16.02.2014 Views

Phylogeography of a pantropical plant with sea-drifted ... - 千葉大学

Phylogeography of a pantropical plant with sea-drifted ... - 千葉大学

Phylogeography of a pantropical plant with sea-drifted ... - 千葉大学

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

( 千 葉 大 学 学 位 申 請 論 文 )<br />

<strong>Phylogeography</strong> <strong>of</strong> a <strong>pantropical</strong> <strong>plant</strong> <strong>with</strong><br />

<strong>sea</strong>‐<strong>drifted</strong> seeds; Canavalia ro<strong>sea</strong> (Sw.) DC.,<br />

(Fabaceae)<br />

汎 熱 帯 海 流 散 布 植 物 ナガミハマナタマメ<br />

(マメ 科 )の 系 統 地 理<br />

2010 年 7 月<br />

千 葉 大 学 大 学 院 理 学 研 究 科<br />

地 球 生 命 圏 科 学 専 攻 生 物 学 コース<br />

Mohammad Vatanparast


<strong>Phylogeography</strong> <strong>of</strong> a <strong>pantropical</strong> <strong>plant</strong> <strong>with</strong><br />

<strong>sea</strong>‐<strong>drifted</strong> seeds; Canavalia ro<strong>sea</strong> (Sw.) DC.,<br />

(Fabaceae)<br />

July 2010<br />

MOHAMMAD VATANPARAST<br />

Graduate School <strong>of</strong> Science<br />

CHIBA UNIVERSITY


TABLE OF CONTENTS PAGES<br />

ABSTRACT 1<br />

GENERAL INTRODUCTION 3<br />

Pantropical <strong>plant</strong>s <strong>with</strong> <strong>sea</strong>-<strong>drifted</strong> seeds species (PPSS) 5<br />

A project on the phylogeography <strong>of</strong> the PPSS 6<br />

A case study <strong>of</strong> PPSS: Hibiscus tiliaceus L. 7<br />

Canavalia ro<strong>sea</strong>: a genuine <strong>pantropical</strong> <strong>plant</strong> <strong>with</strong> <strong>sea</strong>-<strong>drifted</strong> seeds 8<br />

Overview <strong>of</strong> this study 10<br />

CHAPTER 1 12<br />

PHYLOGENETIC RELATIONSHIPS AMONG CANAVALIA ROSEA AND ITS<br />

ALLIED SPECIES 12<br />

1-1 Introduction 12<br />

1-2 Materials and Methods 15<br />

Taxon sampling 15<br />

DNA extraction, PCR, and sequencing 16<br />

Phylogenetic analyses based on cpDNA sequence data 18<br />

Phylogenetic analyses based on ITS sequence data 19<br />

1-3 Results 21<br />

Phylogenetic analyses based on cpDNA sequence data 21<br />

Phylogenetic analyses based on ITS sequence data 22<br />

1-4 Discussion 24<br />

Phylogenetic relationships among C. ro<strong>sea</strong> and its related species 24<br />

The phylogeographic break in the Atlantic Ocean 25<br />

Origin <strong>of</strong> the Hawaiian endemic species 26<br />

Future prospects for the evolutionary studies among C. ro<strong>sea</strong> and its allied<br />

species 27<br />

Tables and figures 29<br />

i


TABLE OF CONTENTS (CONTINUED) PAGES<br />

CHAPTER 2 40<br />

GLOBAL GENETIC STRUCTURE OF CANAVALIA ROSEA; EVIDENCE FROM<br />

CHLOROPLAST DNA SEQUENCES 40<br />

2-1 Introduction 40<br />

2-2 Materials and Methods 44<br />

Sampling 44<br />

DNA extraction, PCR, and sequencing 44<br />

Haplotype Composition and Network <strong>of</strong> C. ro<strong>sea</strong> and its allied species 44<br />

Population differentiation 45<br />

Historical migration rates between oceanic regions 46<br />

Estimates <strong>of</strong> recent migration rates 48<br />

2-3 Results 49<br />

Haplotype Composition and Network <strong>of</strong> C. ro<strong>sea</strong> and its allied species 49<br />

Population differentiation 50<br />

Historical migration rates between oceanic regions 51<br />

Estimates <strong>of</strong> recent migration rates 52<br />

2-4 Discussion 54<br />

Gene flow in Indo-Pacific Ocean through Long Distance Seed Dispersal 54<br />

A strong genetic difference between the Indo-Pacific and Atlantic populations<br />

<strong>of</strong> C. ro<strong>sea</strong> 56<br />

2-5 Conclusion 59<br />

Tables and figures 60<br />

GENERAL DISCUSSION 74<br />

REFERENCES 76<br />

ACKNOWLEDGEMENTS 82<br />

BIOGRAPHY 83<br />

ii


ABBREVIATIONS<br />

AEP<br />

AMOVA<br />

bp<br />

cpDNA<br />

CTAB<br />

ESS<br />

Atlantic East Pacific<br />

Analysis <strong>of</strong> molecular variance<br />

base pair<br />

chloroplast DNA<br />

cetyltrimethyl ammonium bromide<br />

Effective sample size<br />

F ST<br />

Fixation index (F-statistices)<br />

IGS<br />

ITS<br />

IWP<br />

MLE<br />

nrDNA<br />

PCR<br />

PCR-SSCP<br />

PCR-SSP<br />

PPSS<br />

RCA<br />

SAMOVA<br />

Intergenic Spacer<br />

Internal Transcribed Spacers<br />

Indo West Pacific<br />

Maximum likelihood estimates<br />

nuclear ribosomal DNA<br />

Polymerase Chain Reaction<br />

PCR amplification <strong>with</strong> single-strand conformation polymorphism<br />

PCR amplification <strong>with</strong> sequence specific primers<br />

Pantropical Plants <strong>with</strong> Sea-<strong>drifted</strong> Seeds<br />

Rolling Circle Amplification<br />

Spatial Analysis <strong>of</strong> Molecular Variance<br />

iii


ABSTRACT<br />

This study intends to examine the importance <strong>of</strong> long distance seed<br />

dispersal in the recurrent speciation and integration <strong>of</strong> <strong>pantropical</strong> <strong>plant</strong>s <strong>with</strong><br />

<strong>sea</strong>-<strong>drifted</strong> seeds (PPSS). I focused on one <strong>of</strong> genuine member <strong>of</strong> PPSS;<br />

Canavalia ro<strong>sea</strong> (Sw.) DC. and its allied species.<br />

Chapter 1 is concerned <strong>with</strong> the phylogenetic relationships among C.<br />

ro<strong>sea</strong> and its allied species as well as Hawaiian endemic species using<br />

chloroplast DNA (cpDNA) and internal transcribed spacers (ITS) <strong>of</strong> nuclear<br />

ribosomal DNA (nrDNA) sequences. Phylogenetic analyses using nucleotide<br />

sequences <strong>of</strong> 6 cpDNA regions (ca. 6000 bp) as well as nrDNA ITS for C. ro<strong>sea</strong><br />

and its related species suggested that rapid speciation might occurred among<br />

C. ro<strong>sea</strong> and its related species. The phylogenetic results also suggested that<br />

Hawaiian endemic subgenus Maunaloa, was monophyletic and closely related<br />

to subgenus Canavalia than to other subgenera (Wenderothia and Catodonia).<br />

The results suggests that the Hawaiian subgenus originated by single<br />

colonization to Hawaiian archipelagos by <strong>sea</strong>-dispersal.<br />

In chapter 2, spatial genetic structure <strong>of</strong> cpDNA sequences were studied<br />

for C. ro<strong>sea</strong> and its related species. In total 515 individuals from 48 populations<br />

were surveyed based on partial sequences <strong>of</strong> 6 cpDNA regions (ca. 2000 bp).<br />

Statistical analyses (F ST -based and coalescent-based methods) did not show<br />

significant genetic differentiation among the C. ro<strong>sea</strong> populations over whole<br />

Pacific and Indian Oceanic regions and also <strong>with</strong>in Atlantic region. This<br />

suggests that significant gene flow by long distance dispersal <strong>of</strong> <strong>sea</strong>-<strong>drifted</strong><br />

1


seeds occurs among these oceanic regions. On the other hand, the results <strong>of</strong><br />

phylogenetic and population genetic analyses confirm the genetic<br />

differentiation <strong>of</strong> the Atlantic populations. This suggests that African and<br />

American land masses played roles as geographical barriers to gene flow by<br />

<strong>sea</strong>-dispersal. However, partial gene flow was detected between Atlantic and<br />

Indian oceanic regions which suggest that the unity <strong>of</strong> the species in global<br />

scale is kept by long distance seed dispersal over the African continent.<br />

Directional gene flow <strong>with</strong>in Atlantic region might be corresponded to the<br />

variation <strong>of</strong> the strength <strong>of</strong> tropical Atlantic’s major currents which regarded as<br />

transatlantic dispersal in Atlantic region. Moreover, highly differentiated<br />

populations <strong>of</strong> C. ro<strong>sea</strong> were detected in the southern Brazil. The South<br />

Equatorial Current bifurcating at the north-eastern horn <strong>of</strong> Brazil to the<br />

northward and southward appears to be potential barrier to gene flow and<br />

may promote the genetic differentiation <strong>of</strong> the C. ro<strong>sea</strong> populations in<br />

southern Brazil.<br />

2


GENERAL INTRODUCTION<br />

“… mammals have not been able to migrate, whereas some <strong>plant</strong>s, from their<br />

varied means <strong>of</strong> dispersal, have migrated across the wide and broken interspaces.”<br />

(Darwin, 1859)<br />

The term dispersal has two different but interrelated functions in most<br />

species. The first one is, range expansion <strong>of</strong> species, and the second one, gene<br />

flow <strong>with</strong>in and among populations. Range expansion is necessary for almost<br />

all species, so they have various strategies to expand their distribution ranges<br />

(Linhart & Grant, 1996). However, the wider distribution range arisen from<br />

dispersal causes high genetic heterogeneity among populations because <strong>of</strong><br />

increased levels <strong>of</strong> selection <strong>with</strong>in local populations and/or because <strong>of</strong> limited<br />

levels <strong>of</strong> genetic exchange among the local populations (Heywood, 1991;<br />

Hamrick & Nason, 1996; Linhart & Grant, 1996). When genetic heterogeneity<br />

among populations becomes significantly enough, local populations can evolve<br />

and eventually form a distinct species (Wright, 1931; Ennos, 1994; Bohonak,<br />

3


1999). Gene flow is one <strong>of</strong> the most important processes for species to evolve<br />

as cohesive units in their distribution range (Mayr, 1963; Levin, 2000; Morjan &<br />

Rieseberg, 2004). In fact, if levels <strong>of</strong> gene flow <strong>with</strong>in and among populations<br />

become high (e.g. greater than four migrants per generation), it homogenize<br />

the species and prevents genetic divergence <strong>of</strong> local populations.<br />

In many <strong>plant</strong> species, populations are spatially isolated from each other,<br />

<strong>of</strong>ten by hundreds <strong>of</strong> meters or more and seed dispersal represents the only<br />

way by which populations can exchange individuals or to expand the<br />

distribution ranges (Cain et al., 2000). As there are geographical, ecological or<br />

behavioral barriers to seed dispersal, most <strong>plant</strong> species are not distributed<br />

globally (Howe & Smallwood, 1982; Cain et al., 1998; Willson & Traveset, 2000).<br />

The biggest barrier for land <strong>plant</strong>s will be the ocean, so that most <strong>of</strong> the<br />

floristic compositions are generally quite different among continents which are<br />

divided by oceans. However, there are a few <strong>plant</strong> species that characterized by<br />

their extremely wide distribution ranges across littoral areas in tropics and<br />

subtropics worldwide. They are called “<strong>pantropical</strong> <strong>plant</strong>s <strong>with</strong> <strong>sea</strong>-<strong>drifted</strong><br />

seeds” (Takayama et al., 2006; 2008), referred to as PPSS. A few PPSS are known<br />

from various families which can roughly divide into 2 categories. One is<br />

genuine PPSS in which a single species distributes around the globe.<br />

Canavalia ro<strong>sea</strong> (Sw.) DC. (Fabaceae) and Ipomoea pes-caprae (L.) R. Br.<br />

(Convolvulaceae) are in this category. The other one is Sub-PPSS, in which<br />

small numbers <strong>of</strong> closely related species compose the global distribution in<br />

total. Hibiscus tiliaceus L., <strong>with</strong> H. pernambucensis Arruda (Malvaceae), Vigna<br />

marina (Burm.f.) Merr. <strong>with</strong> V. luteola (Jacq.) Benth. (Fabaceae), and species <strong>of</strong><br />

Rhizophora L. and Entada Adans. are in this category.<br />

4


Pantropical <strong>plant</strong>s <strong>with</strong> <strong>sea</strong>-<strong>drifted</strong> seeds species (PPSS)<br />

The main dispersal mode <strong>of</strong> PPSS is <strong>sea</strong>-dispersal. Almost all PPSS have<br />

seeds or fruits that can float in <strong>sea</strong> water for long time. The seed coats <strong>of</strong> these<br />

species are hard <strong>with</strong> lightweight cotyledons and there are air spaces between<br />

the folds <strong>of</strong> the cotyledons which help the seeds to stay impermeable on <strong>sea</strong><br />

water (Nakanishi, 1988; Loewer, 2005; Thiel & Haye, 2006). Nakanishi (1988)<br />

investigated germination and buoyancy <strong>of</strong> seeds and fruits <strong>of</strong> seventeen<br />

maritime species (including most <strong>of</strong> PPSS), after immersion in artificial <strong>sea</strong>water.<br />

He revealed that all seeds and fruits tested in the study continued to float in<br />

<strong>sea</strong> water for at least three months (Nakanishi, 1988). These characteristics help<br />

PPSS to distribute in wide areas in equatorial belt around the globe. Their<br />

distribution ranges are consistent <strong>with</strong> the areas where the average<br />

temperature <strong>of</strong> Ocean water is around 20 ͦC. In the West Atlantic, they are<br />

distributed from Florida to the Uruguay in Southern West Atlantic. In East<br />

Atlantic they are distributed from Senegal to the Angola and in the Indian<br />

Ocean from Northern part <strong>of</strong> Indian Ocean to the East Cape <strong>of</strong> South Africa. In<br />

the West Pacific Ocean their northern limit is Ryukyu Islands in Japan and in the<br />

south west Pacific is the Brisbane in Australia. In central pacific, their<br />

distribution ranges are from Hawaii to Polynesia and in the East Pacific from<br />

Sinaloa in Mexico to the Northern part <strong>of</strong> Peru in the East Pacific <strong>sea</strong>shores (Fig.<br />

1-2A).<br />

The extremely wide distribution range <strong>of</strong> PPSS has been explained<br />

mainly by their high ability <strong>of</strong> seed dispersal (Sauer, 1988; Whistler, 1992). On<br />

the other hand, high dispersal ability <strong>with</strong> <strong>sea</strong>-<strong>drifted</strong> seeds could raise the<br />

5


potential <strong>of</strong> population differentiation and speciation. New species could arise<br />

independently in remote populations following long-distance seed dispersal<br />

and/or adaptation to a new habitat. This kind <strong>of</strong> speciation process is assumed<br />

to occur in PPSS, as the most <strong>of</strong> the PPSS have some closely related species<br />

distributed in limited areas comparing to the mother PPSS (Levin, 2001;<br />

Takayama et al., 2006). However, until recently there were no empirical data to<br />

explain how PPSS keep the extremely wide range <strong>of</strong> distribution.<br />

A project on the phylogeography <strong>of</strong> the PPSS<br />

One <strong>of</strong> the major difficulties that prevent re<strong>sea</strong>rchers from performing<br />

comprehensive study in PPSS was their extremely wide distribution ranges. The<br />

project <strong>of</strong> PPSS was started more than 10 years ago <strong>with</strong> the leading <strong>of</strong> Dr.<br />

Tadashi Kajita and <strong>with</strong> propose <strong>of</strong> global population sampling as well as<br />

performing population genetic analyses using various molecular markers.<br />

Accomplishment <strong>of</strong> this project was facilitated because <strong>of</strong> development <strong>of</strong><br />

aviation and transportation systems in recent two decades. Cooperation <strong>with</strong><br />

over<strong>sea</strong> re<strong>sea</strong>rchers and institutes are also promoted the project to go ahead.<br />

Until now more than several thousand samples were collected from 30<br />

countries and is ongoing project to performing phylogeographic analyses to<br />

reveal the speciation and population differentiation <strong>of</strong> PPSS.<br />

<strong>Phylogeography</strong> is a relatively new discipline that deals <strong>with</strong> the spatial<br />

arrangements <strong>of</strong> genetic lineages, especially <strong>with</strong>in and among closely related<br />

species (Avise, 2009). It utilized molecular markers to reveal the evolutionary<br />

history <strong>of</strong> the species at the geographical scale. Novel analytical methods <strong>with</strong><br />

high output are also recently developed at the population level (e.g.,<br />

6


application <strong>of</strong> coalescent-based approaches and tree-based thinking) (Knowles<br />

& Maddison, 2002; Knowles, 2009). Using these methods will enable us to<br />

study phylogeographic patterns <strong>of</strong> PPSS.<br />

A case study <strong>of</strong> PPSS: Hibiscus tiliaceus L.<br />

Recently, one <strong>of</strong> our colleagues, Koji Takayama studied genetic structure<br />

<strong>of</strong> Hibiscus tiliaceus and its allied species using chloroplast DNA (cpDNA)<br />

polymorphisms and Microsatellite markers (Takayama et al., 2006; 2008).<br />

Hibiscus tiliaceus is distributed in the East Atlantic and Indo-West Pacific<br />

regions and its counterpart species, Hibiscus pernambucensis, is distributed in<br />

the East Pacific-West Atlantic regions. All together distribution range <strong>of</strong> H.<br />

tiliaceus and H. pernambucensis covers almost the entire littoral area <strong>of</strong> the<br />

tropics worldwide, a situation that could have been established through<br />

dispersal by <strong>sea</strong>-<strong>drifted</strong> seeds. Three morphologically similar species to H.<br />

tiliaceus are recognized in both the Old World (Hibiscus hamabo Siebold &<br />

Zucc.) and New World (Hibiscus glaber Matsum. and Hibiscus elatus Sw.). These<br />

three species have limited distribution ranges, which the two island endemic<br />

species, H. glaber and H. elatus grow in parallel at inland habitats <strong>of</strong> islands in<br />

both the Old and New Worlds respectively, and H. hamabo grows in temperate<br />

areas <strong>of</strong> West Pacific area beyond the northern limit <strong>of</strong> distribution <strong>of</strong> H.<br />

tiliaceus (Takayama et al., 2006). The main summary <strong>of</strong> their finding are as<br />

follows:<br />

I. Recurrent speciation from H. tiliaceus has given rise to all <strong>of</strong> its allied<br />

species.<br />

7


II.<br />

Frequent gene flow by long-distance seed dispersal is responsible for<br />

species integration <strong>of</strong> H. tiliaceus in the wide distribution range.<br />

III.<br />

American and African continents may be geographical barriers to gene<br />

flow by <strong>sea</strong>-<strong>drifted</strong> seeds among populations <strong>of</strong> H. pernambucensis and<br />

H. tiliaceus, respectively.<br />

IV.<br />

Introgression between the H. tiliaceus and H. pernambucensis was<br />

occurred in the Atlantic region.<br />

All these results are new for a sub-PPSS species and explain their wide<br />

distribution range; however we still do not know whether these results are<br />

applicable for other PPSS, especially for a genuine PPSS <strong>with</strong> global distribution.<br />

So, to answer this question, I focused on Canavalia ro<strong>sea</strong> (Beach bean), a<br />

genuine member <strong>of</strong> PPSS.<br />

Canavalia ro<strong>sea</strong>: a genuine PPSS<br />

Canavalia ro<strong>sea</strong> (Beach bean) is mostly prostrate herbaceous vine that<br />

trails along beach dunes, coastal strand and rocky shores and sometimes<br />

climbs into low vegetation (Whistler, 1992). The thick and fleshy stem can grow<br />

to 6 m or more in length and more than 2.5 cm in diameter. The stem is rather<br />

woody near the base and several branches radiate outward, forming mats <strong>of</strong><br />

light green semi-succulent foliage. Beach bean has compound leaves <strong>with</strong><br />

three thick, more or less rounded, fleshy leaflets, each about 5 -12 cm long. The<br />

leaflets fold up under the hot sun at midday and are coriaceous, oblong to<br />

nearly circular in outline, obtuse to emarginate, <strong>of</strong>ten minutely apiculate at tip.<br />

Bracteoles are 1.5 mm long. Pedicel is 3 mm long. Calyx 12 mm long;<br />

pubescence short, white, sparse to moderately dense; upper lip much shorter<br />

8


than tube, upper edge constricted behind non-apiculate tip; lowest tooth 2<br />

mm long, acute, slightly exceeding acute laterals. Standard is 3 cm long. The<br />

flowers are typical pea flowers, purplish pink, about 5 cm long and borne in<br />

erect spikes on long stalks. Beach bean blooms most <strong>of</strong> the summer and<br />

sporadically the rest <strong>of</strong> the year. The pods are commonly 10-15 cm long, flat,<br />

moderately compressed, spirally dehiscent; each valve <strong>with</strong> sutural ribs and an<br />

extra rib ca. 3 mm from ventral rib. They are prominently ridged and woody<br />

when mature. Seeds to 18 X 13 X 10 mm, elliptic, slightly compressed, brown<br />

<strong>with</strong> darker marbling, mostly buoyant and indefinitely impermeable to water, at<br />

least for a year; hilum ca. 7 mm long (Sauer, 1964). Occasionally it is cultivated<br />

experimentally as a sand binder or cover crop. Flowering time is in all <strong>sea</strong>sons,<br />

even in the subtropics. The species considered as self-compatible and it can<br />

also be pollinated by carpenter bee (Xylocopa) species (Arroyo, 1981; Gross,<br />

1993).<br />

Canavalia ro<strong>sea</strong> is distributed throughout littoral areas <strong>of</strong> the tropics and<br />

subtropics around the world. It is one <strong>of</strong> the common and most widespread<br />

tropical <strong>sea</strong>coast <strong>plant</strong>s, most commonly trailing on beaches at the outer limits<br />

<strong>of</strong> land vegetation, where it is usually associated <strong>with</strong> Ipomoea pes-caprae,<br />

occasionally climbing on littoral thickets, rarely slightly inland along roadsides<br />

or coastal plain lake shores. The seeds float because <strong>of</strong> a lightweight tissue and<br />

air space (Van Der Pijl, 1969; Loewer, 2005) and remain impermeable in water<br />

for years while drifting in the <strong>sea</strong> (Guppy, 1906; Nakanishi, 1988; Thiel & Gutow,<br />

2004). The extremely wide distribution <strong>of</strong> C. ro<strong>sea</strong> is considered to be a result<br />

<strong>of</strong> long distance seed dispersal by <strong>sea</strong>-<strong>drifted</strong> seeds; however, we do not have<br />

9


any empirical data and don't know whether the gene flow by seed dispersal is<br />

kept throughout the distribution range over the globe.<br />

There are three related species to C. ro<strong>sea</strong>; Canavalia cathartica Thouars,<br />

C. lineata L. and C. sericea A. Gray from the same subgenus Canavalia, which is<br />

distributed in coastal areas <strong>of</strong> Indo-Pacific Oceanic regions. Their<br />

morphological similarity suggests that they might be closely related to each<br />

other. Given the wide distribution <strong>of</strong> C. ro<strong>sea</strong> and the limited distribution <strong>of</strong> the<br />

other three species, the three species <strong>of</strong> more limited distribution might have<br />

diversified from the widely distributed species. Moreover, there are six endemic<br />

Canavalia species in Hawaiian Islands (subgenus Maunaloa). There is<br />

hypothesis that these species might originate from species which reach to the<br />

Hawaiian island by <strong>sea</strong> <strong>drifted</strong> seed dispersal (Sauer, 1964). Although the<br />

presence <strong>of</strong> several closely related species <strong>with</strong> limited distribution are likely to<br />

be explained by recurrent speciation from widely distributed species, as were<br />

shown in H. tiliaceus; the speciation process among C. ro<strong>sea</strong> and these species<br />

is still in question.<br />

The main questions addressed in this thesis are: (1) Have recurrent<br />

speciation occurred in the distribution range <strong>of</strong> C. ro<strong>sea</strong>? and (2) Can a single<br />

species keep gene flow over the extremely wide range <strong>of</strong> distribution?<br />

Overview <strong>of</strong> this study<br />

In chapter 1, I will study the phylogenetic relationships and evolutionary<br />

history among C. ro<strong>sea</strong> and its allied species as well as Hawaiian endemic<br />

species based on cpDNA and internal transcribed spacers (ITS) <strong>of</strong> nuclear<br />

ribosomal DNA (nrDNA) sequences, which enable to investigate multiple lines<br />

10


<strong>of</strong> evidence. In chapter 2, I will study global phylogeography and genetic<br />

structure <strong>of</strong> C. ro<strong>sea</strong> and its allied species based on the cpDNA sequences,<br />

which provide an ideal marker for studying gene flow through seed dispersal.<br />

Finally, based on the results obtained in this study, I will discuss importance <strong>of</strong><br />

<strong>sea</strong>-<strong>drifted</strong> seed dispersal for the speciation and integration <strong>of</strong> C. ro<strong>sea</strong>.<br />

11


CHAPTER 1<br />

PHYLOGENETIC RELATIONSHIPS AMONG CANAVALIA ROSEA AND<br />

ITS ALLIED SPECIES<br />

1-1 Introduction<br />

The genus Canavalia Adans, 1763, is distributed in tropics and subtropics<br />

<strong>of</strong> all over the world (Lackey, 1981; Schrire, 2005). According to the latest<br />

taxonomic revision, the genus are further divided into four subgenera, namely,<br />

Catodonia (7 species) and Wenderothia (16 species) mostly distributed in the<br />

New World, Canavalia (23 species) in both the Old and New World which<br />

includes some crop species, and Maunaloa (6 species) which is endemic to<br />

Hawaii Islands (Sauer, 1964). Sauer (1964) studied the morphological<br />

differences/similarities among species and considered that the most primitive<br />

subgenus would be Wenderothia, and subgenera Catodonia and Canavalia<br />

would be originated from it. He also considered that subgenus Maunaloa<br />

would be originated from subgenus Canavalia because <strong>of</strong> the presence <strong>of</strong> a<br />

<strong>pantropical</strong> species, Canavalia ro<strong>sea</strong>, in the subgenus.<br />

12


Canavalia ro<strong>sea</strong> is a typical member <strong>of</strong> the <strong>plant</strong> group called<br />

“Pantropical Plants <strong>with</strong> Sea-<strong>drifted</strong> Seeds (Takayama et al. 2006, 2008)";<br />

abbreviated as PPSS. These species are distributed in littoral areas <strong>of</strong> the<br />

tropics and subtropics all over the world and their main mode <strong>of</strong> seed dispersal<br />

is <strong>sea</strong>-dispersal. In addition, C. ro<strong>sea</strong> has some closely related species from<br />

subgenus Canavalia which have <strong>sea</strong>-dispersal, namely, C. cathartica which is<br />

distributed over Indo-West Pacific regions, C. lineata in South East Asia and C.<br />

sericea in South West Pacific. These species are distributed <strong>with</strong>in distribution<br />

range <strong>of</strong> C. ro<strong>sea</strong>. The other species which is known to have <strong>sea</strong>-<strong>drifted</strong> seeds is<br />

C. bonariensis <strong>of</strong> subgenus Catodonia. All other species <strong>of</strong> genus Canavalia do<br />

not have <strong>sea</strong>-<strong>drifted</strong> seeds and their main mode <strong>of</strong> seed dispersal is<br />

mechanical (thrown by dehiscent <strong>of</strong> pods) or gravity dispersal. Although the<br />

presence <strong>of</strong> several closely related species <strong>with</strong> limited distribution are likely to<br />

be explained by recurrent speciation from widely distributed species, as were<br />

shown in H. tiliaceus (Takayama et al., 2006); the speciation process among C.<br />

ro<strong>sea</strong> and its related species from subgenus Canavalia is still in question.<br />

Moreover, given the distribution ranges <strong>of</strong> the species and their modes <strong>of</strong> seed<br />

dispersal, Sauer (1964) and Carlquist (1966) suggested that the endemic<br />

species <strong>of</strong> Hawaii Islands might be originated from the species that reach to<br />

the Hawaiian Islands by <strong>sea</strong>-dispersal, and the species would plausibly be the<br />

<strong>pantropical</strong> species, Canavalia ro<strong>sea</strong>. However, this hypothesis has never been<br />

tested using modern molecular phylogenetic methods.<br />

13


To test the hypothesis about the origin <strong>of</strong> Hawaiian endemic species <strong>of</strong><br />

Canavalia and speciation process among C. ro<strong>sea</strong> and its related species, we<br />

employed both nuclear and chloroplast DNA markers. These markers have<br />

been successfully used to study the origin <strong>of</strong> Hawaiian endemic species in<br />

other <strong>plant</strong> groups (Baldwin & Wagner, 2010). Chloroplast DNA is regarded as<br />

single locus and maternally inherited through seeds in most angiosperms<br />

(Mogensen, 1996), and it has been utilized to discover phylogenetic<br />

relationships in many <strong>plant</strong> species (Soltis & Soltis, 1998). In addition, nuclear<br />

DNA markers can also be used in combination <strong>with</strong> cytoplasmic markers to<br />

decipher phylogenetic relationships among closely relates species, as using<br />

multiple lines <strong>of</strong> molecular markers <strong>with</strong> highly polymorphic loci could give<br />

resolved topologies than single locus (Small et al., 1998; Brito & Edwards, 2009;<br />

Calonje et al., 2009). Although a recent phylogenetic study on tribe Diocleinae<br />

based on nrDNA ITS sequences (Varela et al., 2004) showed that subgenus<br />

Canavalia was sister to Catedonia, taxon sampling <strong>of</strong> the study was not enough<br />

to reveal phylogenetic relationships <strong>of</strong> the four subgenera and the origin <strong>of</strong><br />

Hawaiian endemic species. I studied all species <strong>of</strong> subgenus Maunaloa together<br />

<strong>with</strong> other samples <strong>of</strong> C. ro<strong>sea</strong> and its closely related species, in addition to<br />

both representative species <strong>of</strong> subgenera Wenderothia and Catodonia.<br />

14


1-2 Materials and Methods<br />

Taxon sampling<br />

Leaf samples were collected in large-scale, covering almost all<br />

distribution ranges <strong>of</strong> the species. For cpDNA phylogenetic analyses, in total,<br />

62 individuals from Canavalia ro<strong>sea</strong>, 5 individuals from C. cathartica, 4<br />

individuals from C. lineata, and 3 individuals from C. sericea were included in<br />

phylogenetic study (Table 1-1). Moreover, an inland species, Canavalia virosa<br />

(Roxb.) Wight & Arn., two crop species, Canavalia ensiformis (L.) DC. (Common<br />

Jack bean) and Canavalia gladiata (Jacq.) DC. (Sword bean) from the subgenus<br />

Canavalia were added in the phylogenetic analyses (Table 1-1). For nrDNA ITS<br />

phylogenetic analyses selective samples from C. ro<strong>sea</strong> and its related species<br />

were included (Table 1-2). All Hawaiian endemic species from subgenus<br />

Maunaloa including Canavalia molokaiensis Degener & al., C. hawaiiensis<br />

Degener & al., C. napaliensis St. John, C. galeata Gaudich., C. pubescens Hook.<br />

& Arn., and C. kauaiensis J.D. Sauer (St. John H, 1970; Wagner et al., 1999) were<br />

also included in either cpDNA and ITS analyses. Canavalia parviflora Benth. was<br />

added in the phylogenetic analyses as a representative species from subgenus<br />

Catedonia. Canavalia hirsutissima J.D. Sauer and Canavalia villosa Benth. from<br />

15


subgenus Wenderothia were chosen as outgroups according to Sauer (1964)<br />

and Varella (2004). Voucher specimens were deposited in the herbarium <strong>of</strong><br />

University <strong>of</strong> the Ryukyus (RYU), Jardim Botânico, Rio de Janeiro (RBRJ) and<br />

Bishop Museum, Honolulu, Hawaii.<br />

DNA extraction, PCR, and sequencing<br />

Genomic DNA was extracted from dried leaves using modified CTAB<br />

(cetyltrimethyl ammonium bromide) method (Doyle & Doyle, 1987) and DNA<br />

concentration was measured by GeneQuant 100 electrophotometer (GE<br />

Healthcare, Life Sciences). After an initial screening <strong>of</strong> 15 cpDNA candidate<br />

non-coding regions (Shaw et al., 2007), six highly variable regions including<br />

intergenic spacers (IGS) and introns were chosen for further steps (Table 1-3).<br />

For nuclear genome, nuclear rDNA ITS was sequenced. Polymerase chain<br />

reactions (PCR) were performed in 10-25 µL volume reactions containing 1.25<br />

units ExTaq (TaKaRa), 0.2 mM <strong>of</strong> dNTPs, 10x PCR buffer contains 1.5 mM MgCl2,<br />

0.5–1 µM <strong>of</strong> each primer pairs, and 20 ng genomic DNA. The PCR conditions<br />

were as follows: 3 min for initial denaturation at 95 °C, followed by 35<br />

amplification cycles <strong>of</strong> 1 min denaturation at 95 °C, 1-2 min annealing at<br />

fragment-specific temperatures (see Table 1-3), 1-2 min extension at 72 °C, and<br />

a final 10 min extension at 72 °C. The PCR products were visualized by 0.8 %<br />

agarose gel electrophoresis and purified using either GENECLEAN III kit<br />

(Qiagen) or ExoSAP-IT (USB Corp., Cleveland, Ohio, USA) following the<br />

manufacturer's instructions.<br />

The cycle sequencing reactions were carried out using a BigDye<br />

Terminator v. 3.1 Cycle Sequencing Kit (Applied Biosystems) and sequencing<br />

16


eaction products were purified by ethanol precipitation method. All DNA<br />

sequences were determined <strong>with</strong> ABI 377 or ABI 3500 DNA sequencer (Applied<br />

Biosystems). For ITS sequences which direct sequencing yielded unreadable<br />

electropherograms, TOPO-TA cloning kit (Invitrogen) was subsequently used<br />

according to the manufacturer’s instructions. In total for 19 individuals from 14<br />

species cloning method were used. Twelve colonies from each sample were<br />

picked up, purified and amplified via TempliPhi DNA Sequencing Template<br />

Amplification Kits (GE Healthcare). This kit utilize bacteriophage Phi29 DNA<br />

polymerase and rolling circle amplification (RCA) technology (Polidoros et al.,<br />

2006) for rapid amplification <strong>of</strong> circular template DNA. The products were then<br />

sequenced using the ABI Big Dye Terminator v3.1 Cycle Sequencing Ready<br />

Reaction kit (Applied Biosystems). Both forward and reverse strands were<br />

assembled manually using the Autoassembler 2.1 (Applied Biosystems) <strong>with</strong><br />

default setting. Sequences manually edited and aligned <strong>with</strong> Se-Al v2.0a11<br />

(Rambaut, 2002). Seven Indels and inversions from cpDNA sequence data set<br />

were coded as separate binary characters according to Kelchner (2000)<br />

Simmons & Ochoterena (2000) and Ingvarsson (2003). However, length<br />

variations, all gaps containing homopolymers, AT-rich regions and two<br />

homoplasious inversions were excluded from all analyses. For the ITS sequence<br />

data set, cloned products were assembled and singleton mutations were<br />

excluded by comparing to the corresponding direct ITS sequence data <strong>of</strong> this<br />

study. These singletons were presumed as a result <strong>of</strong> Taq or PCR error in<br />

cloning method. I chose a threshold that less than two singletons from aligned<br />

sequence data were removed from the data set. After making final dataset,<br />

haplotype file in Nexus format was made by the program DNAsp v. 5.10.01<br />

17


(Librado & Rozas, 2009). Sequences were deposited to the GenBank under the<br />

accession numbers ### ###.<br />

Phylogenetic analyses based on cpDNA sequence data<br />

As the chloroplast genome is inherited as a single unit <strong>with</strong>out<br />

recombination, combining sequences from multiple cpDNA regions can be<br />

justified (Soltis & Soltis, 1998). Because all six cpDNA sequenced regions used<br />

in this study occur in the haploid chloroplast genome and their histories are<br />

linked, there is no priori reason to infer that their resulting gene trees will differ.<br />

However, their patterns <strong>of</strong> evolution might be different (e.g. differences in<br />

evolutionary rates and/or base compositions), leading to the incongruence<br />

among datasets (Bull et al., 1993; Wiens, 1998). Keeping these in mind, I<br />

concatenate all these regions to a single sequence data set.<br />

Maximum parsimony (MP), maximum likelihood (ML) and Bayesian<br />

phylogenetic inference were employed for phylogenetic analyses using a<br />

concatenate data set (Table 1-3). MP analyses were conducted in PAUP*<br />

v.4.0b10 (Sw<strong>of</strong>ford, 2002) using heuristic <strong>sea</strong>rches, tree-bisection–reconnection<br />

(TBR) branch swapping algorithm, and all characters were unweighted and<br />

unordered. Branch support was evaluated by bootstrapping method <strong>of</strong><br />

Felsenstein (Felsenstein, 1985) based on 1000 replicates.<br />

ML analysis was employed using the program RAxML (Randomized<br />

Axelerated Maximum Likelihood) version 7.2.6 which implements a rapid hill<br />

climbing algorithm (Stamatakis, 2006). The analysis was run <strong>with</strong> indel and<br />

inversion characters removed under the GTR+G model <strong>of</strong> evolution and best-<br />

18


scoring ML tree inferences. Rapid bootstrap analysis was conducted <strong>with</strong> 1000<br />

replications to assess branch support.<br />

Partitioned Bayesian inference were performed <strong>with</strong> MrBayes v.3.1.2<br />

(Ronquist & Huelsenbeck, 2003). The data set was divided into two partitions;<br />

aligned nucleotides and coded indels and inversions. For nucleotide sequences<br />

GTR+I model was selected as a best-fit model according to the Akaike<br />

information criterion (AIC) by MrModel Test v.2 (Nylander, 2004) and binary<br />

model was employed for coded indels and inversions. Two independent<br />

Markov Chain Monte Carlo (MCMC) analyses <strong>with</strong> four simultaneous chains<br />

and 5,000,000 generations were run. Trees were sampled every 100<br />

generations and the first 20000 trees were discarded as burn-in. MCMC chains<br />

convergence was visualized <strong>with</strong> Tracer v. 1.5 (Rambaut & Drummond, 2009)<br />

and likelihood scores for sampled trees were inspected.<br />

Phylogenetic analyses based on ITS sequence data<br />

For nrDNA ITS sequence data, small number <strong>of</strong> samples were surveyed<br />

comparing to the comprehensive cpDNA samples (36 vs. 88). Maximum<br />

parsimony (MP) and neighbor joining (NJ) analyses were performed <strong>with</strong><br />

PAUP* v.4.0b10 (Sw<strong>of</strong>ford, 2002) for nrDNA ITS sequences (Table 1-2). MP<br />

analyses were conducted using heuristic <strong>sea</strong>rches, tree-bisection–reconnection<br />

(TBR) branch swapping algorithm, and all characters were unweighted and<br />

unordered. Branch support was evaluated by bootstrapping method <strong>of</strong><br />

Felsenstein (Felsenstein, 1985) based on 1000 replicates. The NJ tree was<br />

constructed using the distance set to Kimura 2-parameter model (Kimura,<br />

19


1980) and bootstrap analysis based on 100 replicates. Combine sequence data<br />

set <strong>of</strong> cpDNA and nrDNA ITS was performed based on NJ method. In the both<br />

method Canavalia hirsutissima and Canavalia villosa from subgenus<br />

Wenderothia are defined as outgroups.<br />

Network analyses<br />

20


1-3 Results<br />

Phylogenetic analyses based on cpDNA sequence data<br />

The concatenate cpDNA aligned sequence, which consisted <strong>of</strong> 88<br />

sequences from 16 species, yielded 27 haplotypes (H1-27) <strong>with</strong> total length <strong>of</strong><br />

5654 characters after excluding all gaps and ambiguous characters. In total<br />

seven indels and inversions were coded as binary codes. MP analyses retained<br />

16112 most parsimonious trees <strong>with</strong> tree length <strong>of</strong> 145, consistency index (CI)<br />

<strong>of</strong> 0.862, retention index (RI) <strong>of</strong> 0.807 and rescaled consistency index (RC) <strong>of</strong><br />

0.696 (Fig. 1-3). One hundred twenty-two characters were variable, <strong>of</strong> which 57<br />

were parsimony-informative. MP analysis inferred monophyly <strong>of</strong> all ingroup<br />

taxa including members <strong>of</strong> subgenus Canavalia and subgenus Maunaloa all<br />

together <strong>with</strong> 100% bootstrap support. Although, C. ro<strong>sea</strong> is not monophyletic<br />

species, three major clades (I, II and III) were recognized <strong>with</strong> 93, 88 and 83<br />

bootstrap support, respectively (Fig. 1-3). Clade I consisted <strong>of</strong> C. ro<strong>sea</strong><br />

haplotypes, H5 from Java in Indian Ocean and H6 from Tanzania in Indian and<br />

Brazil in Atlantic Ocean. Clade II comprised 3 haplotypes (H16-18) <strong>of</strong> C. ro<strong>sea</strong><br />

which are exclusive to the Atlantic region samples. Haplotypes in this strongly<br />

supported clade are from both West and East Atlantic Ocean <strong>with</strong> longer<br />

21


anch length. Hawaiian endemic species <strong>of</strong> subgenus Maonalua (clade III)<br />

make monophyletic group <strong>with</strong> high bootstrap support (Fig. 1-3). A major<br />

haplotype (H8) was shared among 5 species (C. ro<strong>sea</strong>, C. lineata, C. cathartica,<br />

C. gladiata and C. ensiformis) and also shared among 24 samples <strong>of</strong> C. ro<strong>sea</strong><br />

from very wide distribution range in Indian and Pacific Oceans. Haplotype H2 is<br />

exclusive between C. sericea and C. cathartica from Pacific Ocean and 14<br />

haplotypes <strong>of</strong> C. ro<strong>sea</strong> (H1, H3-4, H7, H9-15 and H19-21) which are mainly<br />

from Atlantic Ocean made polytomies in all phylogenetic analyses. Canavalia<br />

lineata have only an identical haplotype (H8) <strong>with</strong> C. ro<strong>sea</strong>, but C. cathartica<br />

own two haplotypes (H2 and H8) which are shared <strong>with</strong> C. sericea and C. ro<strong>sea</strong>,<br />

respectively (Figs 1-3 and 2-1). Maximum likelihood tree (not shown) and<br />

Bayesian majority role consensus tree (Fig. 1-4) represented an identical<br />

topology for the main clades which was found in MP analyses.<br />

Phylogenetic analyses based on ITS sequence data<br />

Total sequence length <strong>of</strong> nrDNA ITS was 782 for 36 samples and aligned<br />

sequence after excluding all gaps and ambiguous characters yielded 674 bp<br />

(Table 1-2). The analyses inferred monophyly <strong>of</strong> all ingroup taxa including<br />

members <strong>of</strong> subgenus Canavalia and subgenus Maunaloa all together <strong>with</strong><br />

100% bootstrap support same as resultant trees from cpDNA sequences. The<br />

Hawaiian endemic subgenus Maunaloa was grouped <strong>with</strong> subgenus Canavalia<br />

rather than other subgenera (Wenderothia and Catadonia) (Fig. 1-5). MP<br />

analysis results were revealed that members <strong>of</strong> subgenus Maunaloa is<br />

polyphyletic. Although clear phylogenetic relationships among ingroup taxa<br />

22


were not obtained (polytomy), Hawaiian endemic species and some species<br />

have multiple copies <strong>of</strong> ITS sequences. A copy is shared among all species<br />

including C. ro<strong>sea</strong> and others species (haplotype H1).<br />

Resultant tree from combined dataset <strong>of</strong> cpDNA and nrDNA ITS show that<br />

subgenus Maunaloa is a monophyletic group <strong>with</strong> high statistical support<br />

(bootstrap value 88%; Fig. 1-6). Overall, in the nrDNA ITS trees, also same as<br />

cpDNA trees, clear relationships among C. ro<strong>sea</strong> and its allied species was not<br />

resolved.<br />

23


1-4 Discussion<br />

Phylogenetic relationships among C. ro<strong>sea</strong> and its related species<br />

Phylogenetic analyses based on ca. 6000 bp cpDNA sequences are revealed a<br />

monophyletic group including members <strong>of</strong> subgenus Canavalia and subgenus<br />

Maunaloa (Fig. 1-3) which is disclosed by morphological (Sauer, 1964) and<br />

cladistic surveys (De Queiroz et al., 2003). However, our results do not<br />

segregate clear species relationship among C. ro<strong>sea</strong> and its allied species even<br />

using high variable cpDNA regions (Fig. 1-3). Therefore understanding<br />

evolutionary history among these species is rather complicated, and additional<br />

sequences from nuclear genome and ecological surveys are essential to<br />

address such issues (Cronn et al., 2002). On the other hand, the genealogy <strong>of</strong><br />

the cpDNA haplotypes appears somewhat “star phylogeny” (Fig. 2-1), <strong>with</strong> a<br />

common ancestral-like haplotypes (e.g. H7 and H8) lie at the central position<br />

and recent derivatives (rare haplotypes) independently connected to it by short<br />

branches (Avise, 2000). In general, such gene genealogies are interpreted as a<br />

result <strong>of</strong> range expansion (Slatkin & Hudson, 1991; Exc<strong>of</strong>fier, 2004).<br />

Considering phylogenetic trees and haplotype network (Figs 1-3 and 2-1), its<br />

plausible to suppose that recurrent speciation happened among C. ro<strong>sea</strong> and<br />

its allies as reported in numerous <strong>plant</strong> species (Schaal et al., 1998; Kay et al.,<br />

2005; Mort et al., 2007).<br />

Shared haplotypes <strong>of</strong> C. cathartica <strong>with</strong> C. ro<strong>sea</strong> (H8) and C. sericea (H2)<br />

might be sign <strong>of</strong> retention <strong>of</strong> ancestral polymorphisms or traces <strong>of</strong> recent<br />

introgression events. Because I used relatively long cpDNA sequences, this<br />

24


event could not have been caused by lack <strong>of</strong> information on cpDNA regions.<br />

Therefore the possibility <strong>of</strong> introgressive hybridization is more likely, however,<br />

even extensive organelle sequence data sets may have not sufficient power to<br />

conclusively resolve between the ancestral retention and contemporary<br />

introgression (Donnelly et al., 2004).<br />

Interestingly, two well-known crop species Canavalia ensiformis<br />

(Common Jack bean) and Canavalia gladiata (Sword bean) have completely<br />

identical cpDNA sequence haplotype <strong>with</strong> C. ro<strong>sea</strong> (H8) <strong>with</strong> using ca. 6000<br />

highly variable regions. This case shows that cpDNA genome <strong>of</strong> these crop<br />

species is same <strong>with</strong> C. ro<strong>sea</strong> and might suggest that introgression happened<br />

among these species and/or they might be derived from C. ro<strong>sea</strong>, however<br />

further study is necessary.<br />

The phylogeographic break in the Atlantic Ocean<br />

Usually, when there is a sharp geographic boundary between widely<br />

distributed clades, re<strong>sea</strong>rchers assume that such breaks are the result <strong>of</strong><br />

geographic barriers to dispersal, cryptic species boundaries, or recent contacts<br />

between historically allopatric populations (Irwin & Gibbs, 2002). The<br />

haplotypes <strong>of</strong> clade II (H16-18) are specific to Atlantic region <strong>with</strong> high<br />

probability support and Long Branch length (Fig. 1-3). Although possibility <strong>of</strong><br />

cryptic species cannot be completely rejected and the populations in the<br />

African Oceanic regions are not kind <strong>of</strong> allopatric populations, I considered that<br />

long-term geographic barrier to dispersal is responsible to such kind <strong>of</strong><br />

phylogeographic break in the Atlantic Ocean (Fig. 2-2).<br />

25


Origin <strong>of</strong> the Hawaiian endemic species<br />

Volcanic oceanic islands such as Hawaii which have never been<br />

connected to the continents can be colonize by species which have dispersed<br />

to the islands from elsewhere. These islands which separated by thousands <strong>of</strong><br />

kilometers from a source population are usually colonized by water or animaldispersed<br />

<strong>plant</strong> species. According to phylogenetic analyses using nucleotide<br />

sequences <strong>of</strong> 6 chloroplast DNA regions (clade III, Fig. 1-3) and combined<br />

sequence data set <strong>of</strong> cpDNA and nrDNA ITS (Fig. 1-6), Hawaiian endemic<br />

subgenus, Maunaloa, was monophyletic <strong>with</strong> high bootstrap support (84% and<br />

88%, respectively) and it was closely related to subgenus Canavalia than to<br />

other subgenera. These results suggest that the Hawaiian subgenus is a result<br />

<strong>of</strong> single colonization event to the Hawaiian archipelagos.<br />

Although ancestral lineage <strong>of</strong> Hawaiian endemic Canavalia was not<br />

clearly solved but it is plausible to assume that an ancestral species <strong>of</strong> Hawaiian<br />

endemic species, was somewhat similar to C. ro<strong>sea</strong> which has <strong>sea</strong>-<strong>drifted</strong> seed<br />

dispersal ability to reach the Hawaiian Islands. Sauer (1964) and Carlquist<br />

(1966) reported loss <strong>of</strong> seed buoyancy for the members <strong>of</strong> subgenus Maunaloa.<br />

The loss <strong>of</strong> seed buoyancy in these species might be due to decreased air<br />

space in the seeds. In C. ro<strong>sea</strong>, air spaces between the folds <strong>of</strong> the cotyledons<br />

help long-term seed buoyancy on <strong>sea</strong> water (Nakanishi, 1988). Loss <strong>of</strong> seed<br />

buoyancy in oceanic island species is a common phenomenon and occurred in<br />

many species in Hawaiian endemic <strong>plant</strong>s (Carlquist, 1974). Loss <strong>of</strong> seed<br />

dispersal ability had occurred in response to the habitat shift toward inlands<br />

during the speciation process <strong>of</strong> members <strong>of</strong> subgenus Maunaloa.<br />

26


The important attribute <strong>of</strong> reduced seed dispersibility is expected to<br />

restrict gene flow among local populations <strong>with</strong>in the Islands, and may lead to<br />

genetic differentiation among populations. Further studies are required to<br />

assess gene flow and population genetic structure in inland species.<br />

Future prospects for the evolutionary studies among C. ro<strong>sea</strong> and its allied<br />

species<br />

Although <strong>with</strong> using more than 6000 bp cpDNA sequences I could get<br />

overall phylogenetic relationships among C. ro<strong>sea</strong> and its related species,<br />

however, clear evolutionary history among them is not resolved. This is not the<br />

only case which single locous is not able to decipher phylogenetic relationships<br />

among taxa (Cronn et al., 2002). Given the difficulties that closely related taxa<br />

such as the C. ro<strong>sea</strong> present in terms <strong>of</strong> phylogenetic resolution, and the low<br />

sequence divergence found in cpDNA and nrDNA ITS sequences, novel<br />

approaches are needed to integrate not only a combination <strong>of</strong> single copy<br />

genes, but also multiple molecular markers such as Microsatellite or AFLP<br />

markers (Scherson et al., 2005). The use <strong>of</strong> multiple and independent nuclear<br />

loci, promises not only to resolve phylogenetic relationships but also <strong>of</strong>fers a<br />

means by which speciation events may be solved in PPSS.<br />

The results <strong>of</strong> cpDNA phylogeny suggest that recurrent speciation might<br />

occur among C. ro<strong>sea</strong> and its allied species in oceanic islands. However,<br />

complete segregation has not been established between these species and<br />

also some crop species. As theory suggests, long distance dispersal play a role<br />

for range expansion <strong>of</strong> species and because <strong>of</strong> less flow in marginal<br />

27


populations, speciation might occur. Taking account the limited distribution<br />

range <strong>of</strong> allied species <strong>of</strong> C. ro<strong>sea</strong> and also endemic species <strong>of</strong> Hawaiian Islands,<br />

this kind <strong>of</strong> speciation might occurred among C. ro<strong>sea</strong> and its daughter species<br />

(C. cathartica, C. lineata and C. sericea) in Indo-West pacific region and also for<br />

Hawaiian endemic species at oceanic islands which produced an endemic<br />

subgenus. In terms <strong>of</strong> speciation, long distance seed dispersal clearly<br />

contributed in species diversity in oceanic islands.<br />

28


Subgenus Taxon Oceanic Region Locality N.<br />

Canavalia C. ro<strong>sea</strong> (Sw.) DC.<br />

Indian Ocean South Africa Umdloti 2<br />

Tanzania Dar es salaam 2<br />

Sri Lanka Wattala, Negambo 2<br />

Thailand Phuket, Kmala Beach 1<br />

Indonesia Bali 1<br />

Java 1<br />

Sumatra 1<br />

Australia Darwin 2<br />

Headland Harbour 1<br />

West Pacific Thailand Kho Samui 1<br />

Taiwan Houpihu 1<br />

Singapore Singapore 1<br />

Philippine Quezon, Luzon 1<br />

Australia Queensland 2<br />

Japan Iriomote 2<br />

New Caledonia Plage de poe 1<br />

Fiji Korotogo 1<br />

Tonga Sopu 1<br />

Samoa Samoa 1<br />

Marquises Taipivai, Nuku Hiva 1<br />

East Pacific Mexico Nayarit 2<br />

Sinaloa 1<br />

Oaxaca 1<br />

Costa Rica Jaco Beach 1<br />

Panama Veracruz 1<br />

Ecuador Isla Jambel 1<br />

West Atlantic Mexico Coatzcoalcos 2<br />

Costa Rica Puerto Viejo 1<br />

Panama Pina, Colon 4<br />

Cuango, Colon 3<br />

Brazil Para 2<br />

Gaibu Pernanbuco 1<br />

Rio De Janeiro, Arraial do Cabo 1<br />

Rio De Janeiro, Recreio 2<br />

East Atlantic Senegal Joal-Fadiout 8<br />

Ghana Busua beach 4<br />

Angola Musul, Luanda 1<br />

Subtotal 62<br />

Table 1‐1. List <strong>of</strong> Canavalia samples used for cpDNA phylogenetic analyses. N, number<br />

<strong>of</strong> individuals in each population. (to be continued)<br />

29


Subgenus Taxon Oceanic Region Locality N.<br />

C. cathartica Thouars Pacific Philippine Atimona 1<br />

Samoa Samoa 1<br />

Tonga Tonga 1<br />

Tahiti Tahiti 1<br />

Hawaii Kauai 1<br />

Subtotal 5<br />

C. lineata (Thunb.) DC. Pacific Taiwan Maopi Tao 1<br />

Japan Ishigaki 1<br />

Miyazaki 1<br />

Ogasawara 1<br />

Subtotal 4<br />

C. sericea A. Gray Pacific Tonga Haashini-Lavengatonga 2<br />

Hawaii Maui, Bishop Museum 1<br />

Subtotal 3<br />

C. virosa (Roxb.) Wight & Arn. Atlantic Africa Seed purchased 1<br />

C. gladiata (Jacq.) DC. Pacific Japan Seed purchased 1<br />

C. ensiformis (L.) DC. Pacific Japan Seed purchased 1<br />

Maunaloa C. hawaiiensis Degener & al. Pacific Hawaii Bishop Museum 1<br />

C. napaliensis St. John Pacific Hawaii Bishop Museum 1<br />

C. galeata Gaudich. Pacific Hawaii Bishop Museum 1<br />

C. pubescens Hook. & Arn. Pacific Hawaii Bishop Museum 1<br />

C. kauaiensis J.D. Sauer Pacific Hawaii Bishop Museum 1<br />

C. molokaiensis Degener & al. Pacific Hawaii Bishop Museum 1<br />

C. hawaiiensis Degener & al. Pacific Hawaii Center for Conservation Re<strong>sea</strong>rch and Training (CCRT) 1<br />

C. galeata Gaudich. Pacific Hawaii Center for Conservation Re<strong>sea</strong>rch and Training (CCRT) 1<br />

Catodonia Canavalia parviflora Benth. Atlantic Brazil Jardim Botanico (Rio de Janeiro) 1<br />

Wenderothia Canavalia villosa Benth. Atlantic Mexico MEXU 1<br />

Canavalia hirsutissima J.D. Sauer Atlantic Mexico MEXU 1<br />

Total 88<br />

Table 1‐1. continued<br />

30


N. Subgenus Taxon Oceanic region Locality DNA voucher Sequence method<br />

1 Canavalia C. ro<strong>sea</strong> (Sw.) DC. Indian Ocean South Africa Umdloti 98 Cloning<br />

2 Tanzania Dar es salaam 184 Cloning<br />

3 Sri Lanka Wattala, Negambo 31 Cloning<br />

4 Indonesia Sumatra 500 Direct<br />

5 Australia Headland Harbour 1 Direct<br />

6 West Pacific Marquises Taipivai, Nuku Hiva 31 Direct<br />

7 Tonga Sopu 171 Direct<br />

8 East Pacific Mexico Sinaloa 965 Direct<br />

9 Panama Veracruz 202 Direct<br />

10 Ecuador Isla Jambel 1205 Direct<br />

11 West Atlantic Panama Cuango, Colon 110 Direct<br />

12 Panama Pina, Colon 24 Cloning<br />

13 Costa Rica Puerto Viejo 346 Cloning<br />

14 Mexico Coatzcoalcos 71 Cloning<br />

15 Brazil Gaibu Pernanbuco 70 Cloning<br />

16 East Atlantic Senegal Joal-Fadiout 107 & 118 Cloning<br />

17 Ghana Busua beach 5 Direct<br />

18 Angola Musul, Luanda 0 Cloning<br />

Subtotal 19<br />

19 C. cathartica Thouars Pacific Philippine Atimona CH10 Cloning<br />

Subtotal 1<br />

20 C. lineata (Thunb.) DC. Pacific Taiwan Maopi Tao CL1 Direct<br />

21 Japan Miyazaki 74 Cloning<br />

Subtotal 2<br />

22 C. sericea A. Gray Pacific Tonga Haashini-Lavengatonga 42 Direct<br />

23 Pacific Tonga Haashini-Lavengatonga 138 Cloning<br />

Subtotal 2<br />

24 C. virosa (Roxb.) Wight & Arn. Atlantic Africa Seed purchased S3F8 Direct<br />

25 Maunaloa C. hawaiiensis Degener & al. Pacific Hawaii Bishop Museum 6 Clonig<br />

26 C. galeata Gaudich. Pacific Hawaii Bishop Museum 7 Clonig<br />

27 C. kauaiensis J.D. Sauer Pacific Hawaii Bishop Museum 9 Clonig<br />

28 C. molokaiensis Degener & al. Pacific Hawaii Bishop Museum 4 Clonig<br />

29 C. napaliensis St. John Pacific Hawaii Bishop Museum 5 Clonig<br />

30 C. pubescens Hook. & Arn. Pacific Hawaii Bishop Museum 8 Clonig<br />

31 C. hawaiiensis Degener & al. Pacific Hawaii Center for Conservation Re<strong>sea</strong>rch and Training (CCRT) C6 Direct<br />

32 C. galeata Gaudich. Pacific Hawaii Center for Conservation Re<strong>sea</strong>rch and Training (CCRT) C7 Direct<br />

33 Catodonia Canavalia parviflora Benth. Atlantic Brazil Jardim Botanico (Rio de Janeiro) 809 Direct<br />

34 Wenderothia Canavalia villosa Benth. Atlantic Mexico MEXU 23 Direct<br />

35 Canavalia hirsutissima J.D. Sauer Atlantic Mexico MEXU 11 Direct<br />

Total 36<br />

Table 1‐2 List <strong>of</strong> Canavalia samples used for nrDNA ITS phylogenetic analyses.<br />

31


Primer name Length (bp) Primer Pairs Sequence (5'- 3') Annealing Tm. Source<br />

atpB -rbcL IGS 518-742 atpB GTGGAAACCCCGGGACGAGAAGTAGT 52 °C Hodges and Arnold, 1994<br />

rbcL ACTTGCTTTAGTTTCTGTTTGTGGTGA Hodges and Arnold, 1994<br />

ndhD -ndhE 670 ndhD GAAAATTAAGGAACCCGCAA 48 °C Xu et. al 2000<br />

ndhE TCAACTCGTATCAACCAATC Xu et. al 2000<br />

psbA -trnH IGS 264-338 psbA CGAAGCTCCATCTACAAATGG 48 °C Hamilton 1998<br />

trnH ACTGCCTTGATCCACTTGGC Hamilton 1998<br />

rps16 Intron 781-850 rps16 F GTGGTAGAAAGCAACGTGCGACTT 52 °C Oxelman et al., 1997<br />

rps16 2R TCGGGATCGAACATCAATTGCAAC Oxelman et al., 1998<br />

trnD -trnT 1036-1055 trnD ACCAATTGAACTACAATCCC 52 °C Demesure et al. (1995)<br />

trnT CTACCACTGAGTTAAAAGGG Demesure et al. (1995)<br />

trnK 2485-2532 trnK 1L CTCAATGGTAGAGTACTCG 52 °C Lavin et al. (2000)<br />

trnK 685F GTATCGCACTATGTATCATTTGA Wojciechowski et al. (2004)<br />

matK 789R TAGGAAATCCTGGTGGCGAGATC Hu et al. (2000)<br />

matK 1777L TTCAGTGGTACGAAGTCAAATG Hu et al. (2000)<br />

matK 1932R CAGACCGACTTACTAATGGG Hu et al. (2000)<br />

trnK 2R AACTAGTCGGATGGAGTAG Johnson & Soltis (1994)<br />

nrDNA ITS 674 ITS5 GGAAGTAAAAGTCGTAACAAGG 54 °C White et al. 1990<br />

ITS4 TCCTCCGCTTATTGATATGC White et al. 1990<br />

Table 1‐3 List <strong>of</strong> primers used for amplifications and sequencing <strong>of</strong> chloroplast regions.<br />

PCR amplification primers are shown in bold.<br />

32


Figure 1‐1. Canavalia ro<strong>sea</strong> and its related species. A: Canavalia ro<strong>sea</strong>, B: C.<br />

cathartic, C: C. lineata, D: C. sericea, E: C. ensiformis, F: C. gladiata. G: C.<br />

villosa, H: C. pubescens,I: seed <strong>of</strong> C. ensiformis (1), C. gladiata (2), C. ro<strong>sea</strong><br />

(3), C. lineata (4).<br />

33


A<br />

B<br />

Figure 1‐2. A: Distribution range <strong>of</strong> PPSS species. B: Distribution range <strong>of</strong> Canavalia<br />

ro<strong>sea</strong> and its related species. In each <strong>of</strong> the encircled areas, C. ro<strong>sea</strong>, C. cathartica,<br />

C. lineata and C. sericea are distributed in the coastal areas and the members <strong>of</strong><br />

subgenus Maunaloa are distributed in inland areas.<br />

34


Figure 1‐3. Phylogram <strong>of</strong> strict consensus tree <strong>of</strong> 16112 shortest tree based on<br />

Maximum Parsimony (MP) analysis from 6 cpDNA regions <strong>with</strong> ca. 6000 bp for<br />

Canavalia species (Tree length=145, CI= 0.862, RI= 0.807 and RC= 0.696).<br />

Bootstrap values are shown above branches. Canavalia hirsutissima and<br />

Canavalia villosa from subgenus Wenderothia and Canavalia parviflora from<br />

subgenus Catedonia are defined as outgroups. Twenty seven haplotypes (H1‐29)<br />

were detected for C. ro<strong>sea</strong> and its allied species. Abbreviations in parenthesis are:<br />

P (Pacific Ocean); A (Atlantic Ocean); I (Indian Ocean); Numbers in the localities<br />

correspond to the number <strong>of</strong> sequenced individuals in that locality.<br />

36


Figure 1‐4. The Bayesian phylogram using 80 samples <strong>of</strong> Canavalia ro<strong>sea</strong> and other<br />

species based on concatenate cpDNA data set. Support for nodes is estimated<br />

from Bayesian posterior probabilities. Support values ≥70% is given. The legends<br />

are as follows: P (Pacific Ocean); A (Atlantic Ocean); I (Indian Ocean).<br />

37


Figure 1‐5. Phylogram <strong>of</strong> strict consensus tree <strong>of</strong> 28 shortest tree based on Maximum<br />

Parsimony (MP) analysis from nrDNA ITS for members <strong>of</strong> subgenus Canavalia and<br />

Maunaloa (bold) (Tree length=92, CI= 0.9457, RI= 0.9920 and RC= 0.9381).<br />

Bootstrap values are shown near branches. Canavalia hirsutissima and Canavalia<br />

villosa from subgenus Wenderothia are defined as outgroups. Abbreviations in<br />

parenthesis are DNA voucher numbers. Letters A and P correspond to Atlantic<br />

and Pacific Oceanic regions, respectively.<br />

38


Figure 1‐6. Strict consensus tree resulting from combined cpDNA and ITS sequence data<br />

for members <strong>of</strong> subgenus Canavalia and Maunaloa (bold). Numbers near<br />

branches indicate bootstrap values. Canavalia hirsutissima and Canavalia villosa<br />

from subgenus Wenderothia are defined as outgroups. Abbreviations in<br />

parenthesis are DNA voucher numbers. Letters A and P correspond to Atlantic<br />

and Pacific Oceanic regions, respectively.<br />

39


Figure 1‐7. Haplotype network among Canavalia ro<strong>sea</strong> and its related species based on<br />

statistical parsimony method. Each segment <strong>of</strong> branch shows the one step<br />

difference <strong>of</strong> molecular data. C. ro<strong>sea</strong> has the greatest diversity <strong>of</strong> haplotypes<br />

in the Atlantic region than Pacific and Indian Ocean. Hawaiian endemic<br />

species are more closely related to a haplotype from the Atlantic (Panama‐<br />

Angola) which may suggest that these species are diversified from Atlantic before<br />

closure <strong>of</strong> Panama Isthmus.<br />

40


CHAPTER 2<br />

GLOBAL GENETIC STRUCTURE OF CANAVALIA ROSEA; EVIDENCE<br />

FROM CHLOROPLAST DNA SEQUENCES<br />

2-1 Introduction<br />

Gene flow is considered as a main factor responsible for genetic<br />

cohesion among populations <strong>of</strong> a species (Olmstead & Palmer, 1994). In fact,<br />

when levels <strong>of</strong> gene flow <strong>with</strong>in and among populations become high (e.g.<br />

greater than four migrants per generation), it homogenize the species and<br />

prevents genetic divergence <strong>of</strong> populations; otherwise local populations can<br />

evolve and eventually form a distinct species (Wright, 1931; Ennos, 1994;<br />

Bohonak, 1999). Population differentiation can also results from environmental<br />

impediments or intrinsic barriers (Avise, 2000). Therefore sufficient amount <strong>of</strong><br />

genetic exchange among populations is expected to hold a species as a<br />

cohesive unit in spatially continuous populations.<br />

Populations <strong>of</strong> many <strong>plant</strong> species are spatially isolated from each other<br />

due to presence <strong>of</strong> physical or ecological barriers such as land masses or water<br />

barriers (Cain et al., 2000). As a result, the distribution range <strong>of</strong> many <strong>plant</strong><br />

species are structured to the specific localities like Indo West Pacific (IWP) and<br />

Atlantic East Pacific (AEP) in some <strong>plant</strong> species such as mangroves (Duke et al.,<br />

40


2002; Nettel & Dodd, 2007) and PPSS such as Hibiscus tiliaceus (Takayama et al.,<br />

2006). In H. tiliaceus, PCR-SSCP (PCR amplification <strong>with</strong> single-strand<br />

conformation polymorphism) and PCR-SSP (PCR amplification <strong>with</strong> sequence<br />

specific primers) analyses performed on more than 1100 samples from 65<br />

populations worldwide to illustrate genetic structure <strong>of</strong> populations in very<br />

wide distribution range. The results revealed that gene flow occurred among<br />

populations <strong>of</strong> H. tiliaceus in Indo-West Pacific region and also over the African<br />

continent due to <strong>sea</strong>-<strong>drifted</strong> seeds and introgression was happened between H.<br />

tiliaceus and H. pernambucensis populations in the Atlantic region. On the<br />

other hand, the American continent was the barrier to gene flow through <strong>sea</strong><strong>drifted</strong><br />

seed dispersal among populations <strong>of</strong> H. pernambucensis (Takayama et<br />

al., 2006; Takayama et al., 2008). These results suggest the importance <strong>of</strong> gene<br />

flow through <strong>sea</strong>-<strong>drifted</strong> seeds over the barriers to keep the unity <strong>of</strong> the<br />

species and also the role <strong>of</strong> American continent as a strong barrier to gene<br />

flow. For globally distributed single PPSS such as C. ro<strong>sea</strong> the existence <strong>of</strong> these<br />

barriers seems to be critical to keep gene flow over its distribution range as the<br />

possibilities <strong>of</strong> cryptic species would be expected.<br />

Presence <strong>of</strong> cryptic species is popular in many <strong>plant</strong>s species where<br />

morphological diversification might not be happened among lineages. In<br />

chapter 1, the result <strong>of</strong> phylogenetic analyses detected some distinct<br />

haplotypes which were specific to the Atlantic region (clade II, Fig. 1-3). As<br />

discussed in chapter 1, usually when there is such kind <strong>of</strong> sharp geographic<br />

boundary between widely distributed clades, it is assumes that it might be a<br />

result <strong>of</strong> geographic barriers to dispersal or possibility <strong>of</strong> cryptic species<br />

boundaries (Irwin & Gibbs, 2002). However, gross population data is necessary<br />

41


to infer the existence <strong>of</strong> gene flow over African and American continents to<br />

keep the unity <strong>of</strong> C. ro<strong>sea</strong> in distant populations. Therefore, the main question<br />

targeted in this chapter is: How can a single species keep gene flow over the<br />

extremely wide range <strong>of</strong> distribution?<br />

A variety <strong>of</strong> genetic markers can be used to quantify the gene flow<br />

through seed dispersal. As discussed by Parker (1998), Ouborg (1999) and<br />

others, these markers include allozymes, DNA sequences, microsatellites,<br />

restriction fragment length polymorphisms (RFLPs), and DNA randomly<br />

amplified from the genome (e.g., RAPDs and AFLPs). Many <strong>plant</strong>s species<br />

exchange genes among populations by the movement <strong>of</strong> pollen and seed.<br />

Only dispersal via seed directly bears on colonization <strong>of</strong> new populations.<br />

Movement <strong>of</strong> both pollen and seed, however, leaves a genetic signature <strong>with</strong>in<br />

and among populations. Therefore, estimating gene flow among <strong>plant</strong><br />

populations from nuclear DNA, which is transferred via pollen and seed, may<br />

tend to overestimate seed dispersal (Cain et al., 2000). Because <strong>of</strong> this effect,<br />

estimation <strong>of</strong> the seed dispersal requires appropriate markers inherited only<br />

through seeds. In most angiosperms, chloroplast DNA (cpDNA) is regarded as<br />

single locus and maternally inherited through seeds (Corriveau & Coleman,<br />

1988; Mogensen, 1996). So, it has been utilized to discover phylogenetic and<br />

phylogeographic patterns in many <strong>plant</strong> species (Schaal et al., 1998; Petit et al.,<br />

2005; Soltis et al., 2006; Avise, 2009). On the other hand, uniparentally inherited<br />

markers such as cpDNA may be informative at the interspecific and<br />

intraspecific level (Wakeley, 2003).<br />

There are many statistical methods for retrieving information from<br />

molecular markers to reveal evolutionary relationships among species and<br />

42


estimating important population parameters (Exc<strong>of</strong>fier & Heckel, 2006). A<br />

variety <strong>of</strong> analytical methods exists to estimates gene flow through seed<br />

dispersal such as F ST -based methods (Wright, 1951), Likelihood-based methods<br />

(Rannala & Hartigan, 1996) and Coalescent-based methods (Beerli &<br />

Felsenstein, 1999). Although the latter rely on more realistic parameters than<br />

the others (Kuhner, 2009).<br />

For this purpose, I employed molecular markers using short fragments <strong>of</strong><br />

6 cpDNA regions and performing population genetic analysis in global scale.<br />

Population genetic analyses based on pairwise F ST and coalescent-based<br />

methods, were conducted for 515 individuals from 48 populations using more<br />

than 2000 bp nucleotide sequences to elucidate the spatial genetic structure <strong>of</strong><br />

C. ro<strong>sea</strong> populations in its entire distribution range.<br />

43


2-2 Materials and Methods<br />

Sampling<br />

Leaf samples were collected in large-scale, covering almost all<br />

distribution ranges <strong>of</strong> the species. In total, 436 individuals (37 populations)<br />

from Canavalia ro<strong>sea</strong>, 25 individuals (5 populations) from C. cathartica, 42<br />

individuals (4 populations) from C. lineata, and 12 individuals (2 populations)<br />

from C. sericea were included in this study (Table 2-1). Voucher specimens<br />

were deposited in the herbarium <strong>of</strong> University <strong>of</strong> the Ryukyus (RYU), Jardim<br />

Botânico, Rio de Janeiro (RBRJ) and Bishop Museum, Honolulu, Hawaii.<br />

DNA extraction, PCR, and sequencing<br />

DNA extraction, PCR and sequencing methods are same as described<br />

methods in chapter 1.<br />

Haplotype Composition and Network <strong>of</strong> C. ro<strong>sea</strong> and its allied species<br />

In chapter 1, I surveyed more than 6000 bp cpDNA sequences for<br />

Canavalia samples from wide range <strong>of</strong> distribution. As sequencing <strong>of</strong> more<br />

than 6000 bp <strong>of</strong> cpDNA genome for all population samples (515 individuals) is<br />

required extra labor and cost, I targeted partial sequences <strong>of</strong> 6 cpDNA regions<br />

(Table 2-2) based on the results <strong>of</strong> phylogenetic analyses (cf. chapter 1). Several<br />

new internal primers were designed for these 6 short fragments and in total<br />

44


2012 base pairs were sequenced for all individuals. Sequences were edited and<br />

aligned as chapter 1 and haplotype composition <strong>of</strong> each population was<br />

recorded. The haplotypes frequency <strong>of</strong> populations was presented as pie charts<br />

on the world map. The genealogical relationships <strong>of</strong> haplotypes were<br />

constructed <strong>with</strong> TCS program (Clement et al., 2000) using statistical parsimony<br />

method <strong>with</strong> 95% confidence interval (Templeton et al., 1992). For population<br />

analyses, all populations <strong>of</strong> C. ro<strong>sea</strong> were divided into five regional groups<br />

(West-East Atlantic Ocean, Indian Ocean and West-East Pacific Ocean) based<br />

on geography, phylogenetic results (cf. chapter 1) and observed haplotypes.<br />

Populations <strong>of</strong> C. cathartica, C. lineata and C. sericea were excluded from all<br />

population analyses.<br />

Population differentiation<br />

To compare chloroplast genetic diversity among population <strong>of</strong> five oceanic<br />

regional groups, I calculated numbers <strong>of</strong> haplotypes (H), polymorphis site (S),<br />

Haplotype diversity (Hd) and nucleotide diversity (π) using the program DnaSP<br />

version 5.10 (Librado & Rozas, 2009). To determine the amount <strong>of</strong> genetic<br />

differentiation among populations <strong>of</strong> C. ro<strong>sea</strong>, F-statistics (F ST ) analysis (Wright,<br />

1951) performed by the program AREQUIN v. 3.11 (Exc<strong>of</strong>fier et al., 2005).<br />

Pairwise F ST was calculated following the method <strong>of</strong> Weir & Cockerham (1984)<br />

and statistical significance was assessed using 1000 permutations. An exact test<br />

<strong>of</strong> population differentiation was performed to examine the null hypothesis <strong>of</strong><br />

the random distribution <strong>of</strong> haplotypes across populations (Raymond & Rousset,<br />

1995).<br />

To clarify groups <strong>of</strong> C. ro<strong>sea</strong> populations which are geographically<br />

homogeneous and maximally differentiated from each other, spatial analysis <strong>of</strong><br />

45


molecular variance (SAMOVA; Dupanloup et al. 2002) was conducted. To<br />

ensure that the conformation <strong>of</strong> the groups (K) is not affected by the given<br />

initial conditions, the simulated annealing procedure is repeated 100 times<br />

(Dupanloup et al., 2002). The analyses were repeated for user-defined groups<br />

(K) from 2 to 12, and the highest FCT values was maintained as the best<br />

grouping <strong>of</strong> populations in case there are not included single population<br />

(Heuertz et al., 2004). Because SAMOVA analysis uses locality coordination and<br />

sampling scale in this study cover the whole equatorial belt, arbitrary<br />

coordinates were tried to avoid miscalculation <strong>of</strong> analyses.<br />

Historical migration rates between oceanic regions<br />

To assess migration rates, direction <strong>of</strong> gene flow and genetic diversity among<br />

populations <strong>of</strong> C. ro<strong>sea</strong>, I applied coalescent approach (Kingman, 1982a, b)<br />

using the Maximum likelihood and Bayesian inference methods implemented<br />

in MIGRATE-N version 3.1.3 (Beerli & Felsenstein, 1999; 2001; Beerli, 2008).<br />

Coalescent based genealogies provide more realistic estimates <strong>of</strong> population<br />

parameters than other summary statistics methods (Kuhner, 2009). I was<br />

specifically interested in contrasting the migration rate and direction <strong>of</strong> gene<br />

flow between geographic regions rather than between sampled populations.<br />

Hence, all populations were pooled into five regional groups (West-East<br />

Atlantic Ocean, Indian Ocean and West-East Pacific Ocean). Given the linear<br />

distribution <strong>of</strong> C. ro<strong>sea</strong> at coastal lines, the stepping stone migration model<br />

<strong>with</strong> asymmetric rates was employed (Kimura & Weiss, 1964).<br />

For each <strong>of</strong> the five regional groups, the genetic diversity (θ=Neμ, where<br />

Ne is the effective population size and µ is the mutation rate per site per<br />

generation), pairwise migration rate (M=m/μ, where m is the rate <strong>of</strong> migration<br />

46


for each locus) and number <strong>of</strong> migrants per generation (Nm=Mθ) were<br />

estimated. Starting parameters for migrant values and θ were generated from<br />

F ST calculation. For Maximum likelihood analysis, 20 short chains (length 5.0 x<br />

10 4 ) followed by 5 long chains (length 5.0 x 10 5 ) <strong>with</strong> sample increment <strong>of</strong> 100<br />

and for both run were conducted and the first 15000 generations were<br />

discarded as burn-in at the beginning <strong>of</strong> each chain. An adaptive heating<br />

scheme <strong>with</strong> 4 chains and a swapping interval <strong>of</strong> 1 was applied. Maximum<br />

likelihood estimates (MLE) were verified <strong>with</strong> three replicate Markov Chain<br />

Monte Carlo (MCMC) simulation runs to ensure convergence <strong>of</strong> similar values<br />

for θ.<br />

Bayesian MCMC coalescent modeling were also used which is provide<br />

parameter estimates based on full likelihood estimation and decreased<br />

computation time <strong>of</strong> approximation comparing to Maximum likelihood<br />

estimates. Bayesian parameters included an update frequency <strong>of</strong> 0.5, a<br />

Metropolis-Hastings sampling algorithm for both θ and M; uniform priors were<br />

placed on θ from 0 to 0.001 and M from 0 to 50000. Starting parameters for<br />

migrant values and θ were generated from F ST calculation. An adaptive heating<br />

scheme <strong>with</strong> 4 chains and a swapping interval <strong>of</strong> 1 was applied. I used the<br />

Felsenstein 84 model <strong>of</strong> evolution, and set the transition to transversion ratio<br />

to 2 as a default values in the program. Six independent MCMC runs <strong>of</strong> varying<br />

length and burn-in were conducted which produced similar results. Hence, I<br />

present results from the longest run which consisted a long chain <strong>of</strong> 50 million<br />

steps <strong>with</strong> sample increment <strong>of</strong> 100 and the first 5 million steps were discarded<br />

as burn-in. Using tracer v1.5 (Rambaut & Drummond, 2009) convergence <strong>of</strong><br />

the likelihood in MCMC chains and effective sample size (ESS) observed<br />

47


following the burn-in. The analyses were considered as converged upon a<br />

stationary distribution, if the different runs generated similar posterior density<br />

distributions <strong>with</strong> a minimum ESS <strong>of</strong> 100 (Hey, 2005; Kuhner & Smith, 2007).<br />

Estimates <strong>of</strong> recent migration rates<br />

Although estimating <strong>of</strong> gene flow and migration rates using MIGRATE-N based<br />

on coalescent methods have many advantages comparing to the other<br />

conventional methods, it also has drawback in separation <strong>of</strong> recurrent gene<br />

flow from ancestral polymorphism (Carbone et al., 2004; Brito, 2005; Bowie et<br />

al., 2006; Kane et al., 2009). To distinguish whether the estimated migration<br />

rates from MIGRATE-N are the result <strong>of</strong> the retention <strong>of</strong> ancestral<br />

polymorphism or recent gene flow, additional coalescent based analyses were<br />

conducted using the program MDIV which is implements both likelihood and<br />

Bayesian methods using MCMC coalescent simulations for jointly estimating <strong>of</strong><br />

the θ and M (Nielsen & Wakeley, 2001).<br />

Theta and M for pairwise comparisons <strong>of</strong> each <strong>of</strong> the five oceanic<br />

regional groups were estimated. Uniform prior Values for Mmax (maximum<br />

value for the scaled migration rate) and θ was set to the 10 and zero,<br />

respectively. The program ran under the default finite sites mutation model <strong>of</strong><br />

HKY (Hasegawa et al., 1985); <strong>with</strong> Markov chain simulation for 5000000 steps,<br />

where the first 500000 were discarded as burn-in. Multiple runs <strong>with</strong> different<br />

random seeds and prior distributions were conducted to determine if the<br />

analyses were reached to the convergence in the mode <strong>of</strong> the posterior<br />

distribution (Nielsen & Wakeley, 2001).<br />

48


2-3 Results<br />

Haplotype Composition and Network <strong>of</strong> C. ro<strong>sea</strong> and its allied species<br />

For population analyses 2012 bp aligned sequences were attained using partial<br />

sequence <strong>of</strong> 6 cpDNA regions which is including 12 substitutions, 2 indels and<br />

one inversion (Table 2-3). Through 21 haplotypes (H1-21) which were detected<br />

in chapter 1 by full sequence <strong>of</strong> 6 cpDNA regions among samples <strong>of</strong> four<br />

species (C. ro<strong>sea</strong>, C. lineata, C. cathartica and C. sericea), I detected 17<br />

haplotypes (H1-17) by partial sequences at population level (Fig. 2-1; Table 2-<br />

1). Four haplotypes, H18 and H19-21 <strong>of</strong> phylogenetic analyses (chapter 1)<br />

could not be detected by partial sequences and therefore they are identical to<br />

haplotypes H16 and H7 in population analyses (this chapter), respectively.<br />

Haplotype network based on full sequence length were shown in figure 2-1.<br />

Colors in the network correspond to the haplotypes which could detect by<br />

partial sequence <strong>of</strong> 6 short fragments (H1-17). All haplotypes were shared<br />

among C. ro<strong>sea</strong> populations except haplotype H2 which is exclusive to C.<br />

cathartica and C. sericea populations from Pacific Ocean (n= 18; Fig. 2-2; Table<br />

2-1). A major haplotype (H8) was shared among C. ro<strong>sea</strong>, C. lineata and C.<br />

cathartica populations from Indian and Pacific Oceanic regions (n=293), (Fig. 2-<br />

2). Notably, all samples <strong>of</strong> C. lineata (n=42) have an identical haplotype <strong>with</strong> C.<br />

ro<strong>sea</strong> haplotype (H8), however, C. cathartica shares haplotypes <strong>with</strong> C. ro<strong>sea</strong><br />

(H8, n=19) and <strong>with</strong> C. sericea (H2, n=6). Another major haplotypes <strong>of</strong> C. ro<strong>sea</strong><br />

were found in Atlantic and Indian Oceanic region including H7 (n=45), H6<br />

(n=32), H17 (n=29) and H16 (n=25), respectively (Table 2-1). Haplotype H1<br />

(n=7) shared between a population from southern Brazil (n=1) and Panama<br />

(n=6) in Atlantic Ocean. Haplotype H3 shared between a population from<br />

49


Mexico (n=2) and Senegal (n=7) in Atlantic Ocean; haplotype H4 shared<br />

between 2 populations from Mexico (Nayarit and Sinaloa, n=9) in Pacific Ocean<br />

and haplotypes H5, H9-15 are private haplotypes (haplotypes which is found<br />

only in one population) in oceanic regions (Fig. 2-2; Table 2-1).<br />

Population differentiation<br />

Total haplotype diversity (Hd) and nucleotide diversity (π) <strong>with</strong>in C. ro<strong>sea</strong><br />

populations were estimated to be 0.69124 and 0.00086, respectively (Table 2-4).<br />

The populations at West Atlantic and East Atlantic had the highest haplotype<br />

diversity (0.7809 and 0.74.34, Table 2-4) and populations <strong>of</strong> the West Pacific<br />

Ocean share a lowest haplotype diversity and nucleotide diversity (0.03418 and<br />

0.00002, respectively). Pairwise F ST analysis detected genetic differentiation in 6<br />

out <strong>of</strong> 10 pairs <strong>of</strong> C. ro<strong>sea</strong> populations (5 regional groups) from different<br />

oceanic regions (Table 2-5). The F ST values between the West and East Atlantic,<br />

Indian and West Pacific and between West and East pacific populations were<br />

0.13, 0.19 and 0.16 respectively. The F ST value between the Indian Ocean and<br />

East pacific was 0.08. On the other hand the F ST values between the East<br />

Atlantic and Indian and between East Pacific and West Atlantic and between<br />

and East pacific populations were 0.36 and 0.45 respectively. All differentiations<br />

were significant using an exact test (P < 0.05; Table 2-5).<br />

The results <strong>of</strong> SAMOVA are shown in figure 2-3 and Table 2-6. The<br />

SAMOVA method did not allow to unambiguously identifying the groups <strong>of</strong><br />

populations (K) displaying the highest differentiation among groups, F CT . This<br />

was because F CT values increased progressively as K was increased, reaching a<br />

plateau at K = 8 (Fig. 2-3). The number <strong>of</strong> K which has highest F CT <strong>with</strong>out<br />

single population was K=3 which is divide all populations to three groups<br />

50


(Table 2-6). First group comprise populations from Ghana, Mexico (A), Panama<br />

(colon2), Para and Senegal. Second group includes populations from<br />

Pernambuco, Rio De Janeiro (1 and 2) and Java and the last groups<br />

corresponds to the rests <strong>of</strong> populations (Table 2-6). The composition <strong>of</strong> these<br />

three groups is partially corresponding to the geographical distribution <strong>of</strong><br />

haplotypes visually identified on the haplotype frequency map (Fig. 2-2). From<br />

K=4 to K=10 all groups includes at least single population (populations which<br />

have private haplotypes) which is not appropriate to the analysis (Heuertz et al.,<br />

2004).<br />

Historical migration rates between oceanic regions<br />

Coalescent-based maximum likelihood estimates (MLE) from MIGRATE-N<br />

indicate asymmetric gene flow along the distribution range <strong>of</strong> the C. ro<strong>sea</strong><br />

populations (Table 2-7). The MLE showed that genetic diversity was highest in<br />

the West Atlantic region (θ = 0.0163), and lowest in the West Pacific region (θ<br />

= 0.0069). The most probable estimates <strong>of</strong> migration rates (M) ranged from 0<br />

to 395.7, <strong>with</strong> the highest migration observed out <strong>of</strong> the East Atlantic into the<br />

West Atlantic. In contrast, estimates <strong>of</strong> migration in the both side <strong>of</strong> Panama<br />

Isthmus as well as into the Indian Ocean from the East Atlantic were zero (Table<br />

2-7). Comparing <strong>with</strong> high migration rates between West-East Atlantic region<br />

and between Indian-West Pacific region, migration between the West and East<br />

Pacific was low but significantly greater than zero. The direction <strong>of</strong> migration<br />

was asymmetrical, <strong>with</strong> the West Atlantic region experiencing a greater input <strong>of</strong><br />

polymorphism due to migration (M EA to WA = 395.7) than the East Atlantic (M WA<br />

to EA = 166.4). Based on values <strong>of</strong> M and θ, I calculated a mean probability <strong>of</strong><br />

number <strong>of</strong> migrants per generation (Nm) among 5 regional groups (Table 2-7)<br />

51


which the highest was greater than 8 migrant per generation <strong>with</strong>in East-West<br />

Atlantic region and the lowest was zero between Western and eastern side <strong>of</strong><br />

American continent and Indian region to the Atlantic region. The estimated<br />

number <strong>of</strong> migrants between Indian-West Pacific pairs and between West-East<br />

Pacific pairs were 5.5 and near 1, respectively.<br />

Results <strong>of</strong> pairwise F ST and MIGRATE-N among Indian and West Pacific<br />

Oceanic regions show relatively low genetic differentiation (F ST = 0.19, Table 2-<br />

5) and high number <strong>of</strong> migrants (Nm >5 per generation), respectively (Table 2-<br />

7). Rates <strong>of</strong> gene flow between West and East Pacific regions (F ST = 0.16; Nm ≅<br />

1) were lower than that <strong>of</strong> between Indian and West Pacific which may<br />

consequence <strong>of</strong> very long distance between population in pacific Oceans.<br />

Bayesian inference <strong>of</strong> theta and Migration rates among five groups are<br />

plotted on figures 2-4 and 2-5 respectively. Although Bayesian estimation <strong>of</strong><br />

migration rates looks much bigger than Maximum likelihood estimation but in<br />

total estimates <strong>of</strong> number <strong>of</strong> migrants (Nm) are almost equal in ML and<br />

Bayesian estimation (see fig. 2-6 and Table 2-7). Highest number <strong>of</strong> migrants<br />

was from West Pacific to the Indian Ocean (Nm=10.98; fig. 2-6) and East<br />

Atlantic to West Atlantic region (Nm=9.01).<br />

Estimates <strong>of</strong> recent migration rates<br />

Figure 2-7 shows the posterior distributions for migration rates (M) and<br />

mutation rates (θ) obtained from MDIV program among population <strong>of</strong> C. ro<strong>sea</strong><br />

at different regional groups. Results are interestingly comparable <strong>with</strong> the<br />

results <strong>of</strong> MIGRATE-N (Figs. 2-4, 2-5 and Table 2-7). For example, migration<br />

rates between East Atlantic and Indian Oceanic region and also between East<br />

52


Pacific and West Atlantic region was definitely low among populations<br />

comparing to the populations at three other regional groups (Fig. 2-7). These<br />

results are concordant <strong>with</strong> MIGRATE-N results and reveal that observed gene<br />

flow <strong>with</strong>in East and West Atlantic, Indo-West Pacific and also between West<br />

and East Pacific is likely result <strong>of</strong> recent gene flow than retention <strong>of</strong> ancestral<br />

polymorphism.<br />

53


2-4 Discussion<br />

Gene flow in Indo-Pacific Ocean through Long Distance Seed Dispersal<br />

Geographical variation <strong>of</strong> cpDNA haplotypes is expected to indicate gene flow<br />

through seed dispersal as chloroplast genome is maternally inherited<br />

(Mogensen, 1996). When seed-mediated gene flow frequently occurs among<br />

populations, distribution <strong>of</strong> cpDNA haplotypes will be homogeneous in spatial<br />

scale. Population genetic analyses (Fig. 2-2, Tables 2-4 and 2-5) show high<br />

levels <strong>of</strong> gene flow and low levels <strong>of</strong> population structure both <strong>with</strong>in and<br />

between populations <strong>of</strong> C. ro<strong>sea</strong> across Pacific and Indian Oceanic regions. The<br />

major haplotype <strong>of</strong> C. ro<strong>sea</strong> (H8) has an extremely wide distribution range from<br />

South Africa in the West Indian Ocean to Panama in the East Pacific Ocean (Fig.<br />

2-2). Results <strong>of</strong> Fst, MIGRATE-N and MDIV also show low genetic<br />

differentiation among Indian and West Pacific Oceanic regions (F ST = 0.19) and<br />

high number <strong>of</strong> migrants (Nm > 5 per generation, Table 2-7), respectively.<br />

Rates <strong>of</strong> gene flow between West and East Pacific regions (F ST = 0.16; Nm ≅ 1)<br />

were lower than that <strong>of</strong> between Indian and West Pacific which may due to<br />

very long distance between populations in the Pacific Ocean. According to the<br />

theoretical thoughts such amount <strong>of</strong> migrants is enough to keep the unity <strong>of</strong><br />

species in such long distances (>25000 kilometers).<br />

If we consider the results <strong>of</strong> SAMOVA, the populations in the Atlantic<br />

region are differentiated from Indo-Pacific regions (table 2-6). In the number <strong>of</strong><br />

K=3 (Table 2-6), populations <strong>of</strong> most <strong>of</strong> Atlantic and one population from<br />

Indian Ocean (Java) is grouped in two groups and rest <strong>of</strong> populations from<br />

Indo-Pacific regions are making an integrate group. The result <strong>of</strong> coalescent<br />

methods from MIGRATE-N and MDIV programs are also coherent <strong>with</strong><br />

54


SAMOVA results, as significant differentiation between Indo-Pacific oceanic<br />

regions was not detected. Hence, long distance dispersal <strong>of</strong> <strong>drifted</strong> seeds by<br />

oceanic currents appears to be the most probable explanation for the present<br />

distribution patterns <strong>of</strong> C. ro<strong>sea</strong> haplotypes <strong>with</strong>in Indo-Pacific oceanic. Overall,<br />

these results suggest that substantial gene flow occurs among populations <strong>of</strong><br />

Indo-Pacific oceanic regions by <strong>sea</strong>-<strong>drifted</strong> seeds which are impermeable in<br />

water for years (Guppy, 1906; Thiel & Gutow, 2005).<br />

Although overall haplotype diversity and population differentiation is<br />

very low between the West and East Pacific Oceanic region, however migration<br />

rates between these oceanic regions is low comparing to the other regions like<br />

between Indian and West Pacific region (Fig. 2-6, Table 2-7). A possible<br />

explanation to this outcome can be presence <strong>of</strong> very long distance between<br />

West and East Pacific populations which makes difficult dispersal <strong>of</strong> seeds by<br />

oceanic currents (Figs., 2-2, 2-6 and Table 2-7). Another interpretation can be<br />

the role <strong>of</strong> the East Pacific barrier in this region as similar results have seen in<br />

another PPSS, Ipomea pes-caprae (Wakita, unpublished) and Hibiscus tiliaceus<br />

(Takayama et al., 2006). Thus, land masses such as African and American<br />

continents not only could be barriers to the seed dispersal for C. ro<strong>sea</strong> and also<br />

other PPSS but also Ocean currents itself are as intrinsic barriers to seed<br />

dispersal which naturally are main barrier to the terrestrial <strong>plant</strong>s (Murray, 1986;<br />

Sauer, 1988; Cox & Moore, 2005; Thiel & Haye, 2006).<br />

55


A strong genetic difference between the Indo-Pacific and Atlantic populations <strong>of</strong><br />

C. ro<strong>sea</strong><br />

The results <strong>of</strong> phylogenetic (chapter 1) and population genetic analyses<br />

clearly exhibit genetic structure in the populations <strong>of</strong> Atlantic region even very<br />

limited gene flow have occurred between the Atlantic and Indian Oceanic<br />

region (H6 shared between Brazil and Tanzania populations, Figs. 2-1 and 2-2).<br />

Pairwise F ST analysis revealed presence <strong>of</strong> genetic differentiation between the<br />

West Atlantic and East Pacific region (F ST = 0.45) and between the East Atlantic<br />

and Indian Oceanic region (F ST = 0.36). Results <strong>of</strong> migration analyses also unveil<br />

that migration to the in and out <strong>of</strong> Atlantic region is restricted except some<br />

migrants from East Atlantic to the Indian Ocean (Nm=1.6; see Table 2-7). These<br />

results confirm that African and American land masses are geographical<br />

barriers to gene flow through seed dispersal by oceanic currents among C.<br />

ro<strong>sea</strong> populations, as observed in another PPSS Hibiscus tiliaceus (Takayama et<br />

al., 2006) and Ipomea pes-caprae (N. Wakita, unpublished) and also in<br />

mangrove species (Dodd et al., 2002; Duke et al., 2002; Nettel & Dodd, 2007).<br />

Inference <strong>of</strong> patterns <strong>of</strong> current or historical gene flow from gene<br />

genealogies seems straightforward when there is a sharp geographic boundary<br />

between two widely distributed clades (Irwin & Gibbs, 2002). Usually,<br />

re<strong>sea</strong>rchers assume that such boundary is the result <strong>of</strong> geographic barriers to<br />

dispersal or cryptic species boundaries. The haplotypes H16-18 are specific to<br />

Atlantic region <strong>with</strong> high probability support and Long Branch length (Fig. 2-1).<br />

Although possibility <strong>of</strong> cryptic species cannot be completely rejected and the<br />

populations in the African Oceanic regions are not kind <strong>of</strong> allopatric<br />

populations, I considered that long-term geographic barrier to dispersal is<br />

56


esponsible to such kind <strong>of</strong> phylogeographic break in the Atlantic Ocean (Fig.<br />

2-2).<br />

Directional gene flow <strong>with</strong>in Atlantic region was suggested by MIGRATE-<br />

N; as migration rates from the East to the West Atlantic were two times larger<br />

than vice versa (M EA to WA = 395.65; M WA to EA = 166.42). These results might<br />

correspond to the variation <strong>of</strong> the strength <strong>of</strong> tropical Atlantic’s major currents<br />

which regarded as transatlantic dispersal in Atlantic region (Renner, 2004).<br />

Chloroplast DNA sequences detected highly differentiated populations <strong>of</strong> C.<br />

ro<strong>sea</strong> in southern Brazil (H6, Fig. 2-2). Although there are no known barriers,<br />

the bifurcating status <strong>of</strong> the South Equatorial Current at the north-eastern horn<br />

<strong>of</strong> Brazil to the northward and southward (Fratantoni et al., 2000; Renner, 2004),<br />

appears to be potential barrier to gene flow. These opposite ocean current<br />

directions may promote the genetic differentiation <strong>of</strong> the C. ro<strong>sea</strong> populations<br />

in southern Brazil which is also the case in H. pernambucensis populations<br />

(Takayama et al., 2008).<br />

Overall, same haplotypes are distributed in the all range <strong>of</strong> Indo-Pacific<br />

regions and also over Africa (Fig. 2-2). Similar results also were found in other<br />

PPSS such as Hibiscus tiliaceus and H. pernambucensis (Takayama et al., 2006;<br />

Takayama et al., 2008) and Ipomea pes-caprae (N. Wakita, unpublished). The<br />

results suggested that this kind <strong>of</strong> haplotype distribution patterns was occurred<br />

due to long distance dispersal <strong>of</strong> <strong>sea</strong>-<strong>drifted</strong> seeds which keep the unity <strong>of</strong><br />

these species over such a long distances. However, additional data using<br />

molecular markers is necessary to assess the possibility <strong>of</strong> cryptic species in the<br />

Atlantic region for C. ro<strong>sea</strong> populations and also verify presence <strong>of</strong> the gene<br />

57


flow over Isthmus <strong>of</strong> Panama which was restricted by <strong>sea</strong>-<strong>drifted</strong> seed dispersal<br />

according to the cpDNA sequence results.<br />

58


2-5 Conclusion<br />

Chloroplast DNA sequence revealed that frequent gene flow through longdistance<br />

seed dispersal occurs over the Pacific and Indian Oceanic regions and<br />

also <strong>with</strong>in Atlantic region. Chloroplast DNA sequence also revealed partial<br />

gene flow has detected between Atlantic and Indian oceanic regions which<br />

suggest that the unity <strong>of</strong> the species in global scale is kept by long distance<br />

seed dispersal by ocean currents. The results detected highly differentiated<br />

populations <strong>of</strong> C. ro<strong>sea</strong> overall Atlantic region. This suggests that African and<br />

American land masses played as geographical barriers to gene flow by <strong>sea</strong>dispersal.<br />

Gene flow across the extremely wide distribution range <strong>of</strong> C. ro<strong>sea</strong> was<br />

kept by long distance seed dispersal through oceanic currents. In terms <strong>of</strong> long<br />

distance dispersal, this scale <strong>of</strong> gene flow from such widely distributed range <strong>of</strong><br />

<strong>plant</strong> species is a new and unique case. And it shows the significance <strong>of</strong> long<br />

distance seed dispersal to keep species integration in worldwide distributed<br />

populations.<br />

Although <strong>sea</strong> currents enable gene flow over the global distribution<br />

range <strong>of</strong> <strong>plant</strong>s <strong>with</strong> <strong>sea</strong>-<strong>drifted</strong> seeds, our study detected highly differentiated<br />

populations in southern Brazil which might be the result <strong>of</strong> bidirectional<br />

current <strong>of</strong> South Equatorial Current. Finally, the results provide good evidence<br />

for transatlantic long-distance seed dispersal by <strong>sea</strong> currents in Atlantic region.<br />

Seed dispersal prevented in the Isthmus <strong>of</strong> Panama and caused a distinct<br />

genetic difference in the cpDNA haplotype distribution between the Pacific and<br />

Atlantic populations <strong>of</strong> C. ro<strong>sea</strong>.<br />

59


Taxon Oceanic Region Locality N.<br />

C. ro<strong>sea</strong> (Sw.) DC. H1 H2 H3 H4 H5 H6 H7 H8 H9 H10 H11 H12 H13 H14 H15 H16 H17<br />

Indian Ocean South Africa Umdloti ‐29.67 31.12 22 8 14<br />

Tanzania Dar es salaam ‐6.82 39.32 25 4 21<br />

Sri Lanka Wattala, Negambo 6.97 79.89 11 10 1<br />

Thailand Phuket, Kmala Beach 7.99 98.29 10 10<br />

Indonesia Bali ‐1.08 100.42 3 3<br />

Java ‐8.73 115.17 10 10<br />

Sumatra ‐2.91 104.71 10 10<br />

Australia Darwin ‐12.44 130.83 10 10<br />

Headland Harbour ‐20.31 118.58 10 10<br />

West Pacific Thailand Kho Samui 9.51 100.06 9 9<br />

Taiwan Houpihu 38.99 117.56 15 15<br />

Singapore Singapore 1.35 103.99 9 9<br />

Philippine Quezon, Luzon 16.44 120.33 10 10<br />

Australia Queensland ‐16.73 145.66 10 8 2<br />

Japan Iriomote 24.40 123.76 15 15<br />

New Caledonia Plage de poe ‐21.61 165.39 10 10<br />

Fiji Korotogo ‐18.17 177.54 8 8<br />

Tonga Sopu ‐21.13 ‐175.21 10 10<br />

Samoa Samoa ‐13.86 ‐171.71 10 10<br />

Marquises Taipivai, Nuku Hiva ‐8.93 ‐140.09 10 10<br />

East Pacific Mexico Nayarit 21.58 ‐105.28 10 5 5<br />

Sinaloa 25.44 ‐108.73 4 4<br />

Oaxaca 15.94 ‐97.42 10 10<br />

Costa Rica Jaco Beach 9.61 ‐84.63 11 11<br />

Panama Veracruz 8.89 ‐79.62 10 10<br />

Ecuador Isla Jambel 0.74 ‐79.20 10 10<br />

West Atlantic Mexico Coatzcoalcos 18.14 ‐94.41 10 2 8<br />

Costa Rica Puerto Viejo 10.75 ‐83.56 5 5<br />

Panama Pina, Colon 1 9.28 ‐80.05 10 7 2 1<br />

Cuango, Colon 2 9.55 -79.31 10 6 4<br />

Brazil Para ‐1.32 ‐46.40 11 7 4<br />

Pernanbuco ‐8.34 ‐34.95 10 10<br />

Rio De Janeiro 1, Arraial do Cabo ‐22.97 ‐42.03 10 10<br />

Rio De Janeiro 2, Recreio ‐23.02 ‐43.44 9 1 8<br />

East Atlantic Senegal Joal-Fadiout 14.17 ‐16.85 31 7 3 4 17<br />

Ghana Busua beach 4.96 ‐1.73 18 5 10 3<br />

Angola Musul, Luanda ‐8.95 13.06 30 30<br />

Subtotal 436 7 0 9 9 10 32 45 232 5 10 14 1 2 4 2 25 29<br />

C. cathartica Thouars Pacific Philippine Atimona 10 10<br />

Samoa Samoa 2 2<br />

Tonga Tonga 5 5<br />

Tahiti Tahiti 1 1<br />

Hawaii Kauai 7 7<br />

Subtotal 25 6 19<br />

C. lineata (Thunb.) DC. Pacific Taiwan Maopi Tao 11 11<br />

Japan Ishigaki 11 11<br />

Miyazaki 10 10<br />

Ogasawara 10 10<br />

Subtotal 42 42<br />

C. sericea A. Gray Pacific Tonga Haashini-Lavengatonga 10 10<br />

Hawaii Maui, Bishop Museum 2 2<br />

Subtotal 12 12<br />

Table 2‐1. Chloroplast DNA haplotype composition and coordinates <strong>of</strong> populations for<br />

our study group. Haplotypes ranged from H1 to H17 and N, number <strong>of</strong> samples<br />

in each population.<br />

60


Primer name Length (bp) Primer Pairs Sequence (5'- 3') Source<br />

atpB partial 590 atpB-146 TTGGTACCATCCAACCAATTC This study<br />

ndhD partial 396 ndhD ACATCCGTCCCAAGGTATCA This study<br />

ndhE CAACTCGTATCAACCAATCGAA This study<br />

psbA 250 psbA CGAAGCTCCATCTACAAATGG Hamilton 1998<br />

rps16 partial 186 rps16-610 CCTTTGAGTTATCGGGTTGC This study<br />

trnK 5' flanking 250 trnK5' F GAATGGAAAAAGTAGCATGTCG This study<br />

trnK5' R TGCGATACGATCAAAACAGG This study<br />

trnK 3' flanking 340 trnK3' F AAAGGTCGGATTTGGTATTTAGA This study<br />

trnK3' R TCCTGAATCCCAACTCTTATTACAT This study<br />

Table 2‐2. List <strong>of</strong> primers used for sequencing <strong>of</strong> chloroplast regions in population<br />

samples. PCR amplification primers are shown in table 1‐2.<br />

61


Taxon<br />

Population's<br />

haplotypes Haplotype frequency atp B‐rbc L ndh D‐ndh E psb A‐trn H rps 16 trn K 5' trnK 3'<br />

256 510 267 644 23 155 183 308 682 733 748 155 156 2211 2290<br />

C. ro<strong>sea</strong> C1 7 C G G C 0 A T 2= 0 T T T C G A<br />

C. catharca, C. sericea C2 18 T . . . . . . 2= 0 C . . . T .<br />

C. ro<strong>sea</strong> C3 9 T . . . . . . 2= 0 . C . . . .<br />

C. ro<strong>sea</strong> C4 9 T . A . . . . 2= 0 . . . A . .<br />

C. ro<strong>sea</strong> C5 10 T . . . . . . 2= 0 . . G A T T<br />

C. ro<strong>sea</strong> C6 32 T . . . . . . 2= 0 . . . A T T<br />

C. ro<strong>sea</strong> C7 45 T . . . . . . 2= 0 . . . . . .<br />

C. ro<strong>sea</strong>, C. lineata, C. catharca C8 293 T . . . . . . 2= 0 . . . . T .<br />

C. ro<strong>sea</strong> C9 5 T . . T . . . 2= 0 . . . . . .<br />

C. ro<strong>sea</strong> C10 10 T . . . . T . 2= 0 . . . . T .<br />

C. ro<strong>sea</strong> C11 14 T . . . 1$ . . 2= 0 . . . . T .<br />

C. ro<strong>sea</strong> C12 1 T . . . . . . 2+ 0 . C . . . .<br />

C. ro<strong>sea</strong> C13 2 T . . . . . . 2+ 0 . . . . T .<br />

C. ro<strong>sea</strong> C14 4 T . . . . . G 2+ 0 . . . . . .<br />

C. ro<strong>sea</strong> C15 2 T . . . . . G 2= 0 . C . . . .<br />

C. ro<strong>sea</strong> C16 25 T T . . . . G 2= 0 . . . . . .<br />

C. ro<strong>sea</strong> C17 29 T T . . . . G 2= 1# . . . . . .<br />

$ AAAAT; = AGA; + TCT; # TTATT<br />

Table 2‐3. Variable sites <strong>of</strong> short fragments <strong>of</strong> 6 cpDNA regions used to determine<br />

haplotypes in Canavalia ro<strong>sea</strong> and its related species populations. Dots (•)<br />

indicate that character states are same as for C. ro<strong>sea</strong> haplotype C1. Numbers<br />

(0, 1 and 2) in sequences indicate indels, <strong>with</strong> ‘0’ indicating absence, ‘1’<br />

presence and ‘2’ inversion.<br />

62


Regions N (h) (S) (Hd) (π)<br />

West Atlantic 75 7 8 0.7809 0.0014<br />

East Atlantic 79 6 6 0.74034 0.00087<br />

Indian Ocean 111 6 8 0.5507 0.00053<br />

West Pacific 116 2 1 0.03418 0.00002<br />

East Pacific 55 2 3 0.27879 0.00042<br />

Total 436 16* 14 0.69124 0.00086<br />

*, number <strong>of</strong> haplotypes shared between C. ro<strong>sea</strong> populations<br />

Table 2‐4. Number <strong>of</strong> haplotypes (h), Polymorphic sites (S), Haplotype diversity (Hd),<br />

Nucleotide diversity (π).<br />

63


IN WP EP WA EA<br />

IN 0 0.1967 0.0826 0.3331 0.3605<br />

WP 0.1967 0 0.1684 0.6454 0.6577<br />

EP 0.0826 0.1684 0 0.4518 0.47<br />

WA 0.3331 0.6454 0.4518 0 0.1309<br />

EA 0.3605 0.6577 0.47 0.1309 0<br />

Table 2‐5. Pairwise FST comparison between 5 geographic regions. All differentiations<br />

are significant using an exact test (P < 0.05). (WA: West Atlantic, EA: East<br />

Atlantic, In: Indian Ocean, WP: West Pacific, EP: East Pacific).<br />

64


Grouping<br />

K FCT (P < 0.05)<br />

1 2 3 4 5 6 7 8 9 10<br />

Ghana (A)<br />

Mexico (A)<br />

2 0.57801 Panama-A2(A) Rests<br />

Para (A)<br />

Senegal (A)<br />

3 0.67019<br />

Ghana (A)<br />

Mexico (A)<br />

Panama-A2<br />

Para<br />

Senegal (A)<br />

4 0.68682 Mexico-Sin (P)<br />

5 0.71847 Java (I)<br />

Pernanbuco<br />

RDJaneiro1<br />

RDJaneiro2<br />

Java (I)<br />

6 0.73819 Mexico-Nay Mexico-Sin Java (I)<br />

Ghana (A)<br />

Mexico (A)<br />

Panama-A1<br />

Panama-A2<br />

Para Senegal (A)<br />

Ghana (A)<br />

Mexico(A)<br />

Para Senegal (A)<br />

Panama-A2<br />

Rests<br />

Pernanbuco<br />

RDJaneiro1<br />

RDJaneiro2<br />

Java (I)<br />

Angola (A)<br />

Costa-A<br />

Panama-A1<br />

7 0.75362 Darwin (I) Mexico-Nay Mexico-Sin Java (I)<br />

Mexico-Sin<br />

Ghana (A) Mexico(A)<br />

Para Senegal (A)<br />

Panama-A2<br />

Rests<br />

Pernanbuco<br />

RDJaneiro1<br />

RDJaneiro2<br />

Angola (A)<br />

Costa-A<br />

Panama-A1<br />

8 0.76387 Panama-A2 Darwin (I) Mexico-Nay Mexico-Sin Java (I)<br />

Ghana (A) Mexico(A)<br />

Para Senegal (A)<br />

Panama-A2<br />

Rests<br />

Pernanbuco<br />

RDJaneiro1<br />

RDJaneiro2<br />

Angola (A) Costa-A<br />

Panama-A1<br />

Ghana (A) Mexico(A)<br />

Para Senegal (A)<br />

Rests<br />

Pernanbuco<br />

RDJaneiro1<br />

RDJaneiro3<br />

Angola (A) Costa-A<br />

Panama-A1<br />

9 0.76322 Angola (A) Panama-A2 Darwin (I) Mexico-Nay Mexico-Sin Java (I) Costa-A Panama-A1<br />

Rests<br />

Pernanbuco<br />

RDJaneiro1<br />

RDJaneiro3<br />

Ghana (A) Mexico(A)<br />

Para<br />

Senegal (A)<br />

10 0.77158 South Africa (A) Angola (A) Senegal (A) Panama-A2 Darwin (I) Mexico-Nay Mexico-Sin Java (I) Costa-A Panama-A1<br />

Rests<br />

Pernanbuco<br />

RDJaneiro1<br />

RDJaneiro2<br />

Ghana (A)<br />

Mexico(A)<br />

Para<br />

Rests<br />

Pernanbuco<br />

RDJaneiro1<br />

RDJaneiro2<br />

Rests<br />

Table 2‐6. Fixation indices (FCT) <strong>of</strong> C. ro<strong>sea</strong> population groupings obtained from SAMOVA (Dupanloup et<br />

al., 2002) as a function <strong>of</strong> the user‐defined number K <strong>of</strong> groups <strong>of</strong> populations. Bold populations in<br />

the grouping indicate the newly separated populations at a given level <strong>of</strong> K.<br />

65


0.05 MLE 0.95 Nm<br />

θWA 0.0133 0.0163 0.0202<br />

θEA 0.0122 0.0144 0.0171<br />

θIn 0.0131 0.0154 0.0184<br />

θWP 0.006 0.0069 0.0081<br />

θEP 0.0071 0.009 0.0119<br />

MEA to WA 302.43 395.65 506.36 6.4491<br />

MEP to WA 1.11E‐07 1.27E‐07 13.0875 0<br />

MWA to EA 119.79 166.42 223.96 2.3964<br />

M In to EA 1.75E‐08 1.99E‐08 8.0529 0<br />

MEA to In 52.7402 105.53 185.43 1.6252<br />

MWP to In 195.14 296.22 422.56 4.5618<br />

MIn to WP 82.1782 146.84 238.55 1.0132<br />

MEP to WP 14.2155 16.2463 58.7398 0.1121<br />

MWP to EP 41.8149 96.8064 186.78 0.8713<br />

MWA to EP 8.31E‐14 9.36E‐14 26.2257 0<br />

Table 2‐7. Maximum likelihood estimates (MLE) and 95% confidence interval <strong>of</strong> θ and<br />

migration rates (M) and number <strong>of</strong> migrants (Nm) obtained from MIGRATE‐N<br />

for 5 regional groups (WA: West Atlantic, EA: East Atlantic, In: Indian Ocean,<br />

WP: West Pacific, EP: East Pacific).<br />

66


Figure 2‐1. Statistical parsimony networks <strong>of</strong> 21 haplotypes (H1‐21) based on full<br />

length cpDNA sequences among C. ro<strong>sea</strong> and its allied species. Symbols, colors<br />

and size <strong>of</strong> each haplotype correspond to species, haplotypes could detect by<br />

partial sequence <strong>of</strong> 6 short fragments and frequencies <strong>of</strong> the corresponding<br />

haplotypes, respectively. The small white circles indicate the undetected<br />

intermediate haplotypes.<br />

67


Figure 2‐2. Distribution map <strong>of</strong> the cpDNA haplotypes identified by partial sequence<br />

(2012 bp) for 48 populations from five oceanic regional groups <strong>of</strong> C. ro<strong>sea</strong>. Pie<br />

charts represent the proportion <strong>of</strong> haplotypes in each locality and the size <strong>of</strong><br />

pie charts is proportional to sample size. Haplotypes H1‐21 corresponds to the<br />

figure 2‐1.<br />

68


1<br />

0.9<br />

0.8<br />

0.7<br />

0.6<br />

0.5<br />

0.4<br />

0.3<br />

0.2<br />

0.1<br />

0<br />

FST<br />

F<br />

FSC<br />

0 2 4 6 8 10 12<br />

Figure 2‐3. Fixation indices F obtained <strong>with</strong> the SAMOVA program (Dupanloup et al.<br />

2002) as a function <strong>of</strong> the user‐defined number K <strong>of</strong> groups <strong>of</strong> populations. F ST ,<br />

differentiation among populations; F CT , differentiation among groups <strong>of</strong><br />

populations; F SC , differentiation among populations <strong>with</strong>in groups. . All<br />

differentiations were significant (P < 0.01).<br />

69


Figure 2‐4. Posterior density distribution <strong>of</strong> Bayesian analysis from MIGRATE‐N for<br />

mutation rates (θ). Number 1‐5 corresponds to West Atlantic, East Atlantic,<br />

Indian Ocean, West Pacific and East Pacific regions.<br />

70


Figure 2‐5. Posterior density distribution <strong>of</strong> Bayesian analysis obtained from<br />

MIGRATE‐N for migration rates (M) <strong>with</strong>in C. ro<strong>sea</strong> populations.<br />

71


12<br />

Number <strong>of</strong> Migrants (Nm)<br />

10<br />

8<br />

6<br />

4<br />

2<br />

0<br />

Figure 2‐6. Estimated number <strong>of</strong> migrants (Nm) <strong>with</strong>in different Oceanic regions<br />

obtained from Bayesian method implemented in MIGRATE‐N program.<br />

Calculation was based on 95% credible interval <strong>of</strong> θ and M values.<br />

72


Figure 2‐7. The theta and migration rates (M) posterior probability<br />

distributions between five oceanic groups <strong>of</strong> C. ro<strong>sea</strong> populations using MDIV<br />

program (Nielsen & Wakeley, 2001). Legends are as follows: WA=West Atlantic,<br />

EA=East Atlantic, IN=Indian Ocean, WP= West Pacific and EP= East Pacific.<br />

73


GENERAL DISCUSSION<br />

In this study, I reported the results <strong>of</strong> phylogenetic analyses among Canavalia<br />

ro<strong>sea</strong>, a genuine member PPSS and its related species based on cpDNA and<br />

nrDNA ITS sequences and also the result <strong>of</strong> phylogeographic analyses from a<br />

large survey <strong>of</strong> cpDNA variations among populations <strong>of</strong> C. ro<strong>sea</strong>.<br />

This phylogeographic study <strong>of</strong> a genuine PPSS, suggested several<br />

interesting findings on a single species that keeps its distribution range by long<br />

distance seed dispersal by <strong>sea</strong>-<strong>drifted</strong> seeds. Comparing my results <strong>with</strong> the<br />

studies <strong>of</strong> sub PPSS, H. tiliaceus and its allies, will give us a comprehensive idea<br />

about the evolutionary history <strong>of</strong> PPSS.<br />

I. Closely related species are speciated in the wide distribution range<br />

<strong>of</strong> PPSS. Although we did not have strong evidence <strong>of</strong> speciation<br />

in Canavalia ro<strong>sea</strong>, the widest range <strong>of</strong> distribution and high<br />

haplotype diversity <strong>of</strong> this species may suggest that some species<br />

were speciated from C. ro<strong>sea</strong> as mother species. In the case <strong>of</strong> H.<br />

tiliaceus recurrent speciation from H. tiliaceus has given rise to all<br />

<strong>of</strong> its allied species.<br />

II.<br />

Substantial gene flow over wide range <strong>of</strong> distribution is common<br />

in both H. tiliaceus and C. ro<strong>sea</strong>, especially over Indo-Pacific region.<br />

III.<br />

In H. tiliaceus and H. pernambucensis, two species boundary are<br />

present: one is the East Pacific and the other is the Atlantic Ocean.<br />

It is interesting that both <strong>of</strong> them are oceanic barriers. In H.<br />

pernambucensis in addition, American continents are clear barrier<br />

74


to shape a clear geographic structure between Pacific and Atlantic<br />

regions. In C. ro<strong>sea</strong>, gene flow over East Pacific and Atlantic are<br />

observed. These would be caused by the difference <strong>of</strong> seed<br />

dispersibility between the two species group, which suggest that<br />

seed dispersibility is the key factor to be a genuine PPSS.<br />

IV.<br />

The land barriers <strong>of</strong> Africa and American Continents are common<br />

for the both PPSS species which support that <strong>sea</strong>-<strong>drifted</strong> seed<br />

dispersal is limited by land masses.<br />

V. The quite different point between the two studies was the genetic<br />

diversity among populations in Pacific and Atlantic regions.<br />

Haplotype diversity in the Pacific region for the H. tiliaceus was<br />

much higher than that for C. ro<strong>sea</strong>. On the other hand, in the<br />

Atlantic region haplotype diversity in C. ro<strong>sea</strong> populations was<br />

significantly higher than H. tiliaceus populations. Such interesting<br />

results show different evolutionary history <strong>of</strong> these species and<br />

require additional studies to clear up such fascinating patterns.<br />

75


REFERENCES<br />

Arroyo, M.T.K. (1981) Breeding systems and pollination biology in leguminosae. Advances in legume<br />

systematics (ed. by R.M.P.a.P.H. Raven), pp. 723–770. Royal Botanic Gardens, Kew, UK.<br />

Avise, J. (2000) <strong>Phylogeography</strong>: The history and formation <strong>of</strong> species. Harvard University Press.<br />

Avise, J. (2009) <strong>Phylogeography</strong>: Retrospect and prospect. Journal <strong>of</strong> Biogeography, 36, 3‐15<br />

Baldwin, B.G. & Wagner, W.L. (2010) Hawaiian angiosperm radiations <strong>of</strong> north american origin.<br />

Annals <strong>of</strong> Botany, 105, 849‐879<br />

Beerli, P. (2008) Migrate version 3.0: A maximum likelihood and bayesian estimator <strong>of</strong> gene flow<br />

using the coalescent. In: Distributed over the Internet at http://popgen.scs.edu/migrate.html<br />

Beerli, P. & Felsenstein, J. (1999) Maximum‐likelihood estimation <strong>of</strong> migration rates and effective<br />

population numbers in two populations using a coalescent approach. Genetics, 152, 763‐773<br />

Beerli, P. & Felsenstein, J. (2001) Maximum likelihood estimation <strong>of</strong> a migration matrix and effective<br />

population sizes in n subpopulations by using a coalescent approach. Proceedings <strong>of</strong> the<br />

National Academy <strong>of</strong> Sciences <strong>of</strong> the United States <strong>of</strong> America, 98, 4563‐4568<br />

Bohonak, A.J. (1999) Dispersal, gene flow, and population structure. The Quarterly Review <strong>of</strong> Biology,<br />

74, 21<br />

Bowie, R., Fjeldså, J., Hackett, S., Bates, J. & Crowe, T. (2006) Coalescent models reveal the relative<br />

roles <strong>of</strong> ancestral polymorphism, vicariance, and dispersal in shaping phylogeographical<br />

structure <strong>of</strong> an african montane forest robin. Molecular Phylogenetics and Evolution, 38, 171‐<br />

188<br />

Brito, P. & Edwards, S. (2009) Multilocus phylogeography and phylogenetics using sequence‐based<br />

markers. Genetica, 135, 439‐455<br />

Brito, P.H. (2005) The influence <strong>of</strong> pleistocene glacial refugia on tawny owl genetic diversity and<br />

phylogeography in western europe. Molecular Ecology, 14, 3077‐3094<br />

Bull, J.J., Huelsenbeck, J.P., Cunningham, C.W., Sw<strong>of</strong>ford, D.L. & Waddell, P.J. (1993) Partitioning and<br />

combining data in phylogenetic analysis. Systematic Biology, 42, 384‐397<br />

Cain, M., Damman, H. & Muir, A. (1998) Seed dispersal and the holocene migration <strong>of</strong> woodland<br />

herbs. Ecological Monographs, 68, 325‐347<br />

Cain, M.L., Milligan, B.G. & Strand, A.E. (2000) Long‐distance seed dispersal in <strong>plant</strong> populations.<br />

American Journal <strong>of</strong> Botany 87, 1217‐1227<br />

Calonje, M., Martín‐Bravo, S., Dobeš, C., Gong, W., Jordon‐Thaden, I., Kiefer, C., Kiefer, M., Paule, J.,<br />

Schmickl, R. & Koch, M. (2009) Non‐coding nuclear DNA markers in phylogenetic<br />

reconstruction. Plant Systematics and Evolution, 282, 257‐280<br />

Carbone, I., Liu, Y.‐C., Hillman, B.I. & Milgroom, M.G. (2004) Recombination and migration <strong>of</strong><br />

cryphonectria hypovirus 1 as inferred from gene genealogies and the coalescent. Genetics,<br />

166, 1611‐1629<br />

Carlquist, S. (1966) The biota <strong>of</strong> long‐distance dispersal. Iii. Loss <strong>of</strong> dispersibility in the hawaiian flora.<br />

Brittonia, 18, 310‐335<br />

Carlquist, S. (1974) Loss <strong>of</strong> dispersibility in island <strong>plant</strong>s. Island biology, pp. 429‐486. Columbia<br />

University Press, New York.<br />

Clement, M., Posada, D. & Crandall, K. (2000) Tcs: A computer program to estimate gene genealogies.<br />

Molecular Ecology, 9, 1657‐1659<br />

Corriveau, J.L. & Coleman, A.W. (1988) Rapid screening method to detect potential biparental<br />

inheritance <strong>of</strong> plastid DNA and results for over 200 angiosperm species. American Journal <strong>of</strong><br />

Botany, 75, 1443‐1458<br />

76


Cox, C. & Moore, P. (2005) Biogeography: An ecological and evolutionary approach (2005), Seventh<br />

edn. Blackwell Publishing Limited.<br />

Cronn, R.C., Small, R.L., Haselkorn, T. & Wendel, J.F. (2002) Rapid diversification <strong>of</strong> the cotton genus<br />

(gossypium: Malvaceae) revealed by analysis <strong>of</strong> sixteen nuclear and chloroplast genes.<br />

American Journal <strong>of</strong> Botany, 89, 707‐725<br />

Darwin, C. (1859) On the origin <strong>of</strong> species by means <strong>of</strong> natural selection, or the preservation <strong>of</strong><br />

favoured races in the struggle for life.<br />

De Queiroz, L., Fortunato, R. & Giulietti, A. (2003) Phylogeny <strong>of</strong> the diocleinae (papilionoideae:<br />

Phaseoleae) based on morphological characters. Advances in legume systematics, 10, 303–<br />

324<br />

Dodd, R.S., Afzal‐Rafii, Z., Kashani, N. & Budrick, J. (2002) Land barriers and open oceans: Effects on<br />

gene diversity and population structure in avicennia germinans l. (avicenniaceae). Molecular<br />

Ecology, 11, 1327‐1338<br />

Donnelly, M.J., Pinto, J., Girod, R., Besansky, N.J. & Lehmann, T. (2004) Revisiting the role <strong>of</strong><br />

introgression vs shared ancestral polymorphisms as key processes shaping genetic diversity in<br />

the recently separated sibling species <strong>of</strong> the anopheles gambiae complex. Heredity, 92, 61‐68<br />

Doyle, J. & Doyle, J. (1987) A rapid DNA isolation procedure for small quantities <strong>of</strong> fresh leaf tissue.<br />

Phytochemistry, 19, 11‐15<br />

Duke, N., Lo, E. & Sun, M. (2002) Global distribution and genetic discontinuities <strong>of</strong> mangrovesemerging<br />

patterns in the evolution <strong>of</strong> rhizophora. Trees‐Structure and Function, 16, 65‐79<br />

Dupanloup, I., Schneider, S. & Exc<strong>of</strong>fier, L. (2002) A simulated annealing approach to define the<br />

genetic structure <strong>of</strong> populations. Molecular Ecology, 11, 2571‐2581<br />

Ennos, R.A. (1994) Estimating the relative rates <strong>of</strong> pollen and seed migration among <strong>plant</strong><br />

populations. Heredity, 72, 250‐259<br />

Exc<strong>of</strong>fier, L. (2004) Patterns <strong>of</strong> DNA sequence diversity and genetic structure after a range expansion:<br />

Lessons from the infinite‐island model. Molecular Ecology, 13, 853‐864<br />

Exc<strong>of</strong>fier, L. & Heckel, G. (2006) Computer programs for population genetics data analysis: A survival<br />

guide. Nature Reviews Genetics, 7, 745‐758<br />

Exc<strong>of</strong>fier, L., Laval, G. & Schneider, S. (2005) Arlequin (version 3.0): An integrated s<strong>of</strong>tware package<br />

for population genetics data analysis. Evolutionary Bioinformatics Online, 1, 47–50<br />

Felsenstein, J. (1985) Confidence limits on phylogenies: An approach using the bootstrap. Evolution,<br />

39, 783‐791<br />

Fratantoni, D.M., Johns, W.E., Townsend, T.L. & Hurlburt, H.E. (2000) Low‐latitude circulation and<br />

mass transport pathways in a model <strong>of</strong> the tropical atlantic ocean. Journal <strong>of</strong> Physical<br />

Oceanography, 30, 1944‐1966<br />

Gross, C. (1993) The reproductive ecology <strong>of</strong> canavalia ro<strong>sea</strong> (fabaceae) on anak krakatau, indonesia.<br />

Australian Journal <strong>of</strong> Botany, 41, 591‐599<br />

Guppy, H.B. (1906) Observations <strong>of</strong> a naturalist in the pacific between 1896 and 1899 , vol. 2. Plant<br />

dispersal. MacMillan, London.<br />

Hamrick, J. & Nason, J. (1996) Consequences <strong>of</strong> dispersal in <strong>plant</strong>s. Population dynamics in ecological<br />

space and time, pp. 203–236. University <strong>of</strong> Chicago Press.<br />

Hasegawa, M., Kishino, H. & Yano, T. (1985) Dating <strong>of</strong> the human‐ape splitting by a molecular clock <strong>of</strong><br />

mitochondrial DNA. Journal <strong>of</strong> molecular evolution, 22, 160‐174<br />

Heuertz, M., Hausman, J., Hardy, O., Vendramin, G., Frascaria‐Lacoste, N. & Vekemans, X. (2004)<br />

Nuclear microsatellites reveal contrasting patterns <strong>of</strong> genetic structure between western and<br />

southeastern european populations <strong>of</strong> the common ash (fraxinus excelsior l.). Evolution, 58,<br />

976‐988<br />

77


Hey, J. (2005) On the number <strong>of</strong> new world founders: A population genetic portrait <strong>of</strong> the peopling <strong>of</strong><br />

the americas. PLoS Biology, 3, e193<br />

Heywood, J.S. (1991) Spatial analysis <strong>of</strong> genetic variation in <strong>plant</strong> populations. Annual Review <strong>of</strong><br />

Ecology and Systematics, 22, 335‐355<br />

Howe, H.F. & Smallwood, J. (1982) Ecology <strong>of</strong> seed dispersal. Annual Review <strong>of</strong> Ecology and<br />

Systematics, 13, 201‐228<br />

Ingvarsson, P.K., Ribstein, S. & Taylor, D.R. (2003) Molecular evolution <strong>of</strong> insertions and deletion in<br />

the chloroplast genome <strong>of</strong> silene. Molecular Biology and Evolution, 20, 1737‐1740<br />

Irwin, D.E. & Gibbs, H.L. (2002) Phylogeographic breaks <strong>with</strong>out geographic barriers to gene flow.<br />

Evolution, 56, 2383‐2394<br />

Kane, N., King, M., Barker, M., Raduski, A., Karrenberg, S., Yatabe, Y., Knapp, S. & Rieseberg, L. (2009)<br />

Comparative genomic and population genetic analyses indicate highly porous genomes and<br />

high levels <strong>of</strong> gene flow between divergent helianthus species. Evolution, 63, 2061‐2075<br />

Kay, K.M., Reeves, P.A., Olmstead, R.G. & Schemske, D.W. (2005) Rapid speciation and the evolution<br />

<strong>of</strong> hummingbird pollination in neotropical costus subgenus costus (costaceae): Evidence from<br />

nrdna its and ets sequences. American Journal <strong>of</strong> Botany, 92, 1899‐1910<br />

Kelchner, S.A. (2000) The evolution <strong>of</strong> non‐coding chloroplast DNA and its application in <strong>plant</strong><br />

systematics. Annals <strong>of</strong> the Missouri Botanical Garden, 87, 482‐498<br />

Kimura, M. (1980) A simple method for estimating evolutionary rates <strong>of</strong> base substitutions through<br />

comparative studies <strong>of</strong> nucleotide sequences. Journal <strong>of</strong> molecular evolution, 16, 111‐120<br />

Kimura, M. & Weiss, G. (1964) The stepping stone model <strong>of</strong> population structure and the decrease <strong>of</strong><br />

genetic correlation <strong>with</strong> distance. Genetics, 49, 561‐576<br />

Kingman, J. (1982a) The coalescent. Stochastic processes and their applications, 13, 235‐248<br />

Kingman, J. (1982b) On the genealogy <strong>of</strong> large populations. Journal <strong>of</strong> Applied Probability, 19, 27‐43<br />

Knowles, L.L. (2009) Statistical phylogeography. Annual Review <strong>of</strong> Ecology, Evolution, and Systematics,<br />

40, 593‐612<br />

Knowles, L.L. & Maddison, W.P. (2002) Statistical phylogeography. Molecular Ecology, 11, 2623‐2635<br />

Kuhner, M.K. (2009) Coalescent genealogy samplers: Windows into population history. Trends in<br />

Ecology & Evolution, 24, 86‐93<br />

Kuhner, M.K. & Smith, L.P. (2007) Comparing likelihood and bayesian coalescent estimation <strong>of</strong><br />

population parameters. Genetics, 175, 155‐165<br />

Lackey, J. (1981) Phaseoleae. Advances in legume systematics (ed. by R.P.a.P. Raven), pp. 301‐327.<br />

Royal Botanic Gardens, Kew, London.<br />

Levin, D.A. (2000) The origin, expansion and demise <strong>of</strong> <strong>plant</strong> species. Oxford University Press.<br />

Levin, D.A. (2001) The recurrent origin <strong>of</strong> <strong>plant</strong> races and species. Systematic Botany, 26, 197‐204<br />

Librado, P. & Rozas, J. (2009) Dnasp v5: A s<strong>of</strong>tware for comprehensive analysis <strong>of</strong> DNA polymorphism<br />

data. In: Bioinformatics, pp. 1451‐1452<br />

Linhart, Y.B. & Grant, M.C. (1996) Evolutionary significance <strong>of</strong> local genetic differentiation in <strong>plant</strong>s.<br />

Annual Review <strong>of</strong> Ecology and Systematics, 27, 237‐277<br />

Loewer, P. (2005) Seed dispersal. In: Seeds: The definitive guide to growing, history, and lore. Pp. 61‐<br />

74. Timber Press.<br />

Mayr, E. (1963) Animal species and evolution. Belknap Press.<br />

Mogensen, H.L. (1996) The hows and whys <strong>of</strong> cytoplasmic inheritance in seed <strong>plant</strong>s. American<br />

Journal <strong>of</strong> Botany, 83, 383‐404<br />

Morjan, C.L. & Rieseberg, L.H. (2004) How species evolve collectively: Implications <strong>of</strong> gene flow and<br />

selection for the spread <strong>of</strong> advantageous alleles. Molecular Ecology, 13, 1341‐1356<br />

78


Mort, M.E., Archibald, J.K., Randle, C.P., Levsen, N.D., O'leary, T.R., Topalov, K., Wiegand, C.M. &<br />

Crawford, D.J. (2007) Inferring phylogeny at low taxonomic levels: Utility <strong>of</strong> rapidly evolving<br />

cpdna and nuclear its loci. American Journal <strong>of</strong> Botany, 94, 173‐183<br />

Murray, D.R. (1986) Seed dispersal by water. Seed dispersal, pp. 49‐85. Academic Press, Sydney.<br />

Nakanishi, H. (1988) Dispersal ecology <strong>of</strong> the maritime <strong>plant</strong>s in the ryukyu islands, japan. Ecological<br />

Re<strong>sea</strong>rch, 3, 163‐173<br />

Nettel, A. & Dodd, R.S. (2007) Drifting propagules and receding swamps: Genetic footprints <strong>of</strong><br />

mangrove recolonization and dispersal along tropical coasts. Evolution, 61, 958‐971<br />

Nielsen, R. & Wakeley, J. (2001) Distinguishing migration from isolation: A markov chain monte carlo<br />

approach. Genetics, 158, 885‐896<br />

Nylander, J. (2004) Mrmodeltest v2. Program distributed by the author. In: Evolutionary Biology<br />

Centre, Uppsala University<br />

Olmstead, R. & Palmer, J. (1994) Chloroplast DNA systematics: A review <strong>of</strong> methods and data analysis.<br />

American Journal <strong>of</strong> Botany, 81, 1205‐1224<br />

Ouborg, N.J., Piquot Y. & Van Groenendael J. M. (1999) Population genetics, molecular markers and<br />

the study <strong>of</strong> dispersal in <strong>plant</strong>s. Journal <strong>of</strong> Ecology, 87, 551‐568<br />

Parker, P.G., Allison, A.S., Schug, M.D., Booton, G.C. & Fuerst, P.A. (1998) What molecules can tell us<br />

about populations: Choosing and using a molecular marker. Ecology, 79, 361‐382<br />

Petit, R., Duminil, J., Fineschi, S., Hampe, A., Salvini, D. & Vendramin, G. (2005) Comparative<br />

organization <strong>of</strong> chloroplast, mitochondrial and nuclear diversity in <strong>plant</strong> populations.<br />

Molecular Ecology, 14, 689‐701<br />

Polidoros, A., Pasentsis, K. & Tsaftaris, A. (2006) Rolling circle amplification‐race: A method for<br />

simultaneous isolation <strong>of</strong> 5'and 3'cdna ends from amplified cdna templates. BioTechniques,<br />

41, 35<br />

Rambaut, A. (2002) Se‐al: A manual sequence alignment editor v2.0a11.<br />

Rambaut, A. & Drummond, A. (2009) Tracer v1.5. Available<br />

from:. In:<br />

Rannala, B. & Hartigan, J.A. (1996) Estimating gene flow in island populations. Genetics Re<strong>sea</strong>rch, 67,<br />

147‐158<br />

Raymond, M. & Rousset, F. (1995) An exact test for population differentiation. Evolution, 49, 1280‐<br />

1283<br />

Renner, S. (2004) Plant dispersal across the tropical atlantic by wind and <strong>sea</strong> currents. International<br />

Journal <strong>of</strong> Plant Sciences, 165, S23‐S33<br />

Ronquist, F. & Huelsenbeck, J.P. (2003) Mrbayes 3: Bayesian phylogenetic inference under mixed<br />

models. Bioinformatics, 19, 1572‐1574<br />

Sauer, J. (1964) Revision <strong>of</strong> canavalia. Brittonia, 16, 106‐181<br />

Sauer, J. (1988) Plant migration: The dynamics <strong>of</strong> geographic patterning in seed <strong>plant</strong> species.<br />

University <strong>of</strong> California Press, Berkeley, California, USA.<br />

Schaal, B.A., Hayworth, D.A., Olsen, K.M., Rauscher, J.T. & Smith, W.A. (1998) Phylogeographic<br />

studies in <strong>plant</strong>s: Problems and prospects. Molecular Ecology, 7, 465‐474<br />

Scherson, R., Choi, H., Cook, D. & Sanderson, M. (2005) Phylogenetics <strong>of</strong> new world astragalus:<br />

Screening <strong>of</strong> novel nuclear loci for the reconstruction <strong>of</strong> phylogenies at low taxonomic levels.<br />

Brittonia, 57, 354‐366<br />

Schrire, B.D. (2005) Phaseoleae. Legumes <strong>of</strong> the world (ed. by B.S. Gwilym Lewis, Barbara Mackinder,<br />

and Mike Lock), pp. 393‐431. Kew Publishing.<br />

Shaw, J., Lickey, E.B., Schilling, E.E. & Small, R.L. (2007) Comparison <strong>of</strong> whole chloroplast genome<br />

sequences to choose noncoding regions for phylogenetic studies in angiosperms: The tortoise<br />

and the hare iii. American Journal <strong>of</strong> Botany, 94, 275‐288<br />

79


Simmons, M. & Ochoterena, H. (2000) Gaps as characters in sequence‐based phylogenetic analyses.<br />

Systematic Biology, 49, 369<br />

Slatkin, M. & Hudson, R. (1991) Pairwise comparisons <strong>of</strong> mitochondrial DNA sequences in stable and<br />

exponentially growing populations. Genetics, 129, 555<br />

Small, R.L., Ryburn, J.A., Cronn, R.C., Seelanan, T. & Wendel, J.F. (1998) The tortoise and the hare:<br />

Choosing between noncoding plastome and nuclear adh sequences for phylogeny<br />

reconstruction in a recently diverged <strong>plant</strong> group. American Journal <strong>of</strong> Botany, 85, 1301‐1315<br />

Soltis, D.E., Morris, A.B., Mclachlan, J.S., Manos, P.S. & Soltis, P.S. (2006) Comparative<br />

phylogeography <strong>of</strong> unglaciated eastern north america. Molecular Ecology, 15, 4261‐4293<br />

Soltis, D.E. & Soltis, P.S. (1998) Choosing an approach and appropriate gene for phylogenetic analysis.<br />

Molecular systematics <strong>of</strong> <strong>plant</strong>s ii. DNA sequencing (ed. by D.E. Soltis, P.S. Soltis and J.J.<br />

Doyle), pp. 1‐42. Kluwer Academic Publishers.<br />

St. John H (1970) Revision <strong>of</strong> the hawaiian species <strong>of</strong> canavalia (leguminosae). Hawaiian <strong>plant</strong> studies.<br />

Israel Journal <strong>of</strong> Botany, 19, 161‐219<br />

Stamatakis, A. (2006) Raxml‐vi‐hpc: Maximum likelihood‐based phylogenetic analyses <strong>with</strong> thousands<br />

<strong>of</strong> taxa and mixed models. Bioinformatics, 22, 2688‐2690<br />

Sw<strong>of</strong>ford, D. (2002) Paup* 4.0 beta 10: Phylogenetic analysis using parsimony (*and other methods).<br />

In. Sinauer Associates Sunderland, MA<br />

Takayama, K., Kajita, T., Murata, J. & Tateishi, Y. (2006) <strong>Phylogeography</strong> and genetic structure <strong>of</strong><br />

hibiscus tiliaceus‐ speciation <strong>of</strong> a <strong>pantropical</strong> <strong>plant</strong> <strong>with</strong> <strong>sea</strong>‐<strong>drifted</strong> seeds. Molecular Ecology,<br />

15, 2871‐2881<br />

Takayama, K., Tateishi, Y., Murata, J. & Kajita, T. (2008) Gene flow and population subdivision in a<br />

<strong>pantropical</strong> <strong>plant</strong> <strong>with</strong> <strong>sea</strong>‐<strong>drifted</strong> seeds hibiscus tiliaceus and its allied species: Evidence from<br />

microsatellite analyses. Molecular Ecology, 17, 2730‐2742<br />

Templeton, A., Crandall, K. & Sing, C. (1992) A cladistic analysis <strong>of</strong> phenotypic associations <strong>with</strong><br />

haplotypes inferred from restriction endonuclease mapping and DNA sequence data. Iii.<br />

Cladogram estimation. Genetics, 132, 619<br />

Thiel, M. & Gutow, L. (2005) The ecology <strong>of</strong> rafting in the marine environment: I. The floating<br />

substrata. Oceanography and Marine Biology: An Annual Review Volume 42, 42, 181‐264<br />

Thiel, M. & Haye, P.A. (2006) The ecology <strong>of</strong> rafting in the marine environment. Iii. Biogeographical<br />

and evolutionary consequences. Oceanography and marine biology: An annual review (ed. by<br />

R. Gibson, R. Atkinson and J. Gordon), pp. 323‐429. CRC Press.<br />

Varela, E.S., Lima, J.P.M.S., Galdino, A.S., Pinto, L.D.S., Bezerra, W.M., Nunes, E.P., Alves, M.a.O. &<br />

Grangeiro, T.B. (2004) Relationships in subtribe diocleinae (leguminosae; papilionoideae)<br />

inferred from internal transcribed spacer sequences from nuclear ribosomal DNA.<br />

Phytochemistry, 65, 59‐69<br />

Wagner, W., Herbst, D. & Sohmer, S. (1999) Manual <strong>of</strong> the flowering <strong>plant</strong>s <strong>of</strong> hawaii, revised edition.<br />

University <strong>of</strong> Hawaii Press.<br />

Wakeley, J. (2003) Inferences about the structure and history <strong>of</strong> populations: Coalescents and<br />

intraspecific phylogeography. The evolution <strong>of</strong> population biology (ed. by R. Singh and M.<br />

Uyenoyama), pp. 193–215. Cambridge University Press, Cambridge.<br />

Weir, B. & Cockerham, C. (1984) Estimating f‐statistics for the analysis <strong>of</strong> population structure.<br />

Evolution, 38, 1358‐1370<br />

Whistler, W. (1992) Flowers <strong>of</strong> the pacific island <strong>sea</strong>shore. Univ. <strong>of</strong> Hawaii Press.<br />

Wiens, J.J. (1998) Combining data sets <strong>with</strong> different phylogenetic histories. Systematic Biology, 47,<br />

568‐581<br />

Willson, M. & Traveset, A. (2000) The ecology <strong>of</strong> seed dispersal. Seeds the ecology <strong>of</strong> regeneration in<br />

<strong>plant</strong> communities 2nd edition (ed. by M. Fenner), pp. 85‐110. CAB International.<br />

80


Wright, S. (1931) Evolution in mendelian populations. Genetics, 16, 97‐159<br />

Wright, S. (1951) The genetical structure <strong>of</strong> populations. Annals <strong>of</strong> Eugenics, 15, 323‐354<br />

81


ACKNOWLEDGEMENTS<br />

I am deeply grateful to my supervisor, Pr<strong>of</strong>. Dr. Tadashi Kajita who<br />

encouraging me to do a good re<strong>sea</strong>rch, and challenge me to logical thinking.<br />

His efforts, support, and patience in accepting me as a PhD student, will never<br />

forget as my dream (study abroad) came true. I am also grateful to my cosupervisor,<br />

Pr<strong>of</strong>. Dr. Yasuyuki Watano who always remind me how to learn and<br />

Dr. Takeshi Asakawa for his valuable suggestions. I specially thank Pr<strong>of</strong>. Dr.<br />

Takayoshi Tsuchiya and Pr<strong>of</strong>. Dr. Sumiko Kimura for their suggestions and<br />

comments as referees. I also owe a great debt to Dr. Koji Takayama, who<br />

taught me molecular methods and for his valuable comments. I would like to<br />

thank Mr. Alvin Y. Yoshinaga from Center for Conservation Re<strong>sea</strong>rch and<br />

Training, Hawaii for sending materials. I would like to thank Pr<strong>of</strong>. Dr. Yoichi<br />

Tateishi for his helpful comments and allowing me to visit the herbarium <strong>of</strong><br />

University <strong>of</strong> the Ryukyus. I am also grateful to Pr<strong>of</strong>. Dr. Masaki Miya for<br />

valuable classes in phylogenetic analysis.<br />

I specially thank Dr. Akihisa Shirai, Dr. Norihisa Wakita, Dr. Bayu Adjie, for<br />

their help and suggestions. I would also like to thank to all students <strong>of</strong> Watano<br />

and Kajita laboratory for any helps during my re<strong>sea</strong>rches.<br />

And finally, thanks to my family and all friends for their love, support,<br />

and encouragement throughout the years.<br />

My study fully supported by Monbukagakusho (MEXT: Ministry <strong>of</strong><br />

Education, Culture, Sports, Science and Technology) scholarship.<br />

82


BIOGRAPHY<br />

Mohammad Vatanparast was born in Urmia, central city <strong>of</strong> West Azerbaijan<br />

province in the North West <strong>of</strong> Iran in 20 February 1976. He finished his<br />

undergraduate level in <strong>plant</strong> biology from Urmia University in 1999. In 2000 he<br />

enter master course in the <strong>plant</strong> systematic field at Tarbiat Modarres University<br />

in Tehran and graduate in 2003. His master thesis was on biosystematics <strong>of</strong> the<br />

genus Trifolium a part <strong>of</strong> national project about Flora <strong>of</strong> Iran in Persian<br />

language. As his dreams and interest was study abroad to learn novel practical<br />

and analytical methods in the <strong>plant</strong> systematic and evolution, <strong>with</strong> efforts <strong>of</strong><br />

Pr<strong>of</strong>. Dr. Tadashi Kajita, he could enter to the PhD course in his laboratory in<br />

Chiba University, Japan. In October 2006 he entered Japan and after taking 6<br />

month re<strong>sea</strong>rch student course in April 2007 he started his PhD project<br />

focusing on “<strong>Phylogeography</strong> <strong>of</strong> a <strong>pantropical</strong> <strong>plant</strong>s <strong>with</strong> <strong>sea</strong>-<strong>drifted</strong> seeds,<br />

Canavalia ro<strong>sea</strong>” by applying large population sampling using molecular<br />

methods such as chloroplast DNA sequences and population genetic analyses.<br />

83

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!