti

The origins and timeframe of Ae. aegypti’s arrival to Asia and Australasia have been the subject of some ongoing debate, and the results of molecular studies appear to suggest conflicting results. Several molecular studies examining the global population genetics of the species have suggested that the more recent “Asian” invasion of Ae. aegypti (including the broader Asian region and some populations from the South Pacific and Australia) was sourced from the Americas rather than from Africa (Gloria‐Soria et al., 2016; Kotsakiozi et al., 2018b; Powell & Tabachnick, 2013), with the global spread of Ae. aegypti being described as ‘westward’ (from Africa, to the Americas and then into Asia and Australia via the Pacific) (Powell & Tabachnick, 2013). Patterns of reduced genetic diversity in Sri Lanka compared to Mexican and West African populations using exome sequencing also tend to support this notion (Crawford et al., 2017), a cline which has been reported by others (Bennett et al., 2016; Brown et al., 2014). More recently, it has been suggested that the Mediterranean region, rather than the Americas, could have been the source for the spread of ‘domesticated’ forms of Ae. aegypti to Asia, Australia and the South Pacific (Powell et al., 2018), as this coincides with the timing of autochthonous transmission of yellow fever around the Mediterranean Basin (beginning in the early 1800s). It also coincides with the opening of the Suez Canal in 1869 which created a direct shipping passage between the Mediterranean and Red Sea, bridging the North Atlantic and northern Indian oceans. This scenario fits with the timings derived from Gloria‐Soria et al. (2016), which used microsatellite data and approximate Bayesian computation (ABC) to estimate the split between ‘Asia’ and the ‘Americas,’ which occurred around the mid-19th century. In contrast, other results using ABC have supported alternative invasion routes (e.g. from West Africa to Asia to the Americas (Bennett et al., 2016)), although these scenarios had lower confidence in their results (see Table 3 in Bennett et al. (2016)). Overall, the routes of colonisation in Asia and Australasia are unresolved and modern reconstructions tend to over simplify the invasion into this region. For instance, combining Asian, Australian and Pacific populations into a single geographic region (Gloria‐Soria et al., 2016), when the colonisation of the

8 Asian-Pacific would have occurred through multiple jumps, probably with both westward (e.g. via the Americas) and eastward (e.g. via the Mediterranean, Africa, etc.) introductions. While the testing of simplistic invasion scenarios can be useful for extremely broad conclusions regarding the invasion history of Ae. aegypti, it it provides little information about the invasion history at a finer geographic scale, which can lead to overgeneralisations and misinformation about the historical spread of this medically-significant pest. Thus far, no studies have directly tested invasion scenarios at a finer level within the Southeast Asian/Australasian region for Ae. aegypti.

RESEARCH AIMS

The research presented in this thesis aims to shed light on the invasion process in these two key disease vectors, and the interaction between population genetics processes and invasion success more generally. Our knowledge is scant regarding many aspects of the population genetic structure and invasion histories of Ae. albopictus and Ae. aegypti across the vast geographical region of the Indo-Pacific and this thesis addresses this gap. I have also focused specifically on characterising the invasion process, from a genetic standpoint, in the recently established populations of Ae. albopictus in the Torres Strait Islands of Northern Australia. The specific aims of my research was to:

1. Describe the spatio-temporal genetic structure of both Ae. albopictus and Ae. aegypti in the region using microsatellites, targeted gene sequencing, and (for Ae. albopictus) whole-genome sequencing

2. Improve our understanding of the invasion history of both species in this region using genetic and observational data (i.e. time of introductions).

3. Use a genomic approach to explore the demographic histories and evolutionary processes of native and introduced populations of Ae. albopictus and enhance our understanding of the species’ biogeographic history.

4. Compare the traditional approach of population genetics analysis with microsatellite markers to a whole-genome sequencing approach with regards to understanding the evolution of disease vectors and biological invasions.

9 THESIS STRUCTURE

My thesis research has focused on Ae. albopictus and Ae. aegypti because of their importance to human health. Gaining genetic insights into the species’ invasion history and their genetic characteristics serves as a vital tool for their control and for predicting future movements of both species. This can be useful for predicting future genetic incursions between populations, which can have important implications for vector-status or insecticide resistance.

This thesis is presented as five chapters comprising of an introduction (Chapter I), three experimental chapters written in the format of completed or advanced drafts of scientific manuscripts (Chapters II-IV) and a general discussion (Chapter V):

Chapter I provides a general background and rationale to the research undertaken during my PhD.

Chapter II is a published manuscript (see Maynard et al. (2017)) that focuses on uncovering the invasion history of the Asian tiger mosquito in an important region of the species’ native and introduced range in the Indo-Pacific and exploring the spatio-temporal genetic structure of the mosquito to gain insight into the relationships between populations within the region, and how they may be changing over time due to evolutionary pressures such as genetic drift which may strongly shape the genetic structure of invading populations (due to founder effects and population bottlenecks). This work partly proceeds work by Beebe et al. (2013) but includes more rigorous sampling and empirically tests possible invasion scenarios in the region using approximate Bayesian computation whilst also considering the influence of temporal collections on population structures.

Chapter III explores a similar geographic region and uses a similar approach to Chapter II, but focuses on Ae. aegypti, with an additional effort to clarify the invasion history of this species within Southeast Asia and Australasia due to the degree of uncertainty and lack of specific studies that attempt to clarify this at a finer population level. I also more thoroughly explore several populations that have previously not been analysed. This chapter is being prepared to be submitted to a journal. Chapter IV builds upon Chapter II by using whole-genome sequencing to explore 158 individuals of Ae. albopictus from nine populations across Australasia and Southeast Asia (using representative samples of Ae. albopictus from Chapter II). Here, I aimed to further clarify regional population patterns using genome-wide SNPs and whole mitochondrial genomes, but to also compare how the population genetic patterns revealed with whole-genome sequencing differed from the results using

10 nuclear microsatellites and gene sequencing from Chapter II. Additionally, I explore the demographic history of this species across both its native and introduced ranges to elucidate the biogeographic and invasion history of this species. This research will serve as a powerful resource for future studies into Ae. albopictus genomics and will be submitted to a journal soon.

My final chapter (Chapter V) provides a general overview of my research and I discuss future directions and deliberations of this field of study.

REFERENCES

Barrett, R. D., & Schluter, D. (2008). Adaptation from standing genetic variation. Trends in Ecology & Evolution, 23(1), 38-44. Beebe, N. W., Ambrose, L., Hill, L. A., Davis, J. B., Hapgood, G., Cooper, R. D., . . . van den Hurk, A. F. (2013). Tracing the tiger: population genetics provides valuable insights into the Aedes (Stegomyia) albopictus invasion of the Australasian region. PLoS Neglected Tropical Diseases, 7(8), e2361. doi:10.1371/journal.pntd.0002361 Beebe, N. W., Ellis, J. T., Cooper, R. D., & Saul, A. (1999). DNA sequence analysis of the ribosomal DNA ITS2 region for the Anopheles punctulatus group of mosquitoes. Insect Molecular Biology, 8(3), 381-390. doi:10.1046/j.1365-2583.1999.83127.x Bennett, K. L., Shija, F., Linton, Y. M., Misinzo, G., Kaddumukasa, M., Djouaka, R., . . . Hlaing, T. (2016). Historical environmental change in Africa drives divergence and admixture of Aedes aegypti mosquitoes: a precursor to successful worldwide colonization? Molecular Ecology, 25(17), 4337-4354. Bock, D. G., Caseys, C., Cousens, R. D., Hahn, M. A., Heredia, S. M., Hübner, S., . . . Rieseberg, L. H. (2015). What we still don't know about invasion genetics. Molecular Ecology, 24(9), 2277-2297. Bonizzoni, M., Gasperi, G., Chen, X., & James, A. A. (2013). The invasive mosquito species Aedes albopictus: current knowledge and future perspectives. Trends in Parasitology, 29(9), 460- 468. doi:10.1016/j.pt.2013.07.003 Brown, J. E., Evans, B. R., Zheng, W., Obas, V., Barrera-Martinez, L., Egizi, A., . . . Powell, J. R. (2014). Human impacts have shaped historical and recent evolution in Aedes aegypti, the dengue and yellow fever mosquito. Evolution, 68(2), 514-525. doi:10.1111/evo.12281 Brown, J. E., Scholte, E.-J., Dik, M., Den Hartog, W., Beeuwkes, J., & Powell, J. R. (2011). Aedes aegypti mosquitoes imported into the Netherlands, 2010. Emerging Infectious Diseases, 17(12), 2335-2337. doi:10.3201/eid1712.110992

11 Burt, F. J., Rolph, M. S., Rulli, N. E., Mahalingam, S., & Heise, M. T. (2012). Chikungunya: a re- emerging virus. The Lancet, 379(9816), 662-671. Causey, O. R. (1937). Some Anopheline and Culicine mosquitoes of Siam with remarks on malaria control in Bangkok. American Journal of Hygiene, 25(2), 400-420. Chu, D., Guo, D., Tao, Y., Jiang, D., Li, J., & Zhang, Y. (2014). Evidence for rapid spatiotemporal changes in genetic structure of an alien whitefly during initial invasion. Scientific Reports, 4. Cooper, R., Waterson, D., Kupo, M., & Sweeney, A. (1994). Aedes albopictus (Skuse)(Diptera: Culicidae) in the Western Province of Papua New Guinea and the threat of its introduction to Australia. Australian Journal of Entomology, 33(2), 115-116. Cornel, A. J., & Hunt, R. H. (1991). Aedes albopictus in Africa? First records of live specimens in imported tires in Cape Town. Journal of the American Mosquito Control Association, 7(1), 107-108. Crawford, J. E., Alves, J. M., Palmer, W. J., Day, J. P., Sylla, M., Ramasamy, R., . . . Jiggins, F. M. (2017). Population genomics reveals that an anthropophilic population of Aedes aegypti mosquitoes in West Africa recently gave rise to American and Asian populations of this major disease vector. BMC Biology, 15(1), 16. Curco, N., Gimenez, N., Serra, M., Ripoll, A., Garcia, M., & Vives, P. (2008). Asian tiger mosquito bites: perception of the affected population after Aedes albopictus became established in Spain. Actas Dermo-Sifiliogràficas, 99(9), 708-713. Darbro, J., Halasa, Y., Montgomery, B., Muller, M., Shepard, D., Devine, G., & Mwebaze, P. (2017). An economic analysis of the threats posed by the establishment of Aedes albopictus in Brisbane, Queensland. Ecological Economics, 142, 203-213. doi:10.1016/j.ecolecon.2017.06.015 de Lamballerie, X., Leroy, E., Charrel, R. N., Ttsetsarkin, K., Higgs, S., & Gould, E. A. (2008). Chikungunya virus adapts to tiger mosquito via evolutionary convergence: a sign of things to come? Virology Journal, 5(1), 33. Echodu, R., Beadell, J. S., Okedi, L. M., Hyseni, C., Aksoy, S., & Caccone, A. (2011). Temporal stability of Glossina fuscipes fuscipes populations in Uganda. Parasites & Vectors, 4(1), 1- 10. doi:10.1186/1756-3305-4-19 Elliott, S. A. (1980). Aedes albopictus in the Solomon and Santa Cruz Islands, South Pacific. Transactions of the Royal Society of Tropical Medicine and Hygiene, 74(6), 747-748. Endersby-Harshman, N. M., Schmidt, T. L., Chung, J., van Rooyen, A., Weeks, A. R., & Hoffmann, A. A. (2019). Heterogeneous genetic invasions of three insecticide resistance mutations in Indo-Pacific populations of Aedes aegypti (L.). bioRxiv, 768549. doi:10.1101/768549

12 Estoup, A., & Guillemaud, T. (2010). Reconstructing routes of invasion using genetic data: why, how and so what? Molecular Ecology, 19(19), 4113-4130. doi:10.1111/j.1365- 294X.2010.04773.x Failloux, A.-B., Vazeille, M., & Rodhain, F. (2002). Geographic genetic variation in populations of the dengue virus vector Aedes aegypti. Journal of Molecular Evolution, 55(6), 653-663. Fisher, R. A. (1930). The genetical theory of natural selection: a complete variorum edition: Oxford University Press. Fonseca, D. M., Widdel, A. K., Hutchinson, M., Spichiger, S. E., & Kramer, L. D. (2010). Fine- scale spatial and temporal population genetics of Aedes japonicus, a new US mosquito, reveal multiple introductions. Molecular Ecology, 19(8), 1559-1572. doi:10.1111/j.1365- 294X.2010.04576.x Fontenille, D., & Toto, J. C. (2001). Aedes (Stegomyia) albopictus (Skuse), a potential new Dengue vector in southern Cameroon. Emerging Infectious Diseases, 7(6), 1066. Gloria‐Soria, A., Ayala, D., Bheecarry, A., Calderon‐Arguedas, O., Chadee, D. D., Chiappero, M., . . . Kamal, H. A. (2016). Global genetic diversity of Aedes aegypti. Molecular Ecology, 25, 5377-5395. Gould, E. A., de Lamballerie, X., Zanotto, P. M., & Holmes, E. C. (2003). Origins, evolution, and vector/host coadaptations within the genus Flavivirus (Vol. 59). Grard, G., Caron, M., Mombo, I. M., Nkoghe, D., Ondo, S. M., Jiolle, D., . . . Leroy, E. M. (2014). Zika virus in Gabon (Central Africa)–2007: a new threat from Aedes albopictus? PLoS Neglected Tropical Diseases, 8(2), e2681. Guillemaud, T., Beaumont, M. A., Ciosi, M., Cornuet, J. M., & Estoup, A. (2010). Inferring introduction routes of invasive species using approximate Bayesian computation on microsatellite data. Heredity, 104(1), 88-99. Guillemaud, T., Blin, A., Simon, S., Morel, K., & Franck, P. (2011). Weak spatial and temporal population genetic structure in the rosy apple aphid, Dysaphis plantaginea, in French apple orchards. PLoS one, 6(6), e21263. Halasa, Y. A., Shepard, D. S., Fonseca, D. M., Farajollahi, A., Healy, S., Gaugler, R., . . . Clark, G. G. (2014). Quantifying the impact of mosquitoes on quality of life and enjoyment of yard and porch activities in New Jersey. PLoS one, 9(3), e89221. Hanna, E. G., & McIver, L. J. (2018). Climate change: a brief overview of the science and health impacts for Australia. Medical Journal of Australia, 208(7), 311-315. Hawley, W. A. (1988). The biology of Aedes albopictus. Journal of the American Mosquito Control Association, 1, 1-39.

13 Hawley, W. A., Reiter, P., Copeland, R. S., Pumpuni, C. B., & Craig, G. B. (1987). Aedes albopictus in North America: probable introduction in used tires from northern Asia. Science, 236(4805), 1114-1116. Hermisson, J., & Pennings, P. S. (2005). Soft sweeps: molecular population genetics of adaptation from standing genetic variation. Genetics, 169(4), 2335-2352. doi:10.1534/genetics.104.036947 Hill, M. P., Axford, J. K., & Hoffmann, A. A. (2014). Predicting the spread of Aedes albopictus in Australia under current and future climates: Multiple approaches and datasets to incorporate potential evolutionary divergence. Austral Ecology, 39(4), 469-478. Horwood, P. F., Bande, G., Dagina, R., Guillaumot, L., Aaskov, J., & Pavlin, B. (2013a). The threat of chikungunya in Oceania. Western Pacific Surveillance Response Journal, 4(2), 8. Horwood, P. F., McBryde, E. S., Peniyamina, D., & Ritchie, S. A. (2018). The Indo-Papuan conduit: a biosecurity challenge for Northern Australia. Australian and New Zealand. Journal of Public Health, 42, 434-436. Horwood, P. F., Reimer, L. J., Dagina, R., Susapu, M., Bande, G., Katusele, M., . . . Siba, P. M. (2013b). Outbreak of chikungunya virus infection, Vanimo, Papua New Guinea. Emerging Infectious Diseases, 19(9), 1535. Kamal, M., Kenawy, M. A., Rady, M. H., Khaled, A. S., & Samy, A. M. (2018). Mapping the global potential distributions of two arboviral vectors Aedes aegypti and Ae. albopictus under changing climate. PLoS one, 13(12), e0210122. doi:10.1371/journal.pone.0210122 Kambhampati, S., Black IV, W. C., & Rai, K. S. (1991). Geographic origin of the US and Brazilian Aedes albopictus inferred from allozyme analysis. Heredity, 67(1), 85-94. Kamgang, B., Brengues, C., Fontenille, D., Njiokou, F., Simard, F., & Paupy, C. (2011). Genetic structure of the tiger mosquito, Aedes albopictus, in Cameroon (Central Africa). PLoS one, 6(5), e20257. doi:10.1371/journal.pone.0020257 Kamgang, B., Nchoutpouen, E., Simard, F., & Paupy, C. (2012). Notes on the blood-feeding behavior of Aedes albopictus (Diptera: Culicidae) in Cameroon. Parasites & Vectors, 5(1), 57. Kay, B. H., Fanning, I. D., Marks, E. N., Ives, W. A., Whelan, P. I., & Barker‐Hudson, P. (1990). Is Aedes albopictus in Australia? Medical Journal of Australia, 153(1), 31-34. Kolar, C. S., & Lodge, D. M. (2001). Progress in invasion biology: predicting invaders. Trends in Ecology & Evolution, 16(4), 199-204. Kotsakiozi, P., Evans, B. R., Gloria-Soria, A., Kamgang, B., Mayanja, M., Lutwama, J., . . . Powell, J. R. (2018a). Population structure of a vector of human diseases: Aedes aegypti in its ancestral range, Africa. Ecology and Evolution, 8(16), 7835-7848. doi:10.1002/ece3.4278

14 Kotsakiozi, P., Gloria-Soria, A., Schaffner, F., Robert, V., & Powell, J. R. (2018b). Aedes aegypti in the Black Sea: recent introduction or ancient remnant? Parasites & Vectors, 11(1), 396. doi:10.1186/s13071-018-2933-2 Krueger, A., & Hagen, R. M. (2007). first record of Aedes albopictus in Gabon, Central Africa. Tropical Medicine International Health, 12(9), 1105-1107. Lee, C. E. (2002). Evolutionary genetics of invasive species. Trends in Ecology & Evolution, 17(8), 386-391. Lee, D. J., Hicks, M., Griffiths, M., Russell, R., & Marks, E. (1980). The Culicidae of the Australasian Region. Volume IV: Australian Government Publishing Service. Linnaeus, C., & Hasselquist, F. (1762). Zweyter Theil, enthalt Beschreibungen verchiedener wuchtiger Naturalien. Hasselquist, F., Reise nach Palestina in den Jahren von, 1749, 267- 606. Lockwood, J. L., Cassey, P., & Blackburn, T. M. (2009). The more you introduce the more you get: the role of colonization pressure and propagule pressure in invasion ecology. Diversity and Distributions, 15(5), 904-910. Mattingly, P. F. (1957). Genetical aspects of the Aëdes aegypti problem. Annals of Tropical Medicine & Parasitology, 51(4), 392-408. doi:10.1080/00034983.1957.11685829 Maynard, A. J., Ambrose, L., Cooper, R. D., Chow, W. K., Davis, J. B., Muzari, M. O., . . . Beebe, N. W. (2017). Tiger on the prowl: Invasion history and spatio-temporal genetic structure of the Asian tiger mosquito Aedes albopictus (Skuse, 1894) in the Indo-Pacific. PLoS Neglected Tropical Diseases, 11(4), e0005546. doi:10.1371/journal.pntd.0005546 McClelland, G. (1974). A worldwide survey of variation in scale pattern of the abdominal tergum of Aedes aegypti (L.)(Diptera: Culicidae). Transactions of the Royal Entomological Society of London, 126(2), 239-259. McNeill, J. R. (2010). Mosquito empires: ecology and war in the Greater Caribbean, 1620-1914: Cambridge University Press. Mousson, L., Dauga, C., Garrigues, T., Schaffner, F., Vazeille, M., & Failloux, A. B. (2005). Phylogeography of Aedes (Stegomyia) aegypti (L.) and Aedes (Stegomyia) albopictus (Skuse) (Diptera: Culicidae) based on mitochondrial DNA variations. Genetic Research, 86(1), 1-11. doi:10.1017/s0016672305007627 Nicholson, J., Ritchie, S., Russell, R., Zalucki, M., & van den Hurk, A. (2014a). Ability for Aedes albopictus (Diptera: Culicidae) to survive at the climatic limits of its potential range in eastern Australia. Journal of Medical Entomology, 51(5), 948-957.

15 Nicholson, J., Ritchie, S., & van den Hurk, A. (2014b). Aedes albopictus (Diptera: Culicidae) as a potential vector of endemic and exotic arboviruses in Australia. Journal of Medical Entomology, 51(3), 661-669. Nogareda, F., Joshua, C., Sio, A., Shortus, M., Dalipanda, T., Durski, K., . . . Aaskov, J. (2013). Ongoing outbreak of dengue serotype-3 in Solomon Islands, January to May 2013. Western Pacific Surveillance Response Journal, 4(3), 28. Paupy, C., Delatte, H., Bagny, L., Corbel, V., & Fontenille, D. (2009). Aedes albopictus, an arbovirus vector: from the darkness to the light. Microbes and Infection, 11(14–15), 1177- 1185. doi:10.1016/j.micinf.2009.05.005 Paupy, C., Girod, R., Salvan, M., Rodhain, F., & Failloux, A. B. (2001). Population structure of Aedes albopictus from La Reunion Island (Indian Ocean) with respect to susceptibility to a dengue virus. Heredity, 87(3), 273. Paupy, C., Ollomo, B., Kamgang, B., Moutailler, S., Rousset, D., Demanou, M., . . . Simard, F. (2010). Role of Ae. albopictus versus Ae. aegypti in dengue and chikungunya emergences in Central Africa: Laboratory and field evidence. Vector-Borne and Zoonotic Diseases, 10. doi:10.1089/vbz.2009.0005 Powell, J. R., Gloria-Soria, A., & Kotsakiozi, P. (2018). Recent history of Aedes aegypti: vector genomics and epidemiology records. BioScience, 68(11), 854-860. Powell, J. R., & Tabachnick, W. J. (2013). History of domestication and spread of Aedes aegypti - A Review. Memórias do Instituto Oswaldo Cruz, 108(Suppl 1), 11-17. doi:10.1590/0074- 0276130395 Prentis, P. J., Wilson, J. R., Dormontt, E. E., Richardson, D. M., & Lowe, A. J. (2008). Adaptive evolution in invasive species. Trends in Plant Science, 13(6), 288-294. Rašić, G., Filipović, I., Callahan, A. G., Stanford, D., Chan, A., Lam-Phua, S. G., . . . Hoffmann, A. A. (2016). The queenslandensis and the type form of the dengue fever mosquito (Aedes aegypti L.) are genomically indistinguishable. PLoS Neglected Tropical Diseases, 10(11), e0005096. doi:10.1371/journal.pntd.0005096 Reznick, D. N., & Ghalambor, C. K. (2001). The population ecology of contemporary adaptations: what empirical studies reveal about the conditions that promote adaptive evolution. Genetica, 112(1), 183-198. Rezza, G. (2012). Aedes albopictus and the reemergence of Dengue. BMC Public Health, 12(1), 72. Rezza, G., Nicoletti, L., Angelini, R., Romi, R., Finarelli, A., Panning, M., . . . Magurano, F. (2007). Infection with chikungunya virus in Italy: an outbreak in a temperate region. The Lancet, 370(9602), 1840-1846.

16 Richards, S. L., Ponnusamy, L., Unnasch, T. R., Hassan, H. K., & Apperson, C. S. (2006). Host- feeding patterns of Aedes albopictus (Diptera: Culicidae) in relation to availability of human and domestic animals in suburban landscapes of central North Carolina. Journal of Medical Entomology, 43(3), 543-551. Ritchie, S. A., Moore, P., Carruthers, M., Williams, C., Montgomery, B., Foley, P., . . . Cooper, B. (2006). Discovery of a widespread infestation of Aedes albopictus in the Torres Strait, Australia. Journal of the American Mosquito Control Association, 22(3), 358-366. Romi, R. (2001). Aedes albopictus in Italy: an underestimated health problem. Annali Dell'istituto Superiore di Sanita, 37(2), 241-247. Ryan, S. J., Carlson, C. J., Mordecai, E. A., & Johnson, L. R. (2019). Global expansion and redistribution of Aedes-borne virus transmission risk with climate change. PLoS Neglected Tropical Diseases, 13(3), e0007213. doi:10.1371/journal.pntd.0007213 Savage, H. M., Ezike, V. I., Nwankwo, A. C., Spiegel, R., & Miller, B. R. (1992). First record of breeding populations of Aedes albopictus in continental Africa: implications for arboviral transmission. Journal of the American Mosquito Control Association, 8(1), 101-103. Schaffner, F., Medlock, J. M., & Van Bortel, W. (2013). Public health significance of invasive mosquitoes in Europe. Clinical Microbiology and Infection, 19. doi:10.1111/1469- 0691.12189 Schmidt, T. L., Van Rooyen, A. R., Chung, J., Endersby‐Harshman, N. M., Griffin, P. C., Sly, A., . . . Weeks, A. R. (2019). Tracking genetic invasions: genome‐wide SNP s reveal the source of pyrethroid‐resistant Aedes aegypti (yellow fever mosquito) incursions at international ports. Evolutionary Applications. Schuffenecker, I., Iteman, I., Michault, A., Murri, S., Frangeul, L., Vaney, M.-C., . . . Pettinelli, F. (2006). Genome microevolution of chikungunya viruses causing the Indian Ocean outbreak. PLoS Medicine, 3(7), e263. Sherpa, S., Blum, M. G., Capblancq, T., Cumer, T., Rioux, D., & Després, L. (2019a). Unravelling the invasion history of the Asian tiger mosquito in Europe. Molecular Ecology, 28(9), 2360- 2377. Sherpa, S., Guéguen, M., Renaud, J., Blum, M. G. B., Gaude, T., Laporte, F., . . . Després, L. (2019b). Predicting the success of an invader: Niche shift versus niche conservatism. Ecology and Evolution, 9(22), 12658-12675. doi:10.1002/ece3.5734 Shortus, M., Musto, J., Bugoro, H., Butafa, C., Sio, A., & Joshua, C. (2016). Vector-control response in a post-flood disaster setting, Honiara, Solomon Islands, 2014. Western Pacific surveillance and response journal, 7(1), 38-43. doi:10.5365/WPSAR.2015.6.3.004

17 Simberloff, D., Martin, J. L., Genovesi, P., Maris, V., Wardle, D. A., Aronson, J., . . . Pascal, M. (2013). Impacts of biological invasions: what's what and the way forward. Trends in Ecology & Evolution, 28(1), 58-66. Skuse, F. A. A. (1894). The banded mosquito of Bengal. Indian Museum Notes, 3(5). Tabachnick, W. J. (1991). Evolutionary genetics and arthropod-borne disease: the yellow fever mosquito. American Entomologist, 37(1), 14-26. Toto, J. C., Abaga, S., Carnevale, P., & Simard, F. (2003). First report of the oriental mosquito Aedes albopictus on the West African island of Bioko, Equatorial Guinea. Medical Veterinary Entomology, 17(3), 343-346. Tsetsarkin, K. A., McGee, C. E., Volk, S. M., Vanlandingham, D. L., Weaver, S. C., & Higgs, S. (2009). Epistatic roles of E2 glycoprotein mutations in adaption of chikungunya virus to Aedes albopictus and Ae. aegypti mosquitoes. PLoS one, 4(8), e6835. Unlu, I., Farajollahi, A., Healy, S. P., Crepeau, T., Bartlett‐Healy, K., Williges, E., . . . Fonseca, D. M. (2011). Area‐wide management of Aedes albopictus: choice of study sites based on geospatial characteristics, socioeconomic factors and mosquito populations. Pest Management Science, 67(8), 965-974. Urbanelli, S., Bellini, R., Carrieri, M., Sallicandro, P., & Celli, G. (2000). Population structure of Aedes albopictus (Skuse): the mosquito which is colonizing Mediterranean countries. Heredity, 84(3), 331-337. Urbanski, J. M., Aruda, A., & Armbruster, P. (2010a). A transcriptional element of the diapause program in the Asian tiger mosquito, Aedes albopictus, identified by suppressive subtractive hybridization. Journal of Insect Physiology, 56(9), 1147-1154. Urbanski, J. M., Benoit, J. B., Michaud, M. R., Denlinger, D. L., & Armbruster, P. (2010b). The molecular physiology of increased egg desiccation resistance during diapause in the invasive mosquito, Aedes albopictus. Proceedings of the Royal Society B: Biological Sciences, 277(1694), 2683-2692. Urbanski, J. M., Mogi, M., O’Donnell, D., DeCotiis, M., Toma, T., & Armbruster, P. (2012). Rapid adaptive evolution of photoperiodic response during invasion and range expansion across a climatic gradient. The American Naturalist, 179(4), 490-500. Valerio, L., Marini, F., Bongiorno, G., Facchinelli, L., Pombi, M., Caputo, B., . . . della Torre, A. (2010). Host-feeding patterns of Aedes albopictus (Diptera: Culicidae) in urban and rural contexts within Rome province, Italy. Vector-Borne Zoonotic Diseases, 10(3), 291-294. van den Hurk, A. F. (2009). The ‘Tiger’on our doorstep: emergence of Aedes albopictus as an arbovirus vector in northern Australia. Microbiology Australia, 30(4), 142-144. WHO. (2019). Retrieved from www.who.int/emergencies/ten-threats-to-global-health-in-2019

18 Williams, C. R. (2012). The Asian tiger mosquito (Aedes albopictus) invasion into Australia: a review of likely geographic range and changes to vector-borne disease risk. Transactions of the Royal Society of South Australia, 136(2), 128-136. doi:10.1080/03721426.2012.10887167 Worobey, J., Fonseca, D. M., Espinosa, C., Healy, S., & Gaugler, R. (2013). Child outdoor physical activity is reduced by prevalence of the Asian tiger mosquito, Aedes albopictus. Journal of the American Mosquito Control Association, 29(1), 78-81. Zhan, A., Darling, J. A., Bock, D. G., Lacoursière-Roussel, A., MacIsaac, H. J., & Cristescu, M. E. (2012). Complex genetic patterns in closely related colonizing invasive species. Ecology and Evolution, 2(7), 1331-1346. doi:10.1002/ece3.258

19 CHAPTER II – Tiger on the Prowl: Invasion history and spatio-temporal genetic structure of the Asian tiger mosquito Aedes albopictus (Skuse, 1894) in the Indo-

Pacific

Published chapter: Maynard, A.J., Ambrose, L., Cooper, R.D. Chow, W.K., Davis, J.B., Muzari, M.O., van den Hurk, A.F., Hall-Mendelin, S., Hasty, J.M., Burkot, T.R., Bangs, M.J., Reimer, L.J., Butafa, C., Lobo, N.F., Syafruddin, D., Maung, Y.N.M., Ahmad, R. & Beebe, N.W. (2017). Tiger on the Prowl: Spatio-temporal genetic structure of the Asian Tiger Mosquito Aedes albopictus (Skuse, 1894) in the Indo-Pacific. PLoS Neglected Tropical Diseases. 11(4): e0005546.

The population genetic aspect of this study was conceived by myself and my supervisor Nigel Beebe. The invasion history aspect was greatly expanded as a result of reviewer suggestions and the approach conceived by myself. All analyses were carried out by myself. Luke Ambrose scored select microsatellite samples (to check for consistency) and had previously amplified some samples (5%). The entire submission process was carried out by myself (e.g. submitting the manuscript, replying to reviewers, conducting revisions, etc.). The draft and final versions of the manuscript were written by myself, with comments and edits on the draft from co-authors Tom Burkot, Michael Bangs, Nigel Beebe and Luke Ambrose, and fellow colleagues Maddie James, Caitlin Curtis and James Wisdom. All other co-authors were included for providing invaluable samples, which were co-ordinated largely through Nigel Beebe.

Contributor Statement of contribution

Andrew Maynard Conception Data collection: 95% Analysis: 100% Draft: 100% Final version: 100%

Luke Ambrose Collected data: 5% Comments on draft

Tom Burkot Comments on draft

Michael Bangs Comments on draft

Nigel Beebe Conception, co-ordinated sampling and comments on draft

Other co-authors Provided samples and approved of manuscript

20 ABSTRACT

Within the last century, increases in human movement and globalisation of trade have facilitated the establishment of several highly invasive mosquito species in new geographic locations with concurrent major environmental, economic and health consequences. The Asian tiger mosquito, Aedes albopictus, is an extremely invasive and aggressive daytime-biting mosquito that is a major public health threat throughout its expanding range. We used 13 nuclear microsatellite loci (on 911 individuals) and mitochondrial COI sequences to gain a better understanding of the historical and contemporary movements of Ae. albopictus in the Indo-Pacific region and to characterise its population structure. Approximate Bayesian computation (ABC) was employed to test competing historical routes of invasion of Ae. albopictus within the Southeast (SE) Asian/Australasian region. Our ABC results show that Ae. albopictus was most likely introduced to New Guinea via mainland Southeast Asia, before colonizing the Solomon Islands via either Papua New Guinea or SE Asia. The analysis also supported that the recent incursion into northern Australia’s Torres Strait Islands was seeded chiefly from Indonesia. For the first time documented in this invasive species, we provide evidence of a recently colonised population (the Torres Strait Islands) that has undergone rapid temporal changes in its genetic makeup, which could be the result of genetic drift or represent a secondary invasion from an unknown source. There appears to be high spatial genetic structure and high gene flow between some geographically distant populations. The species' genetic structure in the region tends to favour a dispersal pattern driven mostly by human movements. Importantly, this study provides a more widespread sampling distribution of the species’ native range, revealing more spatial population structure than previously shown. Additionally, we present the most probable invasion history of this species in the Australasian region using ABC analysis.

21 INTRODUCTION

Many species of mosquitoes are amongst the most invasive pests in the world. They have a long history of human-mediated introductions [1, 2] that have resulted in the spread of major epidemics (malaria, dengue, Zika, etc.) and the establishment of invading mosquitoes as a biting nuisance. The Asian tiger mosquito, Aedes albopictus (Skuse, 1894) [3], is regarded as one of the most invasive mosquitoes in the world [4]. Native to tropical and subtropical Asia and multiple Western Pacific and Indian Ocean islands, Ae. albopictus now has a pan-global distribution [5-9]. Its initial movement from Southeast (SE) Asia toward the Indo-Malayan Peninsula and Indian Ocean islands may have resulted from the increase in human migration during the 17th and 18th centuries, with international trade (particularly the used-tire and ornamental bamboo trades) further facilitating its global spread in the 20th century [10]. It is among the primary vectors of several globally expanding and medically important arthropod borne viruses (arboviruses) – particularly dengue, chikungunya, yellow fever, and Zika – while also able to transmit at least 23 other arboviruses and canine heartworm [10]. Whilst at present, the yellow fever mosquito Aedes aegypti (Linnaeus, 1762) [11] is responsible for most of the transmission of some of these important arboviruses, the increased cold tolerance of Ae. albopictus relative to Ae. aegypti suggests that it could extend the range of many of these diseases under the right circumstances [12, 13].

Genetic techniques can provide critical information to infer the routes and sources of invasive species as well as informing on the demographic history and genetic composition of founding populations [8, 14]. For mosquito vectors, this knowledge is not only useful for inferring invasion routes in order to focus biosecurity efforts, it can also inform us of the colonizing capacity, adaptability and behaviour of invading mosquito lineages [15, 16]. However, the high dispersibility of Ae. albopictus mediated by human activities can make it challenging to detect genetic variation between populations due to high gene flow facilitated by these activities [17]. Microsatellite markers have been shown to be useful for exploring Ae. albopictus genetic patterns as they evolve rapidly and can often detect subtle population structure [8, 18, 19]. Recently, approximate Bayesian analysis has proven a powerful tool in testing the probability of competing invasion scenarios. This can provide us with crucial information regarding the timing and origin of mosquito introductions, which has been used recently for Aedes mosquitoes [20-22].

Overall, the population genetics of Ae. albopictus through the SE Asian-Indo-Pacific region requires further exploration and samples from this region (particularly Australasia) are often lacking from global population genetic studies, despite the importance of this region for vector research

22 [23-25]. A study by Beebe, Ambrose (18) explored part of the Indo-Australasian invasion by Ae. albopictus and provided an interesting scenario where there is high human connectivity (largely maritime) spanning both oceanic barriers and complex geographic landscapes. The current study expands on this work to include SE Asian native populations (Myanmar, Thailand, Malaysia, Singapore, Indonesia), as well as several younger populations that appear to have been introduced within the last six decades (Papua, Papua New Guinea (PNG), Solomon Islands, Fiji, Christmas Is., Cocos (Keeling) Is., Nauru, Torres Strait Islands (Australia)). Additionally, we included populations from the United States (USA) and northern Asia (only for COI) to see how these populations fit into a broader geographic analysis. We use previously developed nuclear microsatellite markers [18] and mitochondrial cytochrome c oxidase subunit I (COI) sequences to investigate the population genetics of the Ae. albopictus within the Indo-Pacific region. Our primary aims were to uncover the most likely historical invasion route of Ae. albopictus into the Australasian region as well as to detail the genetic connectivity and population structure of Ae. albopictus throughout this broad geographic region that we refer to as the Indo-Pacific (the aforementioned populations). While there are some records of the progressive establishment of Ae. albopictus throughout this region, the origin/s and the precise timing of introductions require testing using genetic methods under a coalescent-based approach such as ABC analysis. Our secondary aim was to further investigate the 2005 colonisation of the Torres Strait Islands, Australia [26]. Many of the Torres Strait Islands have undergone intense spraying efforts since the establishment of Ae. albopictus and the region experiences monsoon-dry seasons leading to regular population bottlenecks [27, 28]. We hypothesised that the genetic changes in neutral alleles may be detectable over time in these newly invaded and small island populations.

MATERIALS & METHODS

Ethics Statement

Mosquito collections involving HLC from the Solomon Islands (Table S2.1) were approved by the Medical Research Ethics Committee in compliance with Australia’s National Statement on Ethical Conduct in Human Research (project no. 2011000603). Collectors involved in HLC took anti- malarial medication and wore long-sleeved, protective clothing.

Mosquito Samples

Both adult and larval samples were collected throughout Australasia, SE Asia, Indian Ocean and Pacific Ocean islands as well as in the United States (Table 2.1, Fig 2.1 (orange dots)). Samples

23 were stored in 70% ethanol or dried (adults) over silica beads. Samples were collected using human landing captures (HLC), human baited sweep netting, egg collections and sampling of aquatic habitats for larvae and pupae (Table S2.1). For identification purposes some samples were reared to adults after field collection (Table S2.1). The logistics of sourcing material across multiple international borders resulted in variability in collection methods and sample sizes (Table 2.1, Table S2.1). Adult mosquitoes were identified morphologically [29] and for larval/pupal samples using either real-time PCR assays [30] or a PCR-restriction digest [31] to differentiate from Aedes scutellaris (Walker 1858) [32].

Table 2.1. Sample information for Aedes albopictus used in the microsatellite study, where n indicates the number of individuals per population (ntotal = 911, npop = 50). Region/description shows broader geographic regions and descriptions referred to in text. Population indicates more specific collection sites and the year of collection in brackets.

Region/description Population (year) n Abbreviation DAPC Abbreviation Torres Strait Islands (invasion) Masig (2007) 21 Mas '07 Mer (2007) 6 Mer '07 Warraber (2007) 8 War '07 Mabuiag (2007) 10 Mab '07 TS Waiben (2010) 3 Wai '10 Ngurupai (2010) 7 Ngu '10 Muralug (2010) 5 Mur '10 Torres Strait Islands (post invasion) Ngurupai (2012) 10 Ngu '12 Ngu '12 Keriri (2012) 23 Ker '12 Ker '12 Keriri (2013) 10 Ker '13 Ker '13 Keriri (2014) 30 Ker '14 Ker '14 Poruma (2015) 30 Por '15 Por '15 Iama (2015) 30 Iam '15 Iam '15 Warraber (2015) 24 War '15 War '15 Southern Fly Region (PNG) Kulalai (2007) 2 Kul '07 Mabaduan (2007) 7 Mab '07 FLY Sigabaduru (2007) 1 Sig '07 Katatai (2008) 2 Kat '08 Papua New Guinea Kiunga (1992) 17 KIU Port Moresby (1996) 2 PM '96 Port Moresby (1997) 2 PM '97 Port Moresby (1998) 17 PM '98 Port Moresby (1999) 8 PM '99 PNG Madang (2011) 33 MAD Daru (1992) 6 DAR '92 Daru (2008) 20 DAR '08 Lihir Is. (2007) 39 LIH Buka Is. (1999) 14 BUK

24 Papua, Indonesia Timika (2015) 20 TIM TIM Timor-Leste Timor-Leste (2001) 10 TL TL Indonesia Jakarta (2012) 177 JAK JAK Sumba (2013) 37 SUM SUM Singapore Singapore (2013) 4 SIN Malaysia Ipoh (2013) 48 IPO SIN/MAL Kota Baru (2013) 7 KOT Kuala Lumpur (2015) 64 KL Thailand Bangkok (2015) 6 BAN Myanmar Yangon (2013) 13 YAN BAN/MYA East Shan State (2013) 5 ESS Christmas Island Christmas Is. (2008) 10 CH CH Cocos (Keeling) Islands Direction Is. (2008) 18 CK CK La Réunion La Réunion (2011) 4 REU REU Solomon Islands Honiara (2013) 23 HON Gizo (2013) 4 GIZ '13 SOL Saeragi village, Gizo (2014) 18 GIZ '14 New Mala (2014) 10 NEW Fiji Fiji (2015) 5 FIJ FIJ Nauru Nauru (2014) 2 NAU NAU Hawaii Hawaii (2015) 22 HAW HAW/ATL USA (mainland) Atlanta (2011) 17 ATL

Fig 2.1. Bayesian STRUCTURE plot (K=4) for 13 microsatellite loci for 911 samples of Aedes albopictus in the study region. Each vertical bar in the plots represents an individual sample, where the colour of the bar indicates the probability of the individual belonging to a genetic cluster. Samples are positioned on the map corresponding to the population’s location (orange dot) and are abbreviated as in Table 2.1. Map insets represent the following: A) Torres Strait Islands and Southern Fly Region of Papua New Guinea; B) Hawaii; C) Atlanta. Insets B and C are to scale with the main map scale. The top-left colour key shows the colour of clusters, as referred to in the main text.

26 Microsatellite analysis

DNA was salt extracted [33] and diluted at 1:10 with 1X TE buffer (Tris, EDTA). Thirteen nuclear microsatellite were used in this study. These markers were previously developed [18] and include two dinucleotide and 11 trinucleotide loci (see Beebe, Ambrose (18) for loci and primers). Some samples included from the previous study were amplified using a variation of the master mix (see Beebe, Ambrose (18)), other samples were amplified in a 15.4μl reaction that consisted of 10.8μl

H2O, 3μl 5X Mytaq buffer (Bioline, containing 5mM dNTPs and 15mM MgCl2), 0.1μl 10μM M13 tagged forward primer, 0.2μl 10μM reverse primer, 0.2μl M13 tagged fluorescent dye (VIC, NED, PET or FAM; Table S2.1), 0.01μl (1U) MyTaq polymerase and 1μl 1:10 DNA template. PCR cycling used the same protocol as in Beebe, Ambrose (18). Amplification was verified by running 1μl of PCR product on a 2% agarose gel stained with either GelRed (Biotium) or MidoriGreen (Bulldog Bio). Samples that amplified successfully were sent to Macrogen Inc. (Republic of Korea) for genotyping.

GeneMarker v.2.4.2 (Hulce, Li (34), SoftGenetics LLC) was used to score alleles for each locus manually after passing the data through the standardisation run wizard using the default animal fragment setting. Random selections of genotyped plates were rescored by a second person to assess consistency in scoring. In addition to the data collected in this study, we included microsatellite scores from samples in Beebe, Ambrose (18). During the present study, it was uncovered that the Beebe, Ambrose (18) study used (in some cases) inconsistent fluorescent dyes for a given locus, which caused a dye-shift [35] resulting in inconsistently scored alleles. We regenotyped a random subset of individuals from each of the populations used in the study by Beebe, Ambrose (18) to ensure consistency with data collected from this study. The predictability of this dye-shift (based on the dyes used previously) enabled shifting of the allele scores from Beebe, Ambrose (18) for use in this study. Samples with fewer than nine scored loci of 13 total were removed before further study as we considered these poor quality samples; thus leaving 911 samples for final analyses (20% of samples were from Beebe, Ambrose (18); Table 2.1). With the remaining dataset, missing values were replaced based on population allele frequencies using GenoDive v. 2.0b27 [36] – based on preliminary analyses this did not significantly alter population structure and relationships between populations. Missing values were not replaced for the STRUCTURE analysis, calculation of HWE and for checking the presence of null alleles.

Scored allele frequencies were checked for the presence of null alleles using MICRO-CHECKER [37] and for Hardy-Weinberg equilibrium using GenAlEx v.6.5 [38, 39]. Additionally, we calculated fixation index (F), allelic richness (Na), number of effective alleles (Ne) and the

27 observed (Ho) and expected (He; unbiased estimate: uHe) values of heterozygosity using GenAlEx v.6.5. Pairwise population indices of genetic variation for Jost’s D, G”ST and FST were also calculated between populations in GenAlEx v.6.5 (Table S2.2). We used 9,999 permutations and an analysis of molecular variance (AMOVA) to assess significance. A Mantel test was also performed in GenAlEx v.6.5 on geographic and genetic distance (pairwise phiPT) using 9,999 permutation [38, 39].

Population structure was investigated using the Bayesian program STRUCTURE v.2.3.4 [40] to infer the most probable number of population clusters (K). Based on our preliminary runs (File S2.1, Supplementary Methods: Preliminary STRUCTURE), final analyses were run at both a lower (K=4) and upper (K=9) K value. For both K values we used a burn-in of 100,000 and runtime of 2,000,000 generations per iteration (20 iterations). For K=9, cluster membership probabilities were somewhat inconsistent across runs due to multimodality; 20 iterations helped to account for this [41]. We assessed whether the burn-in period was adequate by reviewing summary statistics in STRUCTURE [41]. CLUMPP v.1.1.2 [42] was used to compile data from the 20 iterations for the independent values of K using the Greedy algorithm with 1,000 replicates. Final graphs were formed in DISTRUCT v.1.1 [43].

Discriminant analysis of principal components (DAPC) and correspondence analysis (CA) was used to further assess population structure. DAPC was implemented in R Studio v.3.2.2 (RStudio Team 2015) using the adegenet 1.4-1 package [45, 46] using the whole microsatellite dataset, where group membership was defined by the populations outlined in Table 2.1 (see DAPC abbreviation). These populations were more broadly defined and differed slightly from those used in STRUCTURE, to allow for easier interpretation of the results presented here. Specifically, the Torres Strait populations were split into groups based on their genetic relationship to one another and geographic/temporal information to reduce clutter in plots (Table 2.1; DAPC abbreviation). Only populations that were genetically similar were grouped together, which was confirmed using

STRUCTURE, DAPC, CA and pairwise tests for genetic distance (FST, G”ST, Jost’s D) on the full dataset and subsets. Final DAPC analyses were performed on both a full dataset (nind = 911, npop =

23; including all populations) and a reduced dataset (nind=458, npop=18; excluding Jakarta, Sumba, Timor-Leste, the Torres Strait Islands and Southern Fly Region) to help discriminate genetically similar populations.

In adegenet, cross-validation was performed on each of our DAPCs independently, using a training dataset of 90% and a validation set of 10%, using 100 replicates. The number of PCs (n.pca)

28 associated with the lowest root mean squared error (RMSE) was used as this was considered optimum [47]. Cross-validation suggested retaining 60 PCs for the full dataset and 40 PCs (n.pc) for the reduced dataset. We used five discriminant functions (n.da) for each of the analyses, but only the first three are plotted and discussed here as they explained the majority of variance (see Results). A correspondence analysis (CA) was also implemented in adegenet on the full dataset to investigate general trends and to complement DAPC, as visualisation of the data is simplified in CA because within-population genetic diversity is not displayed.

The Garza-Williamson index (M-ratio) was calculated for populations using our microsatellite dataset in Arlequin v.3.5.2.2 [48]. The M-ratio was used to investigate the demographic history of populations and to test for recent bottleneck events; wherein an index statistic closer to 1 suggests the population is in a stationary state whereas very low values suggests a population has gone through a genetic bottleneck in the past (with a critical value of 0.68 indicating a bottleneck) [49, 50]. A Wilcoxon test for heterozygosity excess was also conducted on populations to detect bottlenecks using BOTTLENECK v.1.2.02 [51]. We used a two-phase model (TPM) of mutation with 10% infinite allele model and a 90% single step mutation model with 15% variance for 1000 iterations. A Wilcoxon signed rank test (two-tailed) was used to calculate significance (P < 0.05).

Invasion History – ABC

We tested the invasion history of Ae. albopictus in part of the study region (SE Asia/Australasia) - populations from the Indian Ocean, Fiji, Nauru and United States were not included due to insufficient sampling of these regions. Both COI sequences and the thirteen microsatellites loci were analysed together using ABC analysis in DIYABC v.2.1.0 [52]. Due to the complexity of modelling each population separately in this region, we simplified our invasion scenarios by randomly subsampling individuals from distinct geographic and genetic groups (defined using our other analyses). These representative populations included: mainland SE Asia, Indonesia, Papua, PNG, the Solomon Islands and the Torres Strait Islands/Southern Fly Region. For both the PNG and the Torres Strait/Southern Fly Region populations, we included temporal sampling in our scenarios (asterisks, Fig 2.2). Each of these representative populations/sampling events was made up of 30 individuals except Papua which used all 20 samples from the only sampled population, Timika. In addition, we included an unsampled ancestral population (ANC, Fig 2.2) in our model that split into mainland SE Asian and Indonesian populations.

Fig 2.2. Invasion scenarios of Aedes albopictus in Australasia tested using approximate Bayesian computation (ABC). One unsampled and six sampled populations were modelled, shown as coloured lines in five different invasion scenarios. Time events (t1-t7) are not to scale, but their prior distributions are displayed as the year (except t7 which is shown in years before present

(ybp, present = 2015)). Changes in effective population size (Ne) are represented as differently shaded lines, where db-db4 represent the duration; narrowing lines represent population bottlenecks

30 that were given lower Ne priors ranges; rate of admixture (ra) is also shown. All populations have samples at time = 0 (i.e. 2015) and asterisks represent additional temporal sampling of populations (i.e. for TS and PNG). The posterior probabilities of all scenarios are shown with 95% confidence intervals in square brackets; Scenario 4 was the best-fit scenario. See Table S2.3 for further details and posterior distributions.

Preliminary runs were carried out in accordance with Bertorelle, Benazzo (53) in order to optimise summary statistics, prior estimates and the scenarios tested. Final runs compared five invasion scenarios; the final summary statistics, prior estimates and parameter conditions used are outlined (Table S2.3). Each scenario represents a plausible invasion route into the Australasian region. These were constructed using historical records regarding the timing and suspected sources of the different invasions [26, 54-58]. Priors were sampled from a wide range of distributions based on these records - less certain time priors were given a wider distribution and standard deviation whereas more likely priors were assigned narrower estimates. The upper time bound for the divergence of mainland SE Asia and Indonesia from a common ancestor was based on Porretta, Mastrantonio (59) while the lower bounds allowed for the possibility of a more recent split associated with human migration [10] (Table S2.3). Because DIYABC measures time in terms of the number of generations, we assumed 10 generations per year for Ae. albopictus (which typically ranges from 5- 17 generations in the tropics).

Estimates of effective population size (Ne) ranged from 10 to 1,000,000 individuals (uniform distribution) [20-22] depending on if a population was modelled as going through a change in Ne. For instance, each of the recently introduced populations was modelled to allow for a founder effect after its introduction using lower Ne ranges (Fig 2.2, Table S2.3). We additionally allowed for a change in Ne in the Torres Strait Islands/Fly Region population due to the drastic temporal changes we observed in our other analyses – this would allow us to make a relative comparison of Ne in order to see if the population had undergone any change in Ne (indicative of bottleneck/expansion events) (Fig 2.2, Table S2.3). For COI, we used the HKY mutation model [60] and sampled from a uniform distribution with mutation rates ranging between 7x10-10 - 1x10-7. For microsatellites, both di- and trinucleotide repeats were modelled separately due to the possibility of different repeat lengths having different mutation rates [61]. Both microsatellites used the default generalised stepwise mutation model and were assigned a loguniform distribution with mean mutation rates between 1x10-6 - 1x10-3 (Table S2.3). All mutation rates were based on standard ranges for COI [62] and for Dipteran microsatellites [20-22, 63]. We simulated 15,000,000 datasets and each of the

31 five scenarios was given a uniform probability. The performance of the ABC approach was assessed using multiple methods in DIYABC (File S2.1, Supplementary Methods: Performance of DIYABC).

COI analysis

The mitochondrial protein-coding gene COI was amplified using custom designed primers (albCOIF 5’-TTTCAACAAATCATAAAGATATTGG-3’ and albCOIR 5’- TAAACTTCTGGATGACCAAAAAATCA-3’) for 259 random individuals across different populations. Each 25.3μL reaction consisted of 19µl H2O, 5µl 5X Mytaq buffer (Bioline, with pre- optimised concentrations of dNTPs and MgCl2), 0.1μl 100μM forward primer, 0.1μl 100μM reverse primer, 0.1μl MyTaq polymerase and 1μl 1:10 DNA template. PCR used an initial denature of 94ºC for 3 min, 35 cycles of denaturation at 95ºC for 30 sec, primer annealing at 45ºC for 40 sec, and primer extension at 72ºC for 30 sec. Final elongation lasted 5 min at 72ºC prior to cooling to 4ºC. Amplification was confirmed using gel electrophoresis (as described previously) and PCR products were purified by adding 2µl per sample of a mixture containing equal amounts of Exonuclease I and Antarctic Phosphatase (New England Biolabs, Australia) before incubation at 37ºC for 20 min and denaturation at 80ºC for 10 min. Samples were sequenced by Macrogen Inc. (Republic of Korea) using Sanger sequencing. Additional COI sequences of Ae. albopictus were obtained from other studies and from Genbank (1044 sequences total, 259 produced in this study, Table S2.4).

Sequences were edited and aligned in Geneious v.9.0.4 (http://www.geneious.com, Kearse, Moir (64)) using the MAFFT alignment. The final alignment was trimmed to 445bp to incorporate the large number of COI sequences available from Genbank which were smaller than the ~700bp region sequenced in this study. All sequences were checked for stop codons in Geneious v.9.0.4. TCS haplotype networks [65] were constructed using 1,000 iterations in PopArt v.1.7 (http://popart.otago.ac.nz). In addition, we calculated Tajima’s D for populations with temporal data in PopArt v.1.7 to determine whether sequences were evolving randomly or non-randomly. Haplotype and nucleotide diversity was calculated using DnaSP v.5.10.1 [66].

RESULTS

Microsatellite genetic diversity and population structure

Allelic richness for microsatellites was highest in native populations of Ae. albopictus from Myanmar, Thailand, Malaysia, but also high in recently invaded areas such as some of the Torres

32 Strait Islands and in PNG and the Solomon Islands (See Na in Table S2.5). A Mantel test on the whole dataset showed a significant (P = 0.0001) positive, but weak, correlation (R2 = 0.02) between genetic (phiPT) and geographic distance (y = 0.0002x + 28.1). Pairwise estimates of FST, Jost’s D and G”ST all revealed similar results to each other and recovered mostly significant relationships between populations; here we discuss gene flow and genetic distance in regards to FST estimates but the other measures are shown in Table S2.2, A-C. Lowest FST values were apparent between populations belonging to the same geographical region (for definition of regions see Table 2.1,

Region/description), especially within mainland SE Asia (FST = 0.011 - 0.103) (Table S2.2A). However, some comparisons between regions separated by vast geographical distances also showed low FST scores, such as populations from the Solomon Islands with populations from mainland SE

Asia (FST = 0.050 - 0.114) (Table S2.2A). The relationships between populations were mostly consistent with the results obtained in STRUCTURE and multivariate analyses, which are described in detail below.

Within the study region, four to nine clusters were supported by the Evanno ∆K and log likelihood methods for inferring K. While K=4 (Fig 2.1) represents the simplest summary of the genetic structure of Ae. albopictus in the region, we detected substantial substructure within these four main clusters which are apparent at K=9 (Fig S2.1). We discuss the data in the context of both values of K to avoid underestimating the degree of population structure within the study region.

At K=4, clusters mostly pertained to distinct but broad geographic boundaries although many populations and individuals show signs of admixture, despite the large geographic distances (Fig 2.1). The mainland SE Asian populations of Myanmar, Thailand, Malaysia and Singapore cluster with the USA (Hawaii and Atlanta; Fig 2.1B - C), La Réunion, as well as Fiji and Nauru (light purple; Fig 2.1). The second cluster (light green) contains populations from Indonesia (Jakarta and Sumba), Timor-Leste, the Southern Fly Region of PNG (Fig 2.1A) and several islands of the Torres Strait (especially collections following the first detection of Ae. albopictus in the straits in 2005 (collections between 2006 - 2014)) (Fig 2.1A). An additional cluster (purple) is prominent within the Torres Strait region (Fig 2.1A) and represents populations on the islands collected more recently (2013 - 2015), suggesting temporal shifts in population structure have occurred on some islands (see Fig 2.1A: Ker, War). The fourth cluster (green) is composed of historically-established PNG populations, but note that some of these populations contain admixture with the SE Asian cluster (Fig 2.1). The island of Daru (Fig 2.1A: Dar), which is less than 5km from the Southern Fly Region incursion populations (light green cluster; Fig 2.1A; Sig, Kul, Mbd, Kat), is distinct and clusters with the historically established PNG populations. Timika and the Solomon Islands show genetic

33 affinity to both PNG and SE Asian clusters (Fig 2.1). Indian Ocean islands (Fig 2.1: CK and CH) appear differentiated from each other and contain a notable degree of admixture, but Christmas Is. is more similar to SE Asia whereas Cocos (Keeling) Islands appear as an admixed population made up of the PNG and Indonesian clusters. When K=9, the same broad population patterns are observed but some populations/regions become more distinct, including the USA and Hawaii, Solomon Islands, Timika, the Cocos (Keeling) Islands and Sumba/Torres Strait Island populations (Fig S2.1). Relationships for this K value are described in File S2.1, Supplementary Results, STRUCTURE (K=9).

For multivariate analyses, the DAPC on the full dataset (n.pc = 60, n.da = 5) explained 89% of variance, whereas the reduced dataset (n.pc = 40, n.da = 5) explained 79.3% of the variance in the data. Eigenvalues for these first three PCs are 206.37, 110.78 and 63.21 for the full dataset (Fig S2.2A) and 99.6, 69.2 and 44.99 for the reduced dataset (Fig S2.3); these values correspond to the ratio of between-group over within-group variance for each discriminant function. For the correspondence analysis (CA) we plotted the first three eigenvalues (0.19, 0.15, 0.09), which indicate the proportion of variance explained by the first three PCs (Fig S2.2B).

DAPC and CA results of the full dataset (Fig S2.2) showed similar population differentiation as observed in STRUCTURE at K=4. Due to the large number of populations, we describe population structure based on the broad clustering – populations are colour coded with geographically close populations being more similar in colour. Four major clusters of populations are noticeable when the first three PCs are plotted against each other (C1-C4, Fig S2.2A). However, there is considerable overlap between these clusters, particularly with C4 overlapping C2 and C3, suggesting that these individuals and populations are genetically similar and show signs of admixture (Fig S2.2A). Cluster 1 (C1) represents recent (2012-2015) collections from Torres Strait Islands and is the most distinct from the other clusters. It is most closely related to C2, which contains earlier collections (2007-2014) from the Torres Strait Islands and populations from the Southern Fly Region, Sumba, Timor-Leste and Jakarta (Fig S2.2A). The relationships uncovered in the DAPC of the full dataset were recovered in the CA and are more easily visualised, where the first three principal components (PCs) of the CA are plotted in three-dimensions (Fig S2.2B); however, note that the large amount of within-population variation (as displayed in the DAPC plots) is not shown.

Because of the overlap of C3 and C4 clusters, we separately analysed these clusters by DAPC (referred to in the Materials & Methods as the reduced dataset) that excluded populations from the

34 Torres Strait Islands, Jakarta, Sumba, Timor-Leste and the Southern Fly Region (Fig S2.3A-B) to explore substructure within these clusters (i.e. C3 & C4 in Fig S2.2A). Populations from PNG (Kiunga, Madang, Port Moresby and Daru) were similar to each other but distinct from the other populations (Fig S2.3). The offshore PNG populations (Lihir Is. and Buka Is.) were somewhat differentiated from mainland PNG populations, although Buka Is. shares some overlap with both Port Moresby and Lihir Is. (Fig S2.3). The Solomon Islands also appears similar to Lihir Is. and Buka Is. populations, but is somewhat distinct (Fig S2.3). Mainland SE Asian populations appear genetically similar and tend to exhibit the most genetic overlap with other populations (Fig S2.3). Nauru and Fiji are most similar to mainland SE Asian populations. In contrast, both USA populations (Hawaii and Atlanta) as well as La Réunion appear well differentiated from mainland SE Asian populations. Cocos (Keeling) Island and Timika share some overlap with each other, whereas Christmas Island shares overlap with both PNG and mainland SE Asia populations (Fig S2.3).

Invasion history – ABC

The scenario with the highest posterior probability using the logistic approach was scenario 4 (P = 0.52 [95% CI: 0.44, 0.59], Fig 2.2). In this scenario, Ae. albopictus colonised Papua and PNG in two separate events from mainland SE Asia, established in the Solomon Islands via PNG and more recently colonised the Torres Strait Islands/Southern Fly Region via Indonesia (Fig 2.2). The timing of each introduction event is shown in Table S2.3 and corresponds with historical records for the introduction of Ae. albopictus in the tested populations, although 95% confidence intervals (CI) suggest that some introduction dates could have been earlier than first observed (see Discussion). None of the other scenarios showed overlapping 95% CI with Scenario 4 (Fig 2.2), however, we detected moderate levels of type I (0.45; probability that scenario 4 is rejected given that it is the ‘true’ scenario) and type II error (0.48; probability of deciding scenario 4 is the ‘true’ scenario when it is not) that suggest Scenario 1 (P = 0.34 [95% CI: 0.31, 0.37]) could provide a plausible alternative invasion scenario for our data (Table S2.3). Scenario 1 is identical to scenario 4, except that the Solomon Islands is modelled as originating from mainland SE Asia, rather than from PNG (Fig 2.2). Consequently, we discuss the source of the invasion of the Solomon Islands based on both possible scenarios (i.e. scenarios 1 and 4) and additionally calculated posterior estimates under both scenarios (Table S2.3). A preliminary analysis showed low support (P = 0.001 [95% CI: 0.00, 0.12]) for a scenario where the Solomon Islands introduction was modelled as an admixture event between mainland SE Asia and PNG compared to the five scenarios compared in our final analyses

(but using slightly different Ne prior ranges for all founders (10 - 10,000)).

35 Each introduced population showed no relative change in Ne due to large 95% CIs of posterior distributions, although median Ne values were smaller for founding events (Table S2.3). Likewise, the duration of the modelled bottleneck had large 95% CIs (Table S2.X), but median values generally ranged from 16 - 30 generations (Table S2.3). Overall, the Torres Strait Islands/Southern

Fly Region population showed a stable Ne since its introduction (due to overlapping 95% CIs), although there was a gradual increase in median Ne over time, potentially suggesting growth of the population as a whole. (Table S2.3). Our assessment of the performance of our ABC analysis was supported as fitting our observed data well (Fig S2.4, Fig S2.5).

COI haplotype networks and diversity

A total of 52 COI haplotypes were identified from the 1044 individuals used for generating the TCS haplotype network, with 92% of individuals belonging to nine main haplotypes (H1-5, 11, 15, 39, 43) (Fig 2.3, Table 2.2, Fig S2.6). The distribution of these haplotypes by specific population is shown in Table S2.6. All new sequences generated from this study are available on Genbank (Accession no. KY90719-KY907453; see Table S2.4 for accession numbers of sequences from other studies). The COI haplotype network is less informative in regards to population structure than the microsatellite data, although it does highlight broader geographic relationships that are somewhat consistent with the microsatellite results. Of the nine main haplotypes, H1 has the most individuals, mostly from eastern Asia (China, Taiwan and Japan), USA (mainland USA and Hawaii), Madagascar and La Réunion (Fig 2.3, Table 2.2). Similarly, H39 is distributed in a similar temperate/subtropical region. Haplotype 3 consists primarily of individuals from the Torres Strait Islands, Fly Region, Indonesia, Timor-Leste, PNG and the Philippines. However, the majority of PNG sequences belonged to H5, which also includes individuals from the Solomon Islands, Indian Ocean islands (CK and CH), Singapore and Thailand. Another major haplotype, H4, includes the most diverse range of populations (in terms of geographic spread), although it mostly consists of mainland SE Asian populations and populations from the tropics. Of the additional COI haplotypes, many are exclusive or shared amongst close geographic regions (Fig 2.3, Table 2.2, Table S2.6, Fig S2.6), although others show no apparent geographic pattern.

Haplotype diversity (Hd) for the total dataset was high (Hd = 0.83) as was nucleotide diversity (π = 0.0037); however, population measures of Hd and π varied considerably (Table 2.3). Neutrality tests (Tajima’s D) on all populations with temporal data were not significant (Table S2.7).

36 Table 2.2. Haplotype distribution of mitochondrial COI sequences for 1044 individuals of Aedes albopictus by broad population region. See Table S2.6 for a more specific summary of COI haplotypes by population.

Haplotypes Region H1 H2 H3 H4 H5 H6 H10 H11 H13 H15 H23 H25 H30 H39 H40 H42 H43 H49 Exclusive Haplotypes Torres Strait Islands 1 15 62 40 1 17 4 30 H16, H17, H19, H24 Fly Region 5 15 32 5 1 2 Papua New Guinea 7 23 141 1 1 1 H9, H12, H29, H35 Solomon Islands 14 1 H33, H34 Fiji 4 H14 Nauru 1 Timika 11 Timor-Leste 16 1 Sumba 6 1 1 1 H36, H38 Jakarta 3 25 1 3 1 H20, H21, H22 Singapore 1 11 13 1 20 H37 Malaysia (includes Borneo) 1 1 26 4 2 H7, H26, H27, H28, H31 India 1 H51 Myanmar 10 2 H32 Thailand 4 2 1 Philippines 4 1 Vietnam 1 1 1 Cambodia H47 China 71 17 1 1 H41 Taiwan 26 3 1 Japan 9 6 Hawaii 40 1 H44, H45, H46 USA (mainland) 88 5 12 1 4 H18, H48, H52 Christmas Island 2 2 H8 Cocos (Keeling) Island 12 La Réunion 15 3 Madagascar 54 15 H50 TOTAL (per haplotype) 306 24 137 173 185 2 2 29 5 32 2 5 4 47 5 2 26 5 -

Fig 2.3. Major mitochondrial COI haplotypes for Aedes albopictus in the Indo-Pacific, Asian and USA region, representing 92% of the 1044 individuals analysed. Displayed are the nine most prevalent COI haplotypes (of 52 in total) using data from ours and other studies, where each haplotype is represented as a different colour and the size of the circle represents the number of individuals from a given region (which is plotted on the map). Note, the placement of circles does not correspond to the exact location of haplotypes, but represents the general region they are from; refer to Table S2.6 for the exact location of haplotypes and for additional haplotypes found in the region. Insets show distant regions, but are to scale with the main map: A) Madagascar and La Réunion; B) Hawaii; C) USA.

38 Table 2.3. Estimates of genetic diversity for the mtDNA COI region for populations of Aedes albopictus in the study. The range of collection years is shown. Number of individuals sequenced

(n), number of haplotypes (nH), haplotype diversity (Hd) and nucleotide diversity (π) are displayed along with the total summary for all populations.

Region Year n nH Hd π Torres Strait Islands 2004-2015 175 12 0.780 0.0040 Fly Region 2007 60 6 0.648 0.0034 Papua New Guinea 1992-2011 180 10 0.370 0.0014 Solomon Islands 2013-2014 26 4 0.582 0.0028 Fiji 2015 5 2 0.4 0.0009 Nauru 2014 1 1 N/A N/A Timika, Papua 2015 11 1 0 0 Timor-Leste 2001 17 2 0.118 0.0003 Sumba 2013 11 6 0.727 0.0026 Jakarta 2011-2013 36 8 0.514 0.0016 Singapore 2011-2013 47 6 0.701 0.0034 Malaysia 2013 41 10 0.591 0.0020 India 2012-2014 2 2 1 0.0023 Myanmar 2013 13 3 0.410 0.0010 Thailand 2000-2015 7 3 0.666 0.0017 Vietnam 2000-2004 3 3 1 0.0045 Cambodia 2001 1 1 N/A N/A Philippines 2016 5 2 0.4 0.0009 China 2011 91 5 0.359 0.0009 Taiwan 2011 30 3 0.246 0.0006 Japan 2011 15 2 0.514 0.0012 Hawaii 1971-2015 45 5 0.282 0.0009 USA (mainland) 2011 129 7 0.390 0.0010 Christmas Island 2008 5 3 0.8 0.0023 Cocos (Keeling) Island 2008 12 1 0 0 La Réunion 2000-2011 18 2 0.294 0.0007 Madagascar 2007-2009 70 3 0.364 0.0008 All populations 1044 52 0.834 0.0037

39 Bottleneck tests

Past genetic bottleneck events were not consistently indicated using both Wilcoxon tests for heterozygosity excess and M-ratio, with the exception of the 2010 population from Waiben of the Torres Strait (M = 0.68, P = 0.032 (two-tailed Wilcoxon signed rank test for heterozygosity excess) (Table S2.8). However, multiple populations showed signs of a bottleneck using a single method. The M-ratio indicated a bottleneck for some populations of the Torres Strait Islands (Mabuiag, Waiben, Ngurupai, Poruma, Iama), Port Moresby, Timika and Gizo, whereas the Wilcoxon test was significant for Waiben, Madang, Jakarta, Yangon and the Cocos (Keeling) Islands (Table S2.8).

DISCUSSION

The population structure and genetic connectivity of Ae. albopictus within the Indo-Pacific region has been limited to a few studies that only examined regional structure or had restricted sampling within the species’ range. In addition, the genetic characteristics of some populations examined in this study have been unexplored (e.g. Solomon Islands, many Indian and Pacific Ocean Islands, Papua-Indonesia). Using multiple lines of evidence (and both microsatellite and mitochondrial markers), we show high spatial genetic structure throughout this region. We used coalescent ABC analysis to test for the first time the likely invasion route of Ae. albopictus into the Australasian region and uncovered that the species likely invaded New Guinea from mainland SE Asia and the Solomon Islands via either PNG or SE Asia. We also show the recent invasion of Ae. albopictus into northern Australia’s Torres Strait region and Southern Fly Region of PNG likely originated from Indonesia, as previously suspected [18]. Furthermore, we provide evidence of rapid temporal shifts in population structure occurring less than a decade after the Asian tiger mosquito’s introduction into Australia’s Torres Strait Islands in 2005 [26]. In contrast, historically-introduced and native populations of Ae. albopictus showed less spatial population structure at a regional level, despite large geographic distances and international boundaries between some of these populations. Importantly, this study provides a widespread sampling distribution of the species’ native range and revealed more spatial population structure than previously shown, as well as evidence for rapid temporal genetic change in newly established populations in the Torres Strait Islands.

Population structure in the Indo-Pacific

Multiple population studies have attempted to capture the amount of genetic structure throughout the species’ native range. Allozyme studies have shown that Indonesian and Japanese populations of Ae. albopictus are likely distinct [67] and that SE Asia (Borneo, peninsula Malaysia) and

40 southern Asian populations (India, Sri Lanka) can both be differentiated from northern Asian populations (China, Japan) [68]. While no studies have conducted a comprehensive analysis of the species’ full native range [8], the genetic differentiation of native Asian populations of Ae. albopictus may confer to both a north-south (Korea to Indonesia) and east-west (Japan to India) pattern of genetic differentiation. Our results partly support this pattern, with evidence for genetic differentiation separating northern Asia (COI data only, no microsatellite data available), SE Asia and Indonesia (both COI and microsatellite data). Within mainland SE Asia, our data revealed little to no population structure despite high genetic diversity and COI haplotype diversity, supporting the findings of other studies [59, 69, 70]. Using climatic modelling and two mitochondrial markers, Porretta, Mastrantonio (59) suggested that the low genetic structure across this mainland SE Asian region could be explained by the demographic growth between interconnected populations of Ae. albopictus preceding the last glacial maximum (LGM, occurring ~21,000 ybp), with the species’ ecological flexibility facilitating its success in the ecologically diverse Sundaland (exposed SE Asian landmass) during this period. Whilst their study lacked sampling from Indonesia, their data suggested climatically suitable habitat for Ae. albopictus existed across the southern range of Sundaland which later formed the Indonesian islands after a rise in sea levels. They hypothesised that the emergence of Sundaland during the LGM could have facilitated population connectivity across Indonesia and mainland Asia. In contrast, we found clear genetic differentiation between the Indonesian archipelago and mainland SE Asia with both our microsatellite and COI data, which could be explained by fragmentation and subsequent differentiation of Ae. albopictus populations following a rise in sea levels or potentially driven by human migration from the region [10]. Our ABC analysis supports a more historical split ~9,040 ybp (95% CI = 1,790 – 27,600 ybp) but further investigation with extensive native population sampling would be needed to clarify dating.

The genetic homogeneity and high gene flow (FST = 0.020 - 0.103) in our microsatellite data across mainland SE Asia (including Malaysia, Singapore, Myanmar and Thailand) could also be explained by human-mediated gene flow associated with transportation infrastructure (land, air and sea); aircraft and road networks in particular may be major drivers in the connectivity of Ae. albopictus in this region given the inland location of many of these populations [71-73]. For our full dataset, the relationship between genetic and geographic distance was significant, however, the correlation was weak (R2 = 0.02), highlighting the potential extent that human movements have had on Ae. albopictus population structure; although it does also highlight a minor trend of isolation by distance in our study region.

A recent worldwide study that examined the mitogenome diversity of Ae. albopictus found three major haplogroups, two of which were implicated in the global spread of the species [9]. Some of

41 these haplogroups appeared more prevalent in particular climatic and geographic regions, such as haplogroup A1a which chiefly characterised the tropics and A1a2 which is mostly distributed in temperate regions. Another haplogroup (A2) appeared important in the spread of Ae. albopictus from SE Asia toward Australasia, distinguishing many of the samples from the Philippines, PNG, Indonesia and the Torres Strait Islands [9]. A similar pattern was observed using our COI data when viewing the nine haplotypes that account for 92% of the sequences in our study. Haplotype three (H3) was found to be common in Philippines, Indonesia, the Torres Strait Islands and Southern Fly Region but also present in historically established PNG populations, Vietnam and an individual from Sepilok, Borneo (Malaysia). Consequently, the populations from Vietnam, Borneo, the Philippines and Sulawesi could represent important unsampled populations that may influence ABC results and should be considered in future studies, which could additionally explain why the posterior probability of our most likely invasion scenario was moderate (P = 0.52). We also found that H1 was more prevalent in regions that experience temperate and subtropical climates, while H4 was widespread in the tropics (Fig 2.3). Due to the maternal inheritance of mtDNA, these results could suggest that the movement of females has been somewhat limited to their preadaptation to certain climatic regions. For example, females originating from a temperate region may be more likely to successfully invade other temperate regions. It is possible that these genetic patterns for COI in Ae. albopictus are associated with the photoperiod response of different populations, which could contribute to higher gene flow and invasion success between climatically similar regions [17, 74-76]. However, the photoperiodic response of Ae. albopictus in recently introduced regions within Australasia is lacking (especially for New Guinea, the Solomon Islands, Fiji and Nauru). Within the Torres Strait Islands it appears the population is of tropical, Indonesian origin and egg survival was lower in less humid conditions [77], supporting this conclusion. However, it is worth noting that there are multiple other COI haplotypes that do not conform to any obvious geographic patterns (Fig S2.6), suggesting that there have been multiple introduction events into some locations from mainland Asia or potentially unsampled locations – a similar pattern which was also highlighted by other studies [9, 74]. The possibility of insertion of mtDNA in the nuclear genome [78, 79] was considered in this study and our COI sequences were assessed by examining chromatograms (no double peaks in chromatograms) and by checking for stop codons. It seems unlikely that there is nuclear insertion of mtDNA in samples our study – but it remains a possibility and requires further research.

We included several populations from the USA and Indian Ocean to explore how these populations fit into a broader geographic analysis with our samples, which are chiefly from SE Asia and Australasia. We did not include these in our ABC analysis because of the lack of temperate Asian

42 populations, which have been shown as the source of USA introductions and because of insufficient sample sizes in Indian Ocean populations (Cocos (Keeling) Islands, Christmas Island and La Réunion). We included them in our other analyses to assist in future studies and have discussed them in File S2.1, Supplementary Discussion.

Invasion into Australasia

New Guinea In New Guinea (PNG and Papua (Indonesia)), Ae. albopictus was first detected in Jayapura in the northeastern corner of Papua Province, Indonesia (formerly West New Guinea) in 1962 [55, 57]. By the early 1970s it was reported in northern PNG near Madang and was established in the PNG capital Port Moresby by the 1980s [55, 57, 58]. There were reports of Ae. albopictus from New Guinea earlier but these have been considered doubtful and probably referred to Ae. scutellaris [56]. However, we considered these earlier dates in our ABC prior distributions for testing the timing of introduction events. We found support for a later arrival of Ae. albopictus into Papua (~1959 [95% CI = 1943, 1967]), rather than earlier dates from the 1920s, but likewise, the introduction could have been much earlier than the 1962 detection [55]. Additionally, our ABC analysis show that the Papuan population of Ae. albopictus probably originated from mainland SE Asia, rather than from our sampled Indonesian population. The subsequent spread into PNG was supported as originating from a similar mainland SE Asian source in a separate and later event, rather than from an introduction from Indonesia (including both Papua and non-New Guinea Indonesian populations in our study). The timing of the introduction into PNG in the ABC analysis (~1970 [95% CI = 1963, 1976]) corresponded with first detection dates from 1972 [57], but similar to Papua, could have been earlier than first observed which is plausible given the lack of surveys in PNG at the time.

Previous data [18] from PNG used inconsistent M13 dyes for a given microsatellite locus, causing dye shifts [35] resulting in some incorrectly scored alleles. This was uncovered and corrected in the present study (as outlined in the Materials & Methods) and revealed less substructuring of PNG populations than previously suggested [18]. We found that all historically established PNG populations were a homogenous genetic cluster (whereas in the previous study, Kiunga and Port Moresby both appeared as distinct genetic clusters using microsatellites), but this dye shift had no discernible influence on the general relationships between the other populations in their study [18]. Cooper, Waterson (58) suggested that the dispersal of Ae. albopictus in PNG is primarily driven by aircraft and coastal shipping as there is not an extensive road network in PNG. Surprisingly, the small island of Daru, offshore of southern PNG, showed no detectable genetic change using

43 bottleneck statistics between samples analysed after 16 years (1992-2008), nor did Port Moresby over 4 years (1996-1999); neutrality tests using COI revealed no contraction or expansion of these populations, but results were insignificant.

Solomon Islands, Fiji and Nauru The introduction of Ae. albopictus into the Solomon Islands in 1979 and eventually into Fiji in 1989 was originally suspected to be from a progressive expansion, most likely from PNG via shipping [54, 80]. Our ABC analysis suggested the Solomon Islands introduction originated from either PNG or mainland SE Asia in ~1979 (95% CI = 1973, 1983), corresponding with the first detection date – due to moderate type I and II error it is unclear which region was the primary source of the introduction (see Results). Major commercial shipping routes exist between the Solomon Islands and mainland SE Asia and currently the Solomon Islands has significant trade with Singapore and

Malaysia [81] which could explain the ABC results and high between-region gene flow (FST = 0.05 - 0.11). Large-scale logging operations have been prevalent in the Solomon Islands over the last 40 years, with a four-fold influx of logging industry multinationals, chiefly from SE Asia (especially Malaysia), during the early 1980s [82]. Shipping associated with this industry could have provided a prime opportunity for the introduction of Ae. albopictus from SE Asia. However, our STRUCTURE and DAPC plots also show the Solomon Islands has close genetic affinity to both mainland SE Asia and PNG clusters. Compared to SE Asia, PNG showed less, but relatively high gene flow with the Solomon Islands (FST = 0.08 - 0.20). Papua New Guinea has strong historical and contemporary cultural ties with the Solomon Islands and other islands of the Melanesian region (spanning the islands from PNG to Fiji) and human movements between the regions have been ongoing, which may explain PNG’s genetic ties to the Solomon Islands, particularly Lihir Island (File S2.1, Supplementary Discussion: Lihir and Buka Islands). The majority of Solomon Islands COI sequences were H5, which is widespread throughout PNG; however, this haplotype is also observed in some mainland SE Asian populations (from Singapore and Thailand) as well as Indian Ocean Islands (Cocos (Keeling) Islands and Christmas Is.). Additionally, multiple private COI haplotypes were only found in the Solomon Islands, highlighting some unique structure in the Solomon Islands that was also shown at K=9 in STRUCTURE (Fig S2.1). For microsatellites we found there were no shared private alleles between the Solomon Islands and mainland SE Asia (i.e. alleles that were exclusively found in the two regions), but a single shared private allele was exclusive to the Solomon Islands and PNG. Overall, these results highlight that there have probably been multiple introductions into the Solomon Islands from both mainland SE Asia and PNG.

44 Due to insufficient sampling of Nauru and Fiji populations, we were unable to test these populations in our ABC analysis. Results are discussed in File S2.1, Supplementary Discussion: Fiji and Nauru, but conclusions should be interpreted cautiously given small sample sizes. Aedes albopictus was only recently found on Nauru in April 2014 – first detected and morphologically identified by Michael Bangs, and genetically confirmed using PCR-restriction digest and COI in this study (see Materials and Methods).

Torres Strait Islands and Southern Fly Region (Southern PNG) In 1988, Ae. albopictus was detected in the coastal Southern Fly Region of PNG and on the nearby PNG island of Daru. However, a 1992 survey showed that while Ae. albopictus was still present on Daru, it was not found in/around the coastal villages of the Southern Fly Region. Consequently, Cooper, Waterson (58) highlighted that the species had failed to establish in these coastal locations after its initial introduction, despite an abundance of suitable larval habitats. However, a reintroduction occurred in 2005 at the same time as the Torres Strait incursion, which were suspected to be part of the same invasion wave into the region due to their genetic similarity [18]. Compared to Indonesian and historically established PNG populations, the Torres Strait/Fly Region invasive population was more genetically similar to Indonesian populations and was suspected to have been introduced by illegal, Indonesian fishing vessels [18]. Our ABC analysis supports the Indonesian origin of these introduced populations occurring ~2005 (95% CI = 2003, 2007). It is likely the Torres Strait and Fly Region were either seeded by a similar invasion source, or that local movements (mostly boat (no airstrips in the Fly Region)) after an initial invasion into the region have seeded new populations and homogenised population structure in the region. The presence of multiple shared COI haplotypes with PNG and mainland SE Asian populations also highlight that there have probably been multiple secondary introductions into the region, some potentially from unsampled populations. We found no support for initial founder effects after the Torres Strait/Fly

Region introduction, but median Ne showed a gradual increase after its introduction, potentially suggesting growth of the population over time. Note that in our ABC analysis this region is represented as a random subset of multiple populations (due to the complexity of modelling each population separately), so these results are more reflective of the stabilisation of Ne in the Torres Strait/Fly Region as a whole. In contrast, when we explored population structure on individual islands of the Torres Strait we found variable and sometimes drastic temporal changes in genetic structure (outlined below).

Our microsatellite results show high temporal genetic structure in the Torres Strait region over the past decade. Furthermore, our data indicate either a notable pattern of genetic drift on some islands

45 (such as Keriri, Ngurupai and Warraber) or could represent a secondary introduction into the region from an unsampled population. We found evidence of genetic bottlenecks for several Torres Strait Islands, but only Waiben (2010 population) was supported by both bottleneck statistics used here (M-ratio and heterozygosity excess tests). Williamson-Natesan (83) highlighted that when a bottleneck is supported by the M-ratio, but not heterozygosity excess, it can indicate a more severe and older population bottleneck and vice versa for weaker and more recent bottleneck events. Consequently, we found evidence for older and more severe bottlenecks on Mabuiag, Waiben, Ngurupai from earlier collections after the initial invasion of Ae. albopictus into the region, as well as in more recent collections from Poruma and Iama. The dramatic wet-dry climate in the Torres Strait [77, 84], intense spraying efforts and refocusing of control efforts (see File S2.1, Supplementary Discussion: Control and surveying on the Torres Strait Islands; and [27, 28]) could explain population bottlenecks on certain islands and the differing genetic trajectories of the islands. Additionally, initial founder effects (resulting from small invading populations) could have played a strong role in the pattern of genetic drift on the various islands. Laboratory experiments by Nicholson, Ritchie (77) showed that hotter, dryer conditions significantly reduced egg viability in Ae. albopictus from Masig Is. in the Torres Strait, so it is possible that seasonal changes have also caused genetic bottlenecks, but it is likely there are a multitude of factors at play. Both bottleneck tests used here have been shown to fail to detect recent bottlenecks (within 1-5 generations), particularly given small sample sizes and few loci [49, 85] – to more accurately determine if a recent bottleneck has occurred more extensive temporal sampling will need to be conducted in the future. As previously mentioned, it remains a possibility that a secondary introduction from an unsampled source has caused the rapid temporal genetic changes observed on some islands. Indeed, Urbanelli, Bellini (67) showed that the Sulawesi region harbours highly distinct populations that can be distinguished from other Indonesian populations of Ae. albopictus and the region could be key to fully understanding the species’ invasion of Australasia. Likewise, Southern PNG is highly connected to the Torres Strait Islands via local boating traffic [28] and the temporal genetic change detected could have originated from an unsampled population in this region.

Conclusion and Future Implications

For the first time, we have used ABC analysis to compare various invasion scenarios of Ae. albopictus in the Australasian region and have additionally explored and characterised the genetic structure of a wide range of populations in the Indo-Pacific, which had not been compared under a single study. We uncovered notable temporal population structure in recently introduced Torres Strait Island populations. Importantly, this demonstrates some of the drastic changes that invading populations may undergo within a short time period (i.e. in less than a decade), which has

46 substantial implications on the practicality and accuracy of using such genetic databases for estimating invasion sources. However, we also found that historically established populations of Ae. albopictus displayed stable population structure. Future studies that aim to address the global genetic structure of Ae. albopictus will need to consider the full native range of the species and the influence of temporal collections on population structure, especially newly established populations. Likewise, ABC analyses that account for complex scenarios (requiring thorough spatio-temporal sampling) and gene flow between populations will play a key role in better understanding the population dynamics of Ae. albopictus as well as other mosquito species that are highly associated with humans, such as Ae. aegypti. The standardisation of genotyping methods and sampling efforts will allow for more rigorous assessments of the global population structure of Ae. albopictus, given the scale of its current distribution which makes such population studies logistically challenging. This will prove essential in controlling the spread of Ae. albopictus and for assessing the health risk of different populations, given their variation in vector efficiency, physiology and behaviour [10, 86-88].

ACKNOWLEDGEMENTS

We wish to thank Andre Moura for kindly providing us with script to create 3D plots in R as well as Maddie James, Caitlin Curtis and James Wisdom for providing feedback on drafts. We also thank Din Matias for his assistance with DIYABC and Jeffrey Hii for his assistance in organising regional collections. The opinions expressed herein are those of the authors and do not necessarily reflect those of the Australian Defence Force and/or extant Defence Force Policy.

REFERENCES

1. Bonizzoni M, Gasperi G, Chen X, James AA. The invasive mosquito species Aedes albopictus: current knowledge and future perspectives. Trends in parasitology. 2013;29(9):460-8. doi: 10.1016/j.pt.2013.07.003. PubMed PMID: 23916878; PubMed Central PMCID: PMC3777778.

2. Powell JR, Tabachnick WJ. History of domestication and spread of Aedes aegypti - a review. Memórias do Instituto Oswaldo Cruz. 2013;108(Suppl 1):11-7. doi: 10.1590/0074-

0276130395. PubMed PMID: PMC4109175.

3. Skuse FAA. The banded mosquito of Bengal. Indian Museum Notes. 1894;3(5).

47 4. Medlock JM, Hansford KM, Schaffner F, Versteirt V, Hendrickx G, Zeller H, et al. A review of the invasive mosquitoes in Europe: ecology, public health risks, and control options.

Vector borne and zoonotic diseases. 2012;12(6):435-47. doi: 10.1089/vbz.2011.0814. PubMed

PMID: 22448724; PubMed Central PMCID: PMC3366101.

5. Lounibos LP. Invasions by insect vectors of human disease. Annual Review of Entomology.

2002;47(1):233-66. doi: doi:10.1146/annurev.ento.47.091201.145206. PubMed PMID: 11729075.

6. Campbell LP, Luther C, Moo-Llanes D, Ramsey JM, Danis-Lozano R, Peterson AT.

Climate change influences on global distributions of dengue and chikungunya virus vectors.

Philosophical Transactions of the Royal Society B. 2015;370(1665). doi: 10.1098/rstb.2014.0135.

7. Kraemer MU, Sinka ME, Duda KA, Mylne AQ, Shearer FM, Barker CM, et al. The global distribution of the arbovirus vectors Aedes aegypti and Ae. albopictus. Elife. 2015;4:e08347.

8. Goubert C, Minard G, Vieira C, Boulesteix M. Population genetics of the Asian tiger mosquito Aedes albopictus, an invasive vector of human diseases. Heredity. 2016. doi:

10.1038/hdy.2016.35.

9. Battaglia V, Gabrieli P, Brandini S, Capodiferro MR, Javier PA, Chen X-G, et al. The worldwide spread of the tiger mosquito as revealed by mitogenome haplogroup diversity. Frontiers in Genetics. 2016;7. 10. Paupy C, Delatte H, Bagny L, Corbel V, Fontenille D. Aedes albopictus, an arbovirus vector: from the darkness to the light. Microbes and Infection. 2009;11(14–15):1177-85. doi: 10.1016/j.micinf.2009.05.005. 11. Linnaeus C, Hasselquist F. Zweyter Theil, enthalt Beschreibungen verchiedener wuchtiger Naturalien. Hasselquist, F, Reise nach Palestina in den Jahren von. 1762;1749:267-606. 12. Rochlin I, Ninivaggi DV, Hutchinson ML, Farajollahi A. Climate change and range expansion of the Asian tiger mosquito (Aedes albopictus) in Northeastern USA: implications for public health practitioners. PLoS ONE. 2013;8(4):e60874. doi: 10.1371/journal.pone.0060874. 13. Williams CR. The Asian tiger mosquito (Aedes albopictus) invasion into Australia: a review of likely geographic range and changes to vector-borne disease risk. Transactions of the Royal Society of South Australia. 2012;136(2):128-36. doi: 10.1080/03721426.2012.10887167.

48 14. Guillemaud T, Beaumont MA, Ciosi M, Cornuet JM, Estoup A. Inferring introduction routes of invasive species using approximate Bayesian computation on microsatellite data. Heredity. 2010;104(1):88-99. 15. Simberloff D, Martin J-L, Genovesi P, Maris V, Wardle DA, Aronson J, et al. Impacts of biological invasions: what's what and the way forward. Trends in Ecology & Evolution. 2013;28(1):58-66. 16. Brown JE, Scholte E-J, Dik M, Den Hartog W, Beeuwkes J, Powell JR. Aedes aegypti mosquitoes imported into the Netherlands, 2010. Emerging Infectious Diseases. 2011;17(12):2335- 7. doi: 10.3201/eid1712.110992. PubMed PMID: PMC3311167. 17. Mousson L, Dauga C, Garrigues T, Schaffner F, Vazeille M, Failloux AB. Phylogeography of Aedes (Stegomyia) aegypti (L.) and Aedes (Stegomyia) albopictus (Skuse) (Diptera: Culicidae) based on mitochondrial DNA variations. Genetical research. 2005;86(1):1-11. doi: 10.1017/s0016672305007627. PubMed PMID: 16181519. 18. Beebe NW, Ambrose L, Hill LA, Davis JB, Hapgood G, Cooper RD, et al. Tracing the tiger: population genetics provides valuable insights into the Aedes (Stegomyia) albopictus invasion of the Australasian region. PLoS neglected tropical diseases. 2013;7(8):e2361. doi: 10.1371/journal.pntd.0002361. PubMed PMID: 23951380; PubMed Central PMCID: PMC3738475. 19. Kamgang B, Brengues C, Fontenille D, Njiokou F, Simard F, Paupy C. Genetic structure of the tiger mosquito, Aedes albopictus, in Cameroon (Central Africa). PLoS ONE. 2011;6(5):e20257. doi: 10.1371/journal.pone.0020257. 20. Manni M, Guglielmino CR, Scolari F, Vega-Rúa A, Failloux A-B, Somboon P, et al. Genetic evidence for a worldwide chaotic dispersion pattern of the arbovirus vector, Aedes albopictus. PLoS neglected tropical diseases. 2017;11(1):e0005332. 21. Gloria-Soria A, Ayala D, Bheecarry A, Calderon‐Arguedas O, Chadee DD, Chiappero M, et al. Global genetic diversity of Aedes aegypti. Molecular Ecology. 2016;25(21):5377-95. 22. Bennett KL, Shija F, Linton Y-M, Misinzo G, Kaddumukasa M, Djouaka R, et al. Historical environmental change in Africa drives divergence and admixture of Aedes aegypti mosquitoes: a precursor to successful worldwide colonization? Molecular Ecology. 2016;25(17):4337-54. doi: 10.1111/mec.13762. 23. Duffy MR, Chen T-H, Hancock WT, Powers AM, Kool JL, Lanciotti RS, et al. Zika virus outbreak on Yap Island, federated states of Micronesia. New England Journal of Medicine. 2009;2009(360):2536-43. 24. Ong C. Zika virus: an emerging infectious threat. Internal Medicine Journal. 2016;46(5):525-30.

49 25. Roth A, Mercier A, Lepers C, Hoy D, Duituturaga S, Benyon E, et al. Concurrent outbreaks of dengue, chikungunya and Zika virus infections-an unprecedented epidemic wave of mosquito- borne viruses in the Pacific 2012-2014. Euro Surveillance. 2014;19(41):20929. 26. Ritchie SA, Moore P, Carruthers M, Williams C, Montgomery B, Foley P, et al. Discovery of a widespread infestation of Aedes albopictus in the Torres Strait, Australia. Journal of the American Mosquito Control Association. 2006;22(3):358-65. 27. van den Hurk AF, Nicholson J, Beebe NW, Davis J, Muzari OM, Russell RC, et al. Ten years of the tiger: Aedes albopictus presence in Australia since its discovery in the Torres Strait in 2005. One Health. 2016;2:19-24. 28. Muzari MO, Devine G, Davis J, Crunkhorn B, van den Hurk A, Whelan P, et al. Holding back the tiger: successful control program protects Australia from Aedes albopictus expansion. PLoS neglected tropical diseases. 2017;11(2):e0005286. 29. Huang Y-M. Contributions to the mosquito fauna of Southeast Asia. XIV. The subgenus Stegomyia of Aedes in Southeast Asia. I – The scutellaris group of species. Contributions of the American Entomological Institute. 1972;9(1):109. 30. Hill LA, Davis JB, Hapgood G, Whelan PI, Smith GA, Ritchie SA, et al. Rapid identification of Aedes albopictus, Aedes scutellaris, and Aedes aegypti life stages using real-time polymerase chain reaction assays. The American Journal of Tropical Medicine and Hygiene. 2008;79(6):866-75. 31. Beebe NW, Whelan PI, van den Hurk AF, Ritchie SA, Corcoran S, Cooper RD. A polymerase chain reaction-based diagnostic to identify larvae and eggs of container mosquito species from the Australian region. Journal of Medical Entomology. 2007;44(2):376-80. doi: 10.1603/0022-2585(2007)44[376:apcrdt]2.0.co;2. 32. Walker F. Catalogue of the Dipterous insects collected in the Aru Islands by Mr. A. R. Wallace, with descriptions of new species. Journal of the Proceedings of the Linnean Society of London Zoology. 1858;3(10):77-110. doi: 10.1111/j.1096-3642.1858.tb02413.x. 33. Beebe NW, Whelan PI, van den Hurk AF, Ritchie S, Cooper RD. Genetic diversity of the dengue vector Aedes aegypti in Australia and implications for future surveillance and mainland incursion monitoring. Communicable diseases intelligence quarterly report. 2005;29(3):299-304. PubMed PMID: 16220869. 34. Hulce D, Li X, Snyder-Leiby T, Liu CJ. GeneMarker® genotyping software: tools to increase the statistical power of DNA fragment analysis. Journal of Biomolecular Techniques. 2011;22:S35.

50 35. Sutton JT, Robertson BC, Jamieson IG. Dye shift: a neglected source of genotyping error in molecular ecology. Molecular Ecology Resources. 2011;11(3):514-20. doi: 10.1111/j.1755- 0998.2011.02981.x. PubMed PMID: 21481209. 36. Meirmans PG, Van Tienderen PH. GENOTYPE and GENODIVE: two programs for the analysis of genetic diversity of asexual organisms. Molecular Ecology Notes. 2004;4(4):792-4. doi: 10.1111/j.1471-8286.2004.00770.x. 37. van Oosterhout C, Hutchinson WF, Wills DPM, Shipley P. MICRO-CHECKER: software for identifying and correcting genotyping errors in microsatellite data. Molecular Ecology Notes. 2004;4(3):535-8. doi: 10.1111/j.1471-8286.2004.00684.x. 38. Peakall ROD, Smouse PE. Genalex 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes. 2006;6(1):288-95. doi: 10.1111/j.1471- 8286.2005.01155.x. 39. Peakall R, Smouse PE. GenAlEx 6.5: genetic analysis in Excel. Population genetic software for teaching and research—an update. Bioinformatics. 2012;28(19):2537-9. doi: 10.1093/bioinformatics/bts460. 40. Pritchard JK, Stephens M, Donnelly P. Inference of population structure using multilocus genotype data. Genetics. 2000;155(2):945-59. 41. Pritchard JK, Wen X, Falush D. Documentation for structure software: Version 2.3. 2009. 42. Jakobsson M, Rosenberg NA. CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics. 2007;23(14):1801-6. doi: 10.1093/bioinformatics/btm233. 43. Rosenberg NA. DISTRUCT: A program for the graphical display of population structure. Molecular Ecology Notes. 2004;4(1):137-8. 44. RStudioTeam. RStudio: integrated development for R. RStudio, Inc., Boston, MA. 2015. 45. Jombart T, Ahmed I. adegenet 1.3-1: new tools for the analysis of genome-wide SNP data. Bioinformatics. 2011;27(21):3070-1. 46. Jombart T, Devillard S, Balloux F. Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. BMC Genetics. 2010;11(1):94. 47. Jombart T, Collins C. A tutorial for Discriminant Analysis of Principal Components (DAPC) using adegenet 2.0. 0. 2015. 48. Excoffier L, Lischer HE. Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Molecular Ecology Resources. 2010;10. doi: 10.1111/j.1755-0998.2010.02847.x.

51 49. Peery MZ, Kirby R, Reid BN, Stoelting R, Doucet-Bëer E, Robinson S, et al. Reliability of genetic bottleneck tests for detecting recent population declines. Molecular Ecology. 2012;21(14):3403-18. 50. Garza JC, Williamson EG. Detection of reduction in population size using data from microsatellite loci. Molecular Ecology. 2001;10(2):305-18. doi: 10.1046/j.1365- 294X.2001.01190.x. 51. Piry S, Luikart G, Cornuet JM. BOTTLENECK: A computer program for detecting recent reductions in the effective population size using allele frequency data. Journal of Heredity. 1999;90. doi: 10.1093/jhered/90.4.502. 52. Cornuet J-M, Pudlo P, Veyssier J, Dehne-Garcia A, Gautier M, Leblois R, et al. DIYABC v2. 0: a software to make approximate Bayesian computation inferences about population history using single nucleotide polymorphism, DNA sequence and microsatellite data. Bioinformatics. 2014;30(8):1187-9. 53. Bertorelle G, Benazzo A, Mona S. ABC as a flexible framework to estimate demography over space and time: some cons, many pros. Molecular Ecology. 2010;19(13):2609-25. 54. Elliott SA. Aedes albopictus in the Solomon and Santa Cruz Islands, South Pacific. Transactions of the Royal Society of Tropical Medicine and Hygiene. 1980;74(6):747-8. 55. Colless DH. Aedes albopictus in the Hollandia area. Proceedings of the Linnean Society of New South Wales. 1962;(87):312-5. 56. Lee DJ, Hicks M, Griffiths M, Debenham M, Bryan J, Russell R, et al. The Culicidae of the Australasian Region. Volume IV: Australian Government Publishing Service; 1987. 57. Schoenig E. Distribution of three species of Aedes (Stegomyia) carriers of virus diseases on the main island of Papua New Guinea. The Philippine Scientist. 1972;9:61-82. 58. Cooper R, Waterson D, Kupo M, Sweeney A. Aedes albopictus (Skuse)(Diptera: Culicidae) in the Western Province of Papua New Guinea and the threat of its introduction to Australia. Australian Journal of Entomology. 1994;33(2):115-6. 59. Porretta D, Mastrantonio V, Bellini R, Somboon P, Urbanelli S. Glacial history of a modern invader: phylogeography and species distribution modelling of the Asian tiger mosquito Aedes albopictus. PLoS ONE. 2012;7(9):e44515. doi: 10.1371/journal.pone.0044515. 60. Hasegawa M, Kishino H, Yano T-a. Dating of the human-ape splitting by a molecular clock of mitochondrial DNA. Journal of Molecular Evolution. 1985;22(2):160-74. 61. Schug MD, Hutter CM, Wetterstrand KA, Gaudette MS, Mackay T, Aquadro CF. The mutation rates of di-, tri-and tetranucleotide repeats in Drosophila melanogaster. Molecular Biology and Evolution. 1998;15(12):1751-60.

52 62. Brower A. Rapid morphological radiation and convergence among races of the butterfly Heliconius erato inferred from patterns of mitochondrial DNA evolution. Proceedings of the National Academy of Sciences. 1994;91(14):6491-5. 63. Schug MD, Mackay TF, Aquadro CF. Low mutation rates of microsatellite loci in Drosophila melanogaster. Nature genetics. 1997;15(1):99-102. doi: 10.1038/ng0197-99. PubMed PMID: 8988178. 64. Kearse M, Moir R, Wilson A, Stones-Havas S, Cheung M, Sturrock S, et al. Geneious Basic: An integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics. 2012;28(12):1647-9. doi: 10.1093/bioinformatics/bts199. PubMed PMID: PMC3371832. 65. Clement M, Snell Q, Walker P, Posada D, Crandall K, editors. TCS: estimating gene genealogies. ipdps; 2002: IEEE. 66. Librado P, Rozas J. DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics. 2009;25(11):1451-2. 67. Urbanelli S, Bellini R, Carrieri M, Sallicandro P, Celli G. Population structure of Aedes albopictus (Skuse): the mosquito which is colonizing Mediterranean countries. Heredity. 2000;84(3):331-7. 68. Kambhampati S, Black IV WC, Rai KS. Geographic origin of the US and Brazilian Aedes albopictus inferred from allozyme analysis. Heredity. 1991;67(1):85-94. 69. Ismail N-A, Dom NC, Ismail R, Ahmad AH, Zaki A, Camalxaman SN. Mitochondrial cytochrome oxidase I gene sequence analysis of Aedes albopictus in Malaysia. Journal of the American Mosquito Control Association. 2015;31(4):305-12. 70. Zawani M, Abu H, Sazaly A, Zary S, Darlina M. Population genetic structure of Aedes albopictus in Penang, Malaysia. Genetics and Molecular Research. 2014;13(4):8184. 71. Medley KA, Jenkins DG, Hoffman EA. Human-aided and natural dispersal drive gene flow across the range of an invasive mosquito. Molecular Ecology. 2015;24. doi: 10.1111/mec.12925. 72. Roche B, Léger L, L’Ambert G, Lacour G, Foussadier R, Besnard G, et al. The spread of Aedes albopictus in metropolitan France: contribution of environmental drivers and human activities and predictions for a near future. PLoS ONE. 2015;10(5):e0125600. doi: 10.1371/journal.pone.0125600. 73. Miller MJ, Loaiza JR. Geographic expansion of the invasive mosquito Aedes albopictus across Panama - Implications for control of dengue and chikungunya viruses. PLoS neglected tropical diseases. 2015;9(1):e0003383. doi: 10.1371/journal.pntd.0003383. 74. Zhong D, Lo E, Hu R, Metzger ME, Cummings R, Bonizzoni M, et al. Genetic analysis of invasive Aedes albopictus populations in Los Angeles County, California and its potential public

53 health impact. PLoS ONE. 2013;8(7):e68586. doi: 10.1371/journal.pone.0068586. PubMed PMID: 23861921; PubMed Central PMCID: PMC3702605. 75. Hawley WA. The biology of Aedes albopictus. Journal of the American Mosquito Control Association. 1988;1:1-39. PubMed PMID: 3068349. 76. Birungi J, Munstermann LE. Genetic structure of Aedes albopictus (Diptera: Culicidae) populations based on mitochondrial ND5 sequences: evidence for an independent invasion into Brazil and United States. Annals of the Entomological Society of America. 2002;95. doi: 10.1603/0013-8746(2002)095[0125:gsoaad]2.0.co;2. 77. Nicholson J, Ritchie SA, Russell RC, Zalucki MP, van den Hurk AF. Ability for Aedes albopictus (Diptera: Culicidae) to survive at the climatic limits of its potential range in eastern Australia. Journal of Medical Entomology. 2014;51(5):948-57. doi: 10.1603/me14079. 78. Hlaing T, Tun-Lin W, Somboon P, Socheat D, Setha T, Min S, et al. Mitochondrial pseudogenes in the nuclear genome of Aedes aegypti mosquitoes: implications for past and future population genetic studies. BMC Genetics. 2009;10(1):11. 79. Behura SK, Lobo NF, Haas B, deBruyn B, Lovin DD, Shumway MF, et al. Complete sequences of mitochondria genomes of Aedes aegypti and Culex quinquefasciatus and comparative analysis of mitochondrial DNA fragments inserted in the nuclear genomes. Insect Biochemistry and Molecular Biology. 2011;41(10):770-7. doi: 10.1016/j.ibmb.2011.05.006. 80. Guillaumot L, Ofanoa R, Swillen L, Singh N, Bossin HC, Schaffner F. Distribution of Aedes albopictus (Diptera, Culicidae) in southwestern Pacific countries, with a first report from the Kingdom of Tonga. Parasites & Vectors. 2012;5(1):247. 81. DFAT. Solomon Islands: Department of Foreign Affairs and Trade; 2016 [13 May 2016]. Available from: http://dfat.gov.au/trade/resources/Documents/solo.pdf. 82. Frazer I. The struggle for control of Solomon Island forests. The Contemporary Pacific. 1997:39-72. 83. Williamson-Natesan EG. Comparison of methods for detecting bottlenecks from microsatellite loci. Conservation Genetics. 2005;6(4):551-62. doi: 10.1007/s10592-005-9009-5. 84. Alto BW, Juliano SA. Precipitation and temperature effects on populations of Aedes albopictus (Diptera: Culicidae): implications for range expansion. Journal of Medical Entomology. 2001;38(5):646-56. doi: 10.1603/0022-2585-38.5.646. 85. Putman AI, Carbone I. Challenges in analysis and interpretation of microsatellite data for population genetic studies. Ecology and Evolution. 2014;4(22):4399-428. 86. Vontas J, Kioulos E, Pavlidi N, Morou E, Della Torre A, Ranson H. Insecticide resistance in the major dengue vectors Aedes albopictus and Aedes aegypti. Pesticide Biochemistry and Physiology. 2012;104(2):126-31.

54 87. Vazeille M, Zouache K, Vega-Rúa A, Thiberge J-M, Caro V, Yébakima A, et al. Importance of mosquito “quasispecies” in selecting an epidemic arthropod-borne virus. Scientific Reports. 2016;6. 88. Kamgang B, Nchoutpouen E, Simard F, Paupy C. Notes on the blood-feeding behavior of Aedes albopictus (Diptera: Culicidae) in Cameroon. Parasites & Vectors. 2012;5(1):57. doi: 10.1186/1756-3305-5-57.

55 SUPPLEMTARY FIGURES, TABLES & FILES

Fig S2.1. Bayesian STRUCTURE plot (K=9) for 13 microsatellite loci for 911 samples of Aedes albopictus in the study region. Each vertical bar in the plots represents an individual sample, where the colour of the bar indicates the probability of the individual belonging to a genetic cluster. Samples are positioned on the map corresponding to the population’s location (orange dot) and are abbreviated as in Table 2.1. Map insets represent the following: A) Torres Strait Islands and Southern Fly Region of Papua New Guinea; B) Hawaii; C) Atlanta. Insets B and C are to scale with the main map scale. The top-left colour key shows the colour of clusters, as referred to in the main text.

Fig S2.2. A) DAPC of the full dataset for 13 microsatellite loci for Aedes albopictus in the study region. Three-dimensional plots show the relationship between individuals belonging to 23 different populations (represented by coloured dots, where the colour of a dot corresponds to population) using the first 3 principal components (PC1-3). Each plot shows the same data, but is rotated along the horizontal plane. Four distinct clusters (C1-4) are indicated with dashed ellipses (not confidence intervals) and cluster membership of each population is denoted in the legend. This plot is chiefly to visualise the genetic relationships between the four main clusters and specific relationships are discussed in text. B) Correspondence analysis (CA) of the same data, but presenting population means rather than individual data points. Note that for both plots A and B, population definitions varied slightly from STRUCTURE analyses and are shown in Table 2.1 along with population abbreviations.

Fig S2.3. DAPC analysis for 13 microsatellite loci for Aedes albopictus in the study region. A- B) Scatterplots show the front (A) and top (B) view of a DAPC for the reduced dataset of Ae. albopictus (n=458, containing only C3 and C4 from Fig S2.2A), using the first 3 principal components (PC1-3). This excludes populations from the Torres Strait Islands, Jakarta, Sumba, Timor-Leste and Southern Fly Region. Individuals from each of the 18 populations are colour- coded and labeled with a number (see legend in plot A). Note that Solomon Islands includes Gizo, New Mala and Honiara and that Singapore and Malaysia are treated as one population. Ellipses show the 95% confidence intervals of each population.

Fig S2.4. Principal components analysis (PCA) in the space of summary statistics computed for our DIYABC simulations (across all scenarios). Each coloured dot represents a simulated dataset corresponding to the five scenarios (with 10,000 random prior plots displayed per scenario), while the large yellow dot represents our observed dataset. The first three principal components (PC) are shown with their % variance explained by each PC shown in brackets.

Fig S2.5. Principal components analysis (PCA) in the space of summary statistics used for the most likely Australasian invasion scenario (Scenario 4) in our study. The yellow dot represents the observed Ae. albopictus dataset, solid purple dots represent the simulated dataset with parameters drawn from posterior distributions (1,000 random datasets shown), while hollow purple dots corresponds to the datasets simulated based on prior distributions of parameters (1,000 random datasets shown). The % variance explained by principal components (PC) is displayed in brackets and only the first three PCs are plotted.

Fig S2.6. Mitochondrial COI haplotype network for 52 haplotypes identified for 1044 individuals of Aedes albopictus in the Indo-Pacific, Asian and USA region. Haplotypes are coloured by broad geographic region and the size of circles indicates the number of individuals belonging to a given haplotype. Lines joining haplotypes show genetic distance between haplotypes where each mark indicates a single nucleotide substitution. Small black circles represent unsampled haplotypes.

61 Table S2.1. Specific sample information for Aedes albopictus used in the microsatellite study for 911 individual mosquitoes, including microsatellite allele scores for 13 loci. LC, larval collection; IC, immature (pupal or larval) collection; EC, egg collection; HLC, human landing collection; HBS, human baited sweep netting; LT, light trap; ST, sentinel trap; ASP, battery- powered aspirator; unknown details are shown (-).

Refer to supplementary attachment

Table S2.2. Matrices of pairwise FST (A), G"ST (B) and Jost's D (C) values for populations of Aedes albopictus using 13 microsatellite loci. P values are indicated above the diagonal with insignificant (P>0.05) in bold.

Refer to supplementary attachment

Table S2.3. Details used for approximate Bayesian computation (ABC) analysis for investigating the Australasian invasion of Aedes albopictus. Included are various parameter settings and conditions used in our ABC analysis and the posterior distributions of our most likely scenarios (Scenario 4; Scenarios 1 & 4). Results for the confidence in scenario choice are also displayed (type I and type II error).

Refer to supplementary attachment

Table S2.4. Sample information for mtDNA COI sequences, representing 1044 samples of Ae. albopictus.

Refer to supplementary attachment

62 Table S2.5. Estimates of genetic diversity using all 13 microsatellite loci within populations of Aedes albopictus. Displayed are the mean values per population and the standard error ([SE]): mean population size (N), mean number of alleles (Na), number of effective alleles (Ne), observed heterozygosity (Ho), expected heterozygosity (He), unbiased expected heterozygosity (uHe) and fixation index (F).

Region Population (year) N Na Ne I Ho He uHe F Torres Strait Islands Masig (2007) 21 4.92 2.95 1.22 0.45 0.63 0.65 0.28

[0.35] [0.25] [0.08] [0.05] [0.03] [0.03] [0.07] Mer (2007) 6 2.77 2.05 0.79 0.45 0.48 0.52 0.06

[0.26] [0.15] [0.09] [0.07] [0.05] [0.05] [0.12] Warraber (2007) 8 4.15 2.85 1.15 0.46 0.61 0.65 0.24

[0.3] [0.19] [0.1] [0.06] [0.05] [0.06] [0.07] Mabuiag (2007) 10 3.54 2.46 1.01 0.42 0.57 0.6 0.23

[0.22] [0.17] [0.06] [0.07] [0.03] [0.03] [0.13] Waiben (2010) 3 2.23 2.05 0.65 0.41 0.4 0.48 -0.05

[0.3] [0.28] [0.13] [0.11] [0.07] [0.09] [0.19] Ngurupai (2010) 7 3.54 2.47 0.96 0.51 0.52 0.56 0

[0.29] [0.26] [0.11] [0.07] [0.06] [0.06] [0.09] Muralug (2010) 5 3.23 2.6 0.98 0.37 0.56 0.63 0.32

[0.28] [0.25] [0.1] [0.07] [0.05] [0.05] [0.11] Ngurupai (2012) 10 5.15 3.67 1.36 0.43 0.68 0.71 0.38

[0.48] [0.42] [0.11] [0.06] [0.04] [0.04] [0.06] Keriri (2012) 23 5.62 3.42 1.33 0.53 0.65 0.67 0.19

[0.38] [0.41] [0.1] [0.07] [0.04] [0.04] [0.08] Keriri (2013) 10 4.92 3.22 1.25 0.55 0.63 0.67 0.12

[0.51] [0.32] [0.13] [0.07] [0.06] [0.06] [0.07] Keriri (2014) 30 5.85 3.03 1.25 0.48 0.61 0.62 0.24

[0.46] [0.39] [0.11] [0.06] [0.04] [0.04] [0.08] Poruma (2015) 30 6.31 3.26 1.38 0.57 0.68 0.69 0.16

[0.6] [0.19] [0.06] [0.05] [0.02] [0.02] [0.06] Iama (2015) 30 5.54 2.76 1.18 0.42 0.59 0.6 0.29

[0.64] [0.27] [0.11] [0.05] [0.04] [0.04] [0.05] Warraber (2015) 24 6.23 3.28 1.33 0.51 0.64 0.65 0.23

[0.6] [0.38] [0.12] [0.07] [0.05] [0.05] [0.06] Southern Fly Region, PNG Kulalai (2007) 2 2.31 2.12 0.7 0.42 0.43 0.58 0.04

[0.26] [0.24] [0.13] [0.11] [0.07] [0.1] [0.16] Mabaduan (2007) 7 3.69 2.66 1.04 0.43 0.58 0.62 0.26

[0.29] [0.26] [0.1] [0.06] [0.04] [0.05] [0.08] Sigabaduru (2007) 1 1.54 1.54 0.37 0.54 0.27 0.54 -1

[0.14] [0.14] [0.1] [0.14] [0.07] [0.14] [0] Katatai (2008) 2 2.38 2.2 0.74 0.5 0.45 0.6 -0.11

[0.27] [0.24] [0.13] [0.11] [0.08] [0.1] [0.16] Papua New Guinea Kiunga (1992) 17 3.69 2.57 1.01 0.38 0.56 0.57 0.29

[0.44] [0.3] [0.11] [0.04] [0.04] [0.04] [0.07] Port Moresby (1996-1999) 29 6 3.59 1.42 0.54 0.7 0.72 0.21

[0.41] [0.26] [0.07] [0.05] [0.02] [0.02] [0.08] Londolovit (2007) 39 6.62 3.38 1.4 0.52 0.68 0.69 0.26

[0.53] [0.26] [0.08] [0.05] [0.03] [0.03] [0.05] Daru (1992, 2008) 26 5.38 3.02 1.27 0.5 0.65 0.66 0.24

[0.45] [0.2] [0.07] [0.06] [0.02] [0.02] [0.08]

63 Madang (2011) 33 6.92 3.24 1.4 0.5 0.67 0.68 0.26

[0.5] [0.24] [0.07] [0.05] [0.02] [0.03] [0.06] Buka Is. (2013) 14 4.31 2.72 1.13 0.48 0.6 0.62 0.22

[0.31] [0.23] [0.08] [0.06] [0.03] [0.03] [0.09] Papua, Indonesia Timika (2015) 20 5.38 2.95 1.25 0.44 0.63 0.64 0.29

[0.47] [0.26] [0.09] [0.05] [0.03] [0.04] [0.07] Indonesia Timor-Leste (2001) 10 5.15 3.21 1.33 0.55 0.67 0.71 0.17

[0.27] [0.21] [0.06] [0.07] [0.02] [0.02] [0.1] Jakarta (2012) 177 11 3.76 1.51 0.53 0.68 0.68 0.23

[0.73] [0.41] [0.11] [0.06] [0.05] [0.05] [0.06] Sumba (2013) 37 7 3.49 1.44 0.49 0.68 0.69 0.28

[0.44] [0.3] [0.08] [0.04] [0.03] [0.03] [0.05] Singapore Singapore (2013) 4 3.92 3.25 1.2 0.62 0.64 0.73 0.05

[0.37] [0.33] [0.11] [0.1] [0.05] [0.05] [0.14] Malaysia Ipoh (2013) 48 8.77 4.09 1.61 0.58 0.72 0.73 0.2

[0.6] [0.44] [0.08] [0.05] [0.03] [0.03] [0.05] Kota Baru (2013) 7 4.54 3.24 1.26 0.58 0.66 0.71 0.14

[0.39] [0.32] [0.09] [0.07] [0.03] [0.04] [0.1] Kuala Lumpur (2015) 64 9.77 5.1 1.76 0.5 0.76 0.77 0.36

[0.58] [0.56] [0.1] [0.05] [0.03] [0.03] [0.05] Thailand Bangkok (2015) 6 4.15 3.2 1.22 0.56 0.65 0.71 0.11

[0.32] [0.28] [0.09] [0.06] [0.04] [0.04] [0.08] Myanmar Yangon (2013) 13 4.46 3.4 1.28 0.57 0.67 0.69 0.13

[0.31] [0.3] [0.1] [0.06] [0.04] [0.04] [0.07] East Shan State (2013) 5 3.92 3.06 1.11 0.62 0.59 0.66 -0.03

[0.47] [0.41] [0.14] [0.11] [0.06] [0.07] [0.14] Christmas Is. Christmas Is. (2008) 10 4.85 3.22 1.27 0.42 0.65 0.69 0.35

[0.39] [0.32] [0.09] [0.06] [0.03] [0.04] [0.09] Cocos (Keeling) Islands Direction Is. (2008) 18 4.08 2.86 1.15 0.58 0.64 0.66 0.09

[0.29] [0.15] [0.06] [0.04] [0.02] [0.02] [0.07] La Réunion La Réunion (2011) 4 3.15 2.46 0.92 0.56 0.51 0.59 -0.08

[0.27] [0.26] [0.12] [0.09] [0.07] [0.07] [0.07] Solomon Islands Honiara (2013) 23 5.69 3.44 1.34 0.56 0.66 0.67 0.14

[0.49] [0.39] [0.11] [0.05] [0.04] [0.04] [0.07] Gizo (2013, 2014) 14 5.15 2.82 1.23 0.44 0.62 0.64 0.28

[0.37] [0.2] [0.07] [0.04] [0.03] [0.03] [0.06] New Mala (2014) 18 5 3.07 1.25 0.48 0.63 0.65 0.21

[0.34] [0.28] [0.09] [0.05] [0.04] [0.04] [0.09] Fiji Fiji (2015) 5 3.92 3.02 1.17 0.57 0.64 0.71 0.12

[0.38] [0.29] [0.09] [0.07] [0.03] [0.03] [0.1] Nauru Menen (2014) 2 2.46 2.27 0.75 0.62 0.45 0.6 -0.37

[0.29] [0.28] [0.14] [0.12] [0.08] [0.1] [0.1] USA Hawaii (2015) 22 4.46 2.44 1.01 0.41 0.54 0.55 0.26

[0.45] [0.25] [0.1] [0.05] [0.04] [0.05] [0.08] Atlanta (2011) 17 4.54 3.16 1.2 0.49 0.63 0.64 0.21 [0.4] [0.34] [0.11] [0.05] [0.05] [0.05] [0.06]

64 Table S2.6. Summary of mtDNA COI haplotype distribution across 97 populations of Ae. albopictus, containing a total of 1044 individual sequences and 52 unique haplotypes.

Refer to supplementary attachment

Table S2.7. Tajima’s D and significance value (P value < 0.05) as well as nucleotide diversity (π) using COI sequences for populations of Aedes albopictus that had temporal collections from the present study. Sample sizes varied by year and are indicated. We found no significance (P > 0.05) in all calculations of Tajima’s D.

Sample size per Nucleotide diversity Population Tajima's D P value year (π) Daru (1992, 2008) 25, 10 0.002 -1.48 0.94 Port Moresby (1996-1999) 4, 12, 31, 14 0.001 -0.91 0.81 Keriri (2012-2014) 10, 7, 7 0.001 -0.83 0.78 Poruma (2007, 2015) 9, 10 0.002 -0.13 0.53 Warraber (2007, 2015) 10, 9 0.003 0.5 0.31 Masig (2005, 2007) 3, 11 0.005 0.99 0.18

65 Table S2.8. Garza-Williamson index (M-ratio) and Wilcoxon test for heterozygosity excess (P value) for genetic bottlenecks in populations of Aedes albopictus using 13 microsatellite loci. Values in bold indicate a bottleneck (M-ratio ≤ 0.68; two-tailed Wilcoxon test P < 0.05). Populations from the Southern Fly Region, Nauru and Port Moresby (1996, 1997) were not included due to insufficient data (see Table 2.1 for n).

Population (year) M-ratio Wilcoxon Masig (2007) 0.81 0.542 Mer (2007) 0.79 0.38 Warraber (2007) 0.74 0.85 Mabuiag (2007) 0.68 0.244 Waiben (2010) 0.68 0.032 Muralug (2010) 0.7 0.216 Ngurupai (2010) 0.64 0.685 Ngurupai (2012) 0.75 0.376 Keriri (2012) 0.78 0.946 Keriri (2013) 0.69 0.91 Keriri (2014) 0.77 0.191 Poruma (2015) 0.66 0.34 Iama (2015) 0.66 0.127 Warraber (2015) 0.71 0.455 Kiunga (1992) 0.75 0.068 Port Moresby (1998) 0.76 0.305 Port Moresby (1999) 0.68 0.244 Lihir Is. (2007) 0.79 0.414 Daru (1992) 0.77 0.376 Daru (2008) 0.72 0.839 Madang (2011) 0.72 0.007 Buka Is. (2013) 0.69 0.685 Timika (2015) 0.67 0.244 Timor-Leste (2001) 0.76 0.273 Jakarta (2012) 0.82 0.001 Sumba (2013) 0.8 0.542 Singapore (2013) 0.78 0.588 Ipoh (2013) 0.85 0.146 Kota Baru (2013) 0.83 0.588 Kuala Lumpur (2015) 0.81 0.999 Bangkok (2015) 0.7 0.216 Yangon (2013) 0.71 0.027 East Shan State (2013) 0.77 0.569 Christmas Is. (2008) 0.72 0.839 Cocos (Keeling) Islands (2008) 0.71 0.033 La Réunion (2011) 0.81 0.791 Honiara (2013) 0.82 0.839 Gizo (2013) 0.59 0.91 Gizo (2014) 0.57 0.168 New Mala (2014) 0.72 0.999 Fiji (2015) 0.77 0.542 Hawaii (2015) 0.77 0.542 Atlanta (2011) 0.84 0.068

66 File S2.1

SUPPLEMENTARY METHODS

Preliminary STRUCTURE Preliminary analyses were run with K ranging from 2-20 (5 iterations per value of K) for 200,000 generations per iteration with a burn-in of 50,000. These analyses were run separately with and without location priors to assess its effect on clustering; final analyses were run with these priors as we found no major effect on clustering. Analyses were run with the admixture model to allow for mixed ancestry in individuals [1]. The preliminary STRUCTURE output was run in STRUCTUREHARVESTER [2] to assess the likelihood values across K and to infer the most likely value of K (using the Evanno ∆K method [3] and also by assessing L (K)). Additionally, we considered the value of K based on what we perceived to be the most efficient and biologically relevant summary of the data, as realistically a true K is difficult to determine.

Performance of DIYABC A PCA was performed to check congruency between the observed dataset and the distribution of summary statistics based on priors (Fig S4); one should expect the observed data to fall within the PCA data cloud for priors. We used the logistic regression approach [4] to estimate the posterior probability of our tested scenarios using 1% of the simulated datasets closest to the observed data. Confidence in scenario choice was assessed by estimating type I and type II errors using 500 pseudo-observed datasets drawn from prior distributions of a given scenario [5]. The posterior distributions of parameters were estimated for the most likely scenario (Scenario 4) using 1% of the simulated data closest to the observed, and we computed bias and precision of these parameter estimations using 500 pseudo-observed datasets drawn from the posterior distributions. Model checking was also conducted on the most likely scenario (Scenario 4) using PCA in the space of all one and two sample summary statistics using 1000 datasets simulated from the posterior distributions of parameters. If the model fits our data well, both prior and posterior simulated datasets will show overlap with the observed dataset on the PCA planes (Fig S5) [5].

SUPPLEMENTARY RESULTS

STRUCTURE (K=9) At K=9, mainland SE Asia (pink), USA (dark blue) and the Solomon Islands (dark pink) split into more distinct clusters. Christmas Is. appears more similar to Timika and mainland SE Asia, whereas both Cocos (Keeling) Islands and La Réunion appears most similar to populations from SE Asia (however, the sample size for La Réunion is small (n = 4)). The Indonesian (light green) cluster

67 present at K=4 (Fig 2.1) contains substructure which is visible at K=9 and separates Jakarta from Sumba, Timor-Leste and Torres Strait Islands and Fly Region populations (Fig S1). At K=9, the Jakarta population (light yellow) appears distinct from a group containing populations from Timor- Leste and Sumba (orange) and another cluster mostly containing the PNG Southern Fly Region and Torres Strait Islands (collections between 2006 – 2012) (yellow). Similar to the lower K value results, the more recent collections from the Torres Strait Islands (2013 – 2015) form a distinct cluster (dark orange) from the older collections (yellow). A subset of Torres Strait Island populations appear more similar to the Sumba/Timor-Leste cluster (orange) (see Ker ‘12, Ker ‘14, Ngu ’12) but also contain admixture with the Jakarta cluster (light yellow) as well as Timika/Cocos (Keeling) Islands (blue). The historically-established PNG populations, (excluding Lihir Is.), remain distinct as their own private cluster (red), although some individuals show signs of admixture with the Solomon Islands and mainland SE Asia (pink and dark pink). At K=9, Lihir Is. clusters with the Solomon Islands (dark pink) whereas at K=4 most individuals clustered with PNG. In addition, Timika forms a separate cluster with similar structure to the Cocos (Keeling) Island population (blue); both of these populations also have a notable degree of admixture with mainland SE Asia.

SUPPLEMENTARY DISCUSSION

USA As with many previous studies [6-11], we found that the United States (mainland USA and Hawaii) populations of Ae. albopictus appear more closely related to temperate/northern Asian populations than to more southern, tropical SE Asian populations, as evidenced by their similarity in COI haplotypes. One of the limitations of our study is that it does not include microsatellite samples from temperate Asian populations (e.g. China, Korea and Japan). There are however, multiple shared haplotypes between the USA and mainland SE Asian populations, which could reflect both historic and contemporary movements between these regions. Additionally, microsatellite genetic distance measures show that USA populations are most similar to mainland SE Asian populations

(FST = 0.128 - 0.23) within our study, but are still quite distant. Despite the lack of temperate Asian samples in our microsatellite analyses, we show that at K=4 populations from the USA appear in the same genetic cluster as mainland SE Asia, while at K=9 they form their own distinct cluster.

68 Indian Ocean Islands

Madagascar and La Réunion Populations of Ae. albopictus from Madagascar and La Rèunion are largely thought to have originated from SE Asia as a result of human movements, particularly due to the multiple dispersal opportunities associated with human migration waves (as early as 1,500 - 2,000 ybp based on anthropological data [12] and the spice trade occurred (during the 17th and 18th centauries). Interestingly, we found that Madagascar and La Réunion share most COI haplotypes with the USA and subtropical and temperate regions of Asia (China, Taiwan, Japan). Delatte, Bagny (12) showed that there are multiple distinct genetic groups in Madagascar and La Rèunion that correspond to an ancient lineage and a more widespread contemporary lineage. More recently, Manni, Guglielmino (13) found some support that the introduction into La Rèunion was of SE Asian origin (genetically similar to a Thailand population of Ae. albopictus), but due to overlapping 95% confidence intervals, this could have alternatively been a derivation from an admixture event (between Thailand and Japan) or from China. In addition, the species has recently (since the 1980s) undergone a significant expansion in its distribution in the eco-climatically diverse region of Madagascar [14], where the species’ ecological plasticity (along with changes in rain regimes and human movements) has facilitated its establishment in coastal, tropical and high altitude, temperate habitats. Potential admixture with (and colonisation of) lineages from temperate Asia and the USA may also explain the species’ recent range expansion in Madagascar and ability to exploit temperate conditions. China represents a key import source for Madagascar (20.6% of inter-country trade) and this could explain shared COI haplotypes between these regions [15]. Indeed, it is apparent that COI haplotypes may be associated with the overwintering eggs of Ae. albopictus, with H4 being more prevalent in regions that experience temperate climate.

Cocos (Keeling) Islands and Christmas Island The Cocos (Keeling) Islands and Christmas Island populations are thought to have been introduced in two separate events, both of suspected mainland SE Asian or Indonesian origin (refs). The Cocos (Keeling) Islands introduction represents a more historical invasion (probably introduced between 1879 – 1905 [16]) whereas Christmas Island was colonised by Ae. albopictus in the early 1990s (not detected in 1989, but found after 1996 [17]). Our microsatellite analyses show that both populations are genetically distinct from each other which could reflect their isolation and different introduction sources. Microsatellites revealed that both islands are genetically similar to mainland SE Asian populations (excluding Myanmar: FST = 0.071 - 0.121), supporting their suspected introduction source and corresponding to historical routes taken by British and Dutch vessels for Cocos (Keeling) Islands and more contemporary movements for Christmas Island [16, 17]. Interestingly,

69 COI haplotypes found on the islands (H4 & H5) have a wide distribution in the Indo-Pacific and do not assist in discerning the origin/s of the populations. Note that our sample sizes were small and results are not overly convincing of the introduction source of Ae. albopictus on the islands and further investigation is needed.

Lihir and Buka Islands (PNG) Geographically, Lihir Is. and Buka Is. are part of the northern Solomon archipelago, but administratively part of PNG. Interestingly, the Lihir Is. population clusters mostly with the PNG cluster when K=4, but more so with the Solomon Island populations when K=9. In contrast, Buka Is. clusters consistently with the mainland PNG cluster at both values of K. The most likely explanation for this pattern could be that human connectivity, influenced by geo-political boundaries and commercial activity between the Solomon Islands and PNG, may be influencing the population structure within the PNG-Solomon region. Studies on the malaria mosquitoes (Anopheles) throughout the same region reveal strikingly different population genetic relationships [18, 19], but the dispersal of these species are not as closely associated with humans as Ae. albopictus. However, we cannot reject the idea of temporal variation contributing to the observed genetic structure in these populations, generated by the timing of collections – Buka Is. collections were from 1999 compared to Solomon Is. in 2013-2014 and Lihir Is. in 2007.

Fiji and Nauru

Our results showed that Fiji and Nauru were more similar to mainland SE Asia (FST = 0.024 -

0.116) than to the Solomon Islands (FST = 0.039 - 0.118), but more samples from Fiji and Nauru are needed to clarify this relationship. Fiji has significant import trade with Singapore and Nauru imports primarily from Fiji [20]. In addition, there are weekly flights between Nauru and Fiji. Clear population structure has been recently shown for Ae. aegypti in the South Pacific, which researchers suspected could be driven by a combination of factors including human-mediated movements, island isolation and environmental factors [21]. Aedes aegypti has been established since the early 19th century in the region, so it is likely that the population dynamics of the more recently introduced Ae. albopictus differs, but the two species probably share similar modes of human- mediated dispersal. It is suspected that Fiji was the major source of introductions of Ae. albopictus into nearby regions such as Tonga and Vanuatu through intense sea and air traffic [21, 22], but this is yet to be tested under a coalescent framework (such as ABC analysis) and the chance of long distance introduction events from outside the South Pacific cannot be ruled out at this point.

70 Control and surveying on the Torres Strait Islands More than a decade after its first detection in the Torres Strait, Ae. albopictus is currently well established in large numbers and dominates the mosquito fauna on many Torres Strait islands (especially outer coral cay islands such as Masig, Poruma, Erub and Warraber) [23]. Most of the outer Torres Strait islands (furthest from Australia’s northernmost mainland tip – Cape York) have had Ae. albopictus since its initial detection in 2005. A control program attempted to eradicate Ae. albopictus from all outer islands between 2006-2008, but from 2009 onwards control efforts were shifted to the inner islands (those closer to the Australian mainland with direct transport nodes) Ngurupai and Waiben, along with the establishment of stricter quarantine zones in this inner region to prevent establishment on mainland Australia. Between 2006 and 2010, adult collections (human- baited sweep netting) on Masig, Poruma and Warraber islands revealed evidence of declines in the population sizes of Ae. albopictus in the dry season. However, after control efforts refocused on inner islands post 2009, there was substantial growth in population sizes by 2010 and it is assumed that populations have continued growing due to the lack of control measures [24]. Few surveys have been conducted on these outer islands since 2010, but a recent survey in 2016 found Ae. albopictus in high densities on many outer islands [24]. These surveys were conducted due to an outbreak of dengue on the outer islands of Darnley and Badu in early 2016, supposedly linked to a large outbreak on Daru and other regions of Western Province, PNG [24].

REFERENCES (for File S2.1)

1. Pritchard JK, Wen X, Falush D. Documentation for structure software: Version 2.3. 2009. 2. Earl D, vonHoldt B. STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genet Resour. 2012;4(2):359-61. doi: 10.1007/s12686-011-9548-7. 3. Evanno G, Regnaut S, Goudet J. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology. 2005;14(8):2611-20. Epub 2005/06/23. doi: 10.1111/j.1365-294X.2005.02553.x. PubMed PMID: 15969739. 4. Cornuet J-M, Pudlo P, Veyssier J, Dehne-Garcia A, Gautier M, Leblois R, et al. DIYABC v2. 0: a software to make approximate Bayesian computation inferences about population history using single nucleotide polymorphism, DNA sequence and microsatellite data. Bioinformatics. 2014;30(8):1187-9. 5. Cornuet JM, Pudlo P, Veyssier J, Loire E, Santos F, Dehne-Garcia A, et al. DIYABC version 2.1 - A user-friendly software for inferring population history through Approximate Bayesian Computations using microsatellite, DNA sequence and SNP data. 2015.

71 6. Birungi J, Munstermann LE. Genetic structure of Aedes albopictus (Diptera: Culicidae) populations based on mitochondrial ND5 sequences: evidence for an independent invasion into Brazil and United States. Annals of the Entomological Society of America. 2002;95. doi: 10.1603/0013-8746(2002)095[0125:gsoaad]2.0.co;2. 7. Kambhampati S, Black IV WC, Rai KS. Geographic origin of the US and Brazilian Aedes albopictus inferred from allozyme analysis. Heredity. 1991;67(1):85-94. 8. Urbanelli S, Bellini R, Carrieri M, Sallicandro P, Celli G. Population structure of Aedes albopictus (Skuse): the mosquito which is colonizing Mediterranean countries. Heredity.

2000;84(3):331-7.

9. Zhong D, Lo E, Hu R, Metzger ME, Cummings R, Bonizzoni M, et al. Genetic analysis of invasive Aedes albopictus populations in Los Angeles County, California and its potential public health impact. PLoS ONE. 2013;8(7):e68586. Epub 2013/07/19. doi:

10.1371/journal.pone.0068586. PubMed PMID: 23861921; PubMed Central PMCID:

PMC3702605.

10. Morales Vargas RE, Phumala-Morales N, Tsunoda T, Apiwathnasorn C, Dujardin J-P. The phenetic structure of Aedes albopictus. Infection, Genetics and Evolution. 2013;13:242-51. doi:

10.1016/j.meegid.2012.08.008.

11. Mousson L, Dauga C, Garrigues T, Schaffner F, Vazeille M, Failloux AB. Phylogeography of Aedes (Stegomyia) aegypti (L.) and Aedes (Stegomyia) albopictus (Skuse) (Diptera: Culicidae) based on mitochondrial DNA variations. Genetics Research. 2005;86(1):1-11. Epub 2005/09/27. doi: 10.1017/s0016672305007627. PubMed PMID: 16181519.

12. Delatte H, Bagny L, Brengue C, Bouetard A, Paupy C, Fontenille D. The invaders: phylogeography of dengue and chikungunya viruses Aedes vectors, on the South West islands of the

Indian Ocean. Infection, Genetics and Evolution. 2011;11(7):1769-81.

13. Manni M, Guglielmino CR, Scolari F, Vega-Rúa A, Failloux A-B, Somboon P, et al.

Genetic evidence for a worldwide chaotic dispersion pattern of the arbovirus vector, Aedes albopictus. PLoS Neglected Tropical Diseases. 2017;11(1):e0005332.

72 14. Raharimalala F, Ravaomanarivo L, Ravelonandro P, Rafarasoa L, Zouache K, Tran-Van V, et al. Biogeography of the two major arbovirus mosquito vectors, Aedes aegypti and Aedes albopictus (Diptera, Culicidae), in Madagascar. Parasites & Vectors. 2012;5(1):56. PubMed PMID: doi:10.1186/1756-3305-5-56.

15. DFAT. Madagascar: Department of Foreign Affairs and Trade; 2016 [13 May 2016].

Available from: https://dfat.gov.au/trade/resources/Documents/masy.pdf.

16. Murray M, Marks E. Blood-sucking Diptera of the Cocos (Keeling) Islands. Austral

Entomology. 1984;23(4):265-8.

17. Harrington S, Lindsay M, Douglas A. Christmas Island and Cocos (Keeling) Islands, Indian

Ocean: Mosquito fauna and mosquito-borne disease risk assessment and management recommendations. Western Australia Department of Health, 2009.

18. Ambrose L, Cooper RD, Russell TL, Burkot TR, Lobo NF, Collins FH, et al. Microsatellite and mitochondrial markers reveal strong gene flow barriers for Anopheles farauti in the Solomon

Archipelago: implications for malaria vector control. International Journal for Parasitology.

2014;44(3-4):225-33. Epub 2014/01/21. doi: 10.1016/j.ijpara.2013.12.001. PubMed PMID:

24440418.

19. Seah IM, Ambrose L, Cooper RD, Beebe NW. Multilocus population genetic analysis of the

Southwest Pacific malaria vector Anopheles punctulatus. International Journal for Parasitology.

2013;43(10):825-35. doi: http://dx.doi.org/10.1016/j.ijpara.2013.05.004.

20. DFAT. Nauru: Department of Foreign Affairs and Trade; 2016 [13 May 2016]. Available from: http://dfat.gov.au/trade/resources/Documents/naur.pdf.

21. Calvez E, Guillaumot L, Millet L, Marie J, Bossin H, Rama V, et al. Genetic diversity and phylogeny of Aedes aegypti, the main arbovirus vector in the Pacific. PLoS Neglected Tropical

Diseases. 2016;10(1):e0004374-e.

73 22. Guillaumot L, Ofanoa R, Swillen L, Singh N, Bossin HC, Schaffner F. Distribution of Aedes albopictus (Diptera, Culicidae) in southwestern Pacific countries, with a first report from the

Kingdom of Tonga. Parasites & Vectors. 2012;5(1):247.

23. van den Hurk AF, Nicholson J, Beebe NW, Davis J, Muzari OM, Russell RC, et al. Ten years of the tiger: Aedes albopictus presence in Australia since its discovery in the Torres Strait in

2005. One Health. 2016;2:19-24.

24. Muzari MO, Devine G, Davis J, Crunkhorn B, Van Den Hurk A, Whelan P, et al. Holding back the tiger: successful control program protects Australia from Aedes albopictus expansion.

PLoS Neglected Tropical Diseases. 2017;11(2):e0005286.

74 CHAPTER III – Population structure and invasion origins of Aedes aegypti (Linnaeus, 1762) throughout Southeast

Asia and Australasia

ABSTRACT

The yellow fever mosquito, Aedes aegypti (Linnaeus, 1762), is a highly invasive and medically significant vector of dengue and Zika viruses, whose global spread can be attributed to increased globalisation in the 15th through 20th century. Records of the invasion history of Ae. aegypti across Southeast Asia are sparse and there is little knowledge regarding the invasion routes that the species exploited to gain a foothold in the vast geographic region of the Indo-Pacific. Likewise, the region lacks a thorough and geographically-broad investigation of Ae. aegypti population genetics, yet it is highly impacted by the diseases transmitted by this species. Here I employed 11 nuclear microsatellites and mitochondrial COI sequences coupled with widespread sampling to characterise population structure at a broader geographic level and to investigate the invasion history of Ae. aegypti across this region using approximate Bayesian computation. I found evidence of high spatial genetic structure across the region, with signs of high gene flow and admixture between many geographically distant populations. This is likely the result of human-mediated transport. For the first time in this region, I provide a more specific summary of the most plausible scenario of the historical spread and establishment of Ae. aegypti. This appears to have commenced ~240 years ago and established via multiple independent invasions. Overall, my study enhances our understanding of the invasion history and current population structure of Ae. aegypti in Australasia and Southeast Asia. Understanding the relatedness of populations can assist with predicting mosquito movements which can be useful for some control strategies, particularly those which rely the release and dissemination of mosquitoes (such as sterile insect technique, Wolbachia-based technologies and future gene drive tools).

75 INTRODUCTION

The yellow fever mosquito Aedes aegypi (Linnaeus, 1762) is a highly polymorphic species native to Africa. It can be found in both a ‘domestic’ (associated with human) and sylvan form. The domestic form has a widespread global distribution in the tropics and subtropics where it is a major pest and acts as the principal vector of dengue, chikungunya and Zika viruses. Recent molecular studies examining the global population genetics of the species have suggested that the “Asian” (including some populations from the Pacific and Australia) invasion of Ae. aegypti was most likely seeded from the Americas rather than from an African source. Other such studies present conflicting results and pathways (West Africa to Asia to the Americas; see (Bennett et al., 2016)), albeit with low confidence in results. Consequently, the timing and possible routes that Ae. aegypti exploited to colonise areas of Asia, Australasia and the South Pacific requires more thorough investigation, particularly because investigations at a global scale tend to combine broad geographic regions to simplify invasion scenarios.

Aedes aegypti appears to have established in Asia in the late 19th century coinciding with the first reports of dengue fever from an urban setting, where Ae. aegypti was the suspected vector (Smith, 1956). In Southeast Asia, the species seems to have initially established in major ports around the Malaysian Peninsula in Singapore (Fontaine (1899) quoted by Theobald (1901)), Port Klang (Malaysia) (Daniels, 1908) and Indonesia (Java, Sumatra and Sulawesi (1901-1916; Marlatt quoted by Howard et al. (1917), (Boyce, 1911; Schüffner & Swellengrebel, 1914; Stanton, 1920)) in the early 1900s, before spreading along the coast and then inland. Other regions of Asia showed mostly coastal distributions or only establishment at major ports, suggesting later introductions in regions including Thailand, Vietnam, India, Myanmar and China (Farner et al., 1946; Kumm, 1931; Theobald, 1911). Ports in the Bay of Bengal could have acted as an important introduction pathway into Asia given its strong history of trade; although the first occurrence records were from 1899 in India (Goodrich (1899) and James (1900) quoted by Theobald (1901)) and 1901 in Upper Myanmar (Watson quoted by Theobald (1901)). In Australia, reports of dengue suggest the species established during a similar time period to the Asian region, possibly prior. The first indigenous outbreaks of dengue in Australia occurred in Townsville, Queensland (QLD) in 1879 and later in Rockhampton, QLD in 1885 (Lumley & Taylor, 1943a), with several epidemics later described during the 1890s and early 20th century (Mackenzie et al., 1996). In comparison, urban endemicity of dengue commenced in India and Indonesia from the late-1800s and early-1900s for mainland Southeast Asia (Smith, 1956); although little interest was paid to dengue during this time. The first specimen of Ae. aegypti in Australia was recorded from the remote inland Queensland town of

76 Cunnamulla in 1881 (Lumley & Taylor, 1943b; Taylor, 1915a), shortly followed by a record from Brisbane in 1887 (Skuse, 1889). Once in Australia, Ae. aegypti spread rapidly via rail (Hamlyn- Harris, 1927), both inland and along the coast. Since then, its distribution has decreased substantially due to reduction in rainwater tanks in the second half of the 20th century (Trewin et al., 2017). Today it is only found in Queensland where it is responsible for occasional outbreaks of dengue fever in northern regions. Unlike the recent arrival of Ae. albopictus into northern Australia which appears to have originated from Indonesia (Beebe et al., 2013; Maynard et al., 2017), genetic evidence (SNP and nuclear gene sequences) suggests the older invasion by Ae. aegypti into Australia was possibly from either an Asian, American or Western Pacific source (Gloria‐Soria et al., 2016; Powell & Tabachnick, 2013). The possibility of the Mediterranean acting as a source following the opening of the Suez Canal has also been suggested (Powell et al., 2018). However, past studies have also shown that, in Australia, the Torres Strait Islands’ population of Waiben are distinct from other northern Queensland populations using the mitochondrial gene cytochrome oxidase subunit I (COI) (Beebe et al., 2005; Cooper et al., 2005).

Nearby, in New Guinea, Ae. aegypti has been recorded as present on the island since 1907 by Theobald (1907), but more specifically noted in Friedrich Wilhelmshafen (now Madang, PNG) and Dorey (now Manokwari, West Papua, Indonesia) later in 1910 (Walker & Biro quoted by Theobald (1911)), with further records of the species present on steamers (De Rook quoted by Bonne- Wepster and Brug (1932)) travelling to Tanah Merah (southern Netherlands New Guinea) and present at various locations in the New Guinea region between 1910 - 1930s (Hill, 1925; Howard et al., 1917; Stanton, 1920; Taylor, 1914; Taylor, 1915b; Theobald, 1911). Farner et al. (1946) noted the species’ distribution was somewhat discontinuous in New Guinea and limited to areas connected through sea and river traffic; a pattern which was still apparent in 1987 (Lee et al., 1980), reflecting the species’ strong ties to human movements. To the east, within the Nggella Islands of the Solomon Archipelago, Ae. aegypti was common in the houses of Tulagi (the then capital city) (Garment quoted by Edwards (1925); (Ferguson, 1923)) and the nearby Purvis Bay in 1925 (White quoted by Buxton (1927)). It therefore appears that Ae. aegypti began to establish in the Solomon Islands around 1920-1930s. During World War II, troop movements into the Pacific Islands likely greatly expanded the distribution of Ae. aegypti and contributed to dispersal between geographically distant populations (Calvez et al., 2016; Failloux et al., 2002). For an overview of occurrence records within Southeast Asia and Australasia refer to Fig 3.1 and Table S1. Overall, the records presented in the figure show that the first recorded appearances of Ae. aegypti occurred rapidly and at a similar timeframe at the turn of the 20th century (Fig 3.1). These correspond broadly with urban dengue and chikungunya records in the region (Carey, 1971; Mackenzie et al., 1996; Smith, 1956).

77 By the 1940s, Ae. aegypti was ubiquitous in the tropics of the region (see Farner et al. (1946) distribution map), but remained absent from certain areas (Kraemer et al., 2015a; Kraemer et al., 2015b).

Reconstruction of the invasion history of Ae. aegypti within Southeast Asia and Australasia should consider both dengue outbreak and species occurrence records, which can then be tested using genetic markers under coalescent methods such as approximate Bayesian computation (ABC). Here I employ this approach to investigate the most probable route of invasion for Ae. aegypti in Southeast Asia and Australasia. This is a region that lacks thorough investigation of Ae. aegypti population genetics despite being highly impacted by the diseases transmitted by this vector. I also characterise the population structure of multiple populations of Ae. aegypti in this region to provide a more robust understanding of the genetic structure in the region. I hypothesised that Ae. aegypti’s invasion into Southeast Asia and Australasia would reflect a history of multiple, independent introductions. So far, high spatial structure has been uncovered in this species in various parts of Asia, Australia and Indonesia, but many of the populations that I include here have been unexplored and allow us to investigate whether different countries/political boundaries harbour distinct genetic diversity. Gaining knowledge of population structure can be vital to the successful implementation of some mosquito control techniques and has important implications for biosecurity (Brown et al., 2011; Gloria-Soria et al., 2014; Schmidt et al., 2019). Control technologies such as the sterile/incompatible insect technique (SIT/IIT) and Wolbachia-based population suppression or transformation (replacement) technologies rely heavily on understanding population dynamics and genetic structure in order to ensure efficient release of male mosquitoes to eliminate populations and detect reintroduction from re-emergence (Schmidt et al., 2018). Moreover, in this study I characterise the population genetic structure of more southerly populations in Queensland, Australia. These populations are particularly important as they present a significant threat for the re- establishment of the Ae. aegypti in major Australian cities such as Brisbane (Trewin et al., 2017).

Fig 3.1. Sample sites of the present study and invasion history of Ae. aegypti with regards to historical presence records. This figure does not necessarily reflect the actual date of invasion, but provides a simple overview based on some of the first records of Ae. aegypti in the study region (1880 - 1940). Circles indicate occurrence records and are colour-coded based on timing (corresponding to the timeline (left)). Black triangles represent sample sites in the present study (refer to the key for population names). The yellow shaded area represents an early distribution map of Ae. aegypti by Theobald (1911) which has been modified slightly to fit the current map and to correct Australian records from Southern Australia (which were unreliable: see Lee et al. (1980)). Dashed dark grey lines show major shipping routes as a result of the opening of the Suez and Panama canals, while lighter grey lines show the density of shipping movements between 1784 and 1863 (US Maury Collection; modified from Ben Schmidt). For plotted records see Table S3.1.

79 MATERIALS AND METHODS

Collection sites and methods

My collection sites comprised 20 population samples distributed throughout Southeast Asia and Australasia (Fig 3.1, black triangles). This included populations from Arizona (USA), Australia, New Guinea (PNG and Papua-Indonesia), the Solomon Islands, Indonesia (Bali, Sulawesi and Sumba) and mainland SE Asia (Malaysia, Thailand and Cambodia) (Table 3.1; Table S3.2). Both adult and larval samples were collected using human-baited captures, sweep netting, or the collection of larvae from suitable breeding habitats. Samples were either stored in 70-100% EtOH or desiccated (adults) over silica beads. Species identification was verified through morphology or in difficult cases using PCR-restriction digest diagnostics (Beebe et al., 2007).

Microsatellite processing

DNA was extracted from samples of Ae. aegypti using a salt extraction protocol (Beebe et al. 2005) and diluted at 1:10 in 1X TE buffer (Tris, EDTA). Samples were screened for 11 microsatellite markers (Table 3.2). These markers have been employed in previous population genetic studies on Ae. aegypti (Calvez et al., 2016; Gloria‐Soria et al., 2016). I attempted to use an additional microsatellite marker (AG2) but found that this consistently failed to amplify in many samples and was thus excluded early in the study. Microsatellites were amplified and tagged with fluorescent dye using M13 tails in 15.4μl reactions consisting of 10.8μl H2O, 3μl 5X Mytaq buffer (Bioline, with pre-optimised concentrations of dNTPs and MgCl), 0.1μl 10μM M13 tagged forward primer, 0.2μl 10μM reverse primer, 0.2μl M13 tagged fluorescent dye (VIC, NED, PET or FAM), 0.01μl (1U) MyTaq polymerase and 1μl of 1:10 DNA template. Subsequent PCR involved denaturation at 96ºC for 3 min, followed by 13 cycles of denaturation at 95ºC for 30 sec, annealing at 56ºC for 40 sec (with a gradient decrease of 0.5ºC/cycle) and extension at 72ºC for 30 sec. This was followed by a further 25 cycles of 95ºC for 30 sec, 50ºC for 40 sec and 72ºC for 30 sec. Then a final elongation step of 5 min at 72ºC before cooling to 4ºC. Amplification was confirmed by running 1μl of the PCR product on a 2% agarose gel stained with MidoriGreen (Bulldog Bio) (1μl per 100ml of 2% agarose in 1X TBE buffer). Successfully amplified samples were sent to Macrogen Inc. (Republic of Korea) for fragment analysis on an ABI 3730XL DNA analyser (Applied Biosystems, Waltham, Massachusetts, USA).

80 Table 3.1. Sample information for Aedes aegypti from Southeast Asia and Australia (n=366). Regional and population definitions are shown, as are population abbreviations used in some figures and text. Sample size (n) is also indicated per population. Regional abbreviations that are used in approximate Bayesian computation are shown in brackets in the ‘Region’ column. Further details are in Table S3.2.

Population Region Population n abbreviation Waiben (Thursday Is.), Torres Strait TS 18 Islands Holloways Beach HB 17 Cairns CA 26 Australia Charters Towers CT 22 (AUS) Longreach LR 26 Rockhampton RO 25 Emerald EM 27 Mt Morgan MT 21 Yeppoon YE 12 Honiara, Guadalcanal, Solomon Islands SOL 18 Pacific (PAC) Port Moresby, Papua New Guinea PNG 8 Timika, Papua TIM 16 Amamapare, Mimika, Papua MIM 10 Indonesia Luwuk, Sulawesi SUL 14 (INA) Waitabula, Sumba SUM 8 Kuta, Bali BAL 27 Malaysia Kuala Lumpur MAL 35 (MAL) Mainland Bangkok, Thailand BKK 11 Southeast Asia Tro Pang Sap Village, Cambodia CAM 13 (SEA)

United States of Tucson, Arizona AZ 12 America (USA)

Raw microsatellite data was processed using the standardisation run wizard (default animal fragment settings) in GeneMarker v.2.4.2 (SoftGenetics LLC (Hulce et al., 2011)) and alleles were scored manually. A random selection of genotyped plates were scored by a second person to assess consistency in results. Poor quality samples with weak or messy peaks were removed from the final data set due to an excess of missing data; additionally, those with fewer than eight out of 11 scored

81 loci were removed for being of poor quality. This left 366 individuals for the final analyses (Table S3.2).

Microsatellite characteristics and genetic distance

I replaced missing microsatellite values based on mean population allele frequencies using GenoDive v. 2.0b27 (Meirmans & Van Tienderen, 2004); this was used to conduct DAPC and pairwise genetic distance calculations. Missing values were not replaced for STRUCTURE analyses or for the calculation of Hardy-Weinberg equilibrium (HW), linkage disequilibrium and for testing for null alleles. For each locus I calculated allelic richness (Na), number of effective alleles (Ne), observed (Ho) and unbiased expected values of heterozygosity (uHe) and global FST with and without the exclusion of null alleles using Genepop v.4.2 (Raymond & Rousset, 1995; Rousset, 2008) and FreeNA (Chapuis & Estoup, 2006). I checked for HW (with Bonferroni correction) using GenAlEx v.6.5 (Peakall & Smouse, 2006, 2012) whereas deviations from linkage disequilibrium were calculated in Genepop. Pairwise population indices of genetic variation for FST, G”ST and Jost’s D were calculated between populations in GenAlEx v.6.5 and the significance was tested using an analysis of molecular variance (AMOVA, 9,999 permutations).

Isolation by distance was assessed using a mantel test in the adegenet 2.1.1 package (Jombart, 2008; Jombart & Ahmed, 2011) in R v.3.4.4 (RCoreTeam, 2018) using matrices of Edward’s genetic distance and Euclidean geographic distances based on 9,999 replicates.

STRUCTURE analysis

Population structure was investigated using the program STRUCTURE v.2.3.4 (Pritchard et al., 2009). Preliminary analyses were conducted to investigate the most probable number of population clusters (K) present in the data and to explore the effect of models using the admixture and population prior settings. Based on these preliminary analyses, the final analysis was run with the admixture model and using sampling locations as a prior, with K ranging from 2 – 22 (20 iterations per value of K) with a burn-in of 100,000 followed by 1,000,000 iterations. The output from the STRUCTURE run was processed in STRUCTUREHARVESTER (Earl & vonHoldt, 2012) to infer the most likely value of K using the Evanno ∆K and L(K) methods. I analysed subsets of the data based on these STRUCTURE results (commonly referred to as a hierarchical approach, where distinct clusters are sub analysed in independent STRUCTURE runs to explore any substructure). For these sub-analyses I used the same settings and run time, but the value of K ranged based on the

82 number of populations being analysed. Final plots were made using pophelper (Francis, 2017). CLUMPAK (Kopelman et al., 2015) was used to assess K values for each analysis.

DAPC/K-means clustering

I conducted discriminant analyses as an alternative approach to examine population structure using adegenet. Group membership was predefined based on sampling location (Table 3.1; Population) and DAPC was initially conducted on the whole dataset. I performed cross-validation on the DAPC using a validation set of 10% and training dataset of 90% with 100 replicates. To avoid overfitting the discriminant functions in DAPC, I considered the optimum number of principal components (n.pca = 30) to retain as that being associated with the lowest root mean squared error (RMSE) (Jombart & Collins, 2015). Five discriminant functions were retained but only the first three were plotted as these explain the most variance. To assist with the display of data, I plotted population means from the DAPC to highlight patterns and reduce noise in the plots.

To further explore clustering in the dataset given no prior population information (i.e. assuming populations are unknown), I used K-means clustering where the various clustering outcomes were compared using the Bayesian information criterion (BIC). The K-means clustering was performed using the adegenet package on transformed data (using PCA where all principal components were retained). I used the lowest BIC to infer the optimal K value. Inferred group memberships were plotted against actual group (population) membership. I additionally performed this using regional definitions (Table 3.1; Region) to explore how this affected reassignment of individuals.

COI analysis

An approximately 550bp region of the mitochondrial gene COI was amplified using previously used (Beebe et al. 2005) primers (aegCOI-250F 5’-TAG-TTC-CTT-TAA-TAT-TAG-GAG-C-3’ and aegCOI-800R 5’-TAA-TAT-AGC-ATA-AAT-TAT-TCC-3’) for 117 individuals. Each 16μl reaction contained 10.4μl H2O, 4μl 5X Mytaq buffer (Bioline, with pre-optimised concentrations of dNTPs and MgCl2), 0.2μl 100μM forward primer, 0.2μl 100μM reverse primer, 0.2μl MyTaq polymerase and 1μl DNA template. For PCR, I used an initial denature of 94°C for 3 min, 35 cycles of denaturation at 95°C for 30 sec, primer annealing at 45°C for 40 sec, and primer extension at 72°C for 30 sec. Final elongation was 5 min at 72°C prior to storing at 4°C. Amplification was confirmed using gel electrophoresis (as described previously) and PCR products were purified by adding 2.5μl per sample of a mixture containing 1.4μl H2O, 1μl Exonuclease I and 0.1μl Shrimp Alkaline Phosphatase (rSAP; New England Biolabs, Australia) before incubation at 37°C for 20

83 min and denaturation at 80°C for 10 min. Samples were sequenced in both the forward and reverse directions by Macrogen Inc. (Republic of Korea) using Sanger sequencing. An automated workflow was used in Geneious v.11.1 (http://www.geneious.com, (Kearse et al., 2012)) to first trim ends of the sequences (error rate 0.01%), de novo assemble forward and reverse sequences from the same individuals (at which point alignment and chromatogram quality was visually assessed for all sequences) before extracting a consensus sequence for each individual. Sequences that were not processed in the workflow due to either/both forward/reverse reads being low quality were visually inspected; if one read was of acceptable quality (65%) then this single read was used to generate a sequence. From the 117 individuals sequenced, 111 sequences were of adequate quality.

Additional COI sequences of Ae. aegypti were obtained from Genbank (810 sequences total, 111 produced in this study, Table S3.3). Sequences were aligned in Geneious using MAFFT alignment (Katoh & Standley, 2013). All sequences were trimmed to 335bp to incorporate the large number of COI sequences from Genbank, many of which were smaller than, or did not overlap fully with, the ~550bp region sequenced in this study. Sequences were checked for stop codons and a TCS haplotype network was constructed in PopArt v.1.7 (http://popart.otago.ac.nz) using 1,000 iterations.

Invasion history

I attempted to reconstruct the invasion history of Ae. aegypti in the Southeast Asian and Australasian region by performing analyses in DIYABC v. 2.1.0 (Cornuet et al., 2014) using a subset of my microsatellite and COI data. To simplify the number of invasion scenarios tested, I used more broadly defined populations based on distinct genetic and geographic groups that I considered important for testing invasion pathways in the region and randomly subsampled individuals as representative of those groups. These representative groups included: Australia, Indonesia, SE Asia (Cambodia and Thailand), the Pacific (PNG and the Solomon Islands), the United States of America (Arizona) and Malaysia. Each group had between 10-35 individuals (152 individuals in total). Scenarios also included unsampled populations to model hypothetical source populations (Africa and the Americas). Multiple preliminary runs were conducted following the guidelines of Bertorelle et al. (2010) and based on previous studies (Bennett et al., 2016; Crawford et al., 2017; Gloria‐Soria et al., 2016) to optimise the prior settings, the scenarios tested and the summary statistics utilised in the final run.

84 The final run simulated five invasion scenarios (Fig 3.2, Table S3.4) which represent the most plausible pathways based on my preliminary tests, historical records and other published findings which pair ABC with genetic data (Bennett et al., 2016; Crawford et al., 2017; Gloria‐Soria et al., 2016). Prior ranges were broadly defined based on such records, with less certain introduction timings allocated a broader prior distribution (Table S3.1). As DIYABC measures time in number of generations, I used 10 generations per year for Ae. aegypti which is reasonable in tropical regions where my study took place. The effective population size (Ne) estimates were broadly defined and allocated a uniform distribution with the range of Ne depending on the population. For instance, scenarios that attempted to simulate genetic drift (bottleneck/founder effects) in a recently introduced lineage were modelled using a lower Ne range (Table S3.4). Both di- and tri- microsatellite repeats were modelled separately to allow the possibility of different mutation rates given repeat lengths (Schug et al., 1998). A generalised stepwise mutation model was used and the mean mutation rate (1x10-6 - 1x10-3 (loguniform distribution)) was based on Dipteran microsatellites (Schug et al., 1997). For COI, I used mutation rates ranging between 7x10-10 - 1x10-7 (HKY mutation model, uniform distribution) (Brower, 1994; Papadopoulou et al., 2010). Each scenario was given a uniform probability and 6,000,000 datasets simulated. The performance of the simulations was checked using procedures outlined in detail in File S1 of Maynard et al. (2017) (see Performance of DIYABC (File S2.1)), including model checking and calculation of type I and type II error.

Fig 3.2. Final invasion scenarios (SCN1-5) tested using approximate Bayesian computation (ABC) for Aedes aegypti in the Australasia and Southeast Asian region. Populations are represented by coloured lines, where thinner lines of a lighter shade represent population bottlenecks and thicker lines represent a population expansion. Time events (t1-t7) are not to scale and db-db5 represent duration of bottlenecks. Rate of admixture is also shown in some scenarios (adm). Sampled populations are represented at time = 0. Posterior probabilities and 95% confidence intervals of each scenario are shown and scenario one (SCN1; asterisk) was the most likely scenario. Prior and posterior distributions are outlined in Table S3.4. USA, United States of America; AUS, Australia; MAL, Malaysia; INA, Indonesia; SEA, Mainland Southeast Asia, PAC, Pacific (see Table 3.1, Region); A1, unsampled ancestral population 1; A2, unsampled ancestral population 2; ‘e’ and ‘b’ extensions on populations indicate a change in effective population size for the given population (see Table S3.4 for full abbreviation details).

RESULTS

Genetic diversity and gene flow (FST, Jost’s D & G”st)

I observed the lowest heterozygosity (Ho = 0.428) in the Waiben population despite it having the highest mean number of alleles (15.667) within the Australian region and the highest unbiased

86 expected heterozygosity (0.592) (Table S3.5). Overall, SE Asian populations exhibited high allelic diversity (N, Na and Ne) while most mainland Australian populations showed comparatively lower measures. Within Australia, Cairns showed the highest mean observed heterozygosity (although standard error overlaps with Waiben), while all other QLD populations had lower, although similar, values of unbiased heterozygosity.

The loci AC4 and A9 were removed due to the high frequency of null alleles (> 0.2) and the number of populations that violated HW (Table 3.2). A total of 86 alleles were recorded across all nine remaining loci (not including AC4 and A9). I found no significant evidence for linkage disequilibria between any pairs of loci in each of the 20 populations (after Bonferroni correction (α = 0.05)). Allelic richness (Na) was highest in Kuala Lumpur, Cambodia and Honiara, whereas Port Moresby exhibited the lowest (Table S3.5). Port Moresby and Timika had the lowest observed heterozygosity, whereas Tucson, AZ had the highest of all populations in this study (Table S3.5). Overall, the mantel test showed a significant relationship between geographic distance and genetic distance (y = 0.0001x + 5.2284; P = 0.0002), but the correlation is weak (R² = 0.009), implying the influence of other factors shaping the genetic structure of Ae. aegypti in the study region.

87 Table 3.2. Microsatellite characteristics for 11 loci screened on Aedes aegypti. Mean number of alleles (Na), number of effective alleles (Ne), observed heterozygosity (Ho), unbiased expected heterozygosity (uHe), number of populations deviating from Hardy-Weinberg equilibrium (HW), null alleles, global FST (gFST) and global FST without null alleles (gFST (no null)) following ENA correction (Chapuis & Estoup, 2007) are displayed for each locus.

Locus Na Ne Ho uHe HW Null alleles g FST gFST (no null)

A9 6 3.43 0.29 0.71 4 0.24 0.30 0.25

B2 9 2.05 0.30 0.51 3 0.14 0.24 0.24

AC5 14 3.29 0.55 0.70 3 0.08 0.10 0.09

AG5 9 3.45 0.65 0.71 2 0.04 0.09 0.09

A1 6 2.83 0.55 0.65 1 0.06 0.14 0.13

B3 14 2.87 0.52 0.65 1 0.08 0.13 0.12

AG1 8 4.55 0.60 0.78 1 0.10 0.12 0.11

AC2 6 2.24 0.43 0.56 1 0.07 0.24 0.23

AC4 5 1.73 0.13 0.42 10 0.22 0.16 0.15

AC1 11 3.39 0.56 0.71 3 0.09 0.16 0.16

CT2 9 2.17 0.33 0.54 4 0.13 0.31 0.30

There was no major difference in the relationships observed using multiple measures of genetic distance (Table S3.6; FST, Jost’s D and G”ST), but I report FST as an indirect measure of gene flow and genetic distance/differentiation (Meirmans & Hedrick, 2011; Neigel, 2002) (Fig 3.3). Overall, significant FST ranged from 0.029 - 0.249 (P<0.005) for the entire dataset. Genetic differentiation was lowest within Indonesia and Australia. In pairwise comparisons between Queensland population samples, FST ranged from 0.035 - 0.161. For Waiben of the Torres Strait, Australia, lowest FST scores were observed between nearby Northern Queensland populations (FST =0.054 - 0.079; see HB, CA and CT), whereas more southerly populations were more genetically differentiated (FST: 0.094-0.161; see YE, RO, EM, MT, LR). The New Guinean populations of Port Moresby and Timika displayed high differentiation with the majority of other populations in pairwise comparisons.

Fig 3.3. Pairwise genetic distance (FST) for all populations of Aedes aegypti in the study region.

The range of FST is indicated by a colour scale, where redder values indicate a lower FST (min =

0.017) and greener indicate a higher FST (max=0.249). Insignificant (P > 0.05) comparisons are black bordered. White lines divide broader regional levels for visualisation purposes. Refer to

Table 3.1 for population abbreviations and Table S3.6 for FST values.

STRUCTURE

Aedes aegypti shows clear spatial genetic structure in the study region. Using STRUCTURE (Fig 3.4), populations were more clearly differentiated without the admixture model and using sampling locations as a prior, which assists with clustering when population structure is somewhat weak (Pritchard et al., 2009). However, here I present and discuss results using the admixture model and location prior settings as it provides a more realistic depiction of the population processes occurring within this highly anthropophilic species, which has likely had admixture between populations associated with human-mediated dispersal.

Fig 3.4. STRUCTURE plots of nine microsatellite loci for 366 samples of Aedes aegypti from Southeast Asia and Australasia. Data was analysed as a whole (A) and using a hierarchical approach (B-D, a-e) and various values of K are displayed. Each vertical bar represents an individual where the bar colour is proportional to genetic cluster membership. Region and population abbreviations are shown in Table 3.1. White-dashed lines show sampling site differences within a population (a-e). Arrows show the progressive sub-analysis of clusters used in the hierarchical approach.

For the whole dataset, ∆K suggested two genetic clusters in the study region (Fig 3.4A). At K=2, the clusters generally correspond to an Australian (red) and Indonesian/Malaysian (blue) clusters

90 with several other populations showing varying degrees of admixture between these two clusters (Torres Strait Islands, Solomon Islands, Port Moresby, Bangkok, Cambodia and Arizona). An upper K value of 10 (Fig 3.4A) was suggested using median values of Ln(Pr) in CLUMPAK; at this value of K populations can be more differentiated and specific clusters dominate different geographic regions, leading to the broad differentiation of mainland Australia, the Torres Strait Islands (Waiben), the Solomon Islands, PNG, Indonesia, Malaysia, Southeast Asia and the USA.

To investigate this substructure, a hierarchical approach was used to analyse the red, blue and admixed clusters separately when K=2 on the whole dataset. The Australian cluster (not including Waiben) is made up of approximately 3-6 genetic subgroups (Fig 3.4B). In general, at K=3 these clusters represent Northern (Cairns, Holloways Beach, Charters Towers), Central (Longreach) and Central-Eastern (Rockhampton, Mt Morgan and Yeppoon) Queensland divisions. However, there is significant admixture between many of these populations and further differentiation when K=6. Finer population substructure was also uncovered in Holloways Beach (K=2) and Charters Towers (K=2) (Fig 3.4a-b).

Within the blue cluster (Fig 3.4C), there are approximately 2-4 subgroups. When K=2, Malaysia can be distinguished from the Indonesian populations. Furthermore, when K=4, subgroups appear within Indonesia. These represent Timika, Amamapare (Mimika)/Sulawesi and Bali/Sumba. Malaysia was comprised of three genetic clusters when sub-analysed (Fig 3.4d), corresponding to differences in sampling site.

When admixed populations (<80% overall population membership to either red or blue clusters at K=2) were analysed together, the optimal K value ranged from 4-7 (Fig 3.4D). When K=4, genetic clusters correspond to broad geographic groupings: Torres Strait, PNG/Solomon Islands, Southeast Asia and USA. At K=7 some populations are further separated, but some show signs of admixture across these geographic regions hence populations appear less genetically distinct.

DAPC

DAPC of the full dataset (Fig 3.5) is somewhat consistent with results of the STRUCTURE analysis. In general, there are two main clusters of individuals representing Australia and Indonesia respectively, with other populations such as Arizona and Timika appearing more differentiated from these. To investigate consistency in results with the hierarchical approach in STRUCTURE, I analysed the same three broad sub-datasets using DAPC.

Fig 3.5. Discriminant analysis of principal components (DAPC) for Aedes aegypti (n=366) in the study region using nine microsatellite loci. Principal components 1-3 (PC1-3) are plotted showing individual variation within population (A) and population means (B). Populations are colour coded and abbreviations are in Table 3.1.

The Australian cluster (excluding the Torres Strait Islands) shows that all populations share some overlap in their 95% CI, but that Holloways Beach, Longreach and Emerald are all somewhat distinct from the rest of Queensland (Fig S3.1). Charters Towers and Cairns appear to cluster together as do more southerly populations, more in line with the Australian STRUCTURE sub- analysis results when K=3 (Fig 3.4B).

From the Indonesian/Malaysian cluster, all populations share some overlap in their 95% CI, but Malaysia and Timika are most differentiated from other Indonesian populations (Fig S3.2). If Indonesian populations are analysed separately, Sumba, Sulawesi and Timika are somewhat unique whereas Amamapare (MIM) and Bali shared the most overlap.

From the admixed dataset, which included populations from across the study region, DAPC supported the clustering shown in the STRUCTURE sub-analysis, where the Torres Strait and PNG show genetic clustering as do Cambodia and Bangkok, with Arizona and the Solomon Islands being distinct and the most differentiated from all other populations (Fig S3.3).

92 To summarise, similar patterns of genetic structure were uncovered when undertaking DAPC on the same clusters analysed in the hierarchal structure approach. Compared to STRUCTURE results, genetic structure using DAPC appears more conservative (i.e. less substructure revealed). However, when using no population information, approximately 12 genetic clusters were supported by the BIC, where several regions could be genetically distinguished and comprised of several genetic clusters (e.g. inferred clusters 1, 2, 3, 7 and 11 are mostly made up of Australian populations), whereas some clusters are comprised of a mix of multiple regions (e.g. inferred clusters 8 and 12) (Fig S3.4).

When data was analysed using DAPC with broader regional definitions rather than using population information, individuals were mostly assigned to their original cluster with a few exceptions. The proportion of correct reassignment to original population was 0.41 when populations where used compared to 0.85 when broad regions were used (Fig S3.5).

COI

Overall, the mitochondrial marker COI did not reveal strong genetic structure, but some patterns are worth noting. The thirteen haplotypes plotted in Fig 3.6 represent 83% of the 810 COI sequences analysed, but additional haplotypes (n=99 haplotypes) are plotted in a TCS network in Fig S3.6. Overall nucleotide diversity (π) was 0.0256 with 75 segregating sites in the alignment. Haplotypes 2 and 3 have a global distribution and were the most prevalent haplotypes, making up 50% of all the sequences analysed (Fig 3.6). Haplotype 4 dominates the Asian and Australasian region, while similarly, H13 and H62 were only found within Asia but were less widespread. H8 and H30 are shared between parts of Asia and Africa, while H14 spans the Asian/Australasian region and Americas (Fig 3.6). Several less frequent haplotypes (Fig S3.6) reveal further potential structure with H92 and H95 unique to Malaysia and Indonesia respectively.

Fig 3.6. Thirteen of the most prevalent COI haplotypes (335bp region) for Aedes aegypti on a global scale. Circle size corresponds to the number of sequences (see white dashed scale) from a given locality. The proportion of individuals belonging to a given haplotype are colour-coded in the key (right). Refer to Table S3.3 for specific details and location numbers.

94 Invasion history

Scenario one had the highest support (P=0.67 [95% CI = 0.66 - 0.69]), this scenario simulated multiple, independent introduction events all originating from a similar ancestral source (A2) (Fig 3.2, SCN 1). I detected low levels of type I (0.33) and type II error (0.05 – 0.20). The timing of invasions are consistent with the arrival of Ae. aegypti into the Australasian and Asian region between the late-1700s and early-1900s, however 95% confidence intervals were large and overlapping, suggesting that most invasions occurred at an indistinguishable time period based on the data (Table S3.4). The introduction timeframes (95% confidence intervals shown by year; Fig 3.2, SCN 1) for sampled populations are as follows: Arizona 1527-1810 (t6), Australia 1804-1883 (t5), Indonesia 1782-1882 (t4), Malaysia 1790-1893 (t3), mainland Southeast Asia (Cambodia and Bangkok) 1794-1908 (t2), Pacific Islands (PNG and Solomon Islands) 1832-1944 (t1). These all diverged from a single ancestor (A2) that had diverged from another population (A1; likely representing an African population) ~340-450 years ago, which had undergone a demographic expansion up until 200-250 years ago (Fig 3.2, SCN 1).

Initially, I attempted to replicate some of the ancestral scenarios uncovered by Crawford et al. (2017) and Bennett et al. (2016) in my invasion scenarios, but these fit poorly with the PCA of simulated datasets and generally had lower support values. Overall, more simplistic simulations that focused on the region better fit the observed data. Hence, my analyses are mostly regarding the sources and relationships within the study region rather than more global Ae. aegypti invasion history. Scenarios that simulated an initial change in effective population size (representative of a population bottleneck) following a divergence event also better fit the observed data. I found support for an increase in effective population size following initial introductions, but the length of the initial bottleneck varied between populations (Table S3.4). My ABC analysis fit the observed data well (Fig S3.7, Fig S3.8).

DISCUSSION

Higher than previously shown spatial genetic structure was uncovered in my study on Ae. aegypti populations of the Southeast Asia and Australasia region, when populations were analysed using a hierarchical approach and various methods. I found evidence of differentiation between multiple Queensland populations and clear genetic structure in the Waiben (Thursday Island, Torres Strait Islands) population. This may serve as useful information when considering control measures such as sterile-insect-based eradication/suppression or Wolbachia-based suppression or viral limiting strategies. Understanding genetic differences and gene flow between populations is necessary for

95 the effective roll-out of mosquito field releases in such control measures, where there is potential for connected mosquito populations to reinvade or spread to the controlled areas (which in some cases can reduce the efficiency/long-term stability of these control strategies). Likewise, it can assist in identifying release sites where the dissemination of mosquitoes is ideal or where it may require more effort (i.e. higher release numbers or more release sites across areas that are less connected). Moreover, in regard to the invasion history of Ae. aegypti, which has been understudied at a fine level, my analyses suggest that several populations in Southeast Asia and Australasia were most likely introduced via multiple, independent incursion events rather than from a single event or stepping stone-like models. The most likely invasive origins of Ae. aegypti in the region are extensively discussed to allow for future testing when larger datasets are acquired. This uncertainty in invasion routes is not surprising given the history of frequent and extensive human movements throughout this region, which would have given ample opportunities for mosquitoes to invade other regions.

In previous studies, Endersby et al. (2009) used eight nuclear markers (exon primed intron crossing markers and microsatellites) to explore the genetic structure of multiple populations across Queensland (spanning Charters Towers to Mossman) and Vietnam as well as a single population from Thailand. They uncovered that these regions and some of the populations within them are genetically differentiated, but that Vietnam harboured greater allelic variation than Australia. Likewise, they found that inland populations (Chillagoe and Charters Towers) within Northern Queensland were clearly differentiated from coastal ones, a pattern replicated in my study (see Longreach, Emerald and Charters Towers for example). Here I show at least three clear genetic divisions within Queensland that represent northern, south-eastern and south-central Queensland clusters (Fig 3.4B), with substructure revealed with further analysis. I found a significant positive correlation between geographic and genetic distance that hints at a subtle pattern of isolation by distance, supporting the results of other studies (Endersby et al., 2009). However, the correlation was weak, implying that other drivers such as human movements play an important role in shaping the genetic structure of Ae. aegypti in Queensland. The frequency of human traffic between these Queensland populations likely dictates the amount of gene flow/genetic continuity as this increases the number of potential dispersal events. Such human-mediated dispersal has been documented numerous times including via road, air and shipping networks (Fonzi et al., 2015; Gonçalves da Silva et al., 2012; Hamlyn-Harris, 1927; Huber et al., 2004; Rasheed et al., 2013). Unsurprisingly, further inland locations, such as Longreach, are more genetically isolated from easterly populations which are more connected along Queensland’s major highways. As highways are a dispersal route,

96 inland populations may experience fewer opportunities for dispersal (especially given the harsher, drier climate that may decrease the longevity of new immigrants).

This pattern is also obvious in the Waiben population from the Torres Strait Islands, which is the most disconnected from mainland Australian populations, both in terms of geographic distance and human traffic. Waiben would likely have higher gene flow with other island populations in the Torres Strait and Southern PNG regions, which appears to be the case for Ae. albopictus (Beebe et al., 2013; Maynard et al., 2017). The two regions are highly interconnected due to the implementation of the Torres Strait Treaty which permits cross-border movement (mostly via boat) for indigenous Torres Strait Islanders and coastal communities of PNG and the region is also serviced by regular small aircrafts (on which other Culicids have been observed on board (Maynard pers. obs. 2016)). The globally widespread COI haplotype H14 was prevalent on Waiben but was not detected on mainland Australia. This is probably reflective of the Torres Strait region differing in invasion and demographic history from mainland Australia, reinforcing earlier findings by Beebe et al. (2005). The Torres Strait Islands have a higher incidence of dengue compared to mainland Australian populations. Additionally, Ae. aegypti from Waiben are more competent vectors of dengue serotypes 2 and 4 compared to those from Cairns and Townsville (Knox et al., 2003) highlighting the connection between genetic structure and medically relevent traits such as vector competency. From a historical perspective, the Torres Strait Islands were part of an extensive pearling industry during the 1880s, which sought a workforce primarily from the Pacific Islands (chiefly Fiji, Vanuatu and New Caledonia), Japan, Malaya and the Philippines (Beckett, 1977). Later, during WWII, Waiben served as a military base for the United States and Australia forces. Earlier, from 1800-1850 many sailing ships made the voyage from Brisbane and Sydney to India and other parts of Asia via the Torres Strait, although few stopped. It would be interesting to compare Waiben to a more worldwide dataset as the Torres Strait region likely shows a different invasion history to mainland Australia.

My overall findings are consistent with those of more worldwide studies (Gloria‐Soria et al., 2016), where Australia is a distinct genetic cluster from Asian and Indonesian populations, however I expand on this by analysing previously unexplored populations and uncover finer scale genetic structure. When Asia-Pacific regions have been sub-analysed (using a hierarchical approach in STRUCTURE), Gloria‐Soria et al. (2016) found evidence that Asia, Australia (Cairns and Townsville) and Pacific Ocean Islands (Hawaii and Tahiti) can be differentiated. However, here I have shown further genetic breaks within Australia, Indonesia and Malaysia, in addition to

97 characterising the structure of previously unexplored regions in the context of the broader region. This has revealed some unexpected patterns.

Amongst these unexpected patterns is the genetic structure of Papuan populations. While politically part of Indonesia, Papua is geographically and culturally connected to PNG. The capital of the Mimika regency, Timika (~25km more inland), was highly distinct from Amamapare (MIM) and other parts of Indonesia. In contrast, Amamapare, a more coastal port and cargo facility population, was more genetically similar to Sulawesi and other parts of Indonesia (Bali and Sumba), This finding could reflect shipping movements from other regions of Indonesia, contributing to migration and introgression into the coastal sea port which would explain Amamapare’s genetic continuity with the rest of Indonesia. Whereas, the more inland population in Timika appears to not be influenced by this and is relatively more genetically isolated from the rest of Indonesia. Perhaps Timika shares more genetic similarity to other inland Papuan populations. The signature from the mitochondrial COI data does not support this difference and all of Indonesia shares similar haplotypes, many of which are also found across other countries, especially within Asia (Fig 3.6; see H2, H3, H4).

The Honiara (Solomon Islands) population appeared more genetically distinct from the nearby population of Port Moresby than expected and is somewhat unique within the study area. This is potentially due to mixed ancestry from Australia, the USA and parts of Asia (Cambodia and

Malaysia) which share the smallest pairwise FST measures with Honiara. Human mediated movements during World War II would have had a significant impact on the distribution of mosquitoes in the Pacific area and could have led to admixture between geographically disparate populations (Calvez et al., 2016; Failloux et al., 2002; Powell et al., 2018). One of the most comprehensive genetic studies on Ae. aegypti in the Pacific (including multiple islands from New Caledonia, Fiji, Tonga and French Polynesia, but not the Solomon Islands) found moderate genetic differentiation between islands (FST = 0.05-0.24) and that more isolated islands were more genetically distinct than major towns, which showed a higher degree of mixed ancestry (Calvez et al., 2016). Nevertheless, their microsatellite data revealed differentiation in the Pacific region, broadly corresponding with western, central and eastern genetic divisions, but that further substructure existed within these divisions. Their mtDNA markers (COI and ND4) revealed some geographic patterns of relatedness, but certain haplotypes were more widespread than others. Their study highlighted that different regions likely had multiple introduction origins, both historic and contemporary. I suspect that the Solomon Islands probably shares some similarity to these Pacific Islands (e.g. New Caledonia, Fiji, Tonga, French Polynesia) and that similar demographic processes

98 would have occurred in the Solomon Islands, but unfortunately I did not have samples from these Pacific Islands to draw this comparison.

From a greater global perspective, allozyme studies have shown genetic similarities between Indonesian, Indian and Taiwanese populations (Wallis et al. 1983), which could suggest a similar introduction source, potentially revealing India as an important, yet understudied invasion source, given its importance in early trade and the early history of dengue (Gloria‐Soria et al., 2016; Smith, 1956). Using isoenzymes, Failloux et al. (2002) explored several populations including Ae. aegypti aegypti from French Polynesia (Southern Pacific), French Guiana (South America), SE Asia (Cambodia, Vietnam) and Ae. aegypti formosus from Western Africa and several Indian Ocean Islands. They found three major groups representing 1) sylvan Ae. ae. formosus from West Africa and the Indian Ocean 2) South East Asian and South American Ae. ae. aegypti (domestic form), and 3) the same domestic form but in populations from South Pacific Islands, highlighting potential structure between South Pacific Islands and Asia/America, positioning Africa as more divergent. There was high genetic differentiation within French Polynesia, which was higher than that between Vietnam, Cambodia and French Guiana, and which may have been the result of a past major bottleneck (Failloux et al., 2002), but could equally be suggestive of varying invasive origins in the Pacific. Recently, using genome-wide SNPs Schmidt et al. (2019) showed that at K=3 Australian populations (Townsville and Cairns) cluster with Pacific Island populations (Fiji, Vanuatu, Kiribati, New Caledonia) which are distinct from Asian and Indonesian populations, similar to my findings at low values of K where Australia appears distinct from Asian/Indonesian populations.

Invasion history in Australasia and Southeast Asia

My ABC analyses showed that several populations in the Southeast Asian and Australian region were likely derived from multiple independent invasions from a similar ancestral source (A2) and over a similar timeframe between the late-1700s and early-1900s. What this ancestral source (A2) represents is unknown, but it possibly represents an ancestor from the Americas or a genetically similar region that has had a similar demographic history (the other most likely alternative source being the Mediterranean region); typically referred to as the “New World” genetic group compared to the “Old World” African ancestor. This is opposed to the Asian region being introduced from an African source. My results suggest that A2 diverged from A1 between 340 - 560 years ago, with A2 undergoing a demographic expansion 200 - 380 years ago. It is important that future studies that have a more comprehensive set of worldwide representative populations model their introductions separately to better understand the invasion history of Ae. aegypti.

99 It is difficult to be sure of exact source locations (as there could have been reintroductions from say the Americas back into Africa or the Mediterranean from which Ae. aegypti could have spread to the Asian and Australasian regions), but accumulating genetic evidence suggests that a “New World” genetic source has seeded the invasion into Southeast Asia and Australasia, and this is further supported here. Recently, exome sequencing has shown that some populations in Africa appear to have been reintroduced from the Americas, providing new perspectives on the evolutionary history of populations of Ae. aegypti (Crawford et al., 2017). Indeed, future studies using genome-wide SNPs will provide significant insights the evolutionary history of Ae. aegypti. The 16th century saw European maritime trade rise within Southeast Asia (notably for valuable spices from the Maluku (Moluccas) Islands), a region mostly under Portuguese and Spanish control, via Southern Africa (the Cape Route) and the Indian Ocean. Later in the 17th and 18th centuries, this was overrun by British and Dutch enterprises, with the Dutch monopolising the Indonesian region (under the Dutch East India Company) and possessing a widespread trade network centralised in Batavia (now Jakarta), Java. New imperialism strongly shaped trade in the 19th century, which became more globalised and began to approach today’s form. The finding that the regions modelled here were established from independent introduction events is not surprising given the frequency of human movements (especially maritime trade driven) at the time when Ae. aegypti established in the Asian part of the world. Past notions of the worldwide spread of Ae. aegypti out of Africa included introduction into Asia from Africa itself. Although Africa was a major stop on voyages from British and Dutch ships toward Asia in the early-1800s, genetic data positions Asian populations as more closely related to the Americas. STRUCTURE analyses from Gloria‐Soria et al. (2016) showed that the Middle Eastern region clusters with Asian and Australian populations, highlighting that the Middle Eastern region could have been an important stepping stone into Asia/Australasia. Overall, while there would have been ample opportunity for Ae. aegytpi to spread from Africa to Asia, the current body of genetic evidence does not support this west to east movement.

Early records from the late-1800s and early-1900s, while patchy in some parts, paint a clear picture that Ae. aegypti colonised Asia and Australasia rapidly, and that it occurred in major trading ports at a similar time (see Fig 3.1 and records in Table S3.1). While in my analyses I could not distinguish between individual invasion timings (due to overlapping confidence intervals), it is expected that larger datasets with more genetic markers may be able to elucidate this in the future. James (1913) points out early concerns that the opening of the Suez Canal (in 1869) and Panama Canal (in 1914) would have on the spread of Ae. aegypti and associated diseases, and this was reiterated recently by Powell et al. (2018). Indeed, the opening of both canals dramatically shaped trade routes, resulting

100 in more direct passages between Asia/Australia with the Mediterranean and the Americas. The timing of the opening of the Suez Canal in 1869 and emergence of the first urban outbreaks of chikungunya (Carey, 1971) and dengue (Smith, 1956) shortly after support the hypothesis that this accelerated the spread of Ae. aegypti, but whether it represents the actual route of initial introduction for the populations I modelled is speculative, but possible.

Future work and conclusion

I have used relatively few markers compared to what can be achieved with genome wide SNPs that are becoming more affordable and are more informative. Rašić et al. (2014) found that while microsatellites can differentiate Australian, Indonesian, Vietnamese and Brazilian populations of Ae. aegypti, genome-wide SNPs were far more sensitive, showing strong separation of the populations. They noted however, that pairwise FST values were typically larger than that calculated from microsatellites, but that pairwise FST values calculated using microsatellites were often comparable across studies (Rašić et al., 2014). I expect that future studies employing genome-wide SNPs across a more comprehensive study site will reveal finer scale population genetic structure and reveal more details regarding demographic histories. This high spatial structure has been shown recently by Schmidt et al. (2019) using genome-wide SNPs in Ae. aegypti from various global populations. Whether populations display seasonal differentiation in genetic structure should also be tested with such markers, however others have found stability across the wet-dry seasons in northern Queensland (Endersby et al., 2011) and Indonesia (Rašić et al., 2015), which might be the result of eggs surviving the dry season and hatching at the start of the wet season. As seasons are more pronounced at southerly latitudes it would be reasonable to predict that more southerly populations could undergo greater temporal genetic changes. Future genome-wide data sets will likely uncover clearer spatial divisions within populations of Ae. aegypti in Australia (for instance using landscape genomic approaches (Schmidt et al., 2018)) and potentially temporal structure in regions.

My study demonstrates the need to analyse populations of Ae. aegypti at a finer level to better uncover inter- and intra-continental population dynamics and to better elucidate the invasion history of this important vector species. This has direct implications for identifying invasion pathways for biosecurity (Endersby-Harshman et al., 2019; Schmidt et al., 2019) and for understanding the evolutionary processes that might influence the epidemiology of Ae. aegypti-borne diseases. Such methods could be applied to other invasive insects and would allow for more specific and informed conclusions with important public health and control management outcomes.

101 ACKNOWLEDGEMENTS

The research was supported by the CSIRO Cluster Collaboration Fund ‘Urbanism, Climate Change and Health” and the “Funding Initiatives for mosquito management in Western Australia” (FIMMWA). I thank Michael Bangs, Alicia Perkins, Charles Butafa and Rohani Ahmad for samples.

REFERENCES

Beckett, J. (1977). The Torres Strait Islanders and the pearling industry: a case of internal colonialism. Aboriginal History, 77-104. Beebe, N. W., Ambrose, L., Hill, L. A., Davis, J. B., Hapgood, G., Cooper, R. D., . . . van den Hurk, A. F. (2013). Tracing the tiger: population genetics provides valuable insights into the Aedes (Stegomyia) albopictus invasion of the Australasian region. PLoS Neglected Tropical Diseases, 7(8), e2361. doi:10.1371/journal.pntd.0002361 Beebe, N. W., Whelan, P. I., van den Hurk, A., Ritchie, S., & Cooper, R. D. (2005). Genetic diversity of the dengue vector Aedes aegypti in Australia and implications for future surveillance and mainland incursion monitoring. Communicable Diseases Intelligence Quarterly Report, 29(3), 299-304. Beebe, N. W., Whelan, P. I., van den Hurk, A. F., Ritchie, S. A., Corcoran, S., & Cooper, R. D. (2007). A polymerase chain reaction-based diagnostic to identify larvae and eggs of container mosquito species from the Australian region. Journal of Medical Entomology, 44(2), 376-380. doi:10.1603/0022-2585(2007)44[376:apcrdt]2.0.co;2 Bennett, K. L., Shija, F., Linton, Y. M., Misinzo, G., Kaddumukasa, M., Djouaka, R., . . . Hlaing, T. (2016). Historical environmental change in Africa drives divergence and admixture of Aedes aegypti mosquitoes: a precursor to successful worldwide colonization? Molecular Ecology, 25(17), 4337-4354. Bertorelle, G., Benazzo, A., & Mona, S. (2010). ABC as a flexible framework to estimate demography over space and time: some cons, many pros. Molecular Ecology, 19(13), 2609- 2625. Bonne-Wepster, J., & Brug, S. L. (1932). The subgenus Stegomyia in Netherland India. Geneeskundig Tijdschrift voor Nederlandsche-Indië, 2, 35-119. Boyce, R. W. (1911). Yellow fever and its prevention: E. P. Dutton & Co.

102 Brower, A. (1994). Rapid morphological radiation and convergence among races of the butterfly Heliconius erato inferred from patterns of mitochondrial DNA evolution. Proceedings of the National Academy of Sciences, 91(14), 6491-6495. Brown, J. E., Scholte, E.-J., Dik, M., Den Hartog, W., Beeuwkes, J., & Powell, J. R. (2011). Aedes aegypti mosquitoes imported into the Netherlands, 2010. Emerging Infectious Diseases, 17(12), 2335-2337. doi:10.3201/eid1712.110992 Buxton, P. (1927). Researches in Polynesia and Melanesia, Part III, Medical Entomology. Memoirs of the London School of Hygiene and Tropical Medicine, 1, 113-116. Calvez, E., Guillaumot, L., Millet, L., Marie, J., Bossin, H., Rama, V., . . . Mathieu-Daudé, F. (2016). Genetic diversity and phylogeny of Aedes aegypti, the main arbovirus vector in the Pacific. PLoS neglected tropical diseases, 10(1), e0004374-e0004374. Carey, D. E. (1971). Chikungunya and dengue: a case of mistaken identity? Journal of the History of Medicine and Allied Sciences, 26(3), 243-262. Chapuis, M.-P., & Estoup, A. (2006). Microsatellite null alleles and estimation of population differentiation. Molecular Biology and Evolution, 24(3), 621-631. Cooper, R. D., Whelan, P. I., van den Hurk, A., Beebe, N. W., & Ritchie, S. A. (2005). Genetic diversity of the dengue vector Aedes aegypti in Australia and implications for future surveillance and mainland incursion monitoring. Communicable Diseases Intelligence Quarterly Report, 29(3), 299. Cornuet, J.-M., Pudlo, P., Veyssier, J., Dehne-Garcia, A., Gautier, M., Leblois, R., . . . Estoup, A. (2014). DIYABC v2.0: a software to make approximate Bayesian computation inferences about population history using single nucleotide polymorphism, DNA sequence and microsatellite data. Bioinformatics, 30(8), 1187-1189. doi:10.1093/bioinformatics/btt763 Crawford, J. E., Alves, J. M., Palmer, W. J., Day, J. P., Sylla, M., Ramasamy, R., . . . Jiggins, F. M. (2017). Population genomics reveals that an anthropophilic population of Aedes aegypti mosquitoes in West Africa recently gave rise to American and Asian populations of this major disease vector. BMC Biology, 15(1), 16. Daniels, C. (1908). Notes on the mosquitoes on the river and coast districts on the eastern side of the Peninsula. Studies from the Institute of Medical Research Malaya, 3, 262-266. Earl, D., & vonHoldt, B. (2012). STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genetics Resources, 4(2), 359-361. doi:10.1007/s12686-011-9548-7 Edwards, F. (1925). Mosquito notes.—V. Bulletin of Entomological Research, 15(3), 257-270. Endersby-Harshman, N. M., Schmidt, T. L., Chung, J., van Rooyen, A., Weeks, A. R., & Hoffmann, A. A. (2019). Heterogeneous genetic invasions of three insecticide resistance

103 mutations in Indo-Pacific populations of Aedes aegypti (L.). bioRxiv, 768549. doi:10.1101/768549 Endersby, N. M., Hoffmann, A. A., White, V. L., Lowenstein, S., Ritchie, S., Johnson, P. H., . . . Weeks, A. R. (2009). Genetic structure of Aedes aegypti in Australia and Vietnam revealed by microsatellite and exon primed intron crossing markers suggests feasibility of local control options. Journal of Medical Entomology, 46(5), 1074-1083. Endersby, N. M., Hoffmann, A. A., White, V. L., Ritchie, S. A., Johnson, P. H., & Weeks, A. R. (2011). Changes in the genetic structure of Aedes aegypti (Diptera: Culicidae) populations in Queensland, Australia, across two seasons: implications for potential mosquito releases. J Med Entomol, 48(5), 999-1007. Failloux, A.-B., Vazeille, M., & Rodhain, F. (2002). Geographic genetic variation in populations of the dengue virus vector Aedes aegypti. Journal of Molecular Evolution, 55(6), 653-663. Farner, D. S., Dicke, R. J., Sweet, G., Isenhour, L., & Hsiao, T. (1946). The distribution of mosquitoes of medical importance in the Pacific area. Navmed (US Dept Navy), 983, 1-64. Ferguson, E. (1923). The distribution of Insects capable of carrying disease in eastern Australia. Paper presented at the Proceedings of the Pan-Pacific Science Congress, Australia. Fonzi, E., Higa, Y., Bertuso, A. G., Futami, K., & Minakawa, N. (2015). Human-mediated marine dispersal influences the population structure of Aedes aegypti in the Philippine Archipelago. PLoS Neglected Tropical Diseases, 9(6), e0003829. doi:10.1371/journal.pntd.0003829 Francis, R. M. (2017). POPHELPER: an R package and web app to analyse and visualize population structure. Molecular Ecology Resources, 17(1), 27-32. Gloria-Soria, A., Brown, J. E., Kramer, V., Hardstone Yoshimizu, M., & Powell, J. R. (2014). Origin of the dengue fever mosquito, Aedes aegypti, in California. PLoS Neglected Tropical Diseases, 8(7), e3029. doi:10.1371/journal.pntd.0003029 Gloria‐Soria, A., Ayala, D., Bheecarry, A., Calderon‐Arguedas, O., Chadee, D. D., Chiappero, M., . . . Kamal, H. A. (2016). Global genetic diversity of Aedes aegypti. Molecular Ecology, 25, 5377-5395. Gonçalves da Silva, A., Cunha, I. C. L., Santos, W. S., Luz, S. L. B., Ribolla, P. E. M., & Abad- Franch, F. (2012). Gene flow networks among American Aedes aegypti populations. Evolutionary Applications, 5(7), 664-676. doi:doi:10.1111/j.1752-4571.2012.00244.x Hamlyn-Harris, R. (1927). Notes on the breeding-places of two mosquitos in Queensland. Bulletin of Entomological Research, 17(4), 411-414. Hill, G. F. (1925). The distribution of Anopheline mosquitoes in the Australian region, with notes on some Culicine species. Proceedings of the Royal Society of Victoria, 37(1).

104 Howard, L. O., Dyar, H. G., & Knab, F. (1917). The mosquitoes of North and Central America and the West Indies (Vol. 4): Carnegie Institution of Washington. Huber, K., Loan, L. L., Chantha, N., & Failloux, A. B. (2004). Human transportation influences Aedes aegypti gene flow in Southeast Asia. Acta Tropica, 90(1), 23-29. doi:10.1016/j.actatropica.2003.09.012 Hulce, D., Li, X., Snyder-Leiby, T., & Liu, C. J. (2011). GeneMarker® genotyping software: tools to increase the statistical power of DNA fragment analysis. Journal of Biomolecular Techniques: JBT, 22(Suppl), S35. James, S. (1913). The protection of India from yellow fever. Indian Journal of Medical Research, 1(2). Jombart, T. (2008). adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics, 24(11), 1403-1405. Jombart, T., & Ahmed, I. (2011). adegenet 1.3-1: new tools for the analysis of genome-wide SNP data. Bioinformatics, 27(21), 3070-3071. Jombart, T., & Collins, C. (2015). A tutorial for Discriminant Analysis of Principal Components (DAPC) using adegenet 2.0. 0. Katoh, K., & Standley, D. M. (2013). MAFFT Multiple Sequence Alignment Software Version 7: Improvements in Performance and Usability. Molecular Biology and Evolution, 30(4), 772- 780. doi:10.1093/molbev/mst010 Kearse, M., Moir, R., Wilson, A., Stones-Havas, S., Cheung, M., Sturrock, S., . . . Drummond, A. (2012). Geneious Basic: An integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics, 28(12), 1647-1649. doi:10.1093/bioinformatics/bts199 Knox, T., Kay, B., Hall, R., & Ryan, P. (2003). Enhanced vector competence of Aedes aegypti (Diptera: Culicidae) from the Torres Strait compared with mainland Australia for dengue 2 and 4 viruses. Journal of Medical Entomology, 40(6), 950-956. Kopelman, N. M., Mayzel, J., Jakobsson, M., Rosenberg, N. A., & Mayrose, I. (2015). Clumpak: a program for identifying clustering modes and packaging population structure inferences across K. Molecular Ecology Resourses, 15(5), 1179-1191. doi:10.1111/1755-0998.12387 Kraemer, M. U., Sinka, M. E., Duda, K. A., Mylne, A., Shearer, F. M., Brady, O. J., . . . Carvalho, R. G. (2015a). The global compendium of Aedes aegypti and Ae. albopictus occurrence. Scientific data, 2, 150035. Kraemer, M. U., Sinka, M. E., Duda, K. A., Mylne, A. Q., Shearer, F. M., Barker, C. M., . . . Van Bortel, W. (2015b). The global distribution of the arbovirus vectors Aedes aegypti and Ae. albopictus. Elife, 4, e08347.

105 Kumm, H. W. (1931). The geographical distribution of the yellow fever vectors: a compilation of material recorded in the literature, unpublished communications and certain collections made by the author in Nigeria, West Africa. American Journal of Hygiene, 12, 1-110. Lee, D. J., Hicks, M. H., Griffiths, M., Debenham, M. L., Bryan, J. H., Russell, R. C., . . . Marks, E. N. (1980). The Culicidae of the Australasian region (Vol. 4): Australian Govt. Pub. Service. Lumley, G. F., & Taylor, F. H. (1943a). Dengue. Part I. Medical. School of Public Health and Tropical Medicine (University of Sydney), 3, 1-141. Lumley, G. F., & Taylor, F. H. (1943b). Dengue. Part II. Entomological. School of Public Health and Tropical Medicine (University of Sydney), 3, 142-171. Mackenzie, J., La Brooy, J., Hueston, L., & Cunningham, A. (1996). Dengue in Australia. Journal of Medical Microbiology, 45(3), 159-161. Maynard, A. J., Ambrose, L., Cooper, R. D., Chow, W. K., Davis, J. B., Muzari, M. O., . . . Beebe, N. W. (2017). Tiger on the prowl: Invasion history and spatio-temporal genetic structure of the Asian tiger mosquito Aedes albopictus (Skuse, 1894) in the Indo-Pacific. PLoS Neglected Tropical Diseases, 11(4), e0005546. doi:10.1371/journal.pntd.0005546 Meirmans, P. G., & Hedrick, P. W. (2011). Assessing population structure: FST and related measures. Molecular Ecology Resources, 11(1), 5-18. doi:doi:10.1111/j.1755- 0998.2010.02927.x Meirmans, P. G., & Van Tienderen, P. H. (2004). GENOTYPE and GENODIVE: two programs for the analysis of genetic diversity of asexual organisms. Molecular Ecology Notes, 4(4), 792- 794. doi:10.1111/j.1471-8286.2004.00770.x Neigel, J. E. (2002). Is Fst obsolete? Conservation Genetics, 3(2), 167-173. Papadopoulou, A., Anastasiou, I., Vogler, A. P. (2010). Revisiting the insect mitochondrial molecular clock: The mid-Aegean trench calibration, Molecular Biology and Evolution, 27(7), 1659–1672. doi.org/10.1093/molbev/msq051 Peakall, R., & Smouse, P. E. (2006). GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes, 6. doi:10.1111/j.1471- 8286.2005.01155.x Peakall, R., & Smouse, P. E. (2012). GenAlEx 6.5: genetic analysis in Excel. Population genetic software for teaching and research - an update. Bioinformatics, 28. doi:10.1093/bioinformatics/bts460 Powell, J. R., Gloria-Soria, A., & Kotsakiozi, P. (2018). Recent history of Aedes aegypti: vector genomics and epidemiology records. BioScience, 68(11), 854-860.

106 Powell, J. R., & Tabachnick, W. J. (2013). History of domestication and spread of Aedes aegypti - a review. Memórias do Instituto Oswaldo Cruz, 108(Suppl 1), 11-17. doi:10.1590/0074- 0276130395 Pritchard, J. K., Wen, X., & Falush, D. (2009). Documentation for structure software: Version 2.3. Rasheed, S. B., Boots, M., Frantz, A. C., & Bultin, R. K. (2013). Population structure of the mosquito Aedes aegypti (Stegomyia aegypti) in Pakistan. Medical and Veterinary Entomology, 27(4), 430-440. doi:doi:10.1111/mve.12001 Rašić, G., Endersby-Harshman, N., Tantowijoyo, W., Goundar, A., White, V., Yang, Q., . . . Arguni, E. (2015). Aedes aegypti has spatially structured and seasonally stable populations in Yogyakarta, Indonesia. Parasites & Vectors, 8(1), 610. Rašić, G., Filipović, I., Weeks, A. R., & Hoffmann, A. A. (2014). Genome-wide SNPs lead to strong signals of geographic structure and relatedness patterns in the major arbovirus vector, Aedes aegypti. BMC Genomics, 15(1), 275. Raymond, M., & Rousset, F. (1995). GENEPOP (Version 1.2): Population genetics software for exact tests and ecumenicism. Journal of Heredity, 86. RCoreTeam. (2018). R: A language and environment for statistical computing. In. Vienna, Austria: R Foundation for Statistical Computing. Rousset, F. (2008). GENEPOP'007: a complete re-implementation of the Genepop software for Windows and Linux. Molecular Ecology Resourses, 8. doi:10.1111/j.1471- 8286.2007.01931.x Schmidt, T. L., Filipović, I., Hoffmann, A. A., & Rašić, G. (2018). Fine-scale landscape genomics helps explain the slow spatial spread of Wolbachia through the Aedes aegypti population in Cairns, Australia. Heredity, 120(5), 386. Schmidt, T. L., Van Rooyen, A. R., Chung, J., Endersby‐Harshman, N. M., Griffin, P. C., Sly, A., . . . Weeks, A. R. (2019). Tracking genetic invasions: genome‐wide SNP s reveal the source of pyrethroid‐resistant Aedes aegypti (yellow fever mosquito) incursions at international ports. Evolutionary Applications. Schüffner, W., & Swellengrebel, N. (1914). Stegomyien van Deli. Geneeskundig Tijdschrift voor Nederlandsch-Indië, 54, 204-205. Schug, M. D., Hutter, C. M., Wetterstrand, K. A., Gaudette, M. S., Mackay, T. F., & Aquadro, C. F. (1998). The mutation rates of di-, tri- and tetranucleotide repeats in Drosophila melanogaster. Molecular Biology & Evolution, 15(12), 1751-1760. doi:10.1093/oxfordjournals.molbev.a025901 Schug, M. D., Mackay, T. F., & Aquadro, C. F. (1997). Low mutation rates of microsatellite loci in Drosophila melanogaster. Nature Genetics, 15(1), 99.

107 Skuse, F. A. (1889). Diptera of Australia. Part V - The Culicidae. Proceedings of the Linnean Society of New South Wales, 3, 1717-1764. Smith, C. E. (1956). The history of dengue in tropical Asia and its probable relationship to the mosquito Aedes aegypti. Journal of Tropical Medicine & Hygiene, 59(10), 243-251. Stanton, A. (1920). The mosquitos of far eastern ports with special reference to the prevalence of Stegomyia fasciata, F. Bulletin of Entomological Research, 10(3), 333-344. Taylor, F. (1914). Culicidae from Papua. Transactions of the Entomological Society of London, 1914(1), 185-205. Taylor, F. H. (1915a). Australia and Yellow Fever: Report of survey of the distribution of Stegomyia fasciata in the ports of Queensland. Quarantine Service Publication, 6, 57-91. Taylor, F. H. (1915b). Australia and Yellow Fever: The recorded occurrence of Stegomyia fasciata, Fabr., and S. scutellaris (Walker) in Australia and New Guinea. Quarantine Service Publication, 6, 93-95. Theobald, F. (1901). A monograph of the Culicidae of the world (Vol. 1). Theobald, F. (1911). The distribution of the yellow fever mosquito (Stegomyia fasciata, Fabricius) and general notes on its bionomics. Transactions of the First International Congress of Entomology, 2, 145-170. Theobald, F. V. (1907). A monograph of the Culicidae, or mosquitoes (Vol. 4). London: British Museum (Natural History). Trewin, B. J., Darbro, J. M., Jansen, C. C., Schellhorn, N. A., Zalucki, M. P., Hurst, T. P., & Devine, G. J. (2017). The elimination of the dengue vector, Aedes aegypti, from Brisbane, Australia: The role of surveillance, larval habitat removal and policy. PLoS Neglected Tropical Diseases, 11(8), e0005848. Wallis, G.P., Tabachnick, W.J., Powell, J.R. (1983). Macrogeographic genetic variation in a human commensal: Aedes aegypti, the yellow fever mosquito. Genetetics Research, 41, 241–258.

108 SUPPLEMENTARY FIGURES & TABLES

Fig S3.1. Discriminant analysis of principal components (DAPC) for Australian populations of Aedes aegypti using nine microsatellite loci. Principal components 1-3 (PC1-3) are plotted showing individual variation within populations where each point represents a separate individual and ellipses represent a populations’ 95% confidence interval. Populations are colour-coded and abbreviations are in Table S3.1.

109

Fig S3.2. Discriminant analysis of principal components (DAPC) for Indonesian and Malaysian populations of Aedes aegypti in the study region using nine microsatellite loci. Principal components 1-3 (PC1-3) are plotted showing individual variation within populations where each point represents a separate individual and ellipses represent a populations’ 95% confidence interval. Populations are colour-coded and abbreviations are in Table S3.1.

110

Fig S3.3. Discriminant analysis of principal components (DAPC) for admixed populations of Aedes aegypti when K=2 in STRUCTURE analysis. Principal components 1-3 (PC1-3) are plotted showing individual variation within populations where each point represents a separate individual and ellipses represent a populations’ 95% confidence interval. Populations are colour- coded and abbreviations are in Table S3.1.

111

Fig S3.4. Genetic clusters uncovered using the Bayesian information criterion to infer populations, given no prior population information. Columns show the original populations (see Table S3.1 for population abbreviations) while rows indicate inferred populations (1-12). The scale bar shows the number of individuals assigned to each of the inferred populations.

112

Fig S5. Percentage of reassignment of individuals to original region/population based on discriminant analysis of principal components. Dark grey bars represent a DAPC using populations while light grey bars represents a DAPC using regional definitions (see Table S3.1 for abbreviations). The dashed grey line represents the 80% mark which we consider indicative of strong region/population reassignment.

113

Fig S3.6. Worldwide haplotype network for all COI haplotypes (n=99, 335bp region) for 810 samples of Aedes aegypti. Circle size corresponds to the number of sequences see dashed circle scale (left)) per haplotype. The proportion of individuals belonging to a given haplotype are color- coded. Refer to Table S3.3 for specific details.

114

Fig S3.7. Principal components analysis (PCA) in the space of summary statistics in DIYABC across all invasion scenarios. The large yellow dot represents our observed dataset while colored dots show the simulated datasets for the five invasion scenarios (10,000 random prior plots displayed per scenario). The first three principal components (PC) are shown with their % variance explained by each PC shown in brackets.

115

Fig S3.8. Principal components analysis (PCA) in the space of summary statistics for the most likely invasion scenario (Scenario 1) in our study. The yellow dot represents the observed Ae. aegypti dataset, solid green dots represent the simulated dataset with parameters drawn from posterior distributions (1,000 random datasets shown), while hollow green dots correspond to the datasets simulated based on prior distributions of parameters (1,000 random datasets shown). The first three principal components (PC) are plotted with the % variance explained by each PC in brackets.

116 Table S3.1. Records used for plotting the early occurrence of Aedes aegypti (Fig 3.1). Records lacking a specific year or recorded near 1940 were plotted as indicated in brackets. Reference details are listed below the table.

Refer to supplementary attachment

Table S3.2. Detailed sample information for Aedes aegypti used in the present study (n=366).

Refer to supplementary attachment

Table S3.3. Matrix of worldwide mtDNA COI haplotypes for Aedes aegypti (n=810) representing 99 haplotypes. Sequences were 335bp long. Accession numbers used from other studies are also provided.

Refer to supplementary attachment

Table S3.4. Details used for approximate Bayesian computation (ABC) analysis for investigating the invasion history of Aedes aegypti in Southeast Asia and Australasia. Various parameter settings and conditions are show, as are the posterior distributions of parameters for our most likely scenario (Scenario 1). Type I and II error is also displayed.

Refer to supplementary attachment

117 Table S3.5. Genetic diversity using nine microsatellite loci within populations of Aedes aegypti. Mean values per population and the standard error ([SE]) are displayed: mean population size (N), mean number of alleles (Na), number of effective alleles (Ne), observed heterozygosity (Ho), expected heterozygosity (He), unbiased expected heterozygosity (uHe) and fixation index (F).

Population N Na Ne Ho He uHe F Waiben Mean 15.667 3.667 2.432 0.428 0.572 0.592 0.244 SE [1.027] [0.471] [0.19] [0.053] [0.027] [0.028] [0.098] Holloways Beach Mean 16.778 3.667 2.301 0.437 0.515 0.531 0.149 SE [0.147] [0.373] [0.324] [0.08] [0.049] [0.051] [0.118] Cairns Mean 24.111 4 2.815 0.54 0.619 0.632 0.126 SE [0.824] [0.471] [0.268] [0.035] [0.036] [0.037] [0.031] Charters Towers Mean 21.667 3.556 2.24 0.504 0.511 0.524 0.019 SE [0.236] [0.444] [0.23] [0.056] [0.054] [0.055] [0.044] Longreach Mean 23.556 3.111 2.202 0.501 0.5 0.511 0.022 SE [0.973] [0.309] [0.248] [0.072] [0.053] [0.054] [0.085] Rockhampton Mean 22.667 3.667 2.273 0.5 0.531 0.543 0.063 SE [0.707] [0.373] [0.212] [0.052] [0.04] [0.041] [0.069] Emerald Mean 25.333 3.333 2.305 0.484 0.513 0.524 0.047 SE [0.441] [0.408] [0.275] [0.059] [0.058] [0.059] [0.053] Mt Morgan Mean 19.889 3.222 2.228 0.479 0.502 0.515 0.044 SE [0.484] [0.278] [0.256] [0.076] [0.056] [0.057] [0.101] Yeppoon Mean 11.333 2.889 1.912 0.427 0.439 0.459 0.024 SE [0.333] [0.309] [0.187] [0.063] [0.049] [0.051] [0.093] Honiara Mean 17.444 4.556 2.521 0.494 0.574 0.591 0.148 SE [0.242] [0.475] [0.237] [0.051] [0.04] [0.042] [0.055] Port Moresby Mean 7.889 2.222 1.931 0.339 0.373 0.398 0.014 SE [0.111] [0.324] [0.296] [0.081] [0.093] [0.1] [0.104] Timika Mean 16 3.111 2.074 0.41 0.445 0.459 0.059 SE [0] [0.539] [0.27] [0.07] [0.076] [0.079] [0.048] Mimika Mean 10 3.556 2.649 0.567 0.582 0.612 0.072 SE [0] [0.294] [0.296] [0.096] [0.05] [0.053] [0.12] Sulawesi Mean 13.889 4.333 2.951 0.586 0.608 0.631 0.027 SE [0.111] [0.577] [0.38] [0.055] [0.055] [0.057] [0.045] Sumba Mean 8 3.111 2.243 0.472 0.503 0.537 0.078 SE [0] [0.351] [0.224] [0.083] [0.069] [0.073] [0.092] Bali Mean 26.333 4 2.372 0.542 0.52 0.53 -0.038 SE [0.333] [0.373] [0.328] [0.064] [0.058] [0.059] [0.035] Kuala Lumpur Mean 34.778 5 2.252 0.489 0.488 0.496 0.004 SE [0.147] [0.408] [0.297] [0.081] [0.068] [0.069] [0.062] Bangkok Mean 10.778 4 2.693 0.572 0.568 0.596 0.008 SE [0.147] [0.577] [0.347] [0.079] [0.061] [0.064] [0.068] Cambodia Mean 12.778 4.889 2.705 0.544 0.593 0.617 0.116 SE [0.222] [0.564] [0.266] [0.074] [0.05] [0.052] [0.078] Tucson, AZ Mean 11.889 3.444 2.637 0.635 0.607 0.634 -0.031 SE [0.111] [0.242] [0.176] [0.071] [0.026] [0.027] [0.092]

118 Table S3.6. Matrices of pairwise FST, G"ST and Jost's D values for populations of Aedes aegypti using nine microsatellite loci. Values are color-coded where redder values indicate a lower score and greener values represent higher scores. P-values are indicated above the diagonal (9,999 permutations).

Refer to supplementary attachment

119 CHAPTER IV – Genome-wide SNPs reveal the demographic histories of native and invasive populations of Aedes albopictus (Skuse, 1894) in Australasia and

Southeast Asia

ABSTRACT

The Asian tiger mosquito Aedes albopictus is a hugely successful invasive mosquito that is now established in every continent except Antarctica. Its global spread is recent, having mostly occurred over the last 50 years. Until now, few studies have explored the population genetics of this species using genome-wide single nucleotide polymorphisms (SNPs), or attempted to characterise demographic histories of both native and invasive populations of Ae. albopictus. Here I use whole genomes of 158 individuals of Ae. albopictus from across Australasia and Southeast Asia to explore this. I found clear genetic structure across both the native and invasive ranges of Ae. albopictus, with populations appearing more distinct and with less admixture between populations than previously uncovered using microsatellite markers. All populations showed a strong pattern of post- glacial isolation as sea levels rose rapidly proceeding the last glacial maximum (18-21 kya), followed by a rapid population expansion between 700-2,000 years ago, which could be due to anthropogenic influences such as the uptake of farming and domestication of animals. Over the last 100 years, populations showed varying changes in effective population size, which tend to be characterised by recent bottlenecks followed by gradual demographic expansion in invasive populations. Whole mitochondrial genome analysis including samples from the native range revealed that Ae. albopictus belongs to two predominant clades, with invasive populations belonging to either of these. Overall, these results highlight the promise of genomic approaches for exploring past population histories and population dynamics of Ae. albopictus in an understudied region of its native and invasive range.

120 INTRODUCTION

The Asian tiger mosquito Aedes albopictus (Skuse, 1894) is an emerging and serious health threat worldwide as a major vector for dengue, chikungunya and Zika. It has recently spread around the world (chiefly within the last 50 years), including into the Torres Strait Islands of Australia, a span of 274 islands between Australia’s Northern tip (Cape York Peninsula) and Southern Papua New Guinea (PNG). Following its appearance in the Torres Strait Islands in 2004 (Ritchie et al. 2006), the tiger mosquito was discovered to be widespread in 2005 and was already present on 10 of the 17 human-inhabited islands in the Torres Strait. Previous research has shown temporal and spatial genetic structure in populations from the Torres Strait, which could have been the result of genetic drift or selection (Maynard et al., 2017). Few studies have used genome-wide single nucleotide polymorphisms (SNPs) to explore the population genetics and demographic histories of Ae. albopictus and none have attempted this for the native and invasive populations in Australasia. In this study, I explored several populations across Australasia and Southeast Asia and characterise population genetics at a fine scale. I infer the demographic histories of populations to shed light on their invasion histories and biogeography. I also conducted preliminary investigations into the relative roles of selection, gene flow and genetic drift and how these evolutionary forces may be driving patterns of genetic differentiation following the colonisation of islands. With no effective vaccinations for the most significant arboviruses transmitted by Ae. albopictus, we rely on understanding the mosquito vectors that transmit the diseases to effectively control them. Using genomics, we can better understand mosquito evolution and adaptation, which can have practical applications for vector control efforts and guide future research into these areas.

Aedes albopictus is native to the tropical and subtropical regions of Asia, where the species prefers well-vegetated forest habitats, but often inhabits urban and peri-urban environments (Bonizzoni et al., 2013; Hawley, 1988). Tyres, machinery, bulk steel and lucky bamboo act as primary trade commodities where the hitchhikers lay their eggs and are easily spread as eggs, larvae or pupae - with the eggs readily hatching when flooded by rainfall. Yachts, boats and aircraft have also been implicated as a major mode for the species’ spread between populations, with motor vehicles also facilitating invasion via road networks in some areas (Bennett et al., 2019; Eritja et al., 2017; Medley et al., 2015; Miller & Loaiza, 2015; Roche et al., 2015).

In Southern PNG’s coastal Fly Region, Ae. albopictus has been established since about 1992 (Cooper et al., 1994), with records highlighting several apparently unsuccessful invasions in the surrounding region (Moa, Torres Strait (Lee et al., 1980), Buzi, Fly Region (Kay et al., 1990;

121 Cooper et al., 1994)). In the early 2000s Ae. albopictus was found in several villages of the Fly region and subsequently in the Torres Strait Islands in 2004. It was hypothesised that drought- proofing (increasing water storage infrastructure in response to the 1997-98 El Nino) in southern PNG facilitated the establishment and expansion of Ae. albopictus into nearby villages of the Fly Region and islands of the Torres Strait, trafficked by local boating (Beebe et al., 2013). A strategy for eradicating the species from the Torres Strait Islands was rolled out in the wet season of 2005- 2006, but later shifted to a cordon sanitaire approach in 2008 (van den Hurk et al., 2016) as it was realised that the multiple shared mtDNA COI haplotypes on the islands suggested different females were moving between islands (Beebe et al., 2013). This was to prevent the species from establishing on the Torres Strait’s inner islands (closer to Cape York Peninsula) Ngurupai and Waiben as well as the Northern Cape York Peninsula area of mainland Australia (Muzari et al., 2019), which serve as major gateways between mainland Australia and outer islands of the Torres Strait (more remote islands, further from Cape York Peninsula). By late 2010, Ae. albopictus was detected on several more Torres Strait Islands, including Waiben and Ngurupai. Recent widespread surveys in 2016 suggested that Ae. albopictus was present on all inhabited islands except Waiben, Boigu and Saibai (Muzari et al., 2017).

Genetic analysis of populations in this region has highlighted that the populations in the Torres Strait and coastal Fly region are highly similar and likely derived from a source/s that closely resembles the genetic makeup of some Indonesian populations (Beebe et al., 2013; Maynard et al., 2017). More historically established populations from Daru Island, Southern PNG and other parts of PNG appeared genetically distinct from this invasion, suggesting that these are of a different origin and were unable to establish widely in the Fly Region/Torres Strait (Maynard et al., 2017). Aedes albopictus from the recently invaded Torres Strait Islands exhibited high spatial and temporal structure between some populations (Maynard et al., 2017); however, the evolutionary drivers behind these population patterns are largely speculative.

Native populations across Asia have been shown to have high spatial population genetic structure based on neutral loci (Battaglia et al., 2016; Kotsakiozi et al., 2017; Maynard et al., 2017; Sherpa et al., 2019a; Sherpa et al., 2019b). Such studies have been useful for inferring the sources of invasive populations and for clarifying current and past population structure. However, relatively few studies have explored the population genetics of native and invasive populations of Ae. albopictus using whole-genome approaches, with no such studies exploring the populations scattered across Australasia and Southeast Asia (except Malaysia and Singapore). One of the main advantages of a whole-genome approach is the incredibly high number of single nucleotide polymorphisms (SNPs)

122 available, which provides an extensive amount of genetic information that can allow us to better characterise population structure at a fine scale and elucidate the evolutionary processes (such as selection, drift and gene flow) that shape population genetic patterns across the genome. Using this approach, Sherpa et al. (2019a) showed that preadaptation to cold climates likely precedes successful invasion in temperate areas for Ae. albopictus, specifically showing that several genes associated with cold adaptation were under selection in temperate regions of the species’ native range (comparing several populations from Malaysia, China and Japan). Genome-wide data also offers the opportunity to assess the relative roles of selection, drift and gene flow in shaping patterns of genetic differentiation during the invasion of new island territories by disease vectors.

In the present study, I used whole-genome sequencing to more comprehensively characterise population diversity, structure and biogeography of Ae. albopictus across Australasia and Southeast Asia, while also conducting preliminary analyses to uncover the potential evolutionary drivers of population differentiation. The native and invasive populations included in my study represent several interesting invasion scenarios where: 1) the Torres Strait region likely derived from an Indonesian genetic background and representing a more recent invasion, and 2) the Amamapare, Papua (Indonesia) and Solomon Island populations represent older invasions (>30 years ago) that were potentially derived from Southeast Asian/Malaysian or New Guinean sources. Additionally, by assembling and analysing the mitochondrial genome of individuals I was able to compare my samples to a wider database to provide broader context for the invasive origins of these populations.

MATERIALS AND METHODS

Samples and sequencing Adult and larval samples of Ae. albopictus were used in a previous study (Maynard et al., 2017, Chapter II) and were collected from field sites (Fig 4.1) using various methods and preserved in either EtOH (70-100%) or desiccated over silica beads (Table 4.1, Table S4.1). DNA was extracted using the protocol outlined in Beebe et al. (2005). All samples (n=170) were sequenced on an Illumina HiSeq X Ten, with 150bp paired-end reads. Adapters were removed and sequence ends were quality-trimmed to q10 using BBduk (BBtools package version 38.12) with a kmer of 8.

123

Fig 4.1. Map showing population locations and both principal components analysis (PCA) and ADMIXTURE plots for 114,048 nuclear SNPs for samples of Aedes albopictus from Australasia and Southeast Asia. Dots on the map and inset of the Torres Strait Islands (bottom right) show population collection sites, which are colour-coded according to the key (top left). A) PCA for all populations, where each dot represents a single individual, which is colour-coded based on population. The first two principal components (PC) are plotted with the % of variance explained by each PC shown in brackets. B) ADMIXTURE plots for K=2 and K=3. Each bar represents an individual mosquito, where the proportion coloured indicates cluster assignment. Populations are divided by white vertical lines and labelled below the plot.

124 Table 4.1. Sample information for Aedes albopictus collected from Australasia and Southeast Asia

(ntotal = 170). The suspected endemicity, year of collection and sample size (n) of each population is indicated. Full details are in Table S4.1.

Region Population Range Year n Keriri ‘12 Invasive 2012 12 Keriri ‘14 Invasive 2014 12 Torres Strait Islands, Australia Poruma ‘15 Invasive 2015 12 Iama ‘15 Invasive 2015 12 Warraber ‘15 Invasive 2015 12 Papua, Indonesia Amamapare Invasive 2015 20 Jakarta Endemic 2012 24 Indonesia Sumba Invasive? 2013 24 Malaysia Ipoh Endemic 2013 24 Guadalcanal, Solomon Islands Honiara Invasive 2013 18

Nuclear genome

Reads were mapped to the Ae. albopictus genome Foshan assembly (AaloF1, approximately 1.92 Gbp (Chen et al., 2015)) using BWA mem v0.7 (Li & Durbin, 2009). I used the recommended GATK best practices to call variants (using HaplotypeCaller in gatk v4.0.10.1 (Poplin et al., 2018; Van der Auwera et al., 2013)). Jobs were parallelised by dividing the reference genome into 30 smaller windows composed of complete contigs. Individual g.vcf files were produced for each window, which were later combined and genotyped using GenotypeGVFs (gatk v4.0.10.1). Three individuals (AJM898 (Papua-71, ~2x coverage), AJM242 (Jakarta-81) and AJM90 (Honiara-58) (both ~6x coverage)) were removed due to poor mapping to the reference genome after quality filtering and were excluded from subsequent analyses. Of the 170 individuals initially sequenced, 158 remained that were all high quality with average coverage of 17.5x (ranging from 9-30x).

Single nucleotide polymorphisms were filtered using a quality filter of Q30 in vcftools (minimum depth = 5, maximum depth = 90). Markers that were missing >50% data and any locus that had a minor allele count under three (e.g. one heterozygote and one homozygote) were removed. Applying hard filters for minor allele frequency can cause weak structure to be missed (Linck &

125 Battey, 2019), but most singletons are errors, and also confound analyses. I used the minor allele count (=3) instead, because it is the most conservative way of removing singletons, but not rare alleles (Linck & Battey, 2019). The reference genome was repeat-masked using RepeatMasker v4.0.8 (Smit et al., 2015) using the Dfam and RepBase databases and repetitive regions were converted to a bed file and used to remove SNPs from repetitive regions of the genome with vcftools v0.1.13 (Danecek et al., 2011). Another bed file was created containing all contigs <5kb and SNPs in these short contigs were also removed. I then removed any SNP with higher than 90x coverage, to further remove any repetitive parts of the genome. Indels and non-biallelic SNPs were also removed with vcftools. The dataset was then filtered to remove any SNP with missing data, this dataset is the ‘full SNP dataset’. To make the dataset more manageable for analysing genetic structure and remove any linked markers I then created two ‘thinned’ datasets; the ‘5k-thinned’ dataset contains one SNP per 5,000bp (114,048 SNPs) and the ‘100k-thinned’ dataset contains one SNP per 100,000bp (12,261 SNPs).

To assess overall and between population genetic structure I firstly calculated overall FST and FIS

(Weir & Cockerham, 1984) on the full dataset and secondly calculated pairwise FST’s (using 1,000 bootstraps) for all populations using hierfstat v0.4.22 (Goudet, 2005) in R v.3.5.3 (RCoreTeam, 2018) on the 5k-thinned dataset. To test for isolation by distance across my study region I conducted a Mantel test (Mantel, 1967) using 9,999 replicates on the 5k-thinned dataset using adegenet v2.1.1 (Jombart, 2008; Jombart & Ahmed, 2011) in R.

To explore population structure without using population information, I conducted a principal components analysis (PCA) using the 5k-thinned dataset in adegenet, where missing values were replaced with overall means for a given SNP. I also conducted a discriminant analysis of principal components (DAPC) (Jombart & Collins, 2015) using adegenet, which utilises between-population and within-population variance to maximise discrimination between groups. The optimal number of principal components to retain was determined using cross-validation with 100 replicates, a 90% training dataset and 10% validation dataset. Lastly, to better visualise this multi-dimensional data without assuming population designations, I performed t-distributed stochastic neighbour embedding (t-SNE) in the package Rtsne v.015 (Krijthe, 2015) using two dimensions.

ADMIXTURE was run from K1-10 to investigate admixture proportions (based on maximum likelihood estimation) from the 5k-thinned SNP dataset. Additionally, I ran the Bayesian program STRUCTURE v2.3.4 (Pritchard et al., 2000) on the 100k-thinned dataset employing a hierarchical approach to avoid underestimating genetic structure (Janes et al., 2017). I used the admixture model

126 and population priors. Initially, all samples were analysed from K1-10 using 5 iterations per value of K and with a runtime of 1,000,000 iterations and a burn-in of 50,000. The two clusters uncovered from this full analysis were then sub-analysed in two independent runs using the same settings, but from K1-6. Optimal K was assessed in each analysis by considering ΔK and log-likelihood probabilities LnPr (X/K) using CLUMPAK (Kopelman et al., 2015).

Microsatellite data from the same populations (including some of the same samples) were included from Chapter II (Maynard et al. 2017) to compare to the results using genome-wide SNPs using the same methods to generate STRUCTURE plots, DAPC, t-SNE and pairwise FST.

Sliding-window FST and population histories

To investigate any evidence of selection across the genome, I calculated average FST across 100kb sliding windows every 10kb (windows were made using bedtools for each contig, and pairwise

FST’s calculated using the Weir and Cockerham method (Weir & Cockerham, 1984) in vcftools) for various pairwise population comparisons, excluding contigs and windows <100kb.

Demographic analysis was then conducted to estimate historical effective population sizes (Ne) using SMC++ v1.15.2 (Terhorst et al., 2017), a program designed for analysing large genomic datasets of unphased genomes (Beichman et al., 2018). Unlike other methods such as multiple sequentially Markovian coalescent (MSMC) (Schiffels & Durbin, 2014), SMC++ more accurately estimates recent past demographics. Effective population sizes were estimated using a mutation rate of 3x10-9 per base pair per generation, which falls within the range used for other species of Aedes (Crawford et al., 2017; Sherpa et al., 2019a), Anopheles (The Anopheles gambiae 1000 Genomes Consortium, 2017) and other insects/organisms (Brumfield et al., 2003; Keightley et al., 2014a; Keightley et al., 2014b; Schrider et al., 2013). For the calculation of the time scale I assumed a generation time of 10 per year, which is a conservative estimate for Ae. albopictus in the tropics (which generally ranges 5-17 per year depending on various environmental conditions) and given past climate variability across the study area (Franklin & Whelan, 2009).

Mitochondrial genome

I mapped my sequence data to the mitochondrial genome of Ae. albopictus (Genbank accession KR068634.1) using a pipeline that used BWA for two rounds of mapping (the second one back to the consensus). I called SNPs using GATK in ploidy one mode, filtered to Q30 using vcftools, and

127 made a consensus sequence using bcftools. Regions with low mapping coverage (3x) were masked using bcftools. All mitochondrial genome sequences available on Genbank for Ae. albopictus were downloaded and aligned with the newly generated mitochondrial genomes using MAFFT in Geneious R11 (www.geneious.com). To maximise the number of sequences I could use in the analysis I deleted the control region from the alignment. Additionally, I removed AJM898 (Papua-71) from the analyses as this individual was ~10% divergent from other sequences. I then had an alignment composed of 182 individuals over a 14,837bp region of the mitochondrial genome, using Ae. aegypti (EU352212.1) as an outgroup. For tree building, I calculated the suitability of various evolutionary models with jmodeltest2 (Darriba et al., 2012; Guindon & Gascuel, 2003). The GTR+I+G model had the highest support and was used in subsequent phylogenetic analysis. This model was implemented in MrBayes v2.2.4 (Huelsenbeck & Ronquist, 2001; Ronquist & Huelsenbeck, 2003) (implemented in Geneious) to produce a Bayesian tree with 1,100,000 iterations and 100,000 iterations to burn-in. To check for consistency in relationships across trees, I constructed a maximum-likelihood tree using the same model in RAxML v.8 (Kozlov et al., 2019) with a rapid bootstrap analysis using 1,000 maximum-likelihood bootstrap replicates (both with and without AJM898 (Papua-71). To incorporate a broader geographic spread of samples in my haplotype network, I trimmed the alignment and included shorter mitochondrial sequences from Brazil and China, resulting in a 13,792bp alignment comprised of 180 individuals (with substantially divergent sequences and outgroups removed (Taiwan-NC-006817, AJM898 (Papua- 71), EU352212.1). A TCS haplotype network (Clement et al., 2002) was made in PopArt (Leigh & Bryant, 2015) using 1,000 iterations.

RESULTS

Mitochondrial genome

The mitochondrial genome analyses revealed spatial structure in the distribution of haplotypes, characterised by two major haplogroups (Fig 4.2) that correspond to two strongly-supported clades in the Bayesian and maximum-parsimony trees (Fig S4.1, Fig S4.2). Overall nucleotide diversity of the trimmed region was 0.348, with 71 haplotypes based on 168 segregating sites. Shared haplotypes were found between invasive and native populations and between some geographically proximal populations (such as between Torres Strait populations). Restricting the analysis to the cytochrome oxidase subunit 1 (COI) region revealed similar patterns, albeit with fewer unique haplotypes.

128

Fig 4.2. Haplotype network using a trimmed region of the mitochondrial genomes (13,792bp) of worldwide samples of Aedes albopictus. Haplotypes are coloured by geographic location and the size of circles shows the number of individuals belonging to each haplotype (top right key). Lines connecting haplotypes show genetic distance between haplotypes, where each dash represents a single nucleotide substitution. Small black circles represent unsampled haplotypes.

129 Phylogenetic analysis in MrBayes (Fig S4.1) and RAxML (expanded tree not displayed) highlighted two major clades for Ae. albopictus, with populations in the native range of Ae. albopictus splitting into mainland Asian clade (Clade A) or Indonesian-Philippines (Clade B) clades. Invasive populations fall into both clades, highlighting that both Amamapare and Honiara populations are closely related to Malaysian or Southeast Asian populations, while the Torres Strait is most closely related to Indonesian populations rather than the Philippines or mainland Asian populations. Overall, my analyses of the mitochondrial genomes reveal a high diversity of haplotypes in the Torres Strait Islands compared to other invasive populations, which may be a result of multiple invasions, possibly with rarer haplotypes (e.g. Fig S4.1: Clade B, Warraber-52) from separate invasions. Surprisingly, one individual which morphologically matches Ae. albopictus from Amamapare (Papua-71/AJM898) was found to be approximately 10% divergent across the mitochondrial genome (and mapped poorly to the reference nuclear genome; Fig S4.2). An individual from Taiwan is also distinct from the two predominant Ae. albopictus clades but only approximately 3% divergent across the mitochondrial genome (Fig S4.1). It is possible that both of these individuals represent cryptic species within the Ae. albopictus subgroup, but further sampling and more intensive morphological and molecular investigation would be needed to clarify this. There is also a possibility the more divergent sample from my study could be due to contamination by another culicid mosquito but until further data becomes available for comparison, this is uncertain. This highlights the importance of taxonomic investigations in Ae. albopictus (Guo et al., 2018) and for retaining voucher specimens when feasible in genetic studies.

Nuclear genome

Pairwise FST comparisons between populations using the 5k-thinned nuclear genome dataset reflect strong genetic structure consistent with the mitochondrial genome (Table 4.2). In general, the FST’s support the same splitting of the two major haplogroups revealed in the mitochondrial genome analysis (Fig 4.2), with an overall FST of 0.107 further supporting genetic structure in the dataset.

Mean FST values ranged from 0.0085 - 0.1804. The highest FST scores were observed between the

Sumba and Amamapare populations (FST = 0.1804). Overall, Honiara and Amamapare were the most differentiated from other populations within my study, however, both showed highest similarity to each other (FST = 0.102) and to Ipoh, Malaysia. The lowest FST’s were between temporal samples within Keriri (see Keriri 2012 vs 2014) and between island populations within the

Torres Strait region. The Torres Strait Islands were most similar to Jakarta and Sumba. FST calculated using microsatellites revealed a similar relationship between populations, but populations were more distinct using genome-wide SNPs (Table S4.2).

130 Table 4.2. Pairwise FST values between populations in my study using 114,048 nuclear SNPs.

Values are colour-coded where lower FST scores are redder while greener cells indicate higher values. Population abbreviations use the first characters of population names in Table 4.1 with the year indicated for the Torres Strait Islands.

Ker '12

Ker '14 0.009

Por '15 0.051 0.054

Iam '15 0.061 0.065 0.050

War '15 0.054 0.061 0.061 0.085

AMA 0.146 0.144 0.151 0.160 0.155

JAK 0.065 0.070 0.080 0.089 0.085 0.150

SUM 0.060 0.070 0.081 0.091 0.088 0.180 0.066

IPO 0.117 0.115 0.121 0.130 0.126 0.077 0.121 0.149

HON 0.138 0.136 0.142 0.151 0.147 0.102 0.144 0.172 0.052

Ker Ker Por Iam War AMA JAK SUM IPO HON '12 '14 '15 '15 '15

These patterns are further highlighted using my whole-genome approach. Two genetic clusters were supported in the whole dataset analysis in ADMIXTURE (Fig 4.1) and STRUCTURE (Fig S4.3B). At a broad level, the Indonesian populations of Jakarta and Sumba cluster with the Torres Strait Island populations, while Ipoh, Honiara and Amamapare form a separated cluster (Fig 4.1B, K=2). When K=3, membership was consistent with the PCA (Fig 4.1), where the Torres Strait Islands were distinct from Jakarta and Sumba. Additionally, I uncovered further substructure using a hierarchical approach in STRUCTURE (Fig S4.4) where the two main clusters uncovered at K=2 were split and subsequently analysed from K1-6. CLUMPAK supported K=2 for the Indonesia/Torres Strait Island cluster, which splits these two regions into distinct genetic groups (Fig S4.4A). Whereas the Ipoh, Honiara and Amamapare cluster was recovered as three distinctive

131 populations at K=3 (Fig S4.4B). The same hierarchical approach using microsatellite data revealed different results, where the outer islands of the Torres Strait (Poruma, Warraber and Iama) were distinct from Keriri, Jakarta and Sumba (Fig S4.4A). Furthermore, Honiara and Ipoh clustered whereas Amamapare was distinct (Fig S4.4B).

In summary, cluster membership differed between both microsatellite and whole-genomic approaches, with more obvious clustering and less admixture revealed using a larger number of SNPs (Fig S4.3). Additionally, the clustering relationships between the populations varied somewhat, highlighting that the outer islands of the Torres Strait Islands were not as strongly differentiated from the Keriri samples or Jakarta and Sumba when using microsatellites (Fig S4.3).

DAPC (Fig S4.5), PCA (Fig 4.1) and t-SNE (Fig S4.6) revealed similar population patterns, with only minor variations between analyses. This mirrors the results uncovered using a hierarchical approach in STRUCTURE where, at a broad level, two major genetic clusters are apparent (Torres Strait Islands/Jakarta/Sumba and Ipoh/Honiara/Amamapare) but using multi-dimensional or hierarchical approaches these major genetic groups are further split into independent population clusters. Populations were less genetically structured using microsatellite data than genome-wide SNPs, and the relationships between populations and clusters varied (Fig S4.5; Fig S4.6); most notably, the outer islands of the Torres Strait were not an independent genetic cluster when using the genome-wide datasets. I observed no major temporal differences between the 2012 and 2014 collections from Keriri using both microsatellites and the genome-wide SNPs datasets.

I found significant isolation by distance in the 5k-thinned dataset between geographic and genetic distance (Fig S4.7A), but the correlation was weak (R2=0.12, y=194+0.21x, P=0.0001). Amongst the Torres Strait Islands, the mantel test revealed a weaker correlation but a stronger relationship between geographic and genetic distance (R2=0.05, y=189.18+3.65x, P=0.0001) (Fig S4.7B).

Sliding-window FST and population histories

The sliding-window FST showed variability between some population comparisons and across various genomic regions (as evidenced by various peaks over the genome), highlighting that some regions of the genome may be under selection and that this may be population specific (Fig S4.8); however, this study is primarily concerned with the population genetic structure and biogeography, and selection will be investigated more comprehensively in future work.

132 The populations of Ae. albopictus analysed using SMC++ showed similar variations in Ne between approximately 1 - 800kya (Fig 4.3B). Specifically, all populations appear to have undergone a period of stability in Ne (30-800kya; Fig 4.3B: Ta), followed by a bottleneck (decrease in Ne) approximately 1-30kya (Fig 4.3B: Tb) followed by a sudden expansion event (increase in Ne) at approximately 1,000ybp (Fig 4.3B: Tc). From 1,000ybp to the present day (Fig 4.3B: Td), each population displayed variable changes in Ne, with Jakarta appearing as the most stable over this period (Fig 4.3).

Fig 4.3. Effective population size (Ne) over time assuming 10 generations per year for various endemic and introduced populations of Aedes albopictus from Australasia and Southeast Asia. Line colour represents each population and axes are displayed using a log-log scale. A) Sea level change following the last glacial maximum (LGM, 18-21kya), maps show present day vs. LGM geography; B) Ne over time for all populations, shaded horizontal bars above the plot (Ta-Td) are discussed in the Results; C) populations from the suspected native range of Ae. albopictus; D) historically-introduced populations (introduced >30 years ago); E) recently-introduced populations from the Torres Strait Islands, Australia. The vertical grey-shaded bar represents the last glacial maximum (18-21kya).

133 DISCUSSION

The genome-wide SNPs and mitochondrial genome sequences in my study uncovered clear genetic structure in the native and introduced ranges of Ae. albopictus and revealed a pattern of post-glacial population isolation, followed by rapid expansion, proceeding the last glacial maximum (18-21kya). Although I found evidence of differentiation between Torres Strait inner vs. outer islands, the distinction was not as obvious as previously revealed with microsatellites (Maynard et al., 2017), which highlights some of the limitations of using a low number of loci (in this case 13 microsatellite loci). Importantly, these results highlight that genome-wide SNPs will likely uncover similar broad-scale patterns previously revealed by other methods such as microsatellites, but that populations may be more distinct and with less admixture inferred between population clusters. The apparent extent of admixture can have an important impact on the way we interpret genetic data, and we may find less of it when we look with more markers. Both selection and drift seem to be at play in the invading populations, and isolation by distance also contributes to some of the differentiation between populations in my study.

At a broad level, the endemic populations from the Indonesian archipelago (Jakarta and Sumba) are clearly distinct from Malaysia (Ipoh), using both nuclear and mitogenome datasets, lending support to previous findings (Maynard et al., 2017). However, populations were recovered as distinct clusters using DAPC and t-SNE on genome-wide nuclear SNPs or using hierarchical approaches such as in STRUCTURE (Fig S4.4). The concordance in genetic structure between the mitochondrial and nuclear datasets suggests that a selective or endosymbiont sweep is not driving the observed broad-scale genetic patterns (as these results would otherwise contrast). This may be evident at smaller landscape scales and much work remains to be conducted on potential selective sweeps driven by Wolbachia symbionts of Ae. albopictus; which has been one of the suggested mechanisms for the evolution of a new sympatric cryptic species of Ae. albopictus in China (Guo et al., 2018).

Analysis of the mitochondrial genome in the context of other global populations further clarifies the suspected invasion origins of the introduced populations in my study. The recent invasion (mid- 2000s) into the Torres Strait Islands of Northern Australia has an Indonesian genetic background, however it is possible that there have been multiple invasions into the region as evidenced by the diversity of mitochondrial haplotypes in the region and the clear differentiation from Jakarta/Sumba using the nuclear dataset (which do not appear to be the actual invasion sources). The more historically-introduced (>30 years ago) invasive populations in Honiara (Guadalcanal, Solomon

134 Islands) and Amamapare (Papua, Indonesia) appear to be derived from a Malaysian or possibly mainland Southeast Asian genetic background and showed less mitochondrial diversity, but this may be due to the limited sampling of these populations. It is also possible these populations are closely-related to Papua New Guinea populations (Maynard et al. 2017), however, they were not sequenced in this study. Both Indonesian and Torres Strait Island samples in this study are similar to those from the Philippines and all belong to a well-supported clade (Fig S4.1, Clade B) which is approximately 0.2% divergent from the majority of other sequences of global samples of Ae. albopictus.

The phylogenetic relationships between individuals from the various global populations included here lends some support for previous claims that suggest certain mitochondrial haplotypes are more prevalent across specific geographic regions. Such patterns potentially correspond to temperate vs. subtropical vs. tropical lineages of Ae. albopictus, which may be more able to establish in climatic zones that they are pre-adapted to. Sherpa et al. (2019a) demonstrated that cold-adapted Ae. albopictus were successful colonisers of temperate areas. Indeed, many other successful invasive organisms have shown similar patterns regarding pre-adaptiveness (Hufbauer et al. 2012; Elst et al. 2016), so this appears to be an important factor in determining biosecurity pathway risks. However, my results also suggest that the global spread of Ae. albopictus strongly reflects a pattern of human- mediated dispersal (with several geographically distant countries being closely related. Unlike natural dispersion which occur via a steady increase in distribution or follow a stepping-stone-like pattern of colonisation (a pattern which is apparent for Ae. albopictus at a finer geographic scales (e.g. within the Torres Strait Islands and PNG’s Fly Region)), the pattern of Ae. albopictus’s global genetic relationships shows movement over long distances following established trade pathways. Consequently, the risk of invasion of Ae. albopictus is strongly connected to human trade and movements. Together, the findings of this study and others highlight that global regions that are both strong trading partners and climatically similar represent higher invasion risks.

However, just as importantly, post-introduction evolution appears to play a key role in invasion success, and appears to be shaping the evolution of invasive populations of Ae. albopictus in Europe (Sherpa et al., 2019a; Sherpa et al., 2019c). I found potential signs of selection across the genome, including between the inner and outer islands of the Torres Strait Islands, possibly reflecting differences in selection pressures in these areas of the Torres Strait. This was also apparent between native populations in Jakarta and Ipoh (Malaysia), which I hypothesised would be under less selective pressure than invasive populations. Overall, these findings contrast those from Ae. aegypti,

135 which exhibits an apparently stable FST across the genome/chromosomes (Crawford et al., 2017), but this has not been extensively investigated in this species.

Characterising the dynamics of these invasions helps in predicting the spread of future exotic species in the region. This work also helps our general understanding of the evolutionary forces that shape biological invasions across islands and how this influences the evolutionary trajectory of vector populations.

Demographic history of Aedes albopictus in Australasia and Malaysia

Past reconstructions of the population history of Ae. albopictus have highlighted that the species originated from continental East Asia (Porretta et al., 2012), probably under a tropical, monsoonal climate (Mogi et al., 2017). Historical climatic and associated ecosystem changes have led to the diversification of several species of forest mosquito through geographical isolation across southern Asia’s complex topography (Chen et al., 2011; Mogi et al., 2017; Morgan et al., 2011; O’Loughlin et al., 2008). Rapid marine transgression following the last glacial maximum (LGM, 18-21kya, Fig 4.3A) which submerged vast areas of the periodically exposed Sunda shelf has likely had a strong evolutionary influence on Ae. albopictus. During the LGM, both paleoclimatic and paleoecological data suggest that Southeast Asia was dominated by open grassland savannas and dry forests, which had replaced previously forested areas (Adams & Faure, 1997; Bird et al., 2005; Hope et al., 2004; Morley, 2000; Tamburini et al., 2003; Verstappen, 1997). So far, the inclusion of populations from Indonesia, Borneo, the Philippines and various other parts of Ae. albopictus’s potential native distribution (such as India) is lacking in such investigations of demographic history, as is also the case with invasive populations from the tropics. The genetic differentiation of several populations of Ae. albopictus across Southeast Asia and Indonesia (Beebe et al., 2013; Maynard et al., 2017) tends to support the notion of post-glacial isolation following the marine transgression, in which rising sea levels could have led to the fragmentation of habitat (Cannon et al., 2009) and thus may explain the genetic differentiation of some native populations across this region as they would have become more isolated with the disappearance of land bridges. However, distribution modelling by Porretta et al. (2012) has suggested that much of the area spanning the periodically exposed Sunda shelf would have been suitable for Ae. albopictus (Fig 4.3A, LGM map) at the time of the last glacial maximum. Using two mitochondrial genes, they found that populations of Ae. albopictus across continental East Asia remained connected during the last glacial phase, supporting a scenario of post-glacial population expansion, rather than contraction. Reconstruction of Ne over time using site frequency spectrum by Sherpa et al. (2019a) additionally found support for the post-glacial expansion of populations of Ae. albopictus from China, Japan and Penang Island (Malaysia).

136

In the present study, I attempted to shed light on the biogeographic history of Ae. albopictus from several endemic and invasive populations across Australasia and Malaysia to fill knowledge gaps across this region of Sundaland. In contrast to previous studies, my results lend strong support to a scenario of the post-glacial isolation of populations across Malaysia and Indonesia, with all populations exhibiting a severe population bottleneck commencing at ~20kya that coincides with the last glacial maximum. As sea levels rose and conditions became cooler and more arid, it is conceivable that populations would have become isolated across my study area, resulting in a genetic bottleneck and reduction in effective population size during this period of rising sea levels. The disparity in my results compared to those from other parts of Asia prompts the need for more comprehensive sampling of populations across the native region of Ae. albopictus to better understand the evolutionary biogeography of the species. Indeed, it is possible that more northern native populations of Ae. albopictus experienced a demographic expansion as higher latitudes became more habitable for Ae. albopictus following the last glacial maximum.

My results also indicated that the populations in this study exhibited a stabilisation of effective population size between ~1,000-2,500kya, followed by a drastic increase in effective population size at ~700-2,000ybp. This may have been the result of sea level stabilisation and/or other environmental conditions around the mid-Holocene (Bird et al., 2010; Woodroffe et al., 1985), which could have facilitated the stabilisation and subsequent proliferation of the populations that I inferred. Additionally, it is possible that this pattern was also influenced by the rise of human civilisation, agriculture and the domestication of animals, which could have provided more reliable blood sources and breeding habitats for Ae. albopictus. The sharp increase in Ne could represent the ‘domestication’ (increase in association with humans) of Ae. albopictus, a process often characterised by a sudden expansion of estimated effective population size, as has been associated with the domestication of crops (Beissinger et al., 2016; Cornejo et al., 2018; Cubry et al., 2018; Li et al., 2009; Meyer et al., 2016; Scarcelli et al., 2019) and other animals (Ajmone-Marsan et al., 2010; Bruford et al., 2003; López-Uribe et al., 2016; Mirol et al., 2008; Xiang et al., 2018). For instance, the expansion of a species of dung beetle Gymnopleurus mopsus corresponds to that of both humans and large, domesticated mammal in Mongolia (Kang et al., 2018). Although such hypotheses may be consistent with my results, it is not possible at this time to tease apart the relative contributions of past anthropogenic factors that could have shaped the population histories of Ae. albopictus explored here.

137 Following the signature of rapid expansion, changes in effective population size are more variable between populations towards the present day. I suspect that anthropogenic influences are largely responsible for these variations, with many of the bottlenecks coinciding with suspected introduction dates of the populations in my study. For instance, I suspect that the Jakarta population is a long and well-established native population, characterised by a large and stable Ne over the last ~1,000 years. It is uncertain whether populations of Ae. albopictus have long been present on Sumba, but it is possible that they were introduced more recently due to human movements (possibly by Dutch colonisers in the 17th century, as evidenced by the bottleneck at ~100-300 years ago) as the island lies east of Wallace’s line (separate from the Sunda shelf). A similar scenario is plausible for Ipoh, Malaysia, which grew from a village in the 1880s, as this population also shows a bottleneck at a similar time period followed by gradual growth. Thus, many populations in the native range of Ae. albopictus still appear to show a demographic history of human-mediated colonisation seeded by few individuals that results in a post-introduction genetic bottleneck.

Populations in the Torres Strait showed more variability in Ne in the last 100 years than any of the other populations in my study, likely because these invasive populations have only established in the Torres Strait within the last 10-20 years. Changes in Ne over time on the various islands could be the result of genetic drift, with subtle differences in population structure between islands from my other analyses supporting this. Some islands show a pattern of several, recent bottlenecks of varying severity, followed by expansion which generally correspond to the first appearance of the species in the region (e.g. Keriri, Warraber, Poruma). In contrast, Iama does not show evidence of a recent bottleneck, but instead shows a historical bottleneck approximately 60-90 years ago, after which Ne sharply increases and stabilises. As Ae. albopictus has been absent from Iama until recently (and regular surveys have been conducted to confirm this), it is possible that the invasion into Iama did not experience a genetic bottleneck, which could have been the result of a high genetic diversity of founders – indeed this may explain why the population appears stable over time as high genetic diversity could allow for better resilience of the population to ecological stress (Hughes et al., 2008; Hughes & Boomsma, 2004; Saccheri et al., 1998). Another possibility is that recent admixture and fluctuations in population size could have biased estimates of Ne, which has resulted in the observed pattern of variability (Russel & Fewster, 2009; Waples & England, 2011).

Future work and conclusion

To summarise, this whole-genome study has allowed me to more comprehensively characterise the population structure of Ae. albopictus and uncover some of the evolutionary processes, such as selection, drift and gene flow, that are shaping invasive populations of the species across

138 Australasia and Southeast Asia. Additionally, it has allowed me to investigate the biogeographical history of the species more comprehensively than previous studies which have focused mostly on continental Asian populations. I suspect further genetic breaks, consistent with previous lines of evidence based on the distribution of mitochondrial haplotypes and genetic structure supported by microsatellites (Maynard et al., 2017), will be uncovered from populations of Ae. albopictus from this region using similar genomic approaches. Despite the benefits of whole-genomic datasets, thorough sampling of Ae. albopictus populations remains a crucial aspect for identifying source populations and for elucidating the species’ evolutionary history. It is recommended that future population genetic studies on Ae. albopictus utilise pre-existing SNP databases (for example see Pichler et al. (2019)) to optimise their sampling approaches, as this will provide a more accurate context for conclusions to be drawn regarding this medically-significant species of mosquito. Overall, my work demonstrates how historical processes, as well as modern anthropogenic influences, shape the genomes of insect vectors.

ACKNOWLEDGEMENTS

I wish to thank Jacob Crawford and Verily Life Sciences for providing sequencing in a collaborative effort. Additionally, I thank Michael Bangs for mosquito collections and for valuable information regarding Indonesian populations of Aedes albopictus. Din Matias also provided much appreciated advice regarding bioinformatics and R. Thank you to Maddie James and James Wisdom for providing feedback on drafts. Thank you to others that have provided mosquito samples for this study.

REFERENCES

Adams, J. M., & Faure, H. (1997). Preliminary vegetation maps of the world since the last glacial maximum: an aid to archaeological understanding. Journal of Archaeological Science, 24(7), 623-647. Ajmone-Marsan, P., Garcia, J. F., & Lenstra, J. A. (2010). On the origin of cattle: How aurochs became cattle and colonized the world. Evolutionary Anthropology, 19(4), 148-157. doi:10.1002/evan.20267 Battaglia, V., Gabrieli, P., Brandini, S., Capodiferro, M. R., Javier, P. A., Chen, X.-G., . . . Malacrida, A. R. (2016). The worldwide spread of the tiger mosquito as revealed by mitogenome haplogroup diversity. Frontiers in Genetics, 7, 208.

139 Beebe, N. W., Ambrose, L., Hill, L. A., Davis, J. B., Hapgood, G., Cooper, R. D., . . . van den Hurk, A. F. (2013). Tracing the tiger: population genetics provides valuable insights into the Aedes (Stegomyia) albopictus invasion of the Australasian region. PLoS Neglected Tropical Diseases, 7(8), e2361. doi:10.1371/journal.pntd.0002361 Beebe, N. W., Whelan, P. I., van den Hurk, A., Ritchie, S., & Cooper, R. D. (2005). Genetic diversity of the dengue vector Aedes aegypti in Australia and implications for future surveillance and mainland incursion monitoring. Communicable Diseases Intelligence Quarterly Report, 29(3), 299-304. Beichman, A. C., Huerta-Sanchez, E., & Lohmueller, K. E. (2018). Using genomic data to infer historic population dynamics of nonmodel organisms. Annual Review of Ecology, Evolution, and Systematics, 49(1), 433-456. doi:10.1146/annurev-ecolsys-110617-062431 Beissinger, T. M., Wang, L., Crosby, K., Durvasula, A., Hufford, M. B., & Ross-Ibarra, J. (2016). Recent demography drives changes in linked selection across the maize genome. Nature Plants, 2(7), 16084. doi:10.1038/nplants.2016.84 Bennett, K. L., Gómez Martínez, C., Almanza, A., Rovira, J. R., McMillan, W. O., Enriquez, V., . . . Loaiza, J. R. (2019). High infestation of invasive Aedes mosquitoes in used tires along the local transport network of Panama. Parasites & Vectors, 12(1), 264. doi:10.1186/s13071- 019-3522-8 Bird, M. I., Austin, W. E., Wurster, C. M., Fifield, L. K., Mojtahid, M., & Sargeant, C. (2010). Punctuated eustatic sea-level rise in the early mid-Holocene. Geology, 38(9), 803-806. Bird, M. I., Taylor, D., & Hunt, C. (2005). Palaeoenvironments of insular Southeast Asia during the Last Glacial Period: a savanna corridor in Sundaland? Quaternary Science Reviews, 24(20- 21), 2228-2242. Bonizzoni, M., Gasperi, G., Chen, X., & James, A. A. (2013). The invasive mosquito species Aedes albopictus: current knowledge and future perspectives. Trends in Pasitology, 29(9), 460- 468. Bruford, M. W., Bradley, D. G., & Luikart, G. (2003). DNA markers reveal the complexity of livestock domestication. Nature Reviews Genetics, 4, 900. doi:10.1038/nrg1203 Brumfield, R. T., Beerli, P., Nickerson, D. A., & Edwards, S. V. (2003). The utility of single nucleotide polymorphisms in inferences of population history. Trends in Ecology & Evolution, 18(5), 249-256. doi:10.1016/S0169-5347(03)00018-1 Cannon, C. H., Morley, R. J., & Bush, A. B. G. (2009). The current refugial rainforests of Sundaland are unrepresentative of their biogeographic past and highly vulnerable to disturbance. Proceedings of the National Academy of Sciences, 106(27), 11188. doi:10.1073/pnas.0809865106

140 Chen, B., Pedro, P., Harbach, R., Somboon, P., Walton, C., & Butlin, R. (2011). Mitochondrial DNA variation in the malaria vector Anopheles minimus across China, Thailand and Vietnam: evolutionary hypothesis, population structure and population history. Heredity, 106(2), 241. Chen, X.-G., Jiang, X., Gu, J., Xu, M., Wu, Y., Deng, Y., . . . Vontas, J. (2015). Genome sequence of the Asian tiger mosquito, Aedes albopictus, reveals insights into its biology, genetics, and evolution. Proceedings of the National Academy of Sciences, 112(44), E5907-E5915. Clement, M. J., Snell, Q., Walker, P., Posada, D., & Crandall, K. A. (2002). TCS: estimating gene genealogies. Paper presented at the ipdps. Cooper, R. D., Waterson, D. G. E., Kupo, M., & Sweeney, A. W. (1994). Aedes albopictus (Skuse) (Diptera: Culicidae) in the Western Province of Papua New Guinea and the threat of its introduction to Australia. Australian Journal of Entomology, 33(2), 115-116. doi:10.1111/j.1440-6055.1994.tb00933.x Cornejo, O. E., Yee, M.-C., Dominguez, V., Andrews, M., Sockell, A., Strandberg, E., . . . Motamayor, J. C. (2018). Population genomic analyses of the chocolate tree, Theobroma cacao L., provide insights into its domestication process. Communications Biology, 1(1), 167. doi:10.1038/s42003-018-0168-6 Crawford, J. E., Alves, J. M., Palmer, W. J., Day, J. P., Sylla, M., Ramasamy, R., . . . Jiggins, F. M. (2017). Population genomics reveals that an anthropophilic population of Aedes aegypti mosquitoes in West Africa recently gave rise to American and Asian populations of this major disease vector. BMC Biology, 15(1), 16. Cubry, P., Tranchant-Dubreuil, C., Thuillet, A.-C., Monat, C., Ndjiondjop, M.-N., Labadie, K., . . . Rhoné, B. (2018). The rise and fall of African rice cultivation revealed by analysis of 246 new genomes. Current Biology, 28(14), 2274-2282. e2276. Danecek, P., Auton, A., Abecasis, G., Albers, C. A., Banks, E., DePristo, M. A., . . . Group, G. P. A. (2011). The variant call format and VCFtools. Bioinformatics, 27(15), 2156-2158. doi:10.1093/bioinformatics/btr330 Darriba, D., Taboada, G. L., Doallo, R., & Posada, D. (2012). jModelTest 2: more models, new heuristics and parallel computing. Nature Methods, 9(8), 772. doi:10.1038/nmeth.2109 Elst, E. M., Acharya, K. P., Dar, P. A., Reshi, Z. A., Tufto, J., Nijs, I., & Graae, B. J. (2016). Pre- adaptation or genetic shift after introduction in the invasive species Impatiens glandulifera? Acta Oecologica, 70, 60-66. doi:10.1016/j.actao.2015.12.002 Eritja, R., Palmer, J. R. B., Roiz, D., Sanpera-Calbet, I., & Bartumeus, F. (2017). Direct evidence of adult Aedes albopictus dispersal by car. Scientific Reports, 7(1), 14399. doi:10.1038/s41598- 017-12652-5

141 Franklin, D. C., & Whelan, P. I. (2009). Tropical Mosquito Assemblages Demonstrate ‘Textbook’ Annual Cycles. PLoS one, 4(12), e8296. doi:10.1371/journal.pone.0008296 Goudet, J. (2005). hierfstat, a package for r to compute and test hierarchical F-statistics. 5(1), 184- 186. doi:10.1111/j.1471-8286.2004.00828.x Guindon, S., & Gascuel, O. (2003). A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Systematic Biology, 52(5), 696-704. doi:10.1080/10635150390235520 Guo, Y., Song, Z., Luo, L., Wang, Q., Zhou, G., Yang, D., . . . Zheng, X. (2018). Molecular evidence for new sympatric cryptic species of Aedes albopictus (Diptera: Culicidae) in China: A new threat from Aedes albopictus subgroup? Parasites & Vectors, 11(1), 228. doi:10.1186/s13071-018-2814-8 Hawley, W. A. (1988). The biology of Aedes albopictus. Journal of the American Mosquito Control Association, 1, 1-39. Hope, G., Kershaw, A. P., Kaars, S. v. d., Xiangjun, S., Liew, P.-M., Heusser, L. E., . . . Moss, P. T. (2004). History of vegetation and habitat change in the Austral-Asian region. Quaternary International, 118-119, 103-126. doi:10.1016/S1040-6182(03)00133-2 Huelsenbeck, J. P., & Ronquist, F. (2001). MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics, 17(8), 754-755. Hufbauer, R. A., Facon, B., Ravigné, V., Turgeon, J., Foucaud, J., Lee, C. E., . . . Estoup, A. (2012). Anthropogenically induced adaptation to invade (AIAI): contemporary adaptation to human- altered habitats within the native range can promote invasions. Evolutionary Applications, 5(1), 89-101. doi:10.1111/j.1752-4571.2011.00211.x Hughes, A. R., Inouye, B. D., Johnson, M. T., Underwood, N., & Vellend, M. (2008). Ecological consequences of genetic diversity. Ecology Letters, 11(6), 609-623. Hughes, W. O., & Boomsma, J. J. (2004). Genetic diversity and disease resistance in leaf‐cutting ant societies. Evolution, 58(6), 1251-1260. Janes, J. K., Miller, J. M., Dupuis, J. R., Malenfant, R. M., Gorrell, J. C., Cullingham, C. I., & Andrew, R. L. (2017). The K = 2 conundrum. Molecular Ecology, 26(14), 3594-3602. doi:10.1111/mec.14187 Jombart, T. (2008). adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics, 24(11), 1403-1405. Jombart, T., & Ahmed, I. (2011). adegenet 1.3-1: new tools for the analysis of genome-wide SNP data. Bioinformatics, 27(21), 3070-3071. Jombart, T., & Collins, C. (2015). A tutorial for Discriminant Analysis of Principal Components (DAPC) using adegenet 2.0. 0.

142 Kang, J. H., Lim, C. S., Park, S. H., Seok, S. W., Yoon, T. J., Bayartogtokh, B., & Bae, Y. J. (2018). Historical domestication-driven population expansion of the dung beetle Gymnopleurus mopsus (Coleoptera: Scarabaeidae) from its last refuge in Mongolia. Scientific Reports, 8(1), 3963. doi:10.1038/s41598-018-22182-3 Kay, B. H., Fanning, I. D., Marks, E. N., Ives, W. A., Whelan, P. I., & Barker‐Hudson, P. (1990). Is Aedes albopictus in Australia? Medical Journal of Australia, 153(1), 31-34. Keightley, P. D., Ness, R. W., Halligan, D. L., & Haddrill, P. R. (2014a). Estimation of the spontaneous mutation rate per nucleotide site in a Drosophila melanogaster full-sib family. Genetics, 196(1), 313-320. Keightley, P. D., Pinharanda, A., Ness, R. W., Simpson, F., Dasmahapatra, K. K., Mallet, J., . . . Jiggins, C. D. (2014b). Estimation of the spontaneous mutation rate in Heliconius melpomene. Molecular Biology and Evolution, 32(1), 239-243. Kopelman, N. M., Mayzel, J., Jakobsson, M., Rosenberg, N. A., & Mayrose, I. (2015). Clumpak: a program for identifying clustering modes and packaging population structure inferences across K. Molecular Ecology Resourses, 15(5), 1179-1191. doi:10.1111/1755-0998.12387 Kotsakiozi, P., Gloria-Soria, A., Caccone, A., Evans, B., Schama, R., Martins, A. J., & Powell, J. R. (2017). Tracking the return of Aedes aegypti to Brazil, the major vector of the dengue, chikungunya and Zika viruses. PLoS Neglected Tropical Diseases, 11(7), e0005653. Kozlov, A. M., Darriba, D., Flouri, T., Morel, B., & Stamatakis, A. (2019). RAxML-NG: a fast, scalable and user-friendly tool for maximum likelihood phylogenetic inference. Bioinformatics, 35(21), 4453-4455. doi:10.1093/bioinformatics/btz305 Krijthe, J. H. (2015). Rtsne: T-distributed stochastic neighbor embedding using Barnes-Hut implementation. R package version 0.13, URL: github.com/jkrijthe/Rtsne. Lee, D. J., Hicks, M. H., Griffiths, M., Debenham, M. L., Bryan, J. H., Russell, R. C., . . . Marks, E. N. (1980). The Culicidae of the Australasian region (Vol. 4): Australian Govt. Pub. Service. Leigh, J. W., & Bryant, D. (2015). popart: full-feature software for haplotype network construction. Methods in Ecology and Evolution, 6(9), 1110-1116. doi:10.1111/2041-210x.12410 Li, H., & Durbin, R. (2009). Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics, 25(14), 1754-1760. doi:10.1093/bioinformatics/btp324 Li, X., Dodson, J., Zhou, J., & Zhou, X. (2009). Increases of population and expansion of rice agriculture in Asia, and anthropogenic methane emissions since 5000BP. Quaternary International, 202(1), 41-50. doi:10.1016/j.quaint.2008.02.009 López-Uribe, M. M., Cane, J. H., Minckley, R. L., & Danforth, B. N. (2016). Crop domestication facilitated rapid geographical expansion of a specialist pollinator, the squash bee Peponapis pruinosa. Proceedings of the Royal Society B: Biological Sciences, 283(1833), 20160443.

143 Linck, E., and Battey, C. J. (2019). Minor allele frequency thresholds strongly affect population structure inference with genomic data sets. Molecular Ecology Resources, 19(3), 639–47. Mantel, N. (1967). The detection of disease clustering and a generalized regression approach. Cancer Research, 27(2 Part 1), 209-220. Maynard, A. J., Ambrose, L., Cooper, R. D., Chow, W. K., Davis, J. B., Muzari, M. O., . . . Beebe, N. W. (2017). Tiger on the prowl: Invasion history and spatio-temporal genetic structure of the Asian tiger mosquito Aedes albopictus (Skuse, 1894) in the Indo-Pacific. PLoS Neglected Tropical Diseases, 11(4), e0005546. doi:10.1371/journal.pntd.0005546 Medley, K. A., Jenkins, D. G., & Hoffman, E. A. (2015). Human‐aided and natural dispersal drive gene flow across the range of an invasive mosquito. Molecular Ecology, 24(2), 284-295. Meyer, R. S., Choi, J. Y., Sanches, M., Plessis, A., Flowers, J. M., Amas, J., . . . Purugganan, M. D. (2016). Domestication history and geographical adaptation inferred from a SNP map of African rice. Nature Genetics, 48, 1083. doi:10.1038/ng.3633 Miller, M. J., & Loaiza, J. R. (2015). Geographic expansion of the invasive mosquito Aedes albopictus across Panama - Implications for control of dengue and chikungunya viruses. PLoS Neglected Tropical Diseases, 9(1), e0003383. doi:10.1371/journal.pntd.0003383 Mirol, P. M., Routtu, J., Hoikkala, A., & Butlin, R. K. (2008). Signals of demographic expansion in Drosophila virilis. BMC Evolutionary Biology, 8(1), 59. doi:10.1186/1471-2148-8-59 Mogi, M., Armbruster, P. A., Tuno, N., Aranda, C., & Yong, H. S. (2017). The climate range expansion of Aedes albopictus (Diptera: Culicidae) in Asia inferred from the distribution of Albopictus subgroup species of Aedes (Stegomyia). Journal of Medical Entomology, 54(6), 1615-1625. doi:10.1093/jme/tjx156 Morgan, K., O’Loughlin, S. M., Chen, B., Linton, Y.-M., Thongwat, D., Somboon, P., . . . Prakash, A. (2011). Comparative phylogeography reveals a shared impact of pleistocene environmental change in shaping genetic diversity within nine Anopheles mosquito species across the Indo‐Burma biodiversity hotspot. Molecular Ecology, 20(21), 4533-4549. Morley, R. J. (2000). Origin and evolution of tropical rain forests: John Wiley & Sons. Muzari, M. O., Davis, J., Bellwood, R., Crunkhorn, B., Gunn, E., Sabatino, U., & Gair, R. (2019). Dominance of the tiger: The displacement of Aedes aegypti by Aedes albopictus in parts of the Torres Strait, Australia. Communicable Diseases Intelligence 43, 10. Muzari, M. O., Devine, G., Davis, J., Crunkhorn, B., van den Hurk, A., Whelan, P., . . . Ritchie, S. (2017). Holding back the tiger: Successful control program protects Australia from Aedes albopictus expansion. PLoS Neglected Tropical Diseases, 11(2), e0005286. doi:10.1371/journal.pntd.0005286

144 O’Loughlin, S., Okabayashi, T., Honda, M., Kitazoe, Y., Kishino, H., Somboon, P., . . . Dev, V. (2008). Complex population history of two Anopheles dirus mosquito species in Southeast Asia suggests the influence of Pleistocene climate change rather than human‐mediated effects. Journal of Evolutionary Biology, 21(6), 1555-1569. Pichler, V., Kotsakiozi, P., Caputo, B., Serini, P., Caccone, A., & della Torre, A. (2019). Complex interplay of evolutionary forces shaping population genomic structure of invasive Aedes albopictus in southern Europe. PLoS Neglected Tropical Diseases, 13(8), e0007554. doi:10.1371/journal.pntd.0007554 Poplin, R., Ruano-Rubio, V., DePristo, M. A., Fennell, T. J., Carneiro, M. O., Van der Auwera, G. A., . . . Roazen, D. (2018). Scaling accurate genetic variant discovery to tens of thousands of samples. bioRxiv, 201178. Porretta, D., Mastrantonio, V., Bellini, R., Somboon, P., & Urbanelli, S. (2012). Glacial history of a modern invader: Phylogeography and species distribution modelling of the Asian tiger mosquito Aedes albopictus. PLoS one, 7(9), e44515. doi:10.1371/journal.pone.0044515 Pritchard, J. K., Stephens, M., & Donnelly, P. J. G. (2000). Inference of population structure using multilocus genotype data. Genetics, 155(2), 945-959. RCoreTeam. (2018). R: A language and environment for statistical computing. In. Vienna, Austria: R Foundation for Statistical Computing. Ritchie, S. A., Moore, P., Carruthers, M., Williams, C., Montgomery, B., Foley, P., . . . Cooper, B. (2006). Discovery of a widespread infestation of Aedes albopictus in the Torres Strait, Australia. Journal of the American Mosquito Control Association, 22(3), 358-366. Roche, B., Léger, L., L’Ambert, G., Lacour, G., Foussadier, R., Besnard, G., . . . Fontenille, D. (2015). The spread of Aedes albopictus in metropolitan France: contribution of environmental drivers and human activities and predictions for a near future. PLoS one, 10(5), e0125600. doi:10.1371/journal.pone.0125600 Ronquist, F., & Huelsenbeck, J. P. (2003). MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics, 19(12), 1572-1574. Russell, J. C., & Fewster, R. M. (2009). Evaluation of the linkage disequilibrium method for estimating effective population size. In Modeling demographic processes in marked populations (pp. 291-320). Springer, Boston, MA. Saccheri, I., Kuussaari, M., Kankare, M., Vikman, P., Fortelius, W., & Hanski, I. (1998). Inbreeding and extinction in a butterfly metapopulation. Nature, 392(6675), 491. Scarcelli, N., Cubry, P., Akakpo, R., Thuillet, A.-C., Obidiegwu, J., Baco, M. N., . . . Vigouroux, Y. (2019). Yam genomics supports West Africa as a major cradle of crop domestication. Science Advances, 5(5), eaaw1947. doi:10.1126/sciadv.aaw1947

145 Schiffels, S., & Durbin, R. (2014). Inferring human population size and separation history from multiple genome sequences. Nature Genetics, 46(8), 919. Schrider, D. R., Houle, D., Lynch, M., & Hahn, M. W. (2013). Rates and genomic consequences of spontaneous mutational events in Drosophila melanogaster. Genetics, 194(4), 937-954. doi:10.1534/genetics.113.151670 Sherpa, S., Blum, M. G., Capblancq, T., Cumer, T., Rioux, D., & Després, L. (2019b). Unravelling the invasion history of the Asian tiger mosquito in Europe. Molecular Ecology, 28(9), 2360- 2377. Sherpa, S., Blum, M. G., & Després, L. (2019a). Cold adaptation in the Asian tiger mosquito's native range precedes its invasion success in temperate regions. Evolution, 73(9), 1793- 1808. Sherpa, S., Guéguen, M., Renaud, J., Blum, M. G. B., Gaude, T., Laporte, F., . . . Després, L. (2019c). Predicting the success of an invader: Niche shift versus niche conservatism. Ecology and Evolution, 9(22), 12658-12675. doi:10.1002/ece3.5734 Skuse, F. A. (1889). Diptera of Australia. Part V - The Culicidae. Proceedings of the Linnean Society of New South Wales, 3, 1717-1764. Smit, A., Hubley, R., & Green, P. (2015). RepeatMasker Open-4.0. 2013–2015. In. Tamburini, F., Adatte, T., Föllmi, K., Bernasconi, S. M., & Steinmann, P. (2003). Investigating the history of East Asian monsoon and climate during the last glacial–interglacial period (0– 140 000 years): mineralogy and geochemistry of ODP Sites 1143 and 1144, South China Sea. Marine Geology, 201(1), 147-168. doi:10.1016/S0025-3227(03)00214-7 Terhorst, J., Kamm, J. A., & Song, Y. S. (2017). Robust and scalable inference of population history from hundreds of unphased whole genomes. Nature Genetics, 49(2), 303. The Anopheles gambiae 1000 Genomes Consortium. (2017). Genetic diversity of the African malaria vector Anopheles gambiae. Nature, 552, 96. doi:10.1038/nature24995 van den Hurk, A. F., Nicholson, J., Beebe, N. W., Davis, J., Muzari, O. M., Russell, R. C., . . . Ritchie, S. A. (2016). Ten years of the Tiger: Aedes albopictus presence in Australia since its discovery in the Torres Strait in 2005. One Health, 2, 19-24. Van der Auwera, G. A., Carneiro, M. O., Hartl, C., Poplin, R., del Angel, G., Levy-Moonshine, A., . . . DePristo, M. A. (2013). From FastQ Data to High-Confidence Variant Calls: The Genome Analysis Toolkit Best Practices Pipeline. Current Protocols in Bioinformatics, 43(1), 11.10.11-11.10.33. doi:10.1002/0471250953.bi1110s43 Verstappen, H. T. (1997). The effect of climatic change on southeast Asian geomorphology. Journal of Quaternary Science, 12(5), 413-418.

146 Waples, R.S., England, P.R. (2011). Estimating contemporary effective population size on the basis of linkage disequilibrium in the face of migration. Genetics, 189(2), 633-44. Weir, B. S., & Cockerham, C. C. J. E. (1984). Estimating F‐statistics for the analysis of population structure. Evolution, 38(6), 1358-1370. Woodroffe, C. D., Thom, B. G., & Chappell, J. (1985). Development of widespread mangrove swamps in mid-Holocene times in northern Australia. Nature, 317(6039), 711. Xiang, H., Liu, X., Li, M., Zhu, Y. n., Wang, L., Cui, Y., . . . Zhan, S. (2018). The evolutionary road from wild moth to domestic silkworm. Nature Ecology & Evolution, 2(8), 1268-1279. doi:10.1038/s41559-018-0593-4

147 SUPPLEMENTARY FIGURES & TABLES

Fig S4.1. MrBayes Bayesian tree inferred using a trimmed region of the mitochondrial genomes (14,837bp) of worldwide specimens of Aedes albopictus (n=182). The whole tree (including outgroup) is shown and two main sections of the tree are expanded as subsections showing clades A and B (each with an independent scale shown at the bottom). Only posterior probabilities >0.6 are shown, which are only displayed in the tree subsections. Specimens from my study are colour-coded based on populations in the key, whereas sequences from other studies are in black. Code names are shown in Table S4.1.

148

Fig S4.2. RAxML maximum-likelihood tree inferred using a trimmed region of the mitochondrial genomes of (14,972bp) of worldwide specimens of Aedes albopictus (n=183, includes Papua-71 (AJM 898). The clade containing the majority of samples (n=180) is collapsed to more clearly display the relationship of Papua-71 (AJM898) to the rest of the dataset that was displayed in Fig S4.1. Support values (1,000 maximum-likelihood bootstrap replicates) are shown at each node. The outgroup species (Aedes aegypti) is represented by blue lines.

149

Fig S4.3. STRUCTURE plots for K=2 using both microsatellite (A) and whole-genomic SNP data (B) for Aedes albopictus in Australasia and tropical Southeast Asia. Each bar represents an individual mosquito, where the proportion coloured indicates cluster assignment. Populations are divided by white vertical lines and labelled below each plot.

150

Fig S4.4. STRUCTURE plots using a hierarchical approach with both microsatellite and whole-genomic SNP data for Aedes albopictus in Australasia and tropical Southeast Asia. Each bar represents an individual mosquito, where the proportion coloured indicates cluster assignment. Populations are divided by white vertical lines and labelled below each plot. A) Indonesian and Torres Strait Islands cluster (K=2). B) Honiara, Malaysia and Papua, Indonesia cluster (Microsatellites: K=2, Genome-wide SNPs: K=3).

151

Fig S4.5. Discriminant analysis of principal components (DAPC) for Aedes albopictus in the study region using both microsatellites (A) and genome-wide SNP datasets (B). The first two principal components (PC1-2). Each dot represents a single individual colour-coded based on the key in plot B.

152

Fig S4.6. T-distributed stochastic neighbour embedding (t-SNE) using two dimensions for both microsatellite (A) and genome-wide SNP datasets (B). Each dot represents a single individual colour-coded based on the key in plot B.

153

Fig S4.7. Scatter plots of a mantel test between geographic and genetic distances using 114,048 nuclear SNPs for samples of Aedes albopictus from Australasia and tropical Southeast Asia. Colours show the density of points where warmer colours indicate higher densities. Points represent pairwise comparisons between individual mosquitoes. The lines show a linear regression between geographic distance (Dgeo) and genetic distance (Dgen) for each plot. A) All populations B) Populations from the Torres Strait Islands only.

154 155

Fig S4.8. Sliding-window (100kb) FST across the genome for various comparisons of introduced and endemic populations of Aedes albopictus. Each dot represents a separate 100kb window, occurring every 10kb, for which FST was calculated. Titles above each plot show the populations (or groups of populations) that were compared. ‘Keriri’ represents both 2012 and 2014 collections combined whereas ‘Mimika’ represents Amamapare.

Table S4.1. Specific sample information for Aedes albopictus used in the present study. LC, larval collection; IC, immature (pupal or larval) collection; EC, egg collection; HLC, human landing collection; HBS, human baited sweep netting; LT, light trap; ST, sentinel trap; ASP, battery-powered aspirator; unknown details are shown (-).

Refer to supplementary attachment

156 Table S4.2. Comparison of pairwise FST values using microsatellite (13 loci) and whole- genomic SNP data (114,048 nuclear SNPs) for the same populations of Aedes albopictus.

Values are colour-coded where lower FST scores are redder, while greener cells indicate higher FST values. Population abbreviations use the first characters of population names in Table 4.1 with the year indicated for the Torres Strait Islands. Microsatellite data was from Maynard et al. (2017).

Microsatellite pairwise FST Ker '12 Ker '14 0.027 Por '15 0.117 0.123 Iam '15 0.150 0.159 0.062 War '15 0.126 0.132 0.070 0.119 AMA 0.154 0.179 0.179 0.238 0.217 JAK 0.031 0.058 0.099 0.139 0.094 0.143 SUM 0.068 0.091 0.088 0.129 0.081 0.189 0.056 IPO 0.096 0.120 0.139 0.164 0.140 0.103 0.076 0.099 HON 0.124 0.160 0.170 0.195 0.179 0.178 0.101 0.141 0.073 Ker '12 Ker '14 Por '15 Iam '15 War '15 AMA JAK SUM IPO HON

Genome-wide SNPs pairwise FST

Ker '12 Ker '14 0.009 Por '15 0.051 0.054 Iam '15 0.061 0.065 0.050 War '15 0.054 0.061 0.061 0.085 AMA 0.146 0.144 0.151 0.160 0.155 JAK 0.065 0.070 0.080 0.089 0.085 0.150 SUM 0.060 0.070 0.081 0.091 0.088 0.180 0.066 IPO 0.117 0.115 0.121 0.130 0.126 0.077 0.121 0.149 HON 0.138 0.136 0.142 0.151 0.147 0.102 0.144 0.172 0.052 Ker '12 Ker '14 Por '15 Iam '15 War '15 AMA JAK SUM IPO HON

157 CHAPTER V – GENERAL DISCUSSION

Understanding the evolutionary processes that shape population structure is fundamental to understanding the evolutionary biology of invasive species. However, empirically uncovering these driving forces can be challenging when a species’ evolutionary history is complex. For invasive mosquitoes such as Aedes albopictus and Aedes aegypti, their past population histories are complicated by strong anthropophilic influences, as humans have largely been responsible for spreading these species outside of their native range. Analysing intraspecific genetic structure and population genetics processes can provide crucial information on the evolutionary forces that shape the population structure of both native and invasive populations of mosquitoes. This is relevant to our efforts to eradicate these disease vectors and understand virus transmission routes.

In my dissertation research, I used an array of molecular approaches to explore the genetic structure and invasion history of two medically significant mosquitoes, Ae. albopictus and Ae. aegypti, across multiple populations in the Indo-Pacific. Many of the populations used in the course of this research have received little to no attention in previous studies, nor have they been compared to more global populations. Consequently, this work establishes a strong foundation for the future exploration of the population genetics of both species. Overall, genetic population structure was higher than previously shown for both species, and although both showed significant patterns of isolation by distance, a substantial degree of genetic variance was unexplained by geographic distance. This likely reflects the influence of human-mediated movements as well as the role that other potential biotic and abiotic factors may be having on the overall genetic diversity of both species. By studying a recently invaded population of Ae. albopictus in the Torres Strait Islands of Northern Australia and the nearby Fly Region of Southern Papua New Guinea, I have been able to monitor the population structure of the species both spatially and temporally to shed new light on how the invasion process can influence the genetics of recently invasive populations. I found that the island populations of the Torres Strait showed signs of high spatio-temporal structure that could have been the result of genetic drift or potentially multiple introductions into the recently invaded region. This system was investigated further in Chapter IV using whole-genome sequencing and the various populations in the Torres Strait Islands do indeed appear to have slight differences in their demographic histories which may account for the observed variation in population structure. Overall, these results highlight the complexity of evolutionary processes which underlie the current genetic population structure of Ae. albopictus and Ae. aegypti.

158 Problems with sampling

Capturing an accurate geographical representation of Ae. albopictus and Ae. aegypti remains an ongoing issue and sampling in many population genetics studies is somewhat biased. However, this is often due to logistical constraints and the remoteness of some locations, making it difficult to obtain representative samples (depending on the scope of the study). As our genomic toolset is becoming more advanced, it is imperative to capture representative samples of the global population structure of both Ae. aegypti and Ae. albopictus to more deeply understand the two species’ evolutionary histories. A more directed approach to sampling under-represented populations should be prioritised by the research community. In this dissertation, sampling has been constrained based on collaborations and the material available at hand from our regional collections. This has led to inconsistency in collection methods which could bias analyses, particularly where very few individuals have been collected (e.g. <5 individuals) or when sampling from larval/single sites where samples may be closely related. Additionally, in Chapter II, while I found that the more established populations of Ae. albopictus displayed stable population structure, the more recently invaded areas displayed relatively high spatio-temporal genetic variation. There are a growing number of studies investigating the temporal genetic structure for select populations, but this area remains to be explored comprehensively and temporal effects can have a direct influence on conclusions drawn from genetic findings (Lombaert et al., 2018). Future studies that aim to address broad scale patterns of genetic structure in Ae. albopictus and Ae. aegypti will need to prioritise spatial sampling of the species’ full native ranges and consider the influence of temporal collections on population structure, especially in newly established populations.

Likewise, approximate Bayesian computation (ABC) analyses that account for complex scenarios (requiring thorough spatio-temporal sampling) and gene flow between populations will play a key role in enhancing our understanding of the population dynamics and invasion histories of Ae. albopictus and Ae. aegypti. Using ABC to simulate both species’ invasion histories within the Indo- Pacific, I was also able to show that the spread of both Ae. albopictus and Ae. aegypti follows a pattern of multiple, independent invasions, albeit over different time periods and via different routes for each of the species. This was explored in more detail for Ae. albopictus, for which I had more extensive samples across the species’ native and introduced ranges. However, for Ae. aegypti I highlight the need for future investigations to consider the complexity of the species’ invasion history within the Indo-Pacific and to model population clusters in more detail to gain more insightful knowledge about its establishment in the region.

159 In addition to the ABC results from Chapter III, I compiled a collection of historical records (summarised in Fig 3.1 and Table S1 in Chapter III) for Ae. aegypti which will serve as a useful starting place for future investigations. This exercise highlighted the difficulties with obtaining and critically assessing the reliability of some historical presence records, given both species have complicated taxonomic histories and received varying degrees of research attention in different regions. However, this task also made clear the importance of historical records and how ensuring their preservation and ease of access are crucial factors. Many of these records are often overlooked in modern works on invasions, particularly in the Indo-Pacific. A wealth of knowledge regarding these species’ invasions timings is likely located in museum collections, which are often neglected or underutilised as a resource. Optimistically, such collections and historical literature are becoming more globally accessible through digitisation efforts. However, given that we do not always have accurate records for a given species, nor does this necessarily reflect the introduction timings of a species, genomic data provides a powerful tool for investigating temporal aspects of species distributions. It is important to emphasise that the invasion histories of both species are complex and characterised by multiple invasions and with varying degrees of gene flow (chiefly human- mediated) between some populations, particularly between those situated at major trading ports and transport hubs.

Uncovering invasion histories

A major obstacle with testing competing invasion scenarios is the sheer number of possible invasion scenarios to account for, along with logistically attempting to capture what are deemed important populations to model in such simulations. One trade-off to deal with such complexities is to simplify invasion scenarios in terms of the number of competing scenarios tested and by the populations that are modelled (for example, by combining populations that are geographically or genetically similar to one another - which will depend on the questions being asked). This strategy has been used by many other studies (Bennett et al., 2016; Gloria‐Soria et al., 2016; Kotsakiozi et al., 2018; Sherpa et al., 2019a) that aim to address the invasion histories of major genetic or geographic clusters. With this comes limitations in what one can elucidate from the results and could lead to the propagation of over generalised statements regarding invasive origins, particularly when bias and model checking for ABC are not presented in publications (Manni et al., 2017). In cases where invasion timings and suspected origins are better known, more realistic and more complex scenarios can be tested (Lombaert et al., 2010; Sherpa et al., 2019a). This ABC approach was used in Chapters II and III to test more specific hypotheses regarding the invasion histories of Ae. albopictus and Ae. aegypti in my study region, but population clusters were still grouped to reduce the number of scenarios tested. By characterising the population structure of both species in

160 previously unexplored populations and by conducting more specific ABC analyses on both species, I hope to guide future studies to conduct more robust and realistic analyses, producing higher confidence in results with regards to the invasion histories of both Ae. albopictus and Ae. aegypti.

More recently, with the rise of whole-genome sequencing, innovative inference-based frameworks such as SMC++ (Terhorst et al., 2017) can bring new perspectives on population histories. In Chapter IV, I used this to gain insight into past population histories in Ae. albopictus. A variety of site frequency spectrum (SFS)-based methods are available (Beichman et al., 2017; Nielsen et al., 2012; Terhorst et al., 2017) and have been utilised by others to infer the demographic histories of mosquito populations using genomic data (Crawford et al., 2017; Miles et al., 2017; Sherpa et al., 2019b). The decreasing costs of producing vast genetic databases (Gupta et al., 2008; Hert et al., 2008) has yielded thousands of SNPs (Pichler et al., 2019) that can be used for future investigations of demographic histories, and importantly, the statistical methods for dealing with such data continues to improve (Beaumont & Rannala, 2004; Bertorelle et al., 2010; Csilléry et al., 2010; Terhorst et al., 2017). Improvements in both of these areas will make it possible for future studies to discern between very precise invasion scenarios that are currently indistinguishable (Estoup & Guillemaud, 2010). This has been a limitation of current ABC studies utilising microsatellites and other low resolution markers, particularly when genetic diversity is low (Bennett et al., 2016; Ciosi et al., 2008).

The postgenomic era

A major focus of my dissertation utilised the Torres Strait Island populations of Ae. albopictus as a system to investigate the genetic factors that shape newly invading lineages. Using microsatellites, I uncovered both high temporal and spatial structure within the Torres Strait region (with sampling of various islands between 2007 and 2015) which highlighted the swiftness of genetic change in recently established, invasive populations of Ae. albopictus. Furthermore, by using genome-wide SNPs I was able to preliminarily assess the relative roles that selection, gene flow, and drift have on some of these island populations. This has contributed significantly to our understanding of the genetics of the invasion process in Ae. albopictus. In my fourth chapter, I used a comparative framework to explore how my microsatellite results from Chapter II compared to the results of Chapter IV derived from genome-wide sequencing.

A powerful advantage of using SNPs for whole-genome sequencing is the ability to combine datasets to form a coherent global one, although different genotyping-by-sequencing/restriction- associated digest study designs can create incompatible datasets. Using whole-genome sequencing,

161 as in this thesis, allows the dataset to be combined with any previously conducted SNP datasets, and in future the data generated in this study can be combined with the data published by Pichler et al. (2019) to better infer global invasion pathways and population genetic patterns in this species. A significant disadvantage of technologies such as microsatellites is that they are not readily transferable and can be limited by the smaller number of markers used, resulting in lower resolution for detecting genetic differences that may exist between populations and sometimes obscuring genetic relationships (a trend further supported in my work on Ae. albopictus in Chapters II and IV). This is becoming more apparent as an increasing number of studies employ genome-wide SNPs. Population genetic data is a major tool for identifying the sources of intercepted mosquitoes (or other insects) of biosecurity concern, but conclusions regarding the origin of such specimens has severe limitations depending on the database available to compare samples against, the analyses performed and the amount and quality of interception material (which is generally lacking; often specimens are in poor condition and limited to very few or single individuals).

In work not presented in this dissertation, the microsatellite databases from Chapters II and III were used to assist with the identification of mosquito source populations for samples detected at quarantine borders from around Australia (but primarily for Ae. aegypti). Although Ae. aegypti has long been established throughout much of Queensland, it remains a species of biosecurity concern given the threat it poses to major metropolitan areas of Australia and the risk for the introgression of insecticide resistance genes from non-local populations (Endersby-Harshman et al., 2019; Schmidt et al., 2019). Conclusions regarding the sources of incursions was limited using microsatellite data, hence such an implementation of a microsatellite database for this purpose is mainly constrained by: the number of markers, the number of potential source populations for comparison, and the few individual incursion specimens at hand. Since this investigation, ddRADs have been much more successful for the same purpose (Schmidt et al., 2019). Schmidt et al. (2019) suggested that Ae. albopictus’s low genetic structure would limit the use of such a toolset for this species, from a biosecurity perspective. Others have also suggested that Ae. albopictus lacks genetic structure based on geography (Goubert et al., 2016; Porretta et al., 2012). However, based on the results outlined in Chapters II & IV and genetic structure of numerous populations across several areas of Australasia, Southeast Asia, China and Japan (Battaglia et al., 2016; Kotsakiozi et al., 2017; Maynard et al., 2017; Sherpa et al., 2019a), a similar approach should be feasible for Ae. albopictus, but only after the species global population structure has been captured more widely. This has always been an underpinning theme of inferring invasion sources. Comparison of microsatellite data and genome- wide SNPs (Chapter IV) indicates that this will have higher precision with more markers. Importantly, the results from Chapters II and IV demonstrate that spatio-temporal genetic structure

162 can change within a relatively short time frame within a confined geographic region, which has important implications on the practicality and accuracy of using such genetic databases for estimating invasion sources.

Others have compared microsatellites and SNP datasets, highlighting the improved resolution that SNPs can provide (Rašić et al., 2014). My results from Chapter IV support these trends. In general, cluster membership varied between both approaches, with genome-wide SNPs revealing more distinct clustering and less admixture than shown with microsatellites. Clustering relationships between the populations also differed, revealing that the outer islands of the Torres Strait Islands were less differentiated from the inner Torres Strait Islands, Jakarta and Sumba than when using microsatellites. Consequently, there is a trade-off for resolution vs. number of individuals at a given budget. Microsatellites can reveal similar overall genetic patterns but this depends somewhat on the markers used. However, from a computational perspective, analysing whole genomic data is significantly more technically challenging and time-consuming and can be unachievable if high- performance computing resources are not available (for instance, the data used in Chapter IV used approximately 303,360 cpu hours). Whole-genome sequencing techniques are providing an increasingly-affordable tool for generating large genomic databases that not only contain information target subject, but can also be used to capture genetic information about their microbiota (Bennett et al., 2019; Luis et al., 2019; Rodgers et al., 2017). Studying the microbiomes of mosquitoes and how they interact with their hosts could be crucial to truly understanding mosquito life-history traits and their evolutionary biology. Endosymbionts could also offer promising perspectives for microbiota-based control strategies (Guégan et al., 2018), some of which have already been hugely successful (Flores & O’Neill, 2018). From the data generated in Chapter IV for Ae. albopictus, it will be interesting to assemble reads to the available Wolbachia genomes to see if there are any signs of a cytoplasmic sweep within/among populations (Dobson et al., 2001).

CONCLUSIONS AND FUTURE RESEARCH

This thesis has highlighted the importance of population genetics approaches in clarifying the evolutionary history of invasive disease vectors and informing management priorities. Several populations of Ae. albopictus and Ae. aegypti have been separated for a long time, and future studies should include these distinct populations to test whether a particular effect is observed across the genetic diversity within both species. The patterns of population genetic connectivity and structure outlined in this work represents the likely virus transmission corridors for viruses transmitted by these species. Insecticide resistance is also likely to spread along these pathways.

163 The research over my dissertation has shifted from microsatellite markers to using whole-genome sequencing. A great deal of valuable genomic data was obtained for Ae. albopictus through a collaborative effort with Jacob Crawford for Chapter IV. The data gained through this collaboration will be extremely useful and pertinent for future work and underscores the promise of genomic techniques to answer questions regarding mosquito evolution. However, with such data comes computational limitations, which has limited the scope of Chapter IV, particularly with regards to exploring genomic regions under selection that may convey crucial information about the biology of both endemic and invasive populations of Ae. albopictus.

Much foundation work remains to be conducted on many of the populations of Ae. albopictus and Ae. aegypti in the Indo-Pacific. Several geographic regions are underrepresented in population genetic studies, particularly archipelagic Southeast Asia, Australasia and the South Pacific, and prioritising the characterisation of these populations will eventually fill these gaps in our knowledge. Through leveraging the power of genomic approaches, we will be able to much more rigorously explore the relative contributions of selection, adaptation, drift and gene flow in mosquito populations, which will enable future research opportunities to thoroughly investigate the dynamics of complex evolutionary processes observed at invasion fronts. This area holds enormous and exciting promise for improving our understanding of mosquitoes, particularly the medically- significant species within Aedes.

REFERENCES

Battaglia, V., Gabrieli, P., Brandini, S., Capodiferro, M. R., Javier, P. A., Chen, X.-G., . . . Olivieri, A. (2016). The worldwide spread of the tiger mosquito as revealed by mitogenome haplogroup diversity. Frontiers in Genetics, 7, 208-208. doi:10.3389/fgene.2016.00208 Beaumont, M. A., & Rannala, B. (2004). The Bayesian revolution in genetics. Nature Reviews Genetics, 5(4), 251. Beichman, A. C., Phung, T. N., & Lohmueller, K. E. (2017). Comparison of single genome and allele frequency data reveals discordant demographic histories. G3: Genes, Genomes, Genetics, 7(11), 3605-3620. doi:10.1534/g3.117.300259 Bennett, K. L., Gómez-Martínez, C., Chin, Y., Saltonstall, K., McMillan, W. O., Rovira, J. R., & Loaiza, J. R. (2019). Dynamics and diversity of bacteria associated with the disease vectors Aedes aegypti and Aedes albopictus. Scientific Reports, 9(1), 12160. doi:10.1038/s41598- 019-48414-8

164 Bennett, K. L., Shija, F., Linton, Y. M., Misinzo, G., Kaddumukasa, M., Djouaka, R., . . . Hlaing, T. (2016). Historical environmental change in Africa drives divergence and admixture of Aedes aegypti mosquitoes: a precursor to successful worldwide colonization? Molecular Ecology, 25(17), 4337-4354. Bertorelle, G., Benazzo, A., & Mona, S. (2010). ABC as a flexible framework to estimate demography over space and time: some cons, many pros. Molecular Ecology, 19(13), 2609- 2625. Ciosi, M., Miller, N., Kim, K., Giordano, R., Estoup, A., & Guillemaud, T. (2008). Invasion of Europe by the western corn rootworm, Diabrotica virgifera virgifera: multiple transatlantic introductions with various reductions of genetic diversity. Molecular Ecology, 17(16), 3614- 3627. Crawford, J. E., Alves, J. M., Palmer, W. J., Day, J. P., Sylla, M., Ramasamy, R., . . . Jiggins, F. M. (2017). Population genomics reveals that an anthropophilic population of Aedes aegypti mosquitoes in West Africa recently gave rise to American and Asian populations of this major disease vector. BMC Biology, 15(1), 16. Csilléry, K., Blum, M. G., Gaggiotti, O. E., & François, O. (2010). Approximate Bayesian computation (ABC) in practice. Trends in Ecology & Evolution, 25(7), 410-418. Dobson, S. L., Marsland, E. J., & Rattanadechakul, W. (2001). Wolbachia-induced cytoplasmic incompatibility in single- and superinfected Aedes albopictus (Diptera: Culicidae). Journal of Medical Entomology, 38(3), 382-387. doi:10.1603/0022-2585-38.3.382 Endersby-Harshman, N. M., Schmidt, T. L., Chung, J., van Rooyen, A., Weeks, A. R., & Hoffmann, A. A. (2019). Heterogeneous genetic invasions of three insecticide resistance mutations in Indo-Pacific populations of Aedes aegypti (L.). bioRxiv, 768549. doi:10.1101/768549 Estoup, A., & Guillemaud, T. (2010). Reconstructing routes of invasion using genetic data: why, how and so what? Molecular Ecology, 19(19), 4113-4130. doi:10.1111/j.1365- 294X.2010.04773.x Flores, H. A., & O’Neill, S. L. (2018). Controlling vector-borne diseases by releasing modified mosquitoes. Nature Reviews Microbiology, 16(8), 508-518. Gloria‐Soria, A., Ayala, D., Bheecarry, A., Calderon‐Arguedas, O., Chadee, D. D., Chiappero, M., . . . Kamal, H. A. (2016). Global genetic diversity of Aedes aegypti. Molecular Ecology, 25, 5377-5395. Goubert, C., Minard, G., Vieira, C., & Boulesteix, M. (2016). Population genetics of the Asian tiger mosquito Aedes albopictus, an invasive vector of human diseases. Heredity. doi:10.1038/hdy.2016.35

165 Guégan, M., Zouache, K., Démichel, C., Minard, G., Tran Van, V., Potier, P., . . . Valiente Moro, C. (2018). The mosquito holobiont: fresh insight into mosquito-microbiota interactions. Microbiome, 6(1), 49-49. doi:10.1186/s40168-018-0435-2 Gupta, P. K., Rustgi, S., & Mir, R. R. (2008). Array-based high-throughput DNA markers for crop improvement. Heredity, 101(1), 5-18. doi:10.1038/hdy.2008.35 Hert, D. G., Fredlake, C. P., & Barron, A. E. (2008). Advantages and limitations of next-generation sequencing technologies: A comparison of electrophoresis and non-electrophoresis methods. Electrophoresis, 29(23), 4618-4626. doi:10.1002/elps.200800456 Kotsakiozi, P., Gloria-Soria, A., Schaffner, F., Robert, V., & Powell, J. R. (2018). Aedes aegypti in the Black Sea: recent introduction or ancient remnant? Parasites & Vectors, 11(1), 396. doi:10.1186/s13071-018-2933-2 Kotsakiozi, P., Richardson, J. B., Pichler, V., Favia, G., Martins, A. J., Urbanelli, S., . . . Caccone, A. (2017). Population genomics of the Asian tiger mosquito, Aedes albopictus: insights into the recent worldwide invasion. Ecology & Evolution, 7(23), 10143-10157. doi:10.1002/ece3.3514 Lombaert, E., Guillemaud, T., Cornuet, J.-M., Malausa, T., Facon, B., & Estoup, A. (2010). Bridgehead effect in the worldwide invasion of the biocontrol harlequin ladybird. PLoS one, 5(3), e9743. Lombaert, E., Guillemaud, T., & Deleury, E. (2018). Biases of STRUCTURE software when exploring introduction routes of invasive species. Heredity, 120(6), 485-499. doi:10.1038/s41437-017-0042-1 Luis, P., Vallon, L., Tran, F.-H., Hugoni, M., Tran-Van, V., Mavingui, P., . . . Moro, C. V. (2019). Aedes albopictus mosquitoes host a locally structured mycobiota with evidence of reduced fungal diversity in invasive populations. Fungal Ecology, 39, 257-266. doi:10.1016/j.funeco.2019.02.004 Manni, M., Guglielmino, C. R., Scolari, F., Vega-Rúa, A., Failloux, A.-B., Somboon, P., . . . Gasperi, G. (2017). Genetic evidence for a worldwide chaotic dispersion pattern of the arbovirus vector, Aedes albopictus. PLoS Neglected Tropical Diseases, 11(1), e0005332- e0005332. doi:10.1371/journal.pntd.0005332 Maynard, A. J., Ambrose, L., Cooper, R. D., Chow, W. K., Davis, J. B., Muzari, M. O., . . . Beebe, N. W. (2017). Tiger on the prowl: Invasion history and spatio-temporal genetic structure of the Asian tiger mosquito Aedes albopictus (Skuse, 1894) in the Indo-Pacific. PLoS Neglected Tropical Diseases, 11(4), e0005546. doi:10.1371/journal.pntd.0005546

166 Miles, A., Harding, N. J., Bottà, G., Clarkson, C. S., Antão, T., Kozak, K., . . . Project, c. (2017). Genetic diversity of the African malaria vector Anopheles gambiae. Nature, 552(7683), 96- 100. doi:10.1038/nature24995 Nielsen, R., Korneliussen, T., Albrechtsen, A., Li, Y., & Wang, J. (2012). SNP Calling, Genotype Calling, and Sample Allele Frequency Estimation from New-Generation Sequencing Data. PLoS one, 7(7), e37558. doi:10.1371/journal.pone.0037558 Pichler, V., Kotsakiozi, P., Caputo, B., Serini, P., Caccone, A., & della Torre, A. (2019). Complex interplay of evolutionary forces shaping population genomic structure of invasive Aedes albopictus in southern Europe. PLoS Neglected Tropical Diseases, 13(8), e0007554. doi:10.1371/journal.pntd.0007554 Porretta, D., Mastrantonio, V., Bellini, R., Somboon, P., & Urbanelli, S. (2012). Glacial history of a modern invader: Phylogeography and species distribution modelling of the Asian tiger mosquito Aedes albopictus. PLoS one, 7(9), e44515. doi:10.1371/journal.pone.0044515 Rašić, G., Filipović, I., Weeks, A. R., & Hoffmann, A. A. (2014). Genome-wide SNPs lead to strong signals of geographic structure and relatedness patterns in the major arbovirus vector, Aedes aegypti. BMC Genomics, 15(1), 275. Rodgers, F. H., Gendrin, M., Wyer, C. A. S., & Christophides, G. K. (2017). Microbiota-induced peritrophic matrix regulates midgut homeostasis and prevents systemic infection of malaria vector mosquitoes. PLoS Pathogens, 13(5), e1006391. doi:10.1371/journal.ppat.1006391 Schmidt, T. L., Van Rooyen, A. R., Chung, J., Endersby‐Harshman, N. M., Griffin, P. C., Sly, A., . . . Weeks, A. R. (2019). Tracking genetic invasions: genome‐wide SNP s reveal the source of pyrethroid‐resistant Aedes aegypti (yellow fever mosquito) incursions at international ports. Evolutionary Applications. Sherpa, S., Blum, M. G., Capblancq, T., Cumer, T., Rioux, D., & Després, L. (2019a). Unravelling the invasion history of the Asian tiger mosquito in Europe. Molecular Ecology, 28(9), 2360- 2377. Sherpa, S., Blum, M. G. B., & Després, L. (2019b). Cold adaptation in the Asian tiger mosquito's native range precedes its invasion success in temperate regions. Evolution, 73(9), 1793- 1808. doi:10.1111/evo.13801 Terhorst, J., Kamm, J. A., & Song, Y. S. (2017). Robust and scalable inference of population history from hundreds of unphased whole genomes. Nature Genetics, 49(2), 303-309. doi:10.1038/ng.3748

167 APPENDIX

168