Upload
others
View
5
Download
0
Embed Size (px)
Citation preview
University of MiamiScholarly Repository
Open Access Dissertations Electronic Theses and Dissertations
2015-05-06
Evolution of Polyspora ( = Gordonia; Theaceae) inSri LankaLiyana A. A. H. GunathilakeUniversity of Miami, [email protected]
Follow this and additional works at: https://scholarlyrepository.miami.edu/oa_dissertations
This Embargoed is brought to you for free and open access by the Electronic Theses and Dissertations at Scholarly Repository. It has been accepted forinclusion in Open Access Dissertations by an authorized administrator of Scholarly Repository. For more information, please [email protected].
Recommended CitationGunathilake, Liyana A. A. H., "Evolution of Polyspora ( = Gordonia; Theaceae) in Sri Lanka" (2015). Open Access Dissertations. 1412.https://scholarlyrepository.miami.edu/oa_dissertations/1412
UNIVERSITY OF MIAMI
EVOLUTION OF POLYSPORA ( = GORDONIA; THEACEAE) IN SRI LANKA
By
Liyana Arachchilage Anuradha Himashi Gunathilake
A DISSERTATION
Submitted to the Faculty of the University of Miami
in partial fulfillment of the requirements for the degree of Doctor of Philosophy
Coral Gables, Florida
May 2015
©2015 Liyana Arachchilage Anuradha Himashi Gunathilake
All Rights Reserved
UNIVERSITY OF MIAMI
A dissertation submitted in partial fulfillment of the requirements for the degree of
Doctor of Philosophy
EVOLUTION OF POLYSPORA ( = GORDONIA; THEACEAE) IN SRI LANKA
Liyana Arachchilage Anuradha Himashi Gunathilake Approved: ________________ _________________ Barbara A. Whitlock, Ph.D. John Albert C. Uy, Ph.D. Associate Professor of Biology Associate Professor of Biology ________________ _________________ Jeffrey S. Prince, Ph.D. M. Brian Blake, Ph.D. Associate Professor of Biology Dean of the Graduate School ________________ Carla Hurt, Ph.D. Assistant Professor of Biology Tennessee Technological University Cookeville, Tennessee
GUNATHILAKE, L.A. ANURADHA H. (Ph.D., Biology)
Evolution of Polyspora (=Gordonia; Theaceae) in Sri Lanka. (May 2015) Abstract of a dissertation at the University of Miami. Dissertation supervised by Barbara A. Whitlock, Ph.D. No. of pages in text. (142)
This dissertation examined the evolutionary relationships and evolutionary history of
four endemic species of the genus Polyspora (=Gordonia) in Sri Lanka. Sri Lanka is part
of the Western Ghats-Sri Lanka biodiversity hot spot with extraordinarily high species
richness and endemism. In spite of its diversity and uniqueness, the biogeography of this
region, especially of the flora, remains grossly understudied. My research aimed to fill
this void by using the four endemic species of the genus Polyspora in Sri Lanka in
phylogenetic, biogeographic, and morphological analyses. All four species of Polyspora
in Sri Lanka are restricted to the wetzone of the country and they thereby represent the
distribution pattern of a majority of the endemics of the country. These species have
formerly been assigned to the genus Gordonia; however, recent phylogenetic evidence
indicates that Gordonia is polyphyletic and that all Asian species should be included in
either Polyspora or Laplacea. My research supports their inclusion in Polyspora, and I
follow that nomenclature here. In Chapter 1, I review the geography, vegetation and
floristics of Sri Lankan plants and present three hypotheses for their biogeographic
affinities. I review evidence for each of these hypotheses from published plant molecular
phylogenetic analyses. In Chapter 2, I use Ecological Niche Models (ENM) to test
predictions on the distribution of Polyspora from Sri Lanka and the Western Ghats region
of India, in the present, 65 years into the future, and during the last glacial maximum.
Results show reciprocal areas of suitable habitat for species from Sri Lanka and the
Western Ghats, so that Sri Lankan Polyspora could persist in the Western Ghats and vice
versa. During the last glacial maximum, suitable habitats in the two regions were isolated,
even though they covered greater area, extending into lower elevations, and despite the
landbridge connecting Sri Lanka to the mainland. Projections into the future suggest dire
conservation threats for Polyspora as the climate warms. In Chapter 3, I present a study
of comparative anatomy of seed coat micromorphology in representatives of all three
tribes of Theaceae using scanning electron microscopy. Results suggest fixed differences
in each of the three tribes, supporting the polyphyly of the formerly recognized genus
Gordonia and placement of Sri Lankan species in the tribe Theeae with Laplacea or
Polyspora. Finally, in Chapter 4, I conduct phylogenetic analyses using plastid and
mitochondrial DNA sequence that show species from Sri Lanka and India are most
closely related to species from China in the tribe Theeae. These results support the
renaming of Sri Lankan species from Gordonia to Polyspora. Analyses of three
microsatellite markers from 114 individual plants from all four currently recognized
species of Polyspora in Sri Lanka show differentiation between the morphologically
distinct P. speciosa and sympatric populations of the remaining three species. Genetic
structure also differentiates populations on the three major mountain ranges within Sri
Lanka, but does not separate three species (P. ceylanica, P. dassanayakei, and P.
elliptica) that occur on them.
! iii!
Acknowledgments
First and foremost, I thank my advisor Dr. Barbara Whitlock for her support and
guidance throughout my time as her student. I am very grateful to all the support rendered
by her from the inception of my project to its completion. Barbara was an excellent
mentor who was always available whenever I needed her counsel for academic as well as
personal matters. I was constantly inspired by her knowledge, wisdom and her
dedication. I thank her for giving me the freedom to grow academically and personally
while making sure I stayed on track. I could not have asked for a better advisor.
Next I thank the members of my Dissertation Committee; Dr. Jeff Prince, Dr.
Carla Hurt, Dr. Jun Wen and Dr. Albert Uy for all the advice and support provided. Their
expertise in the relevant fields matched well with my research plans and I thank them for
being available for me when ever I needed help and guidance.
I was lucky to have been able to collaborate with a wonderful group of people
without the help of whom, I would not have been able to complete my work. Dr. Deepthi
Yakandawala, my mentor from my undergraduate days at the University of Peradeniya in
Sri Lanka has been supporting me for a long period of time. I am especially thankful to
her for the assistance rendered in securing the required permits from relevant authorities
for conducting fieldwork in Sri Lanka. I am also very grateful to Ms. Arundhati Das of
the Asoka Trust for Research in Ecology and Environment (ATREE) of Bangalore for
facilitating my fieldwork in India. She was my collaborator, tour guide, hostess (and body
guard) during my field visit to India and made sure that I make the most out of my very
short visit. I also thank Dr. Ravikanth of ATREE for conducting molecular analysis of the
Gordonia obtusa samples collected from the Western Ghats. I also deeply acknowledge
! iv!
the help of Dr. A.H.M.A. Reza of the Delta State University of Cleveland MS from
whom I learnt the basics of Ecological Niche Modeling. I am also thankful to his wife
Selina for her kind hospitality during my stay with their family and their sons Ruhan and
Reehan for tolerating my invasion of their “territory” for one full week. Finally I thank
Dr. Dean Williams for his very generous help and advice with my microsatellite work.
Next I thank all the past and present members of the Whitlock lab for their help
and friendship. A special word of thanks is due to Wyatt Shaber who with his patient,
helpful and tolerating manner is undoubtedly the best lab mate a grad student can wish
for. I will greatly miss his companionship and the countless discussions we had, some of
which were real hard-core science. I also thank the rest of the Biograds for their help and
support.
Last but not the least, I thank my family and friends for the unwavering support,
love and tolerance throughout my career as a grad student. I thank my parents, my
brother and sis-in-law for their unconditional love and for empowering and encouraging
me to pursue my dreams. Whatever success I achieve in life will be largely due to you. I
am very grateful to my husband Hasitha for his love and support. Even though it is quite
natural for the significant others of the graduate students to make scarifies on behalf of
their loved one, I believe Hasitha had to endure much more than his fair share. Thank you
for sticking by me, for giving me freedom to achieve my dreams but never letting me
forget what is more important. I truly appreciate all that you did to keep the family
functioning while I got lost in the craziness. My darling sons Dinuka and Tharuka: thank
you for tolerating my absences during the weekends and all the rushed good byes.
Mommy loves you both to infinity and beyond. To my extended family and friends: I
! v!
could not have survived grad school without your love and support. The constant e-mails,
texts and phone calls across the globe reminded me that I have the love and the support of
the world’s best cheering squad behind me at times I needed it the most. Thank you for
believing in me.
! vi!
Table of Contents
Page
LIST OF FIGURES .......................................................................................................... vii
LIST OF TABLES ............................................................................................................. ix
CHAPTER 1 A review of the biogeographical affinities of Sri Lankan flora ..........................................1 CHAPTER 2 Ecological Niche Models support long-term isolation between the endemic species of Gordonia in Sri Lanka and the Western Ghats of India. ..................................................28 CHAPTER 3 Seed Coat Micromorphology of Gordonia sensu lato ......................................................55
CHAPTER 4 The phylogenetic relationships of Sri Lankan Polyspora (=Gordonia; Theaceae) and the genetic structuring of the genus within the country ..........................................................77
CHAPTER 5 General conclusions .........................................................................................................125 REFERENCES ................................................................................................................128
! vii!
List of Figures
CHAPTER 2 Figure 2.1: Map of southern India and Sri Lanka with elevation .....................................48 Figure 2.2: Location of Gordonia populations included in the analyses for the Western Ghats of India and the central Highlands of Sri Lanka .....................................................49 Figure 2.3: Predicted distribution of Sri Lankan species of Gordonia for Sri Lanka and southern India plus Sri Lanka under current climatic conditions .....................................50 Figure 2.4: Predicted distribution of G. obtusa under current climatic conditions, for Sri Lanka and southern India ..................................................................................................51 Figure 2.5: Predicted distribution of Sri Lankan species of Gordonia during the LGM for Sri Lanka and India plus Sri Lanka ...................................................................................52 Figure 2.6: Predicted distribution of G. obtusa during the LGM for Sri Lanka and India ............................................................................................................................................53 Figure 2.7: Predicted distribution of Sri Lankan species of Gordonia in 2080AD ..........54
CHAPTER 3
Figure 3.1: Seeds of Theaceae ..........................................................................................72 Figure 3.2: Scanning electron micrographs of seed coats of Gordonieae .........................73 Figure 3.3: Scanning electron micrographs of seed coats of Theeae ................................74 Figure 3.4: Scanning electron micrographs showing isodiametric and elongate tests cells in Theeae ...........................................................................................................................75 Figure 3.5: Scanning electron micrographs of seed coats of Stewartieae ..........................76 !CHAPTER 4
Figure 4.1: Gordonia s.l. in Sri Lanka ............................................................................113 Figure 4.2: Location of the sites where Gordonia species from Sri Lanka were collected ..........................................................................................................................................114 Figure 4.3: Distribution of the samples that were allocated to three populations under study within the central highlands of Sri Lanka ..............................................................115 Figure 4.4: Bayesian phylogenetic analysis of trnL-trnLF and matR .............................116
! viii!
Figure 4.5: Bayesian phylogenetic analysis of trnL-trnLF, matR and trnH-psbA ...........117 Figure 4.6: Haplotype diagrams generated for the combined data set of trnL-trnLF and matR and trnL-trnLF, matR and trnH-psbA ....................................................................118 Figure 4.7: Mean allelic patterns across the three populations .......................................119 Figure 4.8: Schematic diagram showing the location of the three populations within the central mountains of Sri Lanka and the pairwise Nei’s genetic distance and Nei’s genetic identity, FST and nm between them ..................................................................................120 Figure 4.9: The difference in the allelic frequencies between G. speciosa populations and the Gordonia populations in the Knuckles region containing G. elliptica and G. ceylanica ..........................................................................................................................................121 Figure 4.10: The PCA and the PCoA analyses using the genotypes of the G. speciosa populations in the Adams Peak region and the Gordonia populations in the Knuckles region ..............................................................................................................................122 Figure 4.11: The neighbor joining tree showing the clustering of the populations from Knuckles and G. speciosa from Adams Peak region as two different groups ................123 Figure 4.10: The UPGMA tree showing the clustering of the populations from Knuckles and G. speciosa from Adams Peak region as two different groups .................................124
! ix!
List of Tables
CHAPTER 1
Table 1.1: Summary of the published studies that were considered in the review ...........24
CHAPTER 3
Table 3.1: Source of seeds examined using SEM .............................................................68
CHAPTER 4
Table 4.1: Description of sequences and data matrices used in phylogenetic analyses .101 Table 4.2: Sample numbers, the groups they were assigned, and lengths of two alleles for the three microsatellite loci M2, M3 and M5. ................................................................103 Table 4.3: Different alleles and their frequencies at each locus for the entire population of Gordonia in Sri Lanka .....................................................................................................104 Table 4.4: heterozygosity, F statistics and Polymorphism at each locus for codominant data ..................................................................................................................................105 Table 4.5: Heterozygosity, F statistics and Polymorphism for the entire population for codominant data ..............................................................................................................106 Table 4.6: Summary table of Chi-square tests for the HWE analysis for the Adams Peak population ........................................................................................................................107 Table 4.7: Summary table of Chi-square tests for the HWE analysis of G. speciosa population and the rest of the Gordonia populations within Sri Lanka ..........................108 Table 4.8: Summary table of Chi-square tests for the HWE analysis for the three main populations (Adams Peak, Knuckles and Nuwara- Eliya) ..............................................109 Table 4.9: Summary AMOVA table for the three populations (Adams Peak, Nuwara-Eliya and Knuckles) .........................................................................................................110 !Table 4.10: Summary table for the results of the STRUCTURE analysis for three main populations (Adams Peak, Nuwara-Eliya and Knuckles) ................................................111 !Table 4.11: Summary table for the results of the STRUCTURE analysis for the G. speciosa samples from Adam’s Peak region and the Gordonia species in the Knuckles region ...............................................................................................................................112
!
1
Chapter 1
INTRODUCTION - A review of the biogeographical affinities of Sri Lankan flora
BACKGROUND
Sri Lanka, known as the “pearl of the Indian Ocean,” is a small island with its
own unique flora and fauna. It lies off of the southern coast of India, close to the Equator
between 5° 55’ - 9° 51’ N latitude and 79°42’ - 81° 53’ E longitude. Sri Lanka together
with the Western Ghats region of India form one of the world’s hottest hotspots: the Sri
Lanka-Western Ghats Biodiversity hotspot (Myers et al., 2000). The total landmass of the
country is only 65,610 km2 (Gunatilleke & Ashton, 1987). In spite of its relatively small
geographic area, the amount of endemism and the species diversity contained within the
island is truly remarkable. Nonetheless, the biogeography of both flora and fauna in Sri
Lanka remains grossly understudied.
Information on the distribution and occurrence of plants in Sri Lanka that is
essential to formulate biogeographic hypotheses is scattered in the literature from very
different disciplines. In this chapter, we first review relevant literature on the geology,
geography, vegetation, and floristics of Sri Lanka, as well as the few studies that
explicitly address the biogeography of its plants. We then present three hypotheses on the
biogeographic relationships of the Sri Lankan flora. Finally, we review evidence for each
of these hypotheses from published molecular phylogenetic analyses of plants.
Geological history: The Indian subcontinent, including Sri Lanka, has a
remarkable geologic history that has generated much speculation on the biogeography of
plants and animals in the region. Sri Lanka consists of three crustal units (the Wanni
!
2
2
complex, the Highland complex and the Vijayan complex) that date back to the
Precambrian and were fused together during the formation of the super-continent
Gondwana (Katz, 2000, Mathaven et al., 1999). Of these, the Highland complex is the
largest and forms a large portion of the modern day Sri Lankan landmass (Dissanayake &
Chandrajith, 1999). Sri Lanka and India together form the Deccan Plate that was part of
southern Gondwana (Ashton & Gunatilleke, 1987) and remained in close contact with
Madagascar, Antarctica and Africa after separating from the Gondwanan landmass in the
early Cretaceous (Ashton & Gunatilleke, 1987; Schatz, 1996). The Deccan Plate then
drifted northwards in isolation for more than 30 million years before it collided with
Laurasia during the Eocene (Ashton & Gunatilleke, 1987).
There is disagreement about the time of initial separation of Sri Lanka from the
mainland. Reeves (2009) state that Sri Lanka was first separated from India in the early
Cretaceous around 136 million years ago due to the westward spread of the new ocean
that arose between India and Antarctica (Reeves, 2009). However, Abeywickrama (1967)
states that the initial separation of Sri Lanka from India did not take place until the
Miocene. In any case, most authors agree that Sri Lanka had a land connection to
modern-day India numerous times through geologic history due to fluctuating sea levels
and as recently as 6000 years ago (McLoughlin, 2001). Currently Sri Lanka is separated
from the Indian peninsula by about 40 km through the narrow and shallow Palk Strait.
Geography and climate of present-day Sri Lanka: The topography of Sri Lanka
has a direct effect on rainfall, temperature and other climatic factors. The Sri Lankan
landmass can be divided into three main topographic zones, based largely on elevation.
!
3
3
The narrow coastal plain surrounds the island and extends from mean sea level (MSL)
0m to MSL 100-130m inland (Cook, 1931, Cooray, 1967). The highlands, formed
through uplifting during the Miocene period (Vithanage, 1972), have an elevation above
365m and occupy the majority of the southwestern end of the island (Cook, 1931). The
highlands contain many complex topographical features such as mountain ridges, peaks,
plateaus, basins and escarpments (Cooray, 1967, Ashton & Gunatilleke, 1987, Ashton et
al., 1997). The highest point of the island is the Piduruthalagala peak at 2524m (Cook,
1953). The intermediate plain, the largest of the three regions, lies between the coastal
plain and the highlands (Cook, 1931, Erdelen, 1984), and is relatively flat and rolling
except for the occasional isolated hills and rocks. The coastal plain and the intermediate
plain together contain nearly 80% of the island; the central highlands encompass the
remaining 20% (Gunatilleke & Ashton, 1987).
Sri Lanka’s drainage system includes nine major rivers and 94 other rivers, with
most watersheds located in the central highlands (Damayanthi & Gamage, 2011),
draining off across the plains and into the Indian Ocean. In addition, there are >1000
man-made lakes in the dry zone (Fernando, 1984), some dating back more than two
millennia (Nilanthi & Jayakumara, 2010), for sustaining human populations in the region
since ancient times (Damayanthi & Gamage, 2011).
The climate of the island can be categorized as tropical and monsoonal (Suppiah
& Yashino, 1984, Webb Jr., 2002). Its close position to the Equator ensures a warm
climate all year around. The small size of the island makes its climate susceptible to
marine influences, with ocean breezes moderating the temperature. The mean annual
rainfall of the country varies from around 970mm to 5000mm (Suppiah & Yahino, 1984,
!
4
4
Burt & Weerasinghe, 2014). The mean annual temperature of the lowlands is around 27°
C and in the central highlands around 15° C (Damayanthi & Gamage, 2011). Variations
in the topography of the island have a direct effect on the climate at a local level
(Holmes, 1956, Suppaiah & Yashino, 1984, Webb, 2002). There are six recognized
climatic zones in Sri Lanka (Gunatilleke & Ashton, 1987). The aseasonal wet zone of the
island (including the wet lowlands and the wet highlands) is found in the southwestern
end of the country (Burt & Weerasinghe, 2014). Most of the rain received by the wet
zone is through the southwestern monsoon during the months of May to September. Mid-
elevation regions on the western slopes of the central highlands receive maximum rainfall
from the southwestern monsoons. The wet zone also receives additional rainfall through
the inter-monsoonal rains during the remaining months of the year. The Sri Lankan wet
zone, with its ample year-round rainfall has been postulated to more closely resemble the
western regions of Malaysia and the eastern coastal regions of Madagascar than the other
mountainous regions of South Asia (Erdelen, 1996). Indeed the part of the Sri Lankan
wet zone with the highest rainfall has been suggested as the “most continuously wet
Asian climate west of Borneo” (Ashton & Gunatilleke, 1987).
In contrast, the dry zone is more seasonal, receiving its rain primarily during the
months of December to February by the northwestern monsoonal rains with a marked dry
period from May to September. The zone in between the wet and the dry zones is known
as the intermediate zone and the climate within this region shows a gradual change from
wet to dry from south to north. The driest parts of the country are in the extreme
northwest and southeast, known as the arid zone (Ashton, 1997).
!
5
5
Vegetation classifications: The Sri Lankan flora has diverse forms of vegetation
within the tiny island, ranging from lowland and montane rainforests to lowland scrub
forests and savanna (Dittus, 1977). Tropical deciduous forests are not found in Sri Lanka
in spite of the existence of regions that seem to have conducive climatic conditions for
them, possibly because of the short duration of the dry season due to two monsoons
(Gunatilleke & Ashton). Some of the inherent features of Sri Lankan vegetation are its
unusual altitudinal zonation, fine scale allopatry and the high number of endemics that
are concentrated in the wet zone of the country (Ashton & Gunatilleke, 1987).
The flora of Sri Lanka has been studied since the 18th century (Gunatilleke &
Gunatilleke, 1990) and vegetation maps have been published since the early 20th century.
Sri Lankan forests were first classified by Trimen (1893-1900) by correlating the mean
annual rainfall and climate (Gunatilleke & Ashton, 1987). Published classification
systems for the Sri Lankan vegetation vary somewhat. In his classification of the Sri
Lankan vegetation, Chapman (1947) relied on the classification system of forest types of
Burma and India by Champion (1935) as a model to maintain uniformity and also
because of the similarity he perceived between some forests of Sri Lanka and Southeast
Asia forests (Chapman, 1947). Chapman was also the first to introduce the basic
framework of using climate-related vegetation names to describe Sri Lankan vegetation
(Muller-Dombois, 1968). The forester de Rosayro (1950), classified the community into
six vegetation types based on Clement’s idea of climax communities, and recognized two
major climaxes within Sri Lanka: the dry mixed evergreen forest formation and the wet
evergreen forest formation. He also recognizes two minor climaxes: the montane
temperate evergreen forest and the montane dry grasslands (or “dry Patana”). Lastly he
!
6
6
described two ecotones: the intermediate wet evergreen forest and tropical savannah.
Cook (1953) appears to concur with de Rosayro but adds soil and light quality to the
climatic variables of rainfall and temperature in classifying the Sri Lankan vegetation;
however, her classification is simplified and mainly depicts areas under cultivation for
the cash crops of the era (tea, coconut and rubber) while lumping most of the regions
with original vegetation under forest and jungle and scrub. Holmes’ description of the Sri
Lankan vegetation agrees with de Rosayro and Cook but differs in recognizing three
climax instead of two: tropical wet evergreen forest, tropical dry evergreen forest and
subtropical wet evergreen forest. Among all of these classification systems the vegetation
types put forth, the one by Gaussen et al. (1964) has been cited as the best vegetation map
of Sri Lanka (Mueller-Dombois, 1968) and seem to be the one that is used more often
(Cruz, 1963, Ashton & Gunatilleke, 1987, Gunatilleke & Gunatilleke, 1991, Gunatilleke
et al., 2005), prepared for the Ceylon sheet of the International Map of the Vegetation and
of Environmental conditions. Gaussen et al. (1964) broadly categorized the natural
vegetation into three main types: 1. Littoral zone and saline soils 2. Dry vegetation types
3. Moist vegetation types. These were then subdivided into all the ecologically related
formations or series that can be derived from the least degraded type of vegetation
occurring the region (Gauseen et al. 1964), named after the dominantly occurring plant
species. Ashton and Gunatilleke (1987) have simplified the classification system of
Gaussen et al. (1964) by matching the vegetation types with the climatic zones. In
addition to recognizing the main physiognomic groups of Sri Lankan forests described
above, Perera (1968) also identified both savannas and grasslands in Sri Lanka. He has
then (1975) correlated the plant communities he recognized with the UNESCO World
!
7
7
Classification of Vegetation in order to facilitate comparisons between the Sri Lankan
communities and the other tropical regions (Asia, Africa and Tropical America) of the
world. Ashton and Gunatilleke (1987) recognized 15 floristic regions of within the Sri
Lankan flora (Ashton & Gunatilleke, 1978) and Gunatilleke and Gunatilleke (1990)
assigned the major vegetation types (as recognized by Gaussen et al., 1964) to each of the
floristic regions. Of these 15 floristic regions, 13 are inland terrestrial regions while one
is inland aquatic and another is coastal (Ashton & Gunatilleke, 1987).
Biodiversity and endemism: Many previous authors have demonstrated the high
rates of diversity and the endemism in the Sri Lankan biota (Cook, 1953, Cruz, 1973,
Ashton & Gunatilleke, 1987, Gunatilleke & Ashton, 1987, Gunatilleke & Gunatilleke,
1990). Of the fauna, 86% of the amphibians, 57% of reptiles, 54% of freshwater fish,
22% of invertebrates, 18% of mammals and 7% of birds are endemic to Sri Lanka
(Weerakoon, 2012). Around 3145-4143 plant species (Senaratna, 2001, IUCN, 2007,
Weerakoon, 2012) representing 1070 genera and 180 families (Gunatilleke &
Gunatilleke, 1991) are found in Sri Lanka. Nearly 25% of these species are endemic
(Gunatilleke & Gunatilleke, 1991). Several genera are endemic to Sri Lanka but there are
no plant families endemic to the country (Abeywickrama, 1956). Nearly all (99%) of the
endemic species of Sri Lanka are restricted to the wet zone of the country (Cruz, 1973)
(Abeywickrama, 1956), in wet lowland forests as well as wet montane forests
(Gunatilleke & Gunatilleke, 1990). Within the wet zone, a small area, roughly 15,000
km2 parallel to the coastline at the foothills of the Southwestern Range is considered to be
the most floristically rich area in the whole of South Asia (Ashton & Gunatilleke, 1987,
!
8
8
Gunatilleke, 2005) with very high endemism. Parallels have been drawn between the wet
montane regions of Sri Lanka and the southern wet temperate forests of India, while the
wet lowland forests have been compared to the lowland rainforests of Malaysia and
Burma (Gunatilleke & Ashton, 1987, Chapman, 1947).
Trees in the family Dipterocarpaceae are a characteristic component of the Sri
Lankan evergreen lowland and montane forests and a good illustration of patterns of
endemism in the country. Of the 56 species of Dipterocarps in Sri Lanka, 55 are endemic
to the island (Balasubramanium, 1985), as is one genus, Stemonoporus. Most species are
restricted to a single mountain in the wet zone, and species believed to be closely related
are allopatric, by mountain, elevation, topography and possibly edaphic conditions
(Ashton, 1988; Ashton and Gunatilleke, 1987).
Fossils and past climates: Fossils that could help to shed light on the history of
the biota are lacking in Sri Lanka. Plant fossils are limited to some Jurassic taxa from the
current dry zone of the island (Seward & Holttum, 1922, Sitholey, 1944, Edirisooriya and
Dharmagunawardhana, 2013), well before the rise of Angiosperms that constitute the
dominant flora today. India has a more extensive fossil record, and given the shared
geological history of India and Sri Lanka, it may be appropriate to use Indian fossils to
extrapolate to Sri Lanka. Ashton and Gunatilleke (1987) refer to fossil evidence in
suggesting that during the late Cretaceous, around the time when the Deccan Plate began
to dissociate, a generalized old world tropical forest covered the whole of southern
Gondwana. Based on the presence of megathermal groups such as Ctenolophonaceae,
Morley (2000) argues that a well developed multistoried rain forest must have been
!
9
9
present on the Deccan Plate during the latter part of the Cretaceous. Some of these Old
World species may have survived on the Deccan Plate during the northward journey to its
current location (Ashton & Gunatilleke, 1987). Indeed some of the plant taxa that
currently exist in Sri Lanka have been referred to as Gondwanan relicts (Abeywickrama,
1956, Gunatilleke & Gunatilleke, 1991) and such an ancient Gondwanan relationship is
one of three hypotheses that we present for the origins of the modern Sri Lankan flora.
As the Deccan Plate was making its northward journey, seasonal climatic
conditions possibly prevailed (Ashton & Gunatilleke, 1987); Abeywickrama (1956)
suggests that climatic changes that ensued after the dissociation of the Deccan Plate from
southern Gondwana were perhaps the biggest contributor towards forming the existing
flora of Sri Lanka. This view is also shared by Axelrod (1974) who questions the
previous identification of fossils as tropical from the Intertrappean rocks of the Deccan
Plate and other rocks from the Paleogene epoch in India, given the location of India (and
Sri Lanka) at the time, and states that temperate climatic conditions may have prevailed
in the Deccan Plate during this period. He points out that these Paleogene fossil plants
include members of families that currently occur in habitats ranging from subtropical to
temperate. Axlerod (1974) suggests that the representatives of these families were either
eliminated or restricted to confined upland areas as the Deccan Plate rafted northwards
into more drier and high pressure regions.
At the time the Deccan Plate collided with Laurasia, between 66 – 55.5 mya
(Briggs, 2003) a humid tropical climate prevailed in the region due to open seas to the
east and the equatorial position of the collision zone (Ashton & Gunatilleke, 1987). There
is much evidence for mixing of the Laurasian flora with the Deccan Plate flora after the
!
10
10
collision. Fossils from the Eocene and the Oligocene of the Assam region of India
provide evidence for the presence of Laurasian plant species (Ashton & Gunatilleke,
1987) during those time periods. There is also evidence that Indo-Malaysian flora had a
wide distribution that extended all the way to Europe and Greenland (Abeywickrama,
1958) during the Eocene. This is evidenced, for example, by the presence of fossils of
genera that are exclusively tropical such as Lauracae and Nipa in the clay flora of London
that belong to the Eocene. Nipa (or Nypa), for example, is currently only found in
Southeast Asia and Sri Lanka; an estimated 73% of the living representatives of the clay
flora are currently found in Malaya (Edwards, 1935). As temperatures cooled towards the
end of the Eocene, these forests became restricted in their distribution. However, there is
some evidence of tropical flora with Malaysian affinities from the Oligocene-Miocene in
India (Axelrod, 1974). Dipterocarpaceae, the ecologically dominant tropical trees in
present-day Sri Lanka, first appear in the Oligocene in the East and Southeast Asia, and
have been identified from the Miocene-Pliocene of India.
During the Pleistocene, temperatures of the South Asian region were not much
lower than the current conditions. However, there was high rainfall, especially during the
pluvial periods. These conditions were conducive to extended distribution of the forests.
The interpluvials were much drier. The drier interpluvial periods caused the tropical rain
forests to become restricted to regions in Malaysia, Malabar Coast of India and southwest
Sri Lanka (Abeywickrama, 1958).
!
11
11
Floristics: Abeywickrama (1956) provided a detailed analysis of the flora of Sri
Lanka. Although this work was outside a phylogenetic framework, it provides important
information about the composition of the flora and distribution of plants in Sri Lanka and
has shaped later studies on the ecology and conservation. He identified six floristic
elements that contribute towards the current flora of Sri Lanka. They are 1. Sri Lankan, 2.
Indo-Sri Lankan, 3. Himalayan, 4. Malayan, 5. African, and 6. Pantropic or
cosmopolitan. (Abeywickrama, 1956).
1. The Sri Lankan element comprises endemics that are restricted to the island.
Most of these endemic species (or more rarely genera) are restricted to the wet zone in
the South that is isolated from similar environments in India by intervening dry habitats.
Many parallels have been drawn between the Sri Lankan wet zone and the Western Ghats
region of India, which contains the closest region of similarly wet, tropical habitats and
houses presumed sister taxa (Subramanyam & Nayar, 1974, Chandran 1997, Webb,
2002). Abeywickrama (1956) suggests that these taxa probably evolved in isolation in Sri
Lanka and India from the relicts that rafted on the Deccan Plate when it moved to the
current position.
2. The Indo-Sri Lankan element includes species that are restricted to the Indian
peninsula and Sri Lanka. This element includes species found in both the wet zone of Sri
Lanka and the Western Ghats, but also species shared with other parts of India. Of the
171 plant families that exist in Sri Lanka all but four exist in India as well. Nearly 65% of
the plant species that are found in Sri Lanka are also found in the Indian peninsula, and
nearly a third of these species are restricted to India and Sri Lanka (Abeywickrama,
1956).
!
12
12
3. The Himalayan element of the Sri Lankan species includes species shared
between the Himalayan region and Sri Lanka. These species are commonly considered
temperate species, for example Berberis aristata or barberry, and in Sri Lanka are
restricted to the coolest montane regions. Only 28 of these temperate species are found in
Sri Lanka, all in the montane regions. Gunatilleke and Ashton (1987) hypothesize that
such temperate montane grassland species must have arrived in South Asia through
corridors or stepping stones bearing a relatively cool, moist climate during glacial periods
of the Pleistocene.
4. The Malayan element of the Sri Lankan flora consists of species hypothesized
to have originated in Malaya and then migrated to Sri Lanka. Some examples include
Clematis gouriana and Drosera peltata. Some authors have noted that equivalent wet,
lowland rainforests habitats found in the wet zone of Sri Lanka do not occur in India or
Southeast Asia; Indo-Malaysia contains the geographically closest such similar habitats
(Ashton & Gunatilleke 1987).
5. The African element as defined by Abeywickrama (1956) includes species
hypothesized to have migrated to Sri Lanka from Africa. These species are restricted to
dry and arid regions of the country and include species such as Salvadora presica and
Cordia gharaf (Abeywickrama, 1956). Ashton and Gunatilleke (1987) has suggested that
after the dissociation of the Deccan Plate from southern Gondwana, it remained close to
the African continent maintaining the distance of about 420 km for a considerable
amount of time. It is possible that at least some of these species with African affinities
dispersed into the Deccan Plate during that period.
!
13
13
6. The pantropic and cosmopolitan element includes the species that show a wide
distribution in the tropics and include many of the weeds that are present in the island
(Abeywickrama, 1956). Cruz (1973) puts the percentage of Sri Lankan species with
pantropical affinities at 6.5% of the total floral population of Sri Lanka.
A similar floristic analysis of the mosses of Sri Lanka was performed by O’Shea
(2003) who concluded that Sri Lanka shared the highest number of taxa with India and
Indomalesia, with fewer taxa shared with sub-Saharan Africa.
Three hypotheses on the biogeographic relationships of the flora of Sri Lanka:
1. The flora of Sri Lanka is of ancient Gondwanan origin, and therefore plant
lineages from Sri Lanka will be closely related to those from southern India, and together
these will be most closely related to lineages in Madagascar and continental Africa. This
vicariance scenario is consistent with the geologic history of the region. However,
because the Deccan Plate separated from Madagascar and Africa during the early
Cretaceous, this hypothesis requires divergence times between lineages from Sri
Lanka/India and Africa/Madagascar to be very old.
2a. The flora of Sri Lanka is the result of dispersal from the Indian peninsula, thus
the closest relatives of Sri Lankan species will occur in India. We expect that long
distance dispersal occurs more frequently between geographically close areas; southern
India is the nearest landmass to Sri Lanka and the two regions were connected by a
landbridge as recently as the late Pleistocene, thus facilitating dispersal. Although
dispersal is expected to proceed primarily from India to Sri Lanka, because of the
former’s greater area and greater diversity, dispersal from Sri Lanka to India cannot be
!
14
14
ruled out. This hypothesis is supported by the many species of plants that occur in both
Sri Lanka and India, especially the Western Ghats region. This hypothesis would also be
relevant for lineages identified as Laurasian as opposed to Gondwanan.
2b. The flora of Sri Lanka is the result of dispersal from southeast Asia and
Australasia, thus Sri Lankan species will be most closely related to species from these
regions. Although these regions are currently more distant from Sri Lanka than the Indian
peninsula and were never connected to the island, they include lowland wet tropical
forest habitats that are more similar to what is found in Sri Lanka than in the Indian
peninsula.
METHODS
We used Google scholar searches and GenBank nucleotide searches with
keywords “Sri Lanka,” “plant” and “phylogeny” to identify peer-reviewed publications in
plant molecular phylogenetics that included sequence data from at least one plant
specimen from Sri Lanka. For each species that we identified in this way, we verified
that it was native to Sri Lanka and not simply cultivated (e.g., from a botanical garden),
and determined its geographic range outside of Sri Lanka, if any.
For each of the identified publications, we calculated what percentage of total
species diversity of the targeted taxon was sampled, and if sampling included
representatives from at least two of the major tropical regions of the world (tropical Asia,
Africa plus Madagascar, and the Americas). For studies that we assessed as having good
taxonomic and geographic coverage, we then used the authors’ best estimate of
phylogenetic relationships to note the following: (1) the number of Sri Lankan lineages.
!
15
15
If more than one species from Sri Lanka was sampled, we noted if they form a single
clade in the phylogeny or if they are unrelated. (2) The sister group of each Sri Lankan
species or lineage. If relationships were not resolved, we noted the origins of all members
of the most recent containing clade. (3) Biogeographic conclusions of the authors
regarding Sri Lankan species or lineages not apparent from the tree. As the authors of
these publications are experts on the taxa under study, they may in some cases take into
consideration additional information in making conclusions on their biogeography. (4)
Divergence times for the Sri Lankan species or lineage, if provided by the authors. (5)
Habitat and habit of species. (6) Inferred mode of dispersal.
RESULTS
A total of 26 published studies were identified that included at least one plant
species from Sri Lanka in a molecular phylogenetic analysis. Of these 10 met our criteria
for reasonable sampling of taxa ranging from 12-98% of species (average=55%) from
wide range of geographic distribution of the targeted taxonomic group (Table 1.1). The
lowest taxonomic coverage was in a study of Impatiens that was also the largest group,
with an estimate 900 species distributed worldwide. The ten well-sampled studies came
from diverse group of plants, including two genera of nonvascular plants, and eight
genera from eight families of flowering plants. They included trees, shrubs, climbers,
herbaceous plants, epiphytes and aquatic plants, with a variety of dispersal modes.
The remaining 16 studies were excluded from further analysis because of low
taxonomic sampling (Chanderbali et al., 2001, Conti et al., 2002, Doyle et al., 2004,
Rutschmann et al., 2004, Renner et al., 2010), because Sri Lankan taxa were used only as
!
16
16
outgroups (Kita & Kato, 2004, Li et al., 2009), because a more recent study of the same
taxon existed with better sampling (Kårehed et al., 2008, Groeninckx et al., 2009),
because of low or biased geographic coverage (Meiers et al., 1999, Kita & Kato 2001,
Thulin et al., 2004, Morley & Dick, 2008, Antonelli, 2008, Oguri, et al., 2013), or
because the analysis was of relationships within a single species (Miryeganeh et al.,
2014). However, these papers all provide additional context for understanding the
biogeography of plants in Sri Lanka, and some are included in the discussion below.
Relationships of plants from Sri Lanka and India: Although one of our original
goals was to review phylogenetic evidence of the close relationship between the floras of
Sri Lanka and India, we found few studies that included sequence data from both regions.
We suspect that this is because many plant species co-occur in both Sri Lanka and the
Western Ghats and because sampling of tropical plants in phylogenetic studies is still
low, with usually only one specimen per species sampled. Furthermore, because it has
been easier for researchers to obtain specimens from Sri Lanka than India in recent times,
many of the studies that we found have sampled populations and species from Sri Lanka
and not southern India.
Of the ten studies with good taxonomic and geographic coverage, seven included
specimens from both Sri Lanka and India (Meimberg et al., 2001, Yuan et al., 2004,
Kress et al., 2005, Yuan et al., 2005, Schaefer & Renner, 2010, Wikstrom et al., 2013,
Chen et al., 2015). Species from Sri Lanka and India form clades in four studies, in
Impatiens (Yuan et al., 2004), Exacum (Yuan et al. 2015), Aponogeton (Chen et al.,
2015), and Alpinia (Kress et al., 2005). Species from Sri Lanka and India are not sister
!
17
17
groups, but are closely related and form clades with species from other parts of Asia in
Mormordica (Schaefer & Renner, 2010), Hedyotis s.s. and Neanotis (Wikstrom et al.,
2013). Species from the two regions are unrelated in Nepenthes (Meimberg, 2001). The
origin of the Indian specimens is not always given, so it is unclear if Sri Lankan species
are more likely to be closely related to southern Indian plants as opposed to plants in the
Himalayas. Furthermore, in some studies, species are included whose range is given as
Sri Lanka and India, although only a single specimen is sequenced and included in the
phylogenetic analysis; the close relationship between the floras of the two regions may
thus be underestimated here. In the results and discussion below, we refer to these as “Sri
Lankan lineages,” even though in many cases they also include species from India.
Relationships of Sri Lankan lineages to species from other parts of Asia:
The predominant biogeographic relationships of Sri Lankan lineages (i.e., including
Indian taxa) are with species from other parts of Asia. Sri Lankan lineages are most
closely related to species from Southeast Asia in Gaertnera (Malcomber, 2002), and
Impatiens (Yuan et al., 2004). The closest relatives to Sri Lankan lineages are more
widely distributed in tropical Asia (including sometimes Australia) in Mormordia
(Schaefer and Renner 2010), Hedyotis s.s. and Neanotis (Wikstrom et al., 2013), one
lineage of Exacum (#1 in Table 1.1; Yuan et al., 2005), and Aponogeton (Chen et al.,
2015). The two liverworts studied, Porella and Scapania, are primarily temperate and the
Sri Lankan lineages within each genus are closely related to temperate Asian species
(Hentschel et al., 2007, Heinrichs et al., 2012). Three of the five sampled species of
!
18
18
Alpinia that occur in Sri Lanka are nested within a large clade of Asian species (#2 and
#3 in Table 1); all three are widespread and occur in India and other parts of mainland
Asia.
Close relationships between species from Sri Lanka and other parts of Asia are
also supported in studies with lower taxonomic and geographic sampling. For example,
Axinandra in Crypteroniaceae has one species from Sri Lanka, that was sampled for a
phylogenetic analysis of the family, and three from Southeast Asia that were not sampled
(Conti et al., 2002, Rutschman et al., 2004). If Axinandra is monophyletic, the Sri Lankan
species must be closely related to the Southeast Asian species. Alseodaphne
semecarpifolia from Sri Lanka was the only species sampled by Chanderbali et al. (2001)
in an analysis of Lauraceae; however, all other species of Alseodaphne are found in
China and Southeast Asia. Species of Gordonia from Sri Lanka are now all placed in
Polyspora, an exclusively Asian taxon (see Chapter 4). One species of Cladopus
(Podostemaceae) was sampled from Sri Lanka in a phylogenetic analysis by Kita and
Kato (2004); although sampling is low, all species in the genus are found in tropics and
subtropics of Asia to northern Australia. As many genera only occur in one continental
region, it is likely that future phylogenetic studies will add support to the close
connection between Sri Lankan lineages and other species from Asia.
Relationships of Sri Lankan lineages to species from Madagascar and Africa:
Some phylogenetic analyses suggest more complicated biogeographic relationships of Sri
Lankan lineages, including possible close relationships with Africa. For example, the sole
species of Nepenthes from Sri Lanka is basal in the genus, in a trichotomy with a species
!
19
19
from the Seychelles and a clade that includes all other species in the genus, from
Madagascar and Asia (Meimberg et al. 2001). The authors suggest that the common
ancestor of Nepenthes occurred in the western Indian Ocean or on the Indian
subcontinent, with dispersal later to the Seychelles, Madagascar and Southeast Asia. In
Alpinia, the Sri Lankan lineage is sister to two species in the genus Renealmia distributed
in tropical Africa and tropical America; this combined clade is sister to Afromomum,
distributed in Africa and Madagascar. Kress et al. (2005) interpret these results as a
disjunction between Sri Lanka/Western Ghats and Africa, and suggest it may be the result
of plants drifting across the Indian Ocean on the Indian subcontinent after the breakup of
Gondwana. A second lineage of Exacum (#2 in Table 1.1) is sister to a clade that includes
several species from Southeast Asia, Socotra and East Africa, and the other Sri Lankan
lineage of Exacum (Yuan et al. 2005). The sister group to this combined clade is in
Madagascar, and the authors propose long dispersal first from Madagascar to Sri Lanka
and the Western Ghats, followed by a second wave of dispersal events to the Himalayas,
Southeast Asia, and west to the Socotra region (Yuan et al. 2005).
Connections to Africa are also suggested by less well-sampled studies. For
example, an African connection to Sri Lanka is revealed in Wajira grahamiana
(Fabaceae) that is widely distributed in eastern Africa and the Arabian Peninsula, but
with disjunct population in Sri Lanka and southern India (Thulin et al., 2004). This
species is nested within a larger African clade. Thulin et al. (2004) state that the disjunct
distribution of W. grahamiana dates to ca. 2 my, indicating long distance dispersal.
In Lobelia, one species L. leschenaultiana from Sri Lanka and southeastern India is in a
clade that includes African, Hawaiian and neotropical species. Although relationships are
!
20
20
not well resolved in this study, the results suggest striking disjunctions, further supported
by the fact that the family Lobeliaceae is not particularly diverse in Asia (Antonelli,
2008). A phylogeographic study of the pantropical Ipomoea pes-caprae
(Convolvulaceae) found that Sri Lankan populations cluster with populations from South
India, the horn of Africa and Indochina (Miryeganeh et al., 2014).
Long distance dispersal vs. Gondwanan vicariance: In order for distributions to
be the result of vicariance due to the break up of Gondwana, we expect phylogenies to
show a close relationship between Sri Lankan lineages and lineages from Madagascar or
southern Africa. In addition, we expect divergence of Sri Lankan and Malagasy/African
lineages to be at least 100my old. Only one study, on Alpinia, showed relationships
consistent with Gondwanan vicariance (Kress et al. 2005); however, the authors of this
study did not provide any estimate of the divergence times of the South Asian species.
Divergence times were estimated from molecular clock analyses in four of the ten well-
sampled studies, and authors frequently provided rough estimates from consideration of
relevant fossil evidence. In all cases, divergence times were far too recent for the
observed disjunctions to be a result of Gondwanan vicariance, regardless of the dispersal
biology or the disjunctions.
!
21
21
DISCUSSION
Our review found a wide range of affinities among Sri Lankan taxa sampled in
molecular phylogenetic analyses. Although relationships were most commonly with
species in East and Southeast Asia, we also found close relationships with species in
India as well as Africa and Madagascar. In terms of the hypotheses presented in the
introduction, we find some support for the role of dispersal from India (hypothesis 2a)
and Asia (hypothesis 2b), but not for Gondwanan vicariance (hypothesis 1), primarily
because lineages with relationships to Madagascar and Africa appear too young.
It is important to note that these three hypotheses are not mutually exclusive. We
expect that different plant lineages that make up the flora of Sri Lanka today will have
differing evolutionary histories. Thus, while dispersal is the key mechanism to explain
disjunctions in Sri Lankan plants, the timing and direction of dispersal may differ across
lineages. Indeed, long dispersal is becoming more appreciated as an important process in
tropical biogeography of both plants and animals (Givnish & Renner, 2004; Li et al.,
2009; de Queiroz, 2005).
We found surprisingly few phylogenetic analyses that included representatives of
species from Sri Lanka or South Asia as a whole. With so few studies, it is not surprising
that we did not find clear patterns between biogeographic relationships, divergence times,
habitat, habit or dispersal mode. We can predict what patterns may appear as sampling
increases across plant diversity. Some lineages will be more likely dispersed across long
distances due to their fruit or seed morphology, for example plants with winged wind-
dispersed seeds or bird-dispersed berries. Dispersal may be more likely to occur in some
directions due to prevailing wind and ocean currents or migratory patterns of birds. The
!
22
22
probability of dispersal between two regions is likely related to the distance between
them. However, the distance between source and sink regions has changed over geologic
time scales, as has available animal dispersers, their migratory patterns, wind and ocean
currents, so we expect the signature of dispersal to vary with the age of the lineage.
The successful establishment of dispersed propagules will depend largely on
climate and edaphic factors. As phylogenetic analyses of tropical Asian plants increase,
we expect that species in the wettest parts of the wet zone of Sri Lanka will be more
likely to be the result of dispersal from (or to) the wet rainforests of Indomalesia, whereas
species that occur in slightly dryer forests will be more likely related to the nearer
Western Ghats region of India.
Few studies of any taxon in any part of the southern hemisphere have found
phylogenetic relationships or divergence times that are consistent with a scenario of
vicariance due to the break up of Gondwana. Yet, the Deccan Plate dissociated from
southern Gondwana well after the origin of Angiosperms, and must have had well-
developed plant communities during the million years of its migration across the Indian
Ocean. The lack of evidence for species in Sri Lanka (or India) as Gondwanan vicariants
may be the result of a temporal bias in studies of plant phylogenetic analyses. Most of the
phylogenetic analyses we reviewed here focused on species in one or a few closely
related genera; while they show fine-scale geographic relationships among closely related
taxa, they may be all too young to reveal any Gondwanan history. Phylogenetic analyses
of older, non-eudict lineages may be more promising in revealing a Gondwanan heritage;
however, these older lineages may have undergone more extinction, thus masking past
connections to Gondwana. If the Deccan Plate experienced season climatic conditions on
!
23
23
its voyage north, following the ideas of Axelrod (1974) and Ashton and Gunatilleke
(1987), then we may expect any Gondwanan relicts to occur in seasonal habitats, if they
have not adapted to new niches. Alternatively, it is possible that such Gondwanan relicts
could have used the Deccan Plate as a raft or a stepping-stone, then later dispersed and
radiated into East or Southeast Asia. Over time, such a scenario would result in close
relationships between plants from Sri Lanka and other parts of Asia, masking the ancient
Gondwanan roots.
We hope that this review will provide an impetus for future molecular
phylogenetic relationship that will include strong taxonomic sampling, especially species
from Sri Lanka and the Western Ghats region of India. Such studies will be needed to
provide robust tests of our biogeographic hypotheses. They will also provide a context for
evolutionary and ecological studies that will elucidate the timing and pattern of
diversification of plants in this biodiversity hotspot.
!
24
T
axon
/fam
ily
Publ
icat
ion
# Sp
ecie
s sa
mpl
ed/to
tal
num
ber
(%)
SL
spec
ies/
linea
ge
Loc
atio
n of
sis
ter
grou
p or
co
ntai
ning
cla
de
Div
erge
nce
times
H
abita
t H
abit
Dis
pers
al
mod
e
Nep
enth
es,
Nep
enth
acea
ee M
eim
berg
et
al. 2
001
71/8
2 (8
7%)
1. N
. dis
talla
tori
a In
bas
al tr
icho
tom
y w
ith a
spe
cies
from
Se
yche
lles.
Nex
t lin
eage
is fr
om
Mad
agas
car,
north
ern
Indi
an
spec
ies
sist
er to
M
alay
sian
spp
.
N/A
but
po
llen
in
Euro
pe,
sugg
estin
g m
ove
to A
sia
in M
ioce
ne.
Wet
tro
pics
, 0-
3500
m
Ever
gree
n,
woo
dy
clim
bers
, sc
ram
blin
g sh
rubs
, ep
iphy
tes
Frui
t are
ca
psul
es,
with
fil
iform
an
d lig
ht
seed
s w
ithou
t ob
viou
s ad
apta
tions
fo
r di
sper
sal
Gae
rtne
ra,
Rub
iace
ae d
Mal
com
ber
2002
28
/68
(41%
) G
. ros
ea,
G. v
agin
ans,
G
. ter
nifo
lia,
G. w
alke
ri
In u
nres
olve
d cl
ade
with
9 s
peci
es fr
om
SE A
sia.
Nex
t cl
oses
t rel
ativ
es in
A
fric
a/M
adag
asca
r
Dis
pers
al a
nd
radi
atio
n in
A
fric
a be
twee
n 5.
68
and
5.21
my,
so
dis
pers
al to
A
sia
mus
t be
mor
e re
cent
.
Wet
trop
ics
Smal
l tre
es,
shru
bs
Frui
t are
da
rk b
lue
drup
es,
bird
di
sper
sed
Impa
tiens
, B
alsa
min
acea
eg Y
uan
et a
l. 20
04
111/
900
(12%
) I.
hook
eria
na,
I. he
nslo
wia
na,
Form
s a
clad
e w
ith
two
spec
imen
s de
scrib
ed a
s So
uth
Indi
an (I
ca
panu
lata
, I.
cord
ata)
and
one
In
dian
(I. l
evin
gei).
To
geth
er, t
hese
are
ne
sted
with
in
Sout
heas
t Asi
an
clad
e
N/A
H
ighl
ands
H
erbs
to
shru
bs?
Cap
sule
s?
Tab
le 1
.1: S
umm
ary
of th
e pu
blis
hed
stud
ies t
hat w
ere
cons
ider
ed in
the
revi
ew
!
25
Exac
um,
Gen
tiana
ceae
f Y
uan
et a
l. 20
05
30/6
4 (4
7%)
1. C
lade
of 6
sp
ecie
s fro
m S
L an
d/or
Sou
th
Indi
a (E
. tr
iner
vium
, E.
mac
rant
hum
, E.
pal
lidum
, E.
at
ropu
rpur
eum
, E.
wig
htia
num
, E.
wal
ker)
Sist
er to
a la
rge
clad
e th
at in
clud
es
linea
ges f
rom
SE
Asi
a (in
cl. #
2 be
low
), an
d Ea
st
Afr
ica.
Nex
t clo
sest
re
lativ
es a
re la
rge
clad
e fr
om A
fric
a an
d M
adag
asca
r.
Bet
wee
n SL
/S. I
ndia
an
d So
crot
an-
Indo
mal
esia
n cl
ade
= 7.
4-31
.9 m
y.
Bet
wee
n A
fr/M
ad a
nd
Soco
tra/A
sian
=
8.2-
35.6
my
H
erba
ceou
s pl
ants
So
me
spec
ies
have
ha
rden
ed
win
gs o
n ca
lyx
in
frui
t
2.
E.
pedu
ncul
atum
(S
L an
d so
uth
Indi
a) a
nd E
. se
ssile
(nor
ther
n In
dia)
Nes
ted
with
in c
lade
fr
om S
E A
sia
and
Soco
tra
See
abov
e
Alpi
nia,
Zi
ngib
erac
eae
Kre
ss e
t al.
2005
c 72
/230
(31%
) 1.
A. f
ax (S
L) a
nd
A. a
bund
ifolia
(S
W In
dia
and
SL)
Sist
er to
Ren
ealm
ia
with
one
spec
ies i
n tro
pica
l Am
eric
a an
d on
e in
trop
ical
A
fric
a, th
is in
turn
is
nes
ted
with
in
Afr
ican
/Mal
agas
y cl
ade
N/A
but
au
thor
s su
gges
t tha
t th
ey d
rifte
d ac
ross
Indi
an
Oce
an w
ith
brea
kup
of
Gon
dwan
a!
Trop
ical
fo
rest
un
ders
tory
Larg
e he
rbs
Inde
hisc
ent
or fl
eshy
fr
uit
2.
A. n
igra
W
ides
prea
d sp
ecie
s, al
so o
ccur
ring
in
Indi
a, in
cla
de w
ith
othe
r wid
espr
ead
Asi
an sp
ecie
s
!
26
3.
A. z
erum
bet,
Z. c
alca
rata
B
oth
wid
espr
ead
spec
ies,
in la
rge
unre
solv
ed A
sian
cl
ade.
May
be tw
o in
depe
nden
t lin
eage
s
Pore
lla,
Pore
llace
aea
Hen
tsch
el e
t al
. 200
7 41
/58
(71%
) 1.
P.
mad
agas
cari
ensi
s C
hina
and
Jap
an
N/A
but
cite
s st
udy
(Hei
nric
hs e
t al
. 200
7) th
at
Pore
lla h
as
very
old
foss
il re
cord
, but
ex
tant
spe
cies
ar
e fr
om
Terti
ary
N/A
Li
verw
ort
Long
di
stan
ce
disp
ersa
l of
spo
res
Mor
mor
ida,
C
ucur
bita
ceae
Sc
haef
er a
nd
Ren
ner 2
010
58/5
9 (9
8%)
1. M
. den
udat
a,
M. d
ioic
a,
Form
a c
lade
with
sp
ecim
ens
from
In
dia,
Chi
na,
Sum
atra
, Tha
iland
3.27
my
for
mrc
a of
this
cl
ade
Var
ious
tro
pica
l ha
bita
ts
Mos
tly
pere
nnia
l cl
imbe
rs,
som
e sh
rubs
, an
nual
s
Bird
di
sper
sed
seed
s
Scap
ania
, Sc
apan
iace
aeb
Hei
nric
hs e
t al
. 201
2 50
/87
(57%
) 1.
S. l
epid
a C
hina
, Jap
an, N
epal
N/A
but
cite
s an
ear
lier
stud
y (H
einr
ichs
et
al. 2
009)
that
Sc
apan
ia
orig
inat
ed in
Eo
cene
and
su
gges
ts S
. le
pida
is
rece
ntly
di
verg
ed
Li
verw
ort
Long
di
stan
ce
disp
ersa
l of
spo
res
!
27
Hed
yotis
s.l.
, R
ubia
ceae
i W
ikst
röm
et
al. 2
013
172/
500-
600
(29-
34%
) 1.
15
spec
ies
of
Hed
yotis
s.s
. fr
om S
ri La
nka
form
cla
de w
ith
thre
e fr
om In
dian
su
bcon
tinen
t and
on
e fr
om
Mal
esia
. M
RC
A
wou
ld p
roba
bly
be re
cons
truct
ed
as S
ri La
nka
Sist
er to
cla
de fr
om
Chi
na, t
ropi
cal
Asi
a, P
NG
, Bor
neo
N/A
Her
bs to
sh
rubs
to
smal
l tre
es
caps
ules
2.
Tw
o sp
ecie
s of
N
eano
tis fr
om S
ri La
nka
Nes
ted
with
in la
rger
N
eano
tis c
lade
with
tw
o sp
ecie
s fr
om
Indi
a, tr
opic
al a
nd
tem
pera
te A
sia,
N
ew G
uine
a
N/A
Ann
ual
herb
s C
apsu
les
Apon
oget
on,
Apo
noge
tona
ceae
C
hen
et a
l. 20
15
42/5
7 (7
4%)
1. A
cla
de
incl
udin
g 8
spec
ies
from
Sri
Lank
a an
d/or
In
dia
( A.
bru
ggen
ii,
A. c
risp
us,
A. e
chin
atus
, A
. jac
obse
nii,
A.
nat
ans,
A
. rig
idifo
lius,
A.
st
achy
ospo
rum
, A.
und
ulat
es)
Sri L
anka
n/In
dian
cl
ade
is s
iste
r to
an
Aus
trala
sian
cla
de.
Toge
ther
, Sou
th
Asi
an/S
E A
sian
cl
ade
is s
iste
r to
Afr
ican
/Mad
agas
car
clad
e
Cro
wn
ca. 1
0 m
y fo
r In
dian
/SL
clad
e
Fres
hwat
er
Aqu
atic
pl
ants
Se
eds
in
mud
stu
ck
on b
irds’
fe
et?
28
Chapter 2
Ecological Niche Models support long-term isolation between the endemic species of Polyspora (= Gordonia; Theaceae) in Sri Lanka and the Western Ghats of India1
SUMMARY
Ecological Niche Models (ENMs) are useful in evaluating the significance of past
climatic conditions in forming the biota of a given region. We present ENMs using the
maximum entropy algorithm (MaxEnt) for Polyspora (=Gordonia) species endemic to
Sri Lanka and the Western Ghats of India to test hypotheses about distribution patterns
under past, present and future climatic conditions. Models constructed using present
climatic conditions indicate that suitable habitat for Sri Lankan species currently exists in
the Western Ghats and vice versa. Paleodistributions during the last glacial maximum of
the Pleistocene were at lower elevations and more extensive, although regions of suitable
habitat in Sri Lanka and India were not contiguous. Together these results are consistent
with a history of isolation and allopatric speciation of lineages in Sri Lanka and the
Western Ghats. Predictions of distributions for 2080AD show a severe reduction of
suitable habitat for Polyspora in Sri Lanka, indicating dire effects of future climate
change on the plant and animal species endemic to montane regions in this biodiversity
hotspot.
[1. Adapted from: Gunathilake LAAH, Reza AHMA, Das A, Yakandawala DMD &
Whitlock BA (2015). Ecological niche models support long-term isolation between the
endemic species of Polyspora (=Gordonia; Theaceae) in Sri Lanka and the Western
Ghats of India. In review at PLoS ONE.
29
BACKGROUND
In this study, we use Ecological Niche Models (ENMs) in plants of the genus
Polyspora (=Gordonia; Theaceae) that has endemic species in both the wet zone of Sri
Lanka and Western Ghats to explore hypotheses on their current, past and future
distributions, and to provide a context for future phylogenetic and phylogeographic
studies in the region. Sri Lanka is a small island off the southern tip of India (Fig. 2.1)
with an extraordinary rich and unique flora. The island has the highest biodiversity per
unit area of Asian countries (Baldwin, 1991) with 3145-4143 flowering tree species
(Senaratna, 2001, IUCN Sri Lanka, 2007, Weerakoon, 2012) and 3112 animal species
spread over a land area of 65, 610 km2 (IUCN Sri Lanka, 2007). In addition to its species
diversity, the degree of endemism for Sri Lanka is remarkable, with 25% of plant species
endemic as well as 86% of amphibians, 57% of reptiles, 54% of freshwater fish, 22% of
invertebrates, 18% of mammals and 7% of birds (IUCN Sri Lanka, 2007). Almost 95% of
endemic plant species are further restricted to the wet zone in the southwestern corner of
the country (Ashton & Gunatilleke, 1987, Gunawardene et al., 2007). Indeed, a small
30km wide arc, roughly 15,000 km2 of land within the wet zone, parallel to the coastline
at the foothills of the Southwestern Range, is considered to be the most floristically rich
area in the whole of South Asia (Ashton & Gunatilleke, 1987, Gunatilleke et al., 2005).
However, despite its high species diversity and endemism, the biogeography and
diversification of plants in Sri Lanka remain understudied ( Gunawardene et al., 2007,
Benjamin, 2011).
30
At a finer geographic scale, the wet zone of Sri Lanka shares similar climate and
patterns of endemism with the Western Ghats region of India (Subramanyam & Nayar,
1974) that is found parallel to the southwestern coast of the Indian subcontinent, between
the latitudes 8°N and 21°N (Fig. 2.1) (Menon & Bawa, 1997). Indeed, the entire country
of Sri Lanka together with the Western Ghats is recognized as a biodiversity hotspot by
the International Union for Conservation of Nature (IUCN) due to the high endemism
(Gunawardene et al., 2007, Gunatilleke et al., 2005, Myers et al., 2000). The area of the
Western Ghats is about 160,000 km2 (ca. 6% of India) (Das et al., 2006). Similar to the
Sri Lankan wet zone, the region has high endemism (Gunawardene et al., 2007,
Subramanyam, 1974, Das et al., 2006), with one third of all Indian plant species found in
the high elevation mountains of the Western Ghats (Robin et al., 2010) while 63% of the
evergreen tree species are endemic to the region (Gimaret-Carpentier et al., 2003).
Species endemism in animals is comparable to what is found in Sri Lanka (Das et al.,
2006). Within the Western Ghats, endemism is higher in the southern half between
roughly 8°N and 15°N (the southwestern Ghats ecoregion), especially in the areas south
of the Palghat gap (Fig. 2.1) (Gunawardene et al., 2007, Ramesh & Pascal, 1997).
The wet zone of Sri Lanka has been claimed to be the only aseasonal wet climate
from western Malaysia to Madagascar (Ashton & Gunatilleke, 1987) receiving its rain
from the southwestern monsoon from May to September (Burt & Weerasinghe, 2014)
and from convectional rainfall throughout the rest of the year ( Erdelen, 1996). Within the
wet zone, the highest rainfall corresponds to areas where endemism is highest, with mean
annual rainfall ranging from 250 cm to 500 cm (Panabokke, 1996). Mean annual rainfall
in the Western Ghats is slightly lower than in the wet zone of Sri Lanka, and more
31
variable within the region. The western slopes of the hills of the Western Ghats receives
annual rainfall of 203 – 254 cm (Subramanyam & Nayar, 1974) from the southwestern
monsoon, while the eastern slopes lie in the rain shadow (Subramanyam & Nayar, 1974,
Menon & Bawa, 1997). Consequently, the vegetation of the Western Ghats appears more
influenced by the rainfall than the temperature (Subramanyam & Nayar, 1974).
Plant and animal species of the wet zone of Sri Lanka are thought to show
similarity to species of Western Ghats of India, with presumed conspecific populations,
sister species or sister genera occurring in the two regions (Subramanyam & Nayar, 1974,
Chandran, 1997, Webb, 2002). However, phylogenetic and phylogeographic analyses that
include species in Sri Lanka and Western Ghats region are limited, especially for plants.
Recent research in animal lineages such as leopards (Miththapala et al., 1996), elephants
(Fernando et al., 2000), lizards (Schulte et al., 2002, Macey et al., 2000) shrub-frogs
(Meegaskumbura et al., 2002) and other groups of animals (Bossuyt et al., 2004) indicate
that presumed conspecific animal populations in India and Sri Lanka are genetically
distinct and should be recognized as different subspecies or even species, with each taxon
forming separate clades endemic to either Sri Lanka or India. However, to our knowledge
there have been no such studies using genetic data of plant lineages disjunctly distributed
in the Sri Lankan wet zone and Western Ghats. It is possible that plants in Sri Lanka
differ in their ecological requirements from closely related populations or species in
Western Ghats. Although the montane rainforests in the two regions are separated by
hundreds of kilometers of intervening lowlands and water, it is not certain if plants would
be as isolated as the animal lineages that have been studied, due to different dispersal
abilities and responses to climate and substrate.
32
The assumption of close relationships of plant and animals species in Sri Lanka
and India is partly based on the shared geologic history of the two regions, their close
proximity, and similar climate. Both Sri Lanka and India are part of a single tectonic
plate, named the Deccan plate, that became detached from southern Gondwana during the
early Cretaceous and drifted in isolation for over 25 million years before finally colliding
with Laurasia during the early tertiary times (Webb, 2002, Katz, 2000). Although the
species of the Deccan plate had the chance to evolve independently while in isolation
during the late Cretaceous and early Tertiary, subsequent mixing with the biota of
Laurasia after the collision have greatly influenced the present-day biota of Sri Lanka and
India (Ashton & Gunatilleke, 1987, Gunatilleke et al., 2005). Currently, Sri Lanka is
separated from mainland India by a narrow body of water, the Palk Strait, that is only 80
km wide at its widest point. However, Sri Lanka and India formed a single landmass
during the Cretaceous and there is evidence indicating that Sri Lanka has been connected
to the mainland by a landbridge as recently as ~10,000 years bp. (Vaz, 2000) that would
have facilitated dispersal and gene flow between the two regions. Remains of this land
bridge still exist in the form of low islands and reef shoals.
The close proximity of the geographic distributions, their restricted distribution
patterns within each region, and apparent dependence on specific climatic conditions
make species of Polyspora in Sri Lanka and the Western Ghats an ideal system to explore
the evolutionary history of plants between the two regions and the effects of climate
change on these ecosystems. Four species of Polyspora occur in Sri Lanka: Polyspora
ceylanica (=Gordonia ceylanica), P. elliptica (=G. elliptica), P. gardneri (= G. speciosa)
and P. dassanayakei (= G. dassanayakei) (Wadhwa, 1996). All are endemic to the
33
country, restricted to the wet montane forests of the prominent mountain ranges in the
southern highlands in the wet zone of the island (Weerasooriya, 1998, Yakandawala &
Gunathilake, 2008) (Fig. 2.1, 2.2B), therefore reflecting the pattern of endemism in Sri
Lanka. Genetic data, including plastid and mitochondrial DNA sequences and genotypes
from three microsatellite loci, indicate that these four species are closely related and
likely represent a single panmictic population (see Chapter 4). Distributions of
Polyspora in Sri Lanka are highly dissected today. Low elevation regions of presumably
unsuitable habitats separate populations on different mountain ranges while populations
on a single range could be separated from each other due to human activities such as
unplanned urbanization and clearing for crop cultivation. There are two species of
Gordonia (=Polyspora) in India; G. obtusa is endemic to montane regions of the Western
Ghats and G. excelsa is found in the Himalayas (Kandu, 2005). Although G. obtusa is a
narrow endemic within India, occurring in the southern Western Ghats, its range spans a
greater distance and it is more locally abundant than Sri Lankan species (L. A. A. H. G.
pers. obs.). Its range is divided by the Palghat Gap, a low mountain pass at ca. 10° 45’N,
and to a lesser extent by lower passes. Distribution of populations in Sri Lanka and
Western Ghats may have been more extensive, more contiguous, and at lower elevations
during the cooler climatic conditions of the Pleistocene, and before extensive human
modification of the landscape.
Climate has been shown to play a key role in determining the species richness of a
given locality (Bellard et al., 2012, Cavieres et al., 2013) and both contemporary
environmental conditions (Hawkins et al., 2003) as well as the effects of paeleoclimatic
conditions (Kissling et al., 2012) have been suggested as a driving force behind species
34
richness. In our study, we use ENMs for these five species of Polyspora to test the
following predictions. 1. Habitats suitable for Sri Lankan species currently exist in the
Western Ghats region (and vice versa). 2. Distribution of Polyspora populations have
been more extensive and at lower elevations during the Last Glacial Maximum (LGM) 3.
If suitable habitats for Sri Lankan Polyspora exist in Western Ghats, those habitats and
the habitats in the Sri Lankan wet zone would have been connected by intervening
suitable habitats during the LGM and 4. Future distributions of endemic species of
Polyspora in Sri Lanka will be reduced due to climate change.
MATERIALS AND METHODS
Location data
Location data of the Sri Lankan Polyspora used in the study include the majority of
currently known populations of Polyspora in Sri Lanka (Fig.2.2B) as identified by
Yakandawala and Gunathilake (2008). We treated all four currently recognized
morphological species from Sri Lanka as one taxonomic unit when generating the
models. All four of these species are found in similar habitats, in many cases
sympatrically (L. A. A. H. G. pers. obs.). Furthermore, morphological analysis indicates
that species limits of Sri Lankan Polyspora are ambiguous (Yakandawala & Gunathilake,
2008), and preliminary genetic analyses suggest they are all closely related within the
Polyspora lineage (see Chapter 4). We assume that combining all populations will not
affect the predictions of the models. Seventeen unique localities were identified by the
model-generating program MaxEnt (Version 3.3.3k), from 89 records of individual trees.
35
Location data for Gordonia obtusa, the species endemic to the Western Ghats
region of India, were obtained from a freely available online database (the India
Biodiversity Portal; www.indiabiodiversity.org) and field observations, for a total of 47
unique locations (Fig. 2.2A). All are consistent with the geographic distribution of G.
obtusa described in the taxonomic literature.
Environmental data: Environmental data used in the study were obtained from
freely available online database WorldClim database (Version 1.4; Release 3)
(http://www.worldclim.org). These data layers have been generated through the
interpolation of average monthly climate data (monthly total precipitation, monthly mean,
minimum, maximum temperature) from weather stations to represent the period from
1950 to 2000 (Hijmans et al., 2005). The layers that were downloaded for this project
were at 2.5 arc minute resolution. All 19 derived bioclimatic variables and elevation data
that are available in WorldClim database were used to generate the initial model for both
the Sri Lankan and the Indian species. Variables that did not meet the criteria (discussed
under model generation) of an “effective variable” for the model for each group (Sri
Lanka and Western Ghats Polyspora) were subsequently discarded.
Model generation: To generate models, we used the freely available ecological
niche modeling program MaxEnt (Version 3.3.3k) (Phillips et al., 2004, Phillips et al.,
2006, Phillips & Dudik, 2008) that predicts the distribution of a given organism using
maximum entropy techniques from environmental data coupled with presence-only
species occurrence data. Selection of this method over other popular modeling methods
36
was based on its demonstrated high performance over other modeling software (Phillips
et al., 2004, Phillips, 2006, Elith et al., 2006), ease of interpretation (Phillips et al., 2004)
and also because this program does not require absence data (Phillips et al., 2004, Phillips
et al., 2006).
Of the presence data, 25% was set aside as test data and the remainder was used
as the training data. Training data are used to build the model while test data, that are
randomly chosen by the software, are used to test the accuracy of the model. To build the
initial model, we used all 19 derived bioclimatic variables (current data) and elevation
data obtained from WorldClim. Data for the past and future climatic conditions were also
obtained from WorldClim. All parameters except test percentage were left at their default
value. The software removed duplicate values for the locality data. We used the area
under the curve (AOC) value for the receiver operation characteristic (ROC) and the
binomial omission test (using the simple threshold rule) to test the accuracy of the model.
MaxEnt has different methods built in to the program that can be used to
determine the importance of the environmental variables that are used in a model. We
used the permutation importance, percent contribution as well as Jackknife support to
evaluate the relevance and the contribution of the environmental variables for the model.
During the Jackknife process, MaxEnt deems a variable important for a given model if,
when used in isolation, it has a gain that is slightly less or almost equal to that of a model
using all variables and if there is a subsequent decrease in the gain when that variable is
removed from the model (Phillips, 2005). The gain is defined as the average log
probability of the presence samples, minus a constant that makes the uniform distribution
have zero gain (Phillips, 2005). Variables that show a high decrease in the gain when
37
excluded from the model contain information that is not present in other variables and
variables with a high gain when used in isolation contain information by themselves.
MaxEnt has the ability to conduct Jack knife tests for the training gain, test gain and
AUC and the comparison of the results of these three tests can provide valuable
information (Phillips, 2005) and a comprehensive idea of the relevance of the variables
that are used. We choose variables that had a percent contribution over 50% Jack knife
value of the total regularized training gain that also had a positive percent contribution
above zero, as those that are most applicable. These variables were then used in further
model building while the rest of the variables were discarded.
Five models were generated for different species for three time periods: during the
LGM (~21,000 bp), present day, and future (2080AD). Model 1: We first generated a
model for the Sri Lankan geographical region, for the current distribution of Polyspora
species using locality data for Sri Lankan Polyspora (model 1A). We then ran a model
with an expanded geographical region that included the Indian subcontinent to test the
prediction that suitable habitat for Sri Lankan taxa currently occurs in the Western Ghats
region (model 1B). Model 2: Reciprocal models for the current distribution of the
Western Ghats species, G. obtusa, were generated for the Indian subcontinent and Sri
Lanka. We next generated models for the LGM using locality data of the Sri Lankan
species, (Model 3A and 3B) and locality data of the Western Ghats species (Model 4).
Model 5: Finally, we projected model 1A to future (2080AD) to predict the distribution
of Sri Lankan species of Polyspora in 2080 AD within Sri Lanka.
38
RESULTS
Model 1: The model generated for the current time period using Sri Lankan
species and 19 climatic variables in BioClim shows a distribution pattern that is
consistent with the distribution pattern that is observed within the country today (Fig. 2.3
A). Training AUC value was 0.988 with a regularized gain of 2.952 and the test AUC
was 0.987 with a standard deviation of 0.004. The test points were classified as
significantly better than by random selection for all common threshold values (p <
0.0001) by the model. Modeled distributions of Polyspora are restricted to the central
hills of the wet zone; regions with highest probability occur along the north-south ridge
of the central highlands. This overlap between observed distribution and projected
distribution indicates the accuracy and the suitability of the model for this study. We then
used the selection criteria described in the methods to eliminate variables that did not
contribute to the model. Accordingly following variables were selected as applicable
variables: 1. Bio 1 (annual mean temperature) 2. Bio 2 (mean diurnal range (Mean of
monthly (max temp - min temp)) 3. Bio5 (max temperature of warmest month) 4. Bio 6
(min temperature of coldest month) 5. Bio 8 (mean temperature of wettest quarter) 6. Bio
10 (mean temperature of warmest quarter) 7. Bio 11 (mean temperature of coldest
quarter). Models generated including and excluding elevation data were identical,
indicating that other layers depict data that are contained in the elevation data layer. We
thus decided to eliminate the elevation data layer from the model. Jackknife values for
test gain without the elevation layer supported our decision. The model that was
generated using only the applicable variables (model 1A) had a training AUC value of
0.985 with a regularized gain of 2.903 while the test AUC was 0.989 with a standard
39
deviation of 0.004. This model also classified the test points as significantly better than
by random selection for all common threshold values (p < 0.0001). The region identified
as suitable habitat for Sri Lankan species of Polyspora using only the seven applicable
climatic variables is nearly the same as when all 19 climatic variables and elevation data
were used. The model developed for the larger geographical region (Sri Lanka and the
Indian sub continent) using distribution data from Sri Lankan species (model 1B),
identified regions in the Western Ghats as having suitable conditions for Sri Lankan
species of Polyspora (Fig. 2.3B), supporting our first prediction.
Model 2: We initially constructed the model with distribution data from Gordonia
obtusa for the Western Ghats region in the present day using all 19 climatic variables.
Similar to the Sri Lankan species, the predicted distribution of suitable habitat reflected
the observed distribution of G. obtusa within the Western Ghats region. The training
AUC value was 0.989 with a regularized gain of 3.042 and the test AUC was 0.981 with
a standard deviation of 0.005. Not all of the environmental variables that were selected as
applicable variables according to our selection criteria for the Sri Lankan Polyspora were
selected as applicable for G. obtusa. The applicable variables for G. obtusa were 1. Bio 1.
(annual mean temperature) 2. Bio 5. (maximum temperature of the warmest month) 3.
Bio 6. (minimum temperature of the coldest month) 4. Bio 8. (mean temperature of the
wettest quarter) 5. Bio 11. (mean temperature of the coldest quarter). Variables Bio 2
(mean diurnal range) and Bio 10 (mean temperature of the warmest quarter) that were
included when modeling the distribution of Sri Lankan species of Polyspora were not
40
included. The model that was generated for the present time using only the applicable
variables had a training AUC value of 0.971 with a regularized gain of 2.572 while the
test AUC was 0.977 with a standard deviation of 0.004.
Regions identified as suitable for the Sri Lankan species in the Western Ghats
region overlap with areas currently occupied by Gordonia obtusa (Fig. 2.4). Conversely,
projection of the model for G. obtusa onto Sri Lanka identified as suitable habitats
regions in the central hills that are currently occupied by Sri Lankan species (Fig. 2.4).
These results are consistent with the prediction that Sri Lankan species of Polyspora have
the ability to survive in the Western Ghats (Fig. 2.3B), while the Western Ghats species
can survive in the central highlands of Sri Lanka (Fig. 2.4), if they were able to propagate
to those regions.
Model 3: The model that was generated for the LGM using distribution data from
the Sri Lankan species of Polyspora for the Sri Lankan land area (model 3A) showed that
the distribution of the populations of Sri Lankan species was more extensive and at lower
elevations during the cooler climatic conditions of the LGM as predicted (Fig. 2.5A).
Projection of the model 1B to the Indian subcontinent and Sri Lanka for the LGM,
showed suitable habitat for the Sri Lankan Polyspora in the Western Ghats region during
the LGM (model 3B). Although more extensive and at lower elevations (Fig. 2.5B),
suitable habitats in Western Ghats and in Sri Lanka were still isolated from each other in
the LGM.
Model 4: Projection of model 2 to predict the distribution of G. obtusa during the
LGM for both Sri Lanka and India also showed two regions of suitable habitat in the
Western Ghats and Sri Lanka during the LGM (Fig. 2.6). Populations in the past covered
41
a greater area than present and were also at lower elevations. However, populations
between Sri Lanka and the Western Ghats were isolated from each other.
Model 5: The projected distribution of populations of Polyspora within Sri Lanka
for 2080 AD indicate drastic reduction by more than 50% in regions that are suitable for
their survival (Fig. 2.7). Within Sri Lanka, the predicted future suitable habitat overlaps
with the current distribution, but is restricted to higher elevations and the suitability in the
areas along the North-South massif is also greatly reduced. High elevation areas that have
a high probability value under current environmental conditions have a significantly
lower probability of occurrence in the future, according to the predictions.
The slight variation between the distribution pattern of Polyspora within Sri
Lanka in the figures 3A (model 1A) and 3B (model 1B) and figures 5A(model 3A) and
5B (model 3B) is due to the change in the extent of the geographic region. Even though
the area that is used for the training is the same, the predictions may change due to the
correlations of the variables (Phillips, 2005).
DISCUSSION
South Asia remains grossly understudied in spite of its high species diversity and
endemism. This is the first study that uses distribution maps generated through ENM to
test predictions about the evolutionary history of plants that span Sri Lanka and India.
Genetic, fossil, and paleoclimatic studies in Europe, North America and East Asia have
concluded that montane plants were at lower elevations during the last glacial maximum,
with less isolated populations (e.g., (Flenley, 1998, Hewitt, 2000, Qian, 2000,Tribsch &
Schönswetter, 2003, Schmitt, 2009). Our results agree with this general pattern; the
42
modeled distributions for the LGM of species from both Sri Lanka and the Western Ghats
extend into lower elevations and cover greater area (Figs. 2.5-2.6). Indeed, conditions
appear more favorable for montane species from both India and Sri Lanka during the
LGM of the Pleistocene than they are now (or will be in the future). However,
populations in the two regions were still isolated during the LGM, despite their more
extensive distribution and the land connection between Sri Lanka and the Indian
subcontinent. Currently, the intervening land in the ~380 km between the two regions of
Polyspora habitat have very dry climatic condition, and host tropical dry forests
unsuitable for the survival of Polyspora species. Vegetation in South Asia during the
LGM was inferred to have been similarly dry (Erdelen & Preu, 1990, Adams & Faure,
1997). Because the distributions of Gordonia reflect common patterns of endemism of
species in the wet zone of Sri Lanka or the Western Ghats, we expect our results will hold
for many lineages of plants and animals.
The long-term isolation of Polyspora populations in Sri Lanka and the Western
Ghats that we recovered has implications for how allopatric speciation could have
occurred in these plants. If species in the two regions are closely related and form a
monophyletic group within the Polyspora lineage, as preliminary genetic data suggest,
then their allopatry may be due to long distance dispersal, in the near or distant past. The
likelihood of such an event is increased by the winged seeds found in all species of
Polyspora that facilitate wind dispersal (Gunathilake et al., 2014). Our results suggest
that a vicariance scenario is unlikely unless the formation of isolation preceded the LGM.
In terms of the average duration of a species, 21,000 years is not long and it is likely to be
less than the divergence time of species in the Western Ghats and Sri Lanka. However,
43
the change in climate of the LGM was extreme and may be representative of earlier
Pleistocene glacial maxima (Moore, 1960). Phylogenetic and molecular clock analyses of
the genus will provide an evolutionary context for interpreting our results.
We found that species currently restricted to Sri Lanka could persist in the
Western Ghats region (Fig. 2.3B), and species currently limited to the Western Ghats
could survive in the wet zone of Sri Lanka (Fig. 2.5A). The reciprocal areas of suitable
habitats are consistent with close relationships among these species and speciation
occurring through isolation and lack of gene flow between these two regions, rather than
local adaptation.
Within Sri Lanka, the distribution of Polyspora populations under current climatic
conditions is divided into two main areas, one around the Central Massif and the other in
the Knuckles Range; a significantly smaller area is also indicated as suitable (albeit at a
lower probability) in the Southwestern Range (Figs. 2.2B, 2.3A). However, our models
show a continuous distribution across all ranges during the LGM (Fig. 2.5A). While all
four described species occur widely in the Central Massif and two are also found in the
Knuckles Range, no species of Polyspora is documented to occur in the Southwestern
Ranges. Ashton and Gunatilleke (Ashton & Gunatilleke, 1987) in studying the
distribution of endemic Angiosperms identified 15 floristic regions within the country
with each mountain in the central highlands assigned its own floristic region, suggesting
still finer-scaled patterns of allopatry. Within the ecologically and economically
important Dipterocarpaceae, all but one of the 45 -55 species are endemic to the
southwestern end of the country (Ashton, 1988), and may be separated by altitude,
44
topography, and edaphic conditions (Ashton & Gunatilleke, 1987). Similar patterns are
observed in the endemic species of Eugenia, Syzygium, Memecylon, Garcinia and
Calophyllum (Ashton & Gunatilleke, 1987).
Within the Western Ghats, our models show disjunct subregions of suitable
habitat for Gordonia obtusa, with a discontinuity of populations at the 40km wide
Palghat gap (Fig. 2.4). This gap in the distributions is recovered under the current
climatic conditions (Fig. 2.4) as well as during the LGM (Fig. 2.6) when populations
occurred at lower elevations. While we do not know if this discontinuity has affected the
genetic diversity of G. obtusa populations, a previous study on another plant species with
populations spanning the Palghat gap, Eurya nitida (Theaceae), showed strong genetic
structure in populations on either side (Bahulikar, 2004) (but see Kuttapetty, 2014).
Genetic divergence and reciprocal monophyly across the Palghat gap has also been found
in diverse animal lineages, including birds (Robin, 2010), Asian elephants (Vidya et al.,
2005), and caecilians (Gower et al., 2007). The Shencottah gap, ca. 150 km south of the
Palghat gap in Tamil Nadu, has also been suggested as a geographical barrier for gene
flow albeit to a lesser degree because of its narrower width (7.5 km). Cryptic genetic
divergence of populations on either side has been documented in bird species (Robin,
2010). Our models also show that although the current populations of G. obtusa are
isolated at either side of the Shencottah gap (Fig. 2.5 A), populations were continuous
across the gap during the LGM (Fig. 2.5 B).
45
Although our models’ predictions appear accurate in comparison to the current
known distributions, there are two caveats for our work. First, we have focused only on
climate when creating these models; however, many factors other than climate can play a
role in determining the suitability of a habitat for a given plant species, including edaphic
factors, biotic interactions such as herbivores and potential competitors, and historical
factors (Brown & Gibson, 1983, Draper et al., 2003). Little is known of the natural
history of Polyspora or its interactions with other species. However, there is some
indication that edaphic conditions may differ between the two regions. The Western
Ghats of India was formed after the breaking away of the microcontinent of Seychelles
from the Deccan plate during the late Cretaceous (Widdowson, 1997) and the subsequent
erosion in the western slope (Widdowson, 1997, Kale & Shejwalker, 2008), and now has
red soils, laterites, black soils and humid soils (Subramanyam & Nayar, 1974). In
contrast, the montane regions of Sri Lanka were formed by uplifting of the central
regions during the Miocene (Katz, 2000) with soils classified as red-yellow podzols,
reddish-brown latosols, immature brown loams, bog and half-bog soils (Panabokke,
1996).
A second limitation of our study relates to shortness of the time scale considered
(~21,000 years), which is less than some estimate for the average time for speciation to
occur (Sepkoksi, 1998). Allopatric speciation at such a time scale may still be possible,
for example with a polyploidy event. If older paleoclimatic layers become available, we
may be able to model distributions over longer time periods to test longer-term isolation.
We expect, however, that populations isolated during the LGM will be similarly isolated
during earlier glacial maxima.
46
Our predictions for the future of Polyspora species in Sri Lanka suggest that most
habitats suitable for its occurrence will be drastically reduced in the near future (Fig. 2.7).
We expect that the many other currently co-occurring plant species will be similarly
affected. These results agree with previous studies that have predicted that future
anthropogenic climate change will have a negative impact on the tropical ecosystems
(Bonebrake & Mastrandrea, 2010) especially the wet zone of Sri Lanka and the Western
Ghats. As the mean temperature in the tropics increases (Dufresne et al., 2013), tropical
dry forests in Sri Lanka will expand in distribution while tropical wet forests will contract
(Somaratne & Dhanapala, 1996). In the Western Ghats, changes of the climate in the
future with higher temperatures have been predicted to have major effects on the forest
margins, species composition and migration, occurrence of pests and the regeneration of
forests (Ravindranath & Sukumar, 1998). Together, these results stress the dire need of
effective conservation strategies and their proper implementation to safeguard endemic
species of the Sri Lankan wet zone, as well as the Western Ghats.
The Western Ghats - Sri Lanka biodiversity hot spot is as one of the most densely
populated of the biological hotspots of the world (Cincotta et al., 2000). During the last
150 years, the wet zone of Sri Lanka has been exploited extensively by humans (Erdelen,
1996, Webb, 2002). Large areas of virgin forests in this region were cleared away during
the colonial times for the cultivation of tea, coffee, rubber and cinchona (Webb, 2002)
and more recently for the cultivation of timber, including eucalyptus, teak and mahogany
(Wijesinghe & De Silva, 2012). The remaining forest patches in the region are rapidly
dwindling due to unplanned urbanization, fragmentation and alteration (Erdelen, 1996)
and currently the remaining percentage of original rain forest area in Sri Lanka is less
47
than 5% (Gunawardene, 2007). Habitats of these species in the Western Ghats region are
facing a similar situation where the original forests are getting fragmented due to
anthropogenic factors similar to those in Sri Lanka. Large areas of forests are cleared
away for plantation of tea, coffee and eucalyptus (Subramanyam & Nayar, 1974, Raman,
2006). Menon and Bawa (1997) found out that between 1920 and 1990 the number of
forest patches in Western Ghats has increased fourfold with the size of each patch
decreasing of 83% (Menon & Bawa, 1997). Our results thus support the consensus that,
unless proper action is taken now in consideration distribution changes due to climate
change during conservation efforts, the majority of the endemic species restricted to these
regions will become extinct in the near future.
ACKNOWLEDGEMENTS:
We thank the staff of the Center for Interdisciplinary Geospatial Information
Technologies at Delta State University, MS, USA, and the staff at the GIS lab of the
Richter Library of the University of Miami for the assistance rendered. We are also
grateful for the assistance in data collected by the Gurukula Botanical Sanctuary in
Wayanad, the Rainforest Retreat Plantation in Coorg and the management of Royal
Valley and Craigmore Tea Estates in Nilgiris.
48
Figure'2.1:'Map$of$southern$India$and$Sri$Lanka,$with$elevation$shaded.''The$red$box$contains$localities$of$Gordonia(obtusa$used$in$analyses$(see$Fig.$2.2A$for$more$detail)$and$the$blue$box$contains$all$known$populations$of$Polyspora$in$Sri$Lanka$(see$Fig.$2.2B).$$
49
Figure'2.2:'Location$of$Polyspora(populations$included$in$analyses$for'A.$Western$Ghats$of$India$and$B.$the$central$highlands$of$Sri$Lanka.$$Shading$indicates$elevation.$
50
Figure'2.3:'Predicted$distribution$of$Sri$Lankan$species$of$Polyspora(for$A.'Sri$Lanka$and$B.$southern$India$plus$Sri$Lanka$under$current$climatic$conditions$(Model(1).$Warmer$colors$represent$regions$with$higher$probabilities$of$occurrence$and$cooler$colors$indicate$regions$lower$probabilities
51
Figure'2.4:$Predicted$distribution$of$G.(obtusa$under$current$climatic$conditions,$for$Sri$Lanka$and$southern$India$(Model(2).$Warmer$colors$represent$regions$with$higher$probabilities$of$occurrence$and$cooler$colors$indicate$regions$lower$probabilities$
52
Figure 2.5: Predicted distribution of Sri Lankan species of Polyspora during the LGM for A. Sri Lanka and B. India plus Sri Lanka (Model 3). Warmer colors represent regions with higher probabilities of occurrence and cooler colors indicate regions lower probabilities
53
Figure'2.6:'Predicted$distribution$of$G.(obtusa$during$the$LGM$(Model(4).$Warmer$colors$represent$regions$with$higher$probabilities$of$occurrence$and$cooler$colors$indicate$regions$lower$probabilities
54
Figure'2.7:'Predicted$distribution$of$Sri$Lankan$species$of$Polyspora(A.'in$2080$AD$(Model(5)$and$for$comparison$B.$during$the$LGM$and$C.'under$present$climatic$conditions.$
55
Chapter 3
Seed coat micromorphology of Gordonia sensu lato (including Polyspora and
Laplacea; Theaceae)2
SUMMARY
Species of Gordonia s.l. are characterized by having seeds with prominent flattened
apical wings. However, recent molecular phylogenetic studies show that this concept of
Gordonia is not monophyletic, with species occurring in two tribes of Theaceae. We
examine seed coat micromophology of 14 species of Gordonia s.l., including
representatives from all proposed lineages, and ten species from six genera from all three
tribes of Theaceae. We observed that seeds from Gordonieae, including two species of
Gordonia s.l., have irregularly protruding groups of cells on the seed coat that appear to
be unique in the family. Seeds of Theeae, including all remaining species of Gordonia
s.l., lack protruding cells and include testa cells that are isodiametric to elongate. Seeds
of Stewartieae lack protrusions and elongate testa cells, and often have sculpting visible
below the seed coat. Seeds of Gordonia s.l. from Gordonieae appear significantly smaller
than species placed in Theeae. These results may help to infer relationships of fossilized
seeds previously identified as Gordonia.
2[Adapted from: Gunathilake LAAH, Prince JS & Whitlock BA (2015). Seed coat
micromorphology of Gordonia sensu lato (including Polyspora and Laplacea; Theaceae).
Brittonia 67 (3): 68 -78.]
56
BACKGROUND
Gordonia J.Ellis is a large genus with 20-65 species distributed in the Americas and Asia,
with the type species G. lasianthus (L.) J.Ellis as the sole representative in the continental
U.S.A. (Stevens et al., 2004; Mabberley, 2008). This concept of the genus dates to Keng
(1980) who combined species from several other taxa, most notably Polyspora Sweet and
Laplacea Kunth, based largely on their shared fruit and seed morphology. Fruit of all
species assigned to Gordonia s.l. are ovoid to subglobose capsules with a columella
present. The basal part of the seed containing the embryo is ovoid and asymmetric, with
a single prominent, flattened apical wing that is usually at least as long as or longer than
the embryo. Seeds of other genera of Theaceae vary but all lack the prominent
characteristic wing seen in Gordonia s.l. (Keng, 1962; Wang et al., 2006).
Recent molecular phylogenetic studies have shown that Gordonia s.l. is not
monophyletic (Prince & Parks, 2001; Yang et al., 2004), with species formerly assigned
to it occurring either in Gordonieae or Theeae, two of three monophyletic tribes of the
family (Prince & Parks, 2001; Yang et al., 2004). The type, G. lasianthus, and another
species from the Americas, G. brandegeei H. Keng, are included in Gordonieae. The
remaining taxa sampled are in Theeae and include species from Asia and tropical
America. Although relationships within Theeae are not well resolved, there appear to be
two lineages of species formerly placed in Gordonia s.l. within this tribe. Prince and
Parks (2001) recommend resurrection of two older names for these taxa, Polyspora and
Laplacea, and some authors have already formally transferred some species to Polyspora
(Yang et al., 2004; Orel et al., 2012). More densely sampled phylogenetic analyses will
be necessary to clarify relationships and place as yet unsampled species. Here, we follow
57
Prince and Parks (2001) and assume that G. brandegeei and G. lasianthus are the only
species of Gordonia s.l. in Gordonieae, and that all remaining species are in Theeae,
closely related to the clades provisionally identified as Polyspora or Laplacea.
The polyphyly of Gordonia s.l. suggests that the prominently winged seeds of
these species may be an example of convergent evolution, and raises the possibility that
subtle difference may exist that distinguish seeds of unrelated species. Here, we examine
the micromorphology of seeds from 48 specimens representing at least 24 species of
Theaceae, with representatives from all three tribes, using Scanning Electron Microscopy
(SEM) in a low vacuum mode with backscatter detector. This mode of SEM has not been
used extensively in botanical studies; however, previous studies of seed coat
micromorphology have revealed detail not observed under the more commonly used high
vacuum mode of SEM (Whitlock et al., 2010; Zona et al., 2012). We test the following
predictions: (1) seeds of Gordonia s.l. in Gordonieae differ in micromorphology from
seeds of Gordonia s.l. (i.e., Polyspora and Laplacea) in Theeae; and (2) members of each
of the three tribes of Theaceae are characterized by similar seed coat micromorphology.
MATERIALS AND METHODS
Seeds were obtained from field collections by the first author or herbarium
specimens from the Arnold Arboretum (A), the Fairchild Tropical Botanic Garden (FTG)
and the New York Botanic Garden (NY). We examined seeds from 48 specimens
representing 24 species of Theaceae (Table I), with representatives of all three tribes and
six of the nine genera recognized by Prince and Parks (2001). Seeds from 14 species of
58
Gordonia s.l. were examined, including both species hypothesized to be in Gordonieae.
All other specimens of Gordonia s.l. examined were assumed to be in Theeae following
Prince and Parks (2001) and Prince (2007), including species previously referred to
Polyspora or Laplacea. Seeds of three specimens were unidentified to species, but are
assumed to be in Theeae based on their geographic origin (Yang et al., 2004;
Yakandawala & Gunathilake, 2007).
Seeds were observed with a light microscope then mounted and left uncoated for
observation under SEM with a Jeol JSM 5600LV in a low vacuum mode (20-100 Pa) at
30KV, with a backscatter detector. A seed of Gordonia s.l. can be divided into two
sections: the part that contains the embryo and the flattened membranous extension of the
seed coat that forms an apical wing (Fig. 3.1). We obtained SEM images at three
locations on the seed: at the proximal end (over the embryo), midway (at the embryo –
wing interface), and at the distal end (over the wing). Images were obtained at X50, X100
and X220.
In order to explore micromorphological variation, the following measurements
were obtained: (1) Length of the seed (in mm), measured to be the longest dimension of
the seed. (2) Width of the seed (in mm), measured as the longest dimension
perpendicular to the length. (3) Number of cells intersected by transects on images at
X220 magnification (“transect cell number”). Two images were used for each seed, one
at the proximal end over the embryo, and one at the distal end over the wing. For each
image, two transects were drawn diagonal to the axes of the image and the number of
cells intersecting each line was counted separately. Cells extending beyond the edge of
59
the image were not counted. The higher number of the two counts was noted as the
transect cell number. (4) The average length of five random cells intersected by each
transect in (3)(“average cell length”), for a total of 20 cells per seed.
RESULTS
Seeds of the 24 species of Theaceae examined vary from globose to compressed,
ovoid, oblong or reniform in profile (Fig. 3.1). Seeds of several specimens have thin,
flattened regions that may be described as wings, although the size, shape, and location of
these wings vary across taxa. All specimens examined of Gordonia s.l. have prominent
apical wings. These wings are longer than the length of the embryo in all examined seeds
of Gordonia s.l. in Theeae and in half of Gordonia s.l. of Gordonieae. Such well-
developed wings were not observed in seeds from any other taxon of Theaceae.
Seed length and width were measured from 35 specimens representing two
species of Gordonia s.l. in Gordonieae and 11 species in Theeae. Seed length of one
sample in Gordonieae could not be obtained due to damage to the wing. Seeds of
Gordonia s.l. assigned to Gordonieae (i.e., G. lasianthus and G. brandegeei) had lengths
of 9.0-14.0 mm (mean = 10.5, SD = 1.7) and widths of 3.2-6.0 mm (mean = 3.9, SD =
0.82). All other species of Gordonia s.l., all assigned to Theeae here had seed lengths of
10.3-25.3 mm (mean = 18.05, SD = 4.5) and widths of 3.0-8.7 mm (mean = 5.4, SD =
1.4). Two-tailed student T-tests indicate significant differences between Gordonieae and
Theeae for both length (p = 6.91E-05) and width (p = 0.0043). Data on seed length and
width, as well as transect cell number and cell length (see below) are available from the
authors on request.
60
Diagonal transects drawn on images of X220 magnification correspond to 750 µm
on the seed. Transect cell number was obtained from 25 seeds of 11 species of Gordonia
s.l. and three samples not identified to species. Due to the ridged and irregular surface of
the seed coat, we obtained measurements for transect cell number and cell length from
only four of the nine samples of Gordonieae included in Table 3.1. Statistical analyses on
relationships among seed length, transect cell number, and transect cell length were thus
only performed on samples from Theeae.
A multiple regression analysis to examine the relationship between total seed
length and transect cell number in Theeae indicates that transect cell number of the wing
is significantly and negatively related to total seed length (bwing = -0.26, SE = 0.1, t (18) = -
2.65, p = 0.01). The relationship between transect cell number over the embryo and the
total seed length was not significant (bembryo = 0.002, SE = 0.14, t (18) = 0.02, p = 0.99).
The overall model used for the analysis was significant (F (2,18) =4.19, p = 0.03, adjusted
R2 =0.24). A multiple regression analysis also indicates a significant correlation between
the cell length in the wing and total seed length in Theeae (bwing = 78.842, SE =21.64, t
(18) = 3.644, p = 0.002), but no significant correlation between cell length over the
embryo and seed length (bembryo = -2.92526, SE = 19. 74, t (18) = -0.14, p = 0.884). The
overall model was significant (F (2,18) = 10.85561, p = 0.0008, adjusted R2 =0.49636).
61
Seedcoat Micromorphology.
Gordonieae - Selected images of seeds from the five species sampled from
Gordonieae are shown in Fig. 3.2. These seeds include the laterally flattened and winged
seeds of Gordonia s.s., including G. lasianthus and G. brandegeei, and angular, reniform
seeds of Franklinia W. Bartram ex Marshall and Stewartia L. (Fig 3.1E-I).
When observed under SEM, seeds of all species examined of Gordonieae show
conspicuous protruding groups of testa cells that form ridges distributed through the
length of the seed. The distribution, density, size and shape of these ridges varied from
seed to seed. In G. brandegeei and G. lasianthus, ridges are more common in the seed
coat covering the embryo than in the part forming the wing. In three out of the seven
seeds observed for G. lasianthus, we did not detect any protrusions in the wing. In G.
brandegeei, the protruding ridges are more regularly sized and spaced, and may be
responsible for the lepidote appearance of the seed visible to the naked eye; the non-
protruding testa cells of G. brandegeei appear collapsed, revealing reticulate sculpting in
the inner periclinal walls (Fig. 3.2A-B). Protruding groups of testa cells were not
observed on the seed coats of any species of Theeae or Stewartiae, including all other
species of Gordonia s.l. sampled. In one specimen of Tutcheria shinkoensis (Hayata)
Nakai of Theeae, slightly elevated cells were observed (Fig. 3.3G); however, these seem
very different from what was observed in seeds of Gordonieae, with only a few single
scattered testa cells appearing elevated and the degree of elevation is very slight.
Theeae - Selected images of seeds from the 15+ species sampled from Theeae are
shown in Figs. 3.3-3.4. These include the globose seeds of Camellia sinensis (L.)
Kuntze, the angular oblong seeds of Tutcheria Dunn, and the conspicuously winged seeds
62
of Gordonia s.l. that likely should all be transferred to Polyspora or Laplacea. Seeds
observed from all species of Theeae have relatively smooth surfaces (Fig. 3.3) and lack
any protrusions seen in species of Gordonieae. The shape of cells varies, ranging from
isodiametric to elongate (Fig. 3.4). In some instances both cell shapes were observed on
the same seed (e.g., Fig. 3.4C,F).
Stewartieae - Selected images of seeds from the four species sampled from
Stewartieae are shown in Fig. 3.5. Seeds of species examined were ovoid to oblong,
angular or compressed, some with a small wing surrounding the entire embryo (e.g.,
Stewartia monadelpha Siebold & Zucc.; Fig. 3.1L). In all seeds observed, testa cells
appear consistently small and isodiametric (Fig. 3.5). In two of the four species (S.
pteropetiolata W.C. Cheng (= Hartia sinensis Dunn) and S. ovata (Cav.) Weath.; Fig.
3.5A,B), testa cells show sculpting on the periclinal surface giving them a plicate
appearance. A third species, S. monadelpha, has less prominent sculpting of even bumps
on the periclinal surface. In contrast, testa cells of S. malacodendron L. appear
exceptionally smooth.
DISCUSSION
The variation in seed coat micromorphology observed here supports the
polyphyly of Gordonia s.l. and the recognition of Theeae, Gordonieae, and Stewartieae as
identified in recent molecular phylogenetic analyses of Theaceae (Prince & Parks, 2001).
Despite variation in macroscopic seed morphology and seed size, all of the seeds
examined from Gordonieae, including G. lasianthus and G. brandegeei, share irregular
ridges formed by groups of protruding testa cells. Seeds of Theeae, including all other
63
species of Gordonia s.l. sampled, have a smooth appearance, without ridges or visible
sculpting, and have both isodiametric and linear testa cells. Seeds examined from
Stewartieae also lack protruding ridges, but appear to have solely isodiametric testa cells,
with sculpting visible in three of the four species examined. These three species form a
clade sister to the fourth species, S. malacodendron (Prince, 2002), suggesting that
periclinal sculpting may be derived within Stewartieae. The generality of our findings
needs to be tested with additional sampling of species and genera within Theaceae,
especially Apterosperma Hung T. Chang, and Pyrenaria Blume of Theeae, and species of
Gordonia s.l. from tropical Asia.
While there is wide variation in seed morphology across Theaceae, seeds of
Gordonia s.l. are all flattened and oblong with a conspicuous apical wing (Fig. 3.1). This
morphology appears unique within the family. Indeed, similarity in fruit and seeds was
cited by Sealy (1958) and Keng (1980) to justify combining older generic concepts of
Laplacea and Polyspora with Gordonia, resulting in the more recent concept of
Gordonia s.l. Other taxa of Theaceae have seeds described as winged but these wings are
not apical or as prominent as in Gordonia s.l. (e.g., Wang et al., 2006) and may have
different ontogenies (Tsou, 1997).
The variation that we observed using SEM lends further support to the results of
molecular phylogenetic analyses indicating that Gordonia s.l. is polyphyletic, with two
American species in Gordonieae and all other species sampled in Theeae. Thus, SEM can
provide evidence to test the hypothesis of homology of seed wings in Gordonia s.l.
Although morphological and anatomical variation has been described that distinguishes
taxa in Theeae and Gordonieae (Airy-Shaw, 1936; Tsou, 1997; Tsou, 1998; Wang et al.,
64
2006), the apically winged seeds appeared uniform across lineages of Gordonia s.l.
(Keng, 1980; Keng, 1984). The micromorphological variation among taxa of Gordonia
s.l. is thus significant. The prominent apical wings in separate lineages of Gordonia s.l.
may thus be an example of convergent evolution as was concluded by morphological
cladistic analyses (Wang et al., 2006), presumably as adaptations for wind dispersal.
Alternatively, the prominent apical wings of seeds may be plesiomorphic within the
combined Gordonieae/Theeae clade of Theaceae. More densely sampled and resolved
phylogenetic analyses are needed to test these alternatives.
In addition to differences in micromorphology, seed size may also distinguish
species of Gordonia s.l. in Gordonieae from those in Theeae (that should be reassigned to
Polyspora or Laplacea). Despite substantial variation, seeds sampled from Gordonia s.l.
of Theeae were significantly larger in length and width that seeds sampled from G.
lasianthus and G. brandegeei of Gordonieae. Since most seeds included here were
obtained from capsules that had already dehisced, we expect that all were mature and that
observed differences in size are not due to developmental variation. Sampling of seeds
from multiple conspecific plants of the same species is needed to determine the extent of
phenotypic plasticity. Our observations appear consistent with reports from the literature.
For example, measurements of seed length of G. lasianthus and G. brandegeei are
reported as less than 20 mm (Kobuski, 1950; Grote & Dilcher, 1992; Prince, 2009) while
seeds of Gordonia s.l. from Theeae are usually reported as 20 mm or greater (Keng,
1984; Ming & Bartholomew, 2007; Orel et al., 2012).
65
An increase in seed size could be the result of an increase in the number of testa
cells, an increase in the size of testa cells, or both. Our measurements of testa cell number
and size are indirect and likely imperfect at capturing variation among these seeds.
However, results of multiple regression analyses suggest that larger seeds in Gordonia
s.l. of Theeae have larger testa cells in the seed wings. The shape and arrangement of
testa cells may also play a role. One question for future investigation is if seed wings in
Theeae have more elongate testa cells that are arranged along the axis of the seed.
Our observations support the hypothesis of Prince and Parks (2001) that Gordonia
lasianthus and G. brandegeei are the only species in Gordonieae, and the only species
that should be retained in Gordonia. The seeds of all remaining species examined here
have a micromorphology more consistent with Theeae. It is important to note that many
of these species have not yet been included in phylogenetic analyses and so their
relationships are still uncertain. For example, the four species of Gordonia s.l. endemic to
Sri Lanka have not been subject to phylogenetic analysis; however, seeds from all four
species have smooth seed coats, lacking protrusions, with isodiametric and elongate cells
(Fig. 3.4), consistent with placement in Theeae and supporting the recent transfer to
Polyspora (Orel et al., 2012).
Two well-supported subclades of Gordonia s.l. have been recovered within
Theeae, that have been recognized provisionally as Polyspora and Laplacea, but
relationships between them are uncertain. We included seeds from two taxa, G. fruticosa
(Schrad.) H.Keng and G. haematoxylon Sw., proposed to be the only species of Laplacea
from the Americas (Weitzman cited in Prince, 2009). All other Gordonia s.l. of Theeae
included here were from Asia and include some species that have been transferred to
66
Polyspora and some of unknown affinity. We did not detect any seed characters that
distinguished the putative species of Laplacea from the other sampled Gordonia s.l. from
Asia.
The variation in seed coat micromorphology that we observed may help to infer
relationships of fossil taxa. Theaceae have a rich fossil record, with fossils from the
Tertiary widespread in the northern hemisphere, including Europe where there are no
extant species (reviewed by Grote & Dilcher, 1989, 1992). Because many fossils are
fruits and seeds, and because fruit and seed characters are important in delimiting genera,
it has often been possible to infer relationships to extant genera. However,
micromorphological details of the seed coat are generally not used. The nonmonophyly
of genera such as Gordonia also has the potential to mislead some of these
identifications.
Of particular interest are several fossil seeds and fruits recovered from the mid-
Eocene Claiborne formation of Kentucky and Tennessee, which includes the earliest
well-documented fossil fruit and seeds assigned to Gordonia (Grote and Dilcher, 1989,
1992). Seeds of G. warmanensis were described by Grote and Dilcher (1992) as
intermediate between G. lasianthus (Gordonieae) and extant Asian species of Gordonia
s.l. (now presumably in Theeae). However, the authors noted that seeds of G.
warmanensis lack the "warty surface" seen in G. lasianthus. If the presence or absence of
protruding ridges on seed coats were a key character separating tribes of Theaceae, as our
data suggest, then G. warmanensis would fit better in Theeae than Gordonieae.
Gordoniopsis was described and suggested as the sister group of Gordonia s.l., although
the isodiametric- to elongate-shaped cells on its outer seed surface are more consistent
67
with a close relationship to lineages in Theeae rather than Gordonieae. Seeds of
Andrewsiocarpon, also from the Claiborne formation, exhibit "ridges and tubercles" on
the outer surface of the seed coat, consistent with a close relationship to Franklinia
suggested by Grote and Dilcher (1989). Clarification of relationships of fossils, especially
European taxa will have implications for understanding the history of diversification and
biogeography of Theaceae.
ACKNOWLEDGEMENTS
The authors wish to thank the curatorial staff of the Arnold Arboretum (A), the Fairchild
Tropical Botanic Garden (FTG), and New York Botanic Garden (NY) for permission to
sample specimens for this study, and the University of Miami for financial support. All
imaging was carried out at the Dauer Electron Microscopy Laboratory of the University
of Miami Biology Department. We thank Linda Prince and an anonymous reviewer for
providing valuable feedback, and Soyeon Ahn of Department of Educational and
Psychological Studies of the University of Miami for advice on statistical analyses.
68
Table 3.1 Source of seeds examined using SEM.
Species Voucher Collection locality
Gordonieae
Franklinia alatamaha
Marshall
Atha 518 (NY) U.S.A. (Florida)
Gordonia brandegeei
H.Keng
Brenes 5357 (NY) Costa Rica
G. brandegeei von Hagen and von Hagen 2127
(NY)
Panama
G. brandegeei Schultes and Reko 798 (A) Mexico
G. lasianthus (L.) J.Ellis Baker 16-5 (Harvard University
Herbaria Fruit and Seed
Collection)
U.S.A. (Florida)
G. lasianthus Sheridan and Telford 1116 (FTG) U.S.A. (Florida)
G. lasianthus Craighead s.n. (FTG) U.S.A. (Florida)
G. lasianthus Godfrey 84777 (NY) U.S.A. (Florida)
G. lasianthus Buckley s.n. (NY) U.S.A. (Florida)
G. lasianthus Hill 22659 (NY) U.S.A. (S. Carolina)
G. lasianthus Britt 3066 (NY) U.S.A. (N. Carolina)
Schima argentea E.Pritz.
ex Diels
Fang 5685 (NY) China
69
S. wallichii (DC.) Korth Soepadmo and Suhaimi s76 (NY) Malaysia
Theeae
Camellia sinensis (L.)
Kuntze
Brach and Palomino 1754 (NY) Peru
Gordonia axillaris
(Roxb. ex Ker) Endl.a
Tsang 25604 (A) China
G. axillarisa Rock 7299 (A) China
G. axillarisa Henry 10398 (A) China
G. axillarisa Sun and Chang 814 (A) China
G. axillarisa Chen 1671 (A) Taiwan
G. balansae Pit.a How 73182 (A) China
G. balansaea Liang 62024 (A) China
G. balansae Aa Lau 91 (A) China
G. balansae Ba Lau 91 (A) China
G. ceylanica Wighta Gunathilake S042 (field
collection)
Sri Lanka
G. ceylanicaa Gunathilake S043 (field
collection)
Sri Lanka
G. ceylanicaa Gunathilake S051 (field
collection)
Sri Lanka
G. ceylanicaa Gunathilake S080 (field
collection)
Sri Lanka
70
G. concentricicatrix
Burkillc
Youn 95078 (A) Malaysia
G. dalglieshiana Craib c Nagamasu T49949 (A) Thailand
G. dassanayakei Wadhwa
& Weeras.1
Gunathilake S012 (field
collection)
Sri Lanka
G. elliptica Gardnera Gunathilake S013 (field
collection)
Sri Lanka
G. ellipticaa Gunathilake S009 (field
collection)
Sri Lanka
G. fruticosa (Schrad.)
H.Kengb
Mexia 7460 (A) Ecuador
G. fruticosab García-Barriga et al. 18587 (A) Colombia
G. fruticosab Fosberg 20068 (A) Colombia
G. haematoxylon Sw.b Alain 264 (A) Cuba
G. haematoxylonb Leon and Victorin 17185 (A) Cuba
G. havilandii Burkillc Stevens et al. 242 (A) Malaysia
G. longicarpa Hung T.
Changa
Li Heng 11546 (A) China
G. penangensis Ridl.c Henderson 32948 (A) Malaysia
Gordonia sp.c Kostermans s.n. (A) Sri Lanka
Gordonia sp.c Li 2755 (A) China
Gordonia sp.c Kouchummen FRI023160 (A) Malaysia
71
aSpecies of Gordonia s.l. transferred to Polyspora (Yang et al., 2004; Orel et al., 2012). bSpecies of Gordonia s.l. hypothesized to be Laplacea (Prince, 2009). cSpecies hypothesized to be Theeae, in either Laplacea or Polyspora.
Tutcheria greeniae Chun Steward and Cheo 1045 (NY) China
T. shinkoensis (Hayata)
Nakai
Boufford and Bartholomew 25109
(NY)
Taiwan
Stewartieae
Stewartia malacodendron
Nakai
Mackenzie 1698 (NY) U.S.A. (Virginia)
S. monadelpha Siebold &
Zucc.
Ahles 35399 (NY) U.S.A. (Florida)
S. ovata (Cav.) Weath. Kearney 548 (NY) U.S.A. (Tennessee)
S. pteropetiolata
W.C.Cheng
Tsang 24021 (NY) China
72
Figure 3.1: Seeds of Theaceae. A. Gordonia lasianthus (Sheridan and Telford 1116); B. Gordonia brandegeei (Schultes and Reko 798); C. Franklinia alatamaha (Atha 518); D. Schima wallichii (Soepadmo and Suhaimi s76); E. Schima argentea (Fang 5685); F. Gordonia (=Polyspora) axillaris, a: embryo, b: wing (Rock 7299); G. Camellia sinensis (Brach and Palomino 1754); H. Tutcheria shinkoensis (Boufford and Bartholomew 25109); I. Tutcheria greeniae (Steward and Cheo 1045); J. Stewartia pteropetiolata (Tsang 24021); K. Stewartia malacodendron (Mackenzie 1698); L. Stewartia monadelpha (Ahles 35399); M. Stewartia ovata (Kearney 548). See Table I for voucher information. Scale bar = 5 mm
73
Figure. 3.2: Scanning electron micrographs of seed coats of Gordonieae. A. Gordonia brandegeei, wing (Schultes and Reko 798); B. G. brandegeei, over embryo (Schultes and Reko 798); C. Gordonia lasianthus, wing showing few protrusions (Sheridan and Telford 1116); D. G. lasianthus, over embryo (Sheridan and Telford 1116); E-F. Schima wallichii (Soepadmo and Suhaimi s76); G. Schima argentea (Fang 5685); H. Franklinia alatamaha (Atha 518). See Table I for voucher information. Scale bar = 500 µm.
74
Figure 3.3: Scanning electron micrographs of seed coats of Theeae. A. Gordonia (=Polyspora?) balansae, middle of seed (How 73182) B. Gordonia (=Polyspora?) sp., over the embryo (Kostermans s.n.) C. Gordonia (=Polyspora) longicarpa, over embryo (Li Heng 11546); D. Gordonia (=Polyspora) axillaris, middle of seed (Tsang 25604); E. Gordonia (=Laplacea?) fruticosa, wing (García-Barriga et al. 18587) F. Camellia sinensis (Brach and Palomino 1754); G. Tutcheria shinkoensis (Boufford and Bartholomew 25109); H. Tutcheria greeniae (Steward and Cheo 1045). See Table I for voucher information. Scale bar = 500 µm.
75
Figure 3.4. Scanning electron micrographs showing isodiametric and elongate testa cells in Theeae. A: Tutcheria shinkoensis (Boufford and Bartholomew 25109); B: Tutcheria greeniae (Steward and Cheo 1045); C: Gordonia (=Polyspora?) sp. (Li 2755) wing; D: Camellia sinensis (Brach and Palomino 1754); E: Gordonia (=Laplacea) fruticosa (García-Barriga et al. 18587) wing, in middle of the seed; F: Gordonia (=Polyspora) axillaris (Rock 7299) wing, in middle of the seed; G: Gordonia havilandii (Stevens et al. 242) wing, in the middle of the seed; H: Gordonia (=Laplacea) haematoxylon (Leon and Victorin 17185) over the embryo; I: Gordonia (=Laplacea) haematoxylon (Alain 264) wing, in the middle of the seed.
76
Figure 3.5: Scanning electron micrographs of seed coats of Stewartieae at two magnifications. A. Stewartia pteropetiolata (Tsang 24021); B. Stewartia ovata (Kearney 548); C. Stewartia monadelpha (Ahles 35399); D. Stewartia malacodendron (Mackenzie 1698). See Table I for voucher information. Scale bars = 100 µm for large images and = 500 µm for smaller inlays.
!
77
Chapter 4
The phylogenetic relationships of Sri Lankan Polyspora (=Gordonia; Theaceae) and the genetic structuring of populations within the country BACKGROUND
The genus Gordonia s.l. of the family Theaceae has ca. 70 species that are
distributed in the tropical and subtropical regions of South, Southeast Asia and the
Americas (Wadhwa, 1996). The type species of the genus, G. lasianthus, is restricted to
the coastal areas of the southeastern United States from North Carolina to Florida
(Kobuski, 1950) and is the only species found in North America (Barker, 1980, Grote &
Dilcher, 1992). There are 10-22 species distributed in South America and the Caribbean.
The majority of the species of the genus is distributed in the tropical regions of Asia and
are mostly narrow endemics (Wadhwa, 1996).
The circumscription of Gordonia is currently in flux. The genus as currently
recognized is a combination of three earlier genera: Polyspora, Laplacea and Gordonia
s.s. Keng (1980) combined them into Gordonia s.l. mainly based on similar fruit and seed
characteristics. However, recent phylogenetic analyses using DNA sequence data suggest
that Gordonia s.l. described above is not monophyletic (Prince & Parks, 2001, Yang et
al., 2004). Accordingly, the two older generic names (Polyspora and Laplacea) have
been resurrected and have been applied to some species (Bartholomew & Tienlu, 2005,
Orel et al., 2012). Prince and Parks using chloroplast rbcL and matK sequences found
that two American species of Gordonia (G. lasianthus and Gordonia brandegeei) formed
a clade with the American monotypic genus Franklinia and the Asian genus Schima,
which they named tribe Gordonieae. The remaining eight species sampled of Gordonia
s.l., from Asia and the Neotropics, formed a clade with Camellia and other genera, named
78
tribe Theeae. Although relationships within Theeae were not well resolved, the eight
species of Gordonia s.l. within it appeared to form two groups corresponding to two older
generic concepts, Polyspora (ca. 40 species from tropical Asia) and Laplacea (ca. 20
species in Asia and the Neotropics) (Prince & Parks, 2001). Yang et al. (2004) included
additional Chinese species of Gordonia (=Polyspora) and confirmed the polyphyly of
Gordonia s.l. using nuclear, mitochondrial and plastid sequences; since then, all Chinese
members of Gordonia s.l. have been transferred to Polyspora (Bartholomew & Tienlu,
2005, 2007). Taxonomic sampling was low for both studies. Species from other parts of
Asia, including all Asian taxa formerly placed in Laplacea and many species from the
Neotropics, were not sampled.
Gordonia s.l. has been viewed as an example of the eastern Asian-eastern North
American disjunction that is well documented in flowering plants (Wen, 1999). However,
the nonmonophyly of the genus makes this interpretation problematic, as does evidence
from the fossil record that indicates a complex biogeographic history. Fossils identified
as Gordonia have been described from the Eocene (Grote & Dilcher, 1992) and the
Miocene (Berry, 1929) of North America; fossils from the Eocene are hypothesized to be
more closely related to extant Asian than American species (Grote & Dilcher, 1992).
Additional fossils of Gordonia have also been discovered from Europe, hinting at a more
widespread distribution in the past (Grote & Dilcher, 1992).
Gordonia in Sri Lanka: There are four species of Gordonia (= Polyspora) s.l. in
Sri Lanka according to the recent revision in the Flora of Ceylon (i.e., Sri Lanka), all
endemic to the island: G. ceylanica, G. dassanayakei, G. elliptica, and G. speciosa
79
(Fig.4.1) (Wadhwa 1992). These four species were recently renamed as Polyspora
ceylanica, P. dassanayakei, P. elliptica and P. gardneri respectively (Orel et al., 2012).
However, in the remainder of this chapter I will continue to refer to these four species by
their older generic names. In addition to these four species, the horticulturally valued
species G. axillaris has been reported from the areas adjoining the Hakgala Botanical
Gardens where it is cultivated (Weerasooriya, 1998), attesting to the naturalization of the
species within Sri Lanka.
All species of Gordonia in Sri Lanka are restricted to montane areas of the wet
zone, ca. 1200-2200m in altitude, reflecting the distribution of the majority of endemic
species on the island. (Weerasooriya, 1998, Yakandawala & Gunathilake, Heart and
Theobald). It is not uncommon to find sympatric populations of two or more species
(pers. obs.), and the total number of sites where any species can be found in Sri Lanka is
fewer than 20 (Fig. 4. 2). G. speciosa and G. dassanayakei are known from only a very
few sites; the two remaining species are more widely distributed within the southern
highlands. The pollination biology of these species has not been studied, although floral
morphology suggests generalist insect pollination (Fig. 4.1). Seeds are winged
(Gunathilake et al., 2015) (Fig. 4. 1), suggesting wind dispersal. The montane habitats of
these species in Sri Lanka are highly threatened due to anthropogenic factors as well as
climate change. India, the closest landmass to Sri Lanka, has two endemic species of
Gordonia; G. obtusa and G. excelsa (Kandu, 2005). Gordonia obtusa is restricted to the
southern part of the Western Ghats region which is said to have climatic conditions
similar to the rainforests of Sri Lanka (Subramanyam & Nayar, 2002) while G. excelsa is
restricted to the Himalayas (Kandu, 2005).
80
Questions have been raised on the recognition of four species within Sri Lanka.
Only Gordonia speciosa and G. ceylanica were recognized as valid species up until very
recent. Gordonia elliptica was listed as a variety of G. ceylanica by Trimen in the
original Flora of Ceylon published in 1847 (Heart & Theobald, 1977, Yakandawala and
Gunathilake, 2008). Gordonia dassanayakei was described only in 1996 (Wadhwa,
1996,Weerasooriya, 1998). Of the four species, only G. speciosa is clearly
morphologically distinct from the other three species based on its flower color and other
morphological characteristics. Flowers of G. speciosa have conspicuous crimson
coloration as opposed to white and pink in the other species, and are also much larger.
Leaves of G. speciosa are more coriaceous and have a highly revolute margin that is
absent in the rest of the species. Also, fruits of G. speciosa have a distinct triangular
pyramidal shape that differs from fruit in the other three species that are globose capsules
(Yakandawala and Gunathilake, 2008). Results of a morphological cladistic analysis of
93 characters with multiple individuals from all four species of Sri Lankan Gordonia
found that only two of the four (G. dassanayakei and G. speciosa) were monophyletic as
currently circumscribed (Yakandawala & Gunathilake, 2008). These results hint at the
possibility that the four morphological species may be a single panmictic lineage with
gene flow among populations of all of them. This idea was further supported during field
visits when I observed members of different morphological species occurring
sympatrically.
Understanding the broader level relationships of Sri Lankan Gordonia to other
lineages of Theaceae and the genetic structure among populations within Sri Lanka are
important for the conservation of these highly threatened endemic species. In this chapter
81
of my dissertation, my aim is to use DNA sequences and microsatellite data to begin to
unravel the relationships of Sri Lankan Gordonia to other lineages within Gordonia s.l.
and relationships of populations within Sri Lanka. I used chloroplast and mitochondrial
DNA sequences as well as microsatellite data to test the following hypotheses.
Hypothesis 1: All four species of Gordonia s.l. endemic to Sri Lanka are more
closely related to Chinese Gordonia s.l. (now Polyspora) and other members of Theeae
than to G. lasianthus in Gordonieae. If supported, this would validate the recent re-
naming of the species of Sri Lankan Gordonia as Polyspora (Orel et al., 2012).
Alternatively, some or all Sri Lankan taxa may be more closely related to the North
American G. lasianthus in Gordonieae, suggesting a geographic disjunction. Hypothesis
2: Gordonia obtusa, the endemic species from the Western Ghats of India will be nested
within Theeae and closely related to Gordonia s.l. from Sri Lanka. Hypothesis 3: Among
the four currently recognized species of Gordonia s.l. in Sri Lanka, Gordonia speciosa is
the only one that is genetically distinct as currently circumscribed, with genetic
structuring between populations of G. speciosa and all other populations sampled. This
hypothesis stems from the morphological distinctness of G. speciosa, the results of the
morphological cladistics analysis (Yakandawala & Gunathilake, 2008) and my own
observations in the field. A lack of genetic structure will indicate that all the currently
recognized morphospecies of Gordonia s.l. in Sri Lanka act as a single panmictic group.
Hypothesis 4: Within Sri Lanka, Gordonia s.l. will show genetic structuring between
geographical regions. I predict that geographically defined populations will have low
diversity within populations, high diversity across populations, with some alleles
geographically confined to individual regions. Hypothesis 5: If hypotheses 3 and 4 are
82
supported, I then hypothesize that G. speciosa populations will be genetically distinct
from the Gordonia populations of the Knuckles region, which is at the northernmost end
of the distribution range. Currently G. speciosa is restricted to the southern and
southwestern end of the distribution range and has not been observed or recorded from
the Knuckles region, which is at the northern most end of the distribution of the genus
within Sri Lanka at present. Given the distinct morphology of G. speciosa, the current
distribution pattern could be an indication that it is either currently undergoing speciation
or an established species that is now hybridizing with the rest of the Gordonia
populations in Sri Lanka.
MATERIALS AND METHODS
Sample collection – Sri Lanka: Samples were collected in Sri Lanka during two
field sessions in December 2010 and June 2011. Localities were identified using previous
collection records as well as communications with local botanists who are currently
updating the IUCN red list of plants in Sri Lanka; nearly all of the currently known
populations of the genus in Sri Lanka were visited. Permits required for collecting
specimens were obtained from the Department of Wildlife and the Forest Department of
Sri Lanka. A total of 123 specimens were collected representing all four currently
recognized species. Each specimen had a leaf sample collected for genetic analyses as
well as a corresponding herbarium sample. The herbarium samples will be deposited in
the herbarium of the National Botanic Garden of Peradeniya in Sri Lanka while the silica-
83
dried samples are stored in the Whitlock Lab at University of Miami, FL USA. GPS
coordinates were obtained for all of the samples collected in June 2011 and when ever
possible for the collections during December 2010.
Sample collection – India: Samples of one of two Gordonia species endemic to
India, Gordonia obtusa, were obtained during May 2012. Samples were collected from
seven locations covering the range of the species in the southern Western Ghats.
Collections were made in privately owned land and a total of 29 samples were collected
in collaboration with Indian scientists. Herbarium material was not collected for these
samples. Leaf material was collected in silica gel for phylogenetic analysis and these
samples were stored and analyzed at Asoka Trust for Ecology and the Environment
(ATREE) in Bangalore, India. GPS coordinates for all locations were recorded. Some
additional samples of Chinese species of Polyspora and outgroups in the family Theaceae
were obtained through collaborators.
Molecular analysis: Total DNA was extracted from all samples with a
FASTDNA Green Spin Kit for plant and animal tissues from MPBio following product
instructions. These DNA samples were then used in downstream analysis for chloroplast
and mitochondrial sequencing and genotyping microsatellite markers.
Chloroplast and mitochondrial markers: I used two markers from the chloroplast
genome and one marker from the mitochondria genome for phylogenetic analyses. Yang
et al. (2004) had used the entire trnL-trnLF region (including the trnL intron, partial
84
sequence of the trnL exon and trnL-trnF intergenic spacer) from the chloroplast, and the
matR gene from the mitochondrial genome in their study of the phylogenetic
relationships of the genus. In order to combine my data with previously published
datasets in Theaceae, I used the same two markers in my study. Accordingly, the primer
sequences for the two regions by Yang et al. (2004) were used for PCR amplification
reactions (trnL c 5’ CGA AAT CGG TAG ACG CTA CG 3’, trnF f 5’ ATT TGA ACT
GGT GAC ACG AG 3’, matR879F 5’ACT AGT TAT CAG GTC AGA GA 3’, matR
1858R 5’ TGC TTG TGG GCY RGG GTG AA 3”). In addition to these two markers,
another chloroplast marker, the trnH-psbA intergenic spacer, was also used (trnH 5’ CGC
GCA TGG TGG ATT CAC AAA 3’, psbA 5’ TGC ATG GTT CCT TGG TAA CTT C
3’). PCR reactions (4.83µl of dH2O, 2µl of 5X buffer, 0.2µl of 10X dNTPS, 0.6µl from
25mM MgCl2, 0.66µl from each primer, 0.05µl of Taq polymerase and 1µl of template
DNA for a total of 10µl per reaction) were run using an MJ Research PTC-200
thermocycler with the following conditions: denaturation at 94°C for 2 min, 30 cycles of
denaturation at 94°C for 30 sec, annealing at 50°C for 30 sec and elongation at 72°C for
1.5 min, with a final extension step at 72°C for 10 min. 2µl from each reaction were then
run in a 1% agarose gel with a 100bp ladder to confirm the amplification of the desired
region. PCR reactions were then cleaned using 1:1 mixture of exonuclease I and shrimp
alkaline phosphatase (USB Corp., Cleveland, Ohio) reactions (5µl of the DNA product,
0.1µl SAP, 0.01µl of Exo 1 and 1.89µl of water for a total volume of 7µl per reaction)
that were incubated at 37°C for 15 min then 80 °C for 15 min. Double-stranded products
were then sequenced in both directions using ABI BigDye terminators. Sequencing
reactions consisted of 4.5µl dH2O, 2µl of 5X sequencing buffer, 1µl of BigDye and 2µl of
85
the clean PCR product. Sequencing reactions were cleaned using sephadex columns and
were sequenced using the Sanger method in 3130xl Genetic Analyzer (Applied
Biosystems) in the Biology Department’s molecular core facility.
Phylogenetic analysis for the chloroplast and mitochondrial markers: Because
of low variation among chloroplast and mitochondrial DNA sequences, only a subset of
samples were included in the phylogenetic analyses, representing all of the currently
recognized morphological species in Sri Lanka. Sequences were added to a matrix that
included previously published sequences from other species Gordonia s.l., and other
representatives of Theaceae, available from GenBank. Sequences were aligned using
Mesquite (Maddison & Maddison, 2015), then manually adjusted, with the ends trimmed
off to minimize missing data. Because plastid and mitochondrial genomes are both
believed to be maternally inherited, I combined datasets without testing for incongruence.
However, matrices for the three markers had different numbers of taxa, primarily due to
problems with amplification of the trnH-psbA region. I thus produced two concatenated
matrices: First I assembled a combined matrix with matR and trnL-trnLF, which had the
largest number of taxa. I also assembled a second combined matrix with all three regions
(matR, trnL-trnLF, trnH-psbA) that had more sequence data but fewer taxa. Detailed
information of the number of samples used in separate matrices is provided in Table 4.1.
Bayesian analyses were carried out on all the matrices with Mr Bayes 3.1
(Huelsenbeck & Ronquest, 2001, Ronquest &Huelsenbeck, 2003). The suitable
nucleotide substitution model for both of the combined data sets were assessed using
jModelTest (Guindon & Gascuel, 2003, Posada, 2012). Accordingly, a general time
86
reversible (GTR) model with gamma shaped rate variation was used for all analyses, with
four rate categories. The priors for the analysis were left at the default values. Two
MCMC runs were conducted simultaneously and each run had three heated and one cold
chain. The first 25% samples from the cold chain were discarded as burnin. Given the
size of the data sets, sample frequency was set to 100 and diagnostics were computed
every 1000th run, The trees remaining after the burnin were used to create a majority rule
consensus tree. Posterior probability percentage values (PP) were calculated as a measure
of support.
Haplotype networks for each combined matrices were constructed using PopART
version 1.7 Beta version for Mac OSX (http://popart.otago.ac.nz ) using sequences in the
Theeae clade identified in the phylogenetic analysis described above, including
sequences from Chinese species of Polyspora, Gordonia obtusa from the Western Ghats
and all of the Sri Lankan species.
Microsatellite markers - amplification and analysis: Microsatellite markers
specific to any lineage of Gordonia s.l. have not been developed previously. However, a
large number of microsatellite markers are available for Camellia (Freeman et al., 2004,
Wen et al., 2012), tea, that is closely related to Polyspora and Laplacea. For my study I
selected 15 markers that had been developed for Camellia sinensis by Freeman et al.
(2004). Of the 15, seven microsatellite loci amplified products (M2, M3, M5, M7, M9,
M12, M13). PCR products for these loci were gel purified using the Wizard® SV Gel
and PCR Clean-Up System (Promega, USA). PCR products were cloned with the pGEM-
T Easy Vector System (Promega USA). Clones were then sequenced with ABI Big Dye
87
Terminator Cycle Sequencing v 3.1 chemistry (Applied Biosystems USA) using pGEM
vector primers. Sequences were electrophoresed on an ABI 3130XL Genetic Analyzer
(Applied Biosystems USA). Sequences were trimmed, edited, and contiged using
Sequencher v. 5.0. Five of the loci (M2, M3, M5, M12) contained microsatellite loci that
aligned with the original cloned sequences of Camellia sinensis. Primers were then re-
designed to make them Gordonia-specific using Primer3 (Rozen & Skaletsky 2000).
Only three of these loci (M2, M3, M5) produced products of the expected length without
non-specific amplification. One primer of each of these pairs was then labeled with the
fluorescent dye 6-FAM.
These three loci were then amplified in all the samples and some outgroups using
the QIAGEN Multiplex Reaction Kit (Qiagen USA) (Table 4.1). Polymerase chain
reactions (PCR) (10 µl) contained 2–50 ng DNA, 0.5 mM of each primer, 1X QIAGEN
Multiplex PCR Master Mix with HotStarTaq, Multiplex PCR buffer with 3 mM MgCl2
pH 8.7, and dNTPs. The cycling parameters were one cycle at 95°C for 15 min, followed
by 30 cycles of 30 sec at 94°C, 90 sec at 60°C, 90 sec at 72°C, then a final extension at
60°C for 30 min on an ABI 2720 thermal cycler. The resulting PCR products were
diluted 20X with dH2O. For each sample, 0.5 µl diluted product was loaded in 10 µl
HIDI formamide with 0.1 µl LIZ-500 size standard (Applied Biosystems USA) and
electrophoresed on an ABI 3130XL Genetic Analyzer (Applied Biosystems USA).
Genotypes were scored using GENEMAPPER v5.0 (Applied Biosystems USA). The
scored genotypes were then analyzed using several freeware that are explained in detail
below.
88
Checking for errors in amplification: The data were first checked through
Micro-Checker v 2.2.3 (Van Oosterhout et al. 2004) for errors in amplification. Micro-
Checker is able to identify errors in genotyping due to null alleles, large allele drop out
and stuttering. If null alleles are detected, the program is able to calculate the null allele
frequency as well as the adjusted allele and genotype frequency, and these allele
frequencies are then used to estimate if the population of interest is in Hardy-Weinberg
equilibrium. However, since Micro-Checker calculates true allele frequencies for the
population and not an individual, the corrected allele frequencies as well as the corrected
genotypes cannot be used in multi-locus analysis. If the analysis by the Micro-Checker
indicated that null alleles should not be present in a given locus, then the samples with
null alleles for those loci were considered as erroneous data and were discarded. The
populations were checked for deviations from HWE using the adjusted data through
Micro-Checker.
Assessing the genetic variation within populations: In order to test my
hypotheses about genetic structuring of Sri Lankan Gordonia, I first grouped the data
according to geographical location. In other words, samples from one location were
considered as one population, irrespective of their species assignment. The three groups
correspond to mountain ranges and surrounding area and were 1. Adams Peak 2. Nuwara-
Eliya (the area around Piduruthalagala mountain range) 3. Knuckles. (Fig. 4.3). Samples
that were allocated to each of the three regions are given in Table 4.2. Sample number
S012 from Thangamali Plains, Haputhalae that had valid data for all three loci was not
included into this analysis as it was located in a region that made it questionable to
89
include it under any of the three locations given above. The Adams Peak population
included all the samples of G. speciosa as well as some samples from the remaining three
species. The Nuwara-Eliya and Knuckles groups include only plants assigned to G.
dassanayakei, G. ceylanica, and G. elliptica. In order to test Hypothesis 3 on restricted
geneflow between G. speciosa and populations of the remaining three species in Sri
Lanka, I analyzed the population parameters for the Adams Peak group separately and
compared G. speciosa to all other plants. To test Hypothesis 4, I calculated population
parameters for the three groups (Adams Peak, Nuwara- Eliya and Knuckles) to study the
variation among and within populations.
All analyses for the population structure were carried out using GenAlEx v 6.5
(Peakall & Smouse, 2006, 2012). The calculations included checking deviations from
HWE, analysis of allelic patterns of the populations including mean number of alleles and
private alleles, pairwise F statistics, Nei’s genetic distance (Nei, 1972) and genetic
identity, Analysis of Molecular Variance (AMOVA) (Excoffier et al., 1992). The
AMOVA was run with 999 permutations and at a confidence level of 0.05. It has been
shown that high levels of genetic diversity can lead to low levels of FST values (Hedrick,
2005). Therefore, in my analysis I focused more on the F’ST (standardized FST)
(Meirmans, 2006) than FST. FST has been defined as F’ST = FST / FMAX. F’ST is useful in
comparisons of samples with different effective population sizes and or between markers
of different mutational rates. As the results of these calculations indicated the three
populations to show genetic differentiation among them, I tested the isolation by distance
(IBD) hypothesis (Wright, 1943) for the populations through a Mantel test (Mantel, 1967)
using a genetic distance and a geographic distance matrix. This test was run using 999
90
permutations and a confidence level of 0.05. Finally to test Hypothesis 5 that G. speciosa
is a group that is currently undergoing speciation as opposed to a group that is beginning
to hybrize with other populations of Gordonia, I carried out another analysis between the
population G. speciosa in the Adams Peak region and the Gordonia population in the
Knuckles region. The calculations and tests that were carried out for this part are similar
to the analyses that were carried out for the three populations. In addition PAST v 3.05
(Hammer et al., 2001) was used to run a principal component analysis (PCA) and a
principal coordinate analysis (PCoA). PCA allows the maximum representation of the
original data set using a new system of complementary coordinates and may sometimes
show the variation that could not be observed in the original data set (Wold et al., 1987).
PCoA also uses Euclidean distances to represent the data. However, a PCoA also has the
possibility to work with any dissimilarity measures, which is not possible in a PCA
(Ramette, 2007). Therefore I decided to run both the analysis for my data set. In addition
a neighbor joining tree and a UPGMA tree was created to describe the genetic
relationships between the G. speciosa population and Gordonia populations of the
Knuckles region. Finally, STRUCTURE (Pritchard et al., 2000) analyses were run for
the three main populations (Adams Peak, Nuwara-Eliya and Knuckles) as well as for the
P. gardneri (G. speciosa) population from the Adams Peak and the rest of the Polyspora
(Gordonia) species in the Knuckles region. For both analyses the burn-in period was set
to 10,000 and the number of MCMC replicates after the burning was also set to 10,000.
The other variables were left at default values when defining the parameter sets. Both the
91
methods by Pritchard et al., 2000 (and Pritchard et al., 2010) and Evanno et al., 2005
were used to calculate the optimum K. The analyses was replicated for 15 times at each
pre-defined K value and the resulting Ln P (D) values were then used in the calculations
to find the best K.
RESULTS
Phylogeneitc analysis: In all analyses, Sri Lankan species were nested within the
tribe Theeae. In the combined analysis of trnL-trnLF and matR, Sri Lankan species were
resolved as a clade with 99% PP support that was nested within another clade also with
99% PP that included all other sequences of Polyspora and Gordonia obtusa from India
(Fig. 4.4); relationships within this Polyspora clade are not well resolved and the multiple
accessions of G. obtusa do not form a clade. In the analysis of the combined matrix of all
three markers, the Sri Lankan species were monophyletic with 97% PP and nested within
a clade of Polyspora from China (Fig. 4.5); no trnH-psbA sequences were obtained from
G. obtusa. Sri Lankan samples had only one base difference from the rest of the Theeae
sampled in trnL-trnLF and matR. There were two base differences in trnH-psbA that
distinguished the Sri Lankan species from the Chinese species of Polyspora. One
additional base change was shared by four of the samples that were included in the
analysis. These four samples belonged to different locations and different species. None
of the analyses showed sufficient variation to resolve relationships among Sri Lankan
samples due to low levels of variation.
92
The haplotype network from the combined trnL-trnLF- matR dataset is shown in
Fig. 4.6A. One haplotype occurs in specimens from Sri Lanka, Gordona obtusa, and
species of Polyspora from China, and is inferred to be the ancestral haplotype due to its
high frequency, widespread geographic distribution and the number of connections to
other haplotypes. Two other haplotypes from Sri Lanka are shown as derived from this
ancestral haplotype, each by one base. The haplotype network for all three markers (Fig.
4.6B) separates haplotypes from Sri Lanka and China. One haplotype inferred to be
ancestral occurs within Sri Lanka and samples of G. ceylanica, G. speciosa and G.
elliptica are shown to contain this haplotype.
Microsatellite analyses: A total of 114 samples were scored for the three loci M2,
M3 and M5. A total of 34 alleles were observed across all three loci. All loci were
polymorphic with 10 alleles at M2, 14 at M3 and 10 at M5. The alleles that were present
at a specific locus and their frequencies are given in Table 4. 3 and Figure 4.7. Only 50
samples had all three loci amplified while 64 samples had null alleles in one or more of
the three loci. The Micro-Checker analysis indicated that loci M2 and M3 should not
have any null alleles present while null alleles were indicated in the locus M5. There was
no evidence for large allele dropout for all three loci. Presence of null alleles has been
shown to cause overestimation of population parameters such as FST (Chapuis & Estoup,
2007). Comparison of the FST and other population parameters of initial analyses for the
entire population (including the samples that had the null alleles) and another with just
93
the genotypes that had amplified alleles at all three loci indicated it to be the case for this
data set as well (Table 4.4 & Table 4.5). Therefore, based on the results of Table 4.4 and
4.5 and also the Micro-Checker analysis, samples with null alleles in all three loci were
discarded for downstream analysis.
Genetic variation between Gordonia speciosa and the other Gordonia
populations: Analysis for the HWE analysis for the Adams Peak population indicated no
departures from random mating expectations for all three alleles (Table 4.6). The FST of
the population was 0.035 (SE 0.009). The mean number of migrants for all three loci was
7.969 (SE 1.799). The inbreeding coefficient was -0.169 indicating a high number of
heterozygotes. Indeed the observed number of heterozygotes (Ho) for the population was
0.833 (SE 0.077) while the expected number of heterozygotes (He) was 0.732 (SE 0.046).
Nei’s genetic distance for the two sub populations was 0.202 while the Nei’s genetic
identity was 0.817. For the analysis that was conducted treating G. speciosa as a separate
group from all the rest of the Sri Lankan population of Gordonia, the results were much
similar. No deviations from the HWE were detected apart from the allele M5 for the
populations other than G. speciosa (Table 4.7). The mean FST was 0.045 (SE 0.016) and
the mean number of immigrants was 8.205 (SE 4.165). The inbreeding co-efficient was -
0.100 (SE 0.015). The HO was 0.878 (SE 0.042) while the HE was 0.764 (SE 0.051).
Nei’s genetic distance between the two groups was 0.289 while Nei’s genetic identity
was 0.749.
94
Genetic variation among populations of three areas (Adams peak, Nuwara-Eliya
and Knuckles): The HWE analysis carried out for the population using all three loci
indicated the M5 locus to deviate from the HWE in the Adams Peak population (Table
4.8). However, the HWE analysis carried out by Micro-Checker using the adjusted allele
frequencies indicated all three loci to behave according to HWE. Allelic pattern analysis
for the populations across all loci indicated Nuwara-Eliya population to have the highest
mean number of alleles (9.00, SE 2.000) as well as private alleles (1.667, SE 0.882) while
the Knuckles population had the lowest mean number of alleles (7.667, SE 1.453) and
private alleles (1.00, SE0.000) (Figure 4.7). Pairwise FST values between the populations
indicated the highest amount of differentiation between Knuckles and Adams Peak
(0.048) and the lowest amount of differentiation between Knuckles and Nuwara-Eliya
(0.019). Interestingly the lowest number of migrants was indicated between Knuckles and
Adams Peak (4.996) while the highest number of migrants was between Nuwara-Eliya
and Knuckles (0.019) (Figure 4.8). The pairwise values for Nei’s genetic distance
analysis indicated the highest genetic distance between Knuckles and Adams Peak with
the lowest pairwise value between Knuckles and Nuwara-Eliya (Figure 4.8). The Nei’s
genetic identity values displayed the inverse of genetic distance by having the lowest
identity between Knuckles and Adams Peak populations and the highest genetic identity
between Knuckles and Nuwara Eliya. The AMOVA analysis indicated the FST for the
population as 0.036 and the standardized F’ST (Meirmans, 2006) value was at 0.199. The
summary of the AMOVA for the three populations is given in Table 4.9. The results of
the Mantel indicated a significant relationship (Rxy = 0.093, p = 0.023) between genetic
distance and geographical distance. However, the analyses by STRUCTURE did not
95
identify population structure between the three main populations. The optimum K when
calculated using the method by Evanno et al. (2005) was 2. However, we also used the
posterior probability method described by Pritchard et al., (2000 & 2010) as the method
by Evanno et al. does not have the ability to find the best K if K=1 (Evanno et al., 2005).
Indeed the best K according to the posterior probability calculations turned out to be 1
(Table 4.10) and therefore the results of the STRUCTURE analysis did not support
population structuring of the genus Polyspora (= Gordonia) within Sri Lanka.
Genetic differentiation between Gordonia speciosa and Gordonia populations in
the Knuckles region: The allele frequencies of the three loci showed a considerable
variation between the populations of G. speciosa in the Adams Peak population in
southwestern end of the range and Gordonia populations in the Knuckles region (Fig.
4.9). The mean number of alleles for G. speciosa population was 6.000 (SE 1.000) while
it was estimated to be 7.667 (SE 1.453) for the Knuckles region population. Number of
private alleles in G. speciosa population was 1.000 (SE 0.577) while the Knuckles region
populations had 2.667 (SE 0.667) private alleles. The mean FST for the two populations
was 0.069 (SE 0.022) while the mean number of migrants was 4.307 (SE 1.570). The
pairwise Nei’s genetic distance was 0.522 while the Nei genetic identity was 0.593. The
results of the Mantel test for IDB showed a positive correlation between genetic and
geographical distance but was not significant at an alpha level of 0.05 (Rxy = 0.108, p =
0.073). The results of both the PCA and PCoA showed clustering of the two populations
(Fig. 4.10) as well as the neighbor-joining tree (Fig. 4.11) and the UPGMA tree (Fig.
4.12). In the STRUCTURE analysis, the number of populations with the highest ΔK value
96
was 4 (Evanno et al., 2005) while the number of populations with the highest posterior
probability was 2 (Pritchard et al., 2000, Pritchard et al, 2010) (Table 4.11).
DISCUSSION
Phylogenetic analysis: My hypotheses one and two were supported by the results
of the phylogenetic analyses. All of the Sri Lankan samples nested within tribe Theeae in
both of the analyses and their closest relatives were the Chinese species of Polyspora and
the Gordonia obtusa from the Western Ghats of India. This result provides support for
the recent renaming of the Sri Lankan species of Gordonia as Polyspora (Orel et al.,
2012) and also justifies the renaming of G. obtusa as Polyspora. However, my
phylogenetic analyses were not able to resolve relationships among species in the
Polyspora clade. Furthermore, sampling of species of Polyspora was low, especially
from Australasia. A close relationship between species in Sri Lanka and the Western
Ghats region of India is not recovered, but cannot be rejected.
The haplotype analysis with the combined matrix of trnL-trnLF indicates the
presence of a persistent ancestral haplotype with a widespread distribution. These results
suggest that the current populations of Gordonia in Sri Lanka are descended from a
species in mainland Asia. This scenario is consistent with the geologic history of the
region. Fossil from India testify to the migration of Laurasian plant species dating to after
the Deccan plate (consisting of modern-day India and Sri Lanka) collided with the
Laurasian landmass, from the Eocene (Ashton & Gunathilake, 1987) and also during the
Oligocene-Miocene (Axlerod, 1974). Some evidence suggests that the Indo-Malaysian
flora was widely distributed that extended all the way into the Europe and Greenland
97
(Abeywickrama, 1958) at the time of the collision. Even though fossils of Gordonia have
not been identified from Sri Lanka or neighboring India from these time periods, fossils
assigned to Gordonia s.l. have been identified from the Eocene (Grote & Dicher, 1992)
and the Miocene (Berry, 1929) of Europe. Therefore, it is a reasonable hypothesis that
Gordonia dispersed into Sri Lanka from Laurasia through India. These widely distributed
Indo-Malaysian forests may then have become more restricted in distribution during the
latter part of the Eocene as the climate cooled.
Although the phylogenetic analysis do not identify a close relationship of Sri
Lankan Gordonia with their Western Ghat counterparts, ecological niche models (ENMs)
developed for the populations of Gordonia in Sri Lanka and the Western Ghats of India
(see chapter 2) show reciprocal areas of suitable habitat for both species in the two
regions, but isolation since the last glacial maximum. The current habitat of Gordonia in
the wet zone of Sri Lanka has served as an isolated refugium for the genus during the
glacial times.
The results of the analysis of microsatellite markers support Hypotheses 4 and 5
but does not support Hypothesis 3. I was not able to identify genetic structuring between
the populations of G. speciosa and the populations of the remaining three species in the
Adams Peak region. The FST values and the Nei’s genetic distance within the Adams Peak
population and between the population of G. speciosa and remaining populations of
Gordonia in Sri Lanka was quite low and was therefore an indication of geneflow. Also
the STRUCTURE analysis for the three main population groups that were studied
98
clustered samples from all three populations as one panmictic group. This result, taken
together with the previous morphological study as well as the low variation of the genetic
markers that were used in this study challenge the current identification of four species of
Gordonia within Sri Lanka.
The results show that there is genetic structuring within Sri Lanka for Gordonia
by geography. The populations in the north (Knuckles region) were more distant from the
populations at the Southeastern edge of the distribution (Adams Peak) while the
populations in Nuwara-Eliya played an intermediary role. Also the results of the Mantel
test show that this structuring can be attributed to the isolation of these populations due to
geographical distance. However, even though the parameters indicate the presence of
structure, they are at the lower end of the spectrum. The central highlands in Sri Lanka
where populations of Gordonia are currently restricted were formed through vertical
uplifting during the Miocene period (Vithanage, 1972) and there has been considerable
erosion and also downwarping in the region (Vithanage, 1972) since then and these
processes have given rise to the formation of the current terrain. Therefore, if the
founding populations of Gordonia had arrived in Sri Lanka during the Miocene or prior
to that after the collision of the Deccan plate with the rest of the Asian landmass, then the
changing topography of the region as well as the changes in the climatic conditions could
have affected the distribution of these populations. These changes in the climate and
topography could have lead to the expansion and contraction of the populations, which in
turn would have enabled the populations to come together and become isolated
interchangeably. Indeed ENMs that were developed for the populations of Gordonia in
Sri Lanka for the last glacial maximum, when the climatic conditions were much cooler,
99
do show that the populations during that period could have been at elevations lower than
at present. In that case the chances of dispersal among the populations within Sri Lanka
would have been quite high given the winged nature of the seeds. The constant mingling
of the populations could have prevented long-term isolation which could have lead to
speciation. However, it has been suggested through studies of geographical distribution
of angiosperms in Sri Lanka, that fine scale allopatry does exist with 15 floristic regions
identified within Sri Lanka and each mountain range assigned its own floristic region
(Gunatilleke & Ashton, 1987). The best example for fine scale alloparty in Sri Lanka is
its wide variety of Dipterocarps. Dipterocarpaceae is most diverse in the submontane
forests of southern Sri Lanka and all but one species of the ca. 50 dipterocarp species in
Sri Lanka are endemic (Ashton, 1988). Unlike Gordonia, most species of
Diptercarpaceae are restricted to a single mountain and species believed to be closely
related are allopatric, by mountain, elevation, topography and possibly edaphic
conditions (Ashton, 1988, Ashton & Gunathilake, 1987). As was mentioned previously
G. speciosa is morphologically very distinct from the rest of the Gordonia “species” in
Sri Lanka. My results show that even though there is gene flow between populations of
G. speciosa and the rest of the Gordonia in the same locality as well as the other
localities, they are genetically most distinct from the populations in the Knuckles region.
This result was well supported by the PCA, PCoA, neighbor joining and UPGMA. The
STRUCTURE analysis results indicated that K>1. This could be an indication that G.
speciosa represent a group that had deviated from the rest and is now mixing in again, or
a group that is currently undergoing speciation and is deviating from the rest of the
population.
100
While the results of my study does not provide enough information about the
species limit of the genus within Sri Lanka, it does provide some important and useful
insight into the past history of this group of plants within the country and also aids in
finding its placement within Theaceae. This information can be used as a foundation for
future studies of plant biogeography in Sri Lanka.
101
Mat
rix
Prev
ious
ly
publ
ishe
d se
quen
ces
of
Thea
ceae
New
Se
quen
ces
(oth
er th
an
Sri L
anka
n an
d In
dian
)
Sequ
ence
s fr
om S
ri La
nkan
G
ordo
nia
sp.
sam
ples
Sequ
ence
s fr
om In
dian
G
ordo
nia
sp.
sam
ples
Tota
l nu
mbe
r of
sequ
ence
s
Alig
ned
Leng
th o
f th
e m
atrix
trnL
-trnL
F 13
(3)
4 (1
) 24
5
46
940
mat
R 13
(3)
8 (5
) 15
5
41
874
trnH
-psb
A 0
7 (5
) 14
0
21
758
trnL
-trnL
F &
mat
R co
ncat
enat
ed
13 (3
) 4
(1)
15
5 37
18
14
trnL
-trnL
F, m
atR
and
trnH
-psb
A co
ncat
enat
ed
0 5
(3)
12
0 17
25
72
Tab
le 4
.1: D
escr
iptio
n of
seq
uenc
es a
nd d
ata
mat
rices
use
d in
phy
loge
netic
ana
lyse
s. T
he n
umbe
rs w
ithin
par
enth
eses
und
er b
oth
prev
ious
ly p
ublis
hed
and
new
seq
uenc
es in
dica
te s
eque
nces
from
Chi
nese
spe
cies
of P
olys
pora
.
102
Sample Population M2 M2 M3 M3 M5 M5 S001 Adam's Peak 144 146 115 121 87 87 S002 Adam's Peak 144 146 103 111 87 91 S003 Adam's Peak 142 142 92 115 87 89 S004 Adam's Peak 142 146 107 113 87 91 S005 Adam's Peak 144 146 107 115 87 99 S006 Adam's Peak 139 144 113 121 87 87 S007 Adam's Peak 139 146 103 117 97 97 S009 Adam's Peak 142 144 117 119 87 89 S015 Nuwara-Eliya 139 146 107 115 93 93 S019 Knuckles Range 142 146 117 119 85 85 S020 Knuckles Range 142 146 111 111 91 97 S021 Knuckles Range 139 142 107 109 83 87 S023 Knuckles Range 142 146 111 119 91 97 S024 Knuckles Range 144 146 113 119 83 89 S025 Knuckles Range 146 146 113 119 87 89 S026 Knuckles Range 142 144 107 113 87 95 S028 Nuwara-Eliya 146 146 113 119 87 91 S029 Nuwara-Eliya 137 142 103 119 89 89 S031 Nuwara-Eliya 142 144 115 121 85 85 S032 Nuwara-Eliya 142 144 107 115 85 89 S034 Nuwara-Eliya 144 146 119 121 87 97 S036 Nuwara-Eliya 137 142 107 119 89 95 S037 Nuwara-Eliya 142 146 107 111 85 93 S038 Nuwara-Eliya 142 146 111 117 85 91 S040 Knuckles Range 146 155 111 115 85 89 S042 Knuckles Range 144 146 113 119 83 89 S045 Knuckles Range 142 146 111 113 85 91 S052 Knuckles Range 139 146 107 121 85 87 S053 Knuckles Range 139 146 119 121 87 91 S056 Knuckles Range 139 142 109 119 91 97 S058 Knuckles Range 139 146 103 109 93 97 S070 Knuckles Range 139 146 100 111 87 93 S078 Nuwara-Eliya 141 144 115 125 87 87 S079 Nuwara-Eliya 142 149 94 119 87 93 S080 Nuwara-Eliya 142 146 105 111 85 93 S082 Nuwara-Eliya 144 149 113 115 85 95 S087 Nuwara-Eliya 142 144 109 115 85 87 S089 Nuwara-Eliya 137 142 107 115 85 91 S097 Nuwara-Eliya 144 146 115 119 91 93 S099 Nuwara-Eliya 144 146 113 113 87 87 S100 Adam's Peak 139 142 103 115 87 87 S101 Adam's Peak 144 146 109 113 87 91 S102 Adam's Peak 144 153 109 113 87 89
103
S103 Adam's Peak 144 151 115 117 87 91 S104 Adam's Peak 144 144 111 115 87 93 S106 Adam's Peak 144 144 111 117 87 91 S107 Adam's Peak 144 153 113 117 123 123 S108 Nuwara-Eliya 139 146 92 117 89 93 S120 Adam's Peak 144 149 109 121 87 91
Table 4.2: Sample numbers, the groups they were assigned, and lengths of two alleles for three microsatellite loci M2, M3 and M5. Groups are color coded.
104
Allele Frequency
M2
137 0.030 139 0.110 141 0.010 142 0.230 144 0.250 146 0.290 149 0.040 151 0.010 153 0.020 155 0.010
M3
92 0.020 94 0.010
100 0.010 103 0.050 105 0.010 107 0.100 109 0.080 111 0.130 113 0.140 115 0.150 117 0.080 119 0.140 121 0.070 125 0.010
M5
83 0.034 85 0.132 87 0.298 89 0.143 91 0.126 93 0.115 95 0.023 97 0.063 99 0.006
123 0.012 Table 4.3: Different alleles and their frequencies at each locus for the entire population of Gordonia in Sri Lanka (Frequencies at M5 are the adjusted frequencies according to Micro-Checker -Brooksfield1 estimation)
105
Population parameters for the
entire population (including the Null alleles)
Population parameters for the samples without null
alleles Locus M2 M3 M5 M2 M3 M5 N 75 110 83 50 50 50 Na 10 16 10 10 14 10 Ne 5.097 9.490 5.601 4.655 9.124 5.734 I 1.810 2.398 1.919 1.737 2.333 1.953 Ho 0.907 0.964 0.735 0.900 0.960 0.780 He 0.804 0.895 0.821 0.785 0.890 0.826 uHe 0.809 0.899 0.826 0.793 0.899 0.834 F -0.128 -0.077 0.105 -0.146 -0.078 0.055
Table 4.4: Heterozygosity, Fstatistics and Polymorphism at each Locus for Codominant Data
Na = No. of Different Alleles Ne = No. of Effective Alleles = 1 / (Sum pi^2) I = Shannon's Information Index = -1* Sum (pi * Ln (pi)) Ho = Observed Heterozygosity = No. of Hets / N He = Expected Heterozygosity = 1 - Sum pi^2 uHe = Unbiased Expected Heterozygosity = (2N / (2N-1)) * He F = Fixation Index = (He - Ho) / He = 1 - (Ho / He) Where pi is the frequency of the ith allele for the population & Sum pi^2 is the sum of the squared population allele frequencies.
106
N
Na
Ne
I H
o H
e uH
e F
All
sam
ples
M
ean
89.3
33
12.0
00
6.72
9 2.
042
0.86
8 0.
840
0.84
5 -0
.033
SE
!10
.588
2.
000
1.38
8 0.
181
0.06
9 0.
028
0.02
7 0.
071
Sam
ples
with
out
null
alle
les
Mea
n 50
.000
11
.333
6.
505
2.00
8 0.
880
0.83
4 0.
842
-0.0
56
SE!
0.00
0 1.
333
1.34
6 0.
174
0.05
3 0.
031
0.03
1 0.
059
T
able
4.5
: Het
eroz
ygos
ity, F
stat
istic
s and
Pol
ymor
phis
m fo
r th
e en
tire
popu
latio
n fo
r Cod
omin
ant D
ata
Na
= N
o. o
f Diff
eren
t Alle
les
Ne
= N
o. o
f Eff
ectiv
e A
llele
s = 1
/ (S
um p
i^2)
I =
Sha
nnon
's In
form
atio
n In
dex
= -1
* Su
m (p
i * L
n (p
i))
Ho
= O
bser
ved
Het
eroz
ygos
ity =
No.
of H
ets /
N
He
= Ex
pect
ed H
eter
ozyg
osity
= 1
- Su
m p
i^2
uHe
= U
nbia
sed
Expe
cted
Het
eroz
ygos
ity =
(2N
/ (2
N-1
)) *
He
F =
Fixa
tion
Inde
x =
(He
- Ho)
/ H
e =
1 - (
Ho
/ He)
W
here
pi i
s the
freq
uenc
y of
the
ith a
llele
for t
he p
opul
atio
n &
Sum
pi^
2 is
the
sum
of t
he sq
uare
d po
pula
tion
alle
le
freq
uenc
ies.
107
Population Locus DF ChiSq. Probability Other M2 15 7.867 0.929 Other M3 28 23.111 0.727 Other M5 10 18.000 0.055 G_speciosa M2 10 5.136 0.882 G_speciosa M3 28 21.333 0.811 G_speciosa M5 10 4.840 0.902
Table 4.6: Summary table of Chi-square tests for the HWE analysis of the Adams Peak population. The population was broken down into two groups with G. speciosa samples in one and the rest of the samples of Gordonia (named as “other” in the table). As the p> 0.05 for both populations at each locus, I failed to reject the null hypothesis (Ho = Departure from random mating expectations, HA= Departure from random mating expectations).
108
Pop Locus DF ChiSq Prob Significance G.speciosa M2 10 5.136 0.882 ns G.speciosa M3 28 21.333 0.811 ns G.speciosa M5 10 4.840 0.902 ns Other Gordonia M2 36 33.383 0.594 ns Other Gordonia M3 91 56.926 0.998 ns Other Gordonia M5 36 69.964 0.001 ***
Key: ns=not significant, * P<0.05, ** P<0.01, *** P<0.001 Table 4.7: Summary table of Chi-square tests for the HWE analysis of G. speciosa population and the rest of the Gordonia populations within Sri Lanka. p> 0.05 for both populations at each locus except for the other populations of Gordonia at M5.
109
Table 4.8: Summary table of Chi-square tests for the HWE analysis of the three main populations (Adams Peak, Knuckles and Nuwara_Eliya). As the p> 0.05 for both populations at each locus, I failed to reject the null hypothesis (Ho = Departure from random mating expectations, HA= Departure from random mating expectations) for all populations at each locus except at M5 in the Adams Peak population (indicated by *).
Population Locus DF ChiSq Probability Adams_Peak M2 21 9.301 0.987 Adams_Peak M3 45 31.324 0.939 Adams_Peak M5 21 38.699 0.011*
Knuckles M2 10 6.454 0.776 Knuckles M3 45 34.286 0.877 Knuckles M5 28 29.767 0.374
Nuwara_Eliya M2 21 20.610 0.483 Nuwara_Eliya M3 78 63.825 0.877 Nuwara_Eliya M5 21 13.506 0.890
110
Source df SS MS Estimated Variance
% of the total
variance Among Populations 2 5.275 2.637 0.046 3% Among Individuals 46 52.481 1.141 0.000 0% Within Individuals 49 64.500 1.316 1.316 97% Total 97 122.255 1.362 100%
Table 4.9: Summary AMOVA table for the three populations (Adams Peak, Nuwara-Eliya and Knuckles). Fst = 0.036, Fst max= 0.181, F’st = 0.199, Nm = 6.686. (df= degrees of freedom, SS = sum of squares, MS= mean sum of squares)
111
K
Ln P (D) Averaged over 15 replicate
runs
Delta K (Evanno
et al., 2005)
Posterior Probability (Prichard
et al., 2000)
1 -610.313 _ 0.879 2 -612.944 0.855 0.063 3 -613.044 0.03 0.057
Table 4.10: Summary table for the results of the STRUCTURE analysis for three main populations (Adam’s Peak, Nuwara – Eliya and Knuckles)
112
K
Ln P (D) Averaged over 15 replicate
runs
Delta K (Evanno et al., 2005)
Posterior Probability
(Pritchard et al., 2000)
1 -273.044 - 0.256 2 -272.825 0.372 0.319 3 -273.06 0.123 0.315 4 -273.44 0.405 0.172
Table 4.11: Summary table for the results of the STRUCTURE analysis for the G. speciosa samples from Adam’s Peak region and the Gordonia species in the Knuckles region.
113
Fig. 1. Gordonia s.l. in Sri Lanka. A. Habit of G. ceylanica. B-D. Flowers of G. elliptica, G. speciosa, and G. dassanayakei. E. Capsule of G. ceylanica. F. Winged seed of G. brenesii. Photo credit: A. Gunathilake (A &F), D. Yakandawala (B, C, & D), P. Karunathilake (E).
Figure 4.1: Gordonia s.l. in Sri Lanka. A. Habit of G. ceylanica B-D. Flowers of G. elliptica, G. speciosa and G. dassanayakei E. Capsule of G. ceylanica F. Winged seed of G. brandegeei Photo credit: A Gunathilake (A &F), D. Yakandawala (B, C & D), P. Karunathilake (E)
114
Figure 4.2: Locations of the sites where Gordonia species from Sri Lanka were collected. All of these locations are in the central highlands of Sri Lanka (area within the red box in the figure on left).
115
Figure 4.3: Distribution of the samples that were allocated to three populations under study with in the central highlands of Sri Lanka. Purple - Adams Peak, Orange – Nuwara-Eliya and blue – Knuckles. Samples from southeastern populations were not included in this section of the analysis due to problems with amplification or the presence of null alleles.
116
Figure 4.4: Results of the Bayesian phylogenetic analysis of trnLc-f and matR regions of Gordonia sp. From Sri Lanka and other members of Gordonia s.l. Bayesian probabilities are shown above the branches. Sri Lankan species are shown in red, while the Chinese species are in blue and G. obtusa samples are in orange
117
Figure 4.5: Results of the Bayesian phylogenetic analysis of trnLc-f, matR and trnH-psbA regions of Gordonia sp. from Sri Lanka (red), Chinese species of Polyspora (blue) and two more members of Theaceae. Bayesian probabilities are shown above the branches
118
Figure 4.6: Haplotype diagrams generated for the combined data set of trnL-trnLF and matR (A) and for the data set of trnL-trnLF, matR and trnH-psbA. G. obtusa samples were included only in the combined data set of trnL-trnF and matR. The * sign in both A and B indicate the ancestral haplotype.
A
*
B
*
119
0.000#0.100#0.200#0.300#0.400#0.500#0.600#0.700#0.800#0.900#
0.000#
2.000#
4.000#
6.000#
8.000#
10.000#
12.000#
Adams_Peak# Knuckles# Nuwara_Eliya#
Heterozygosity+
Mean+
Popula3ons+
Allelic+Pa6erns+across+Popula3ons+
Na#
Ne#
No.#Private#Alleles#
He#
Figure 4.7: Mean allelic patterns across the three populations (Na: Mean number of alleles, Ne: Mean number of effective alleles He: Expected heterozygosity)
120
0.423
0.655 0.831
0.185
Knuckles
0.019
0.048
Nuwara Eliya
4.996 12.678
Adams Peak
0.035
6.900
0.742
0.299
Figure 4.8: Schematic diagram showing the location of the three populations within the central mountains of Sri Lanka and the pairwise Nei’s genetic distance (in black) and Nei’S genetic identity (in red), FST (blue) and nm (orange) between them.
Adams Peak
121
Figure 4.9: The difference in the allele frequencies between G. speciosa populations and the Gordonia populations in the Knuckles region containing mainly G. elliptica and G. ceylanica. G. speciosa is restricted to the southwestern end of the distribution range of the genus with in Sri Lanka while the Knuckles population is at the northern most end of the range.
0.000#
0.100#
0.200#
0.300#
0.400#
0.500#
0.600#
139# 144# 151# 155# 103# 109# 113# 117# 121# 85# 89# 93# 97#
M2# M3# M5#
Freq
uency+
Locus+
Allele+Frequency+Difference+between+G.#speciosa#popula3on+and+the+Knuckles+Gordonia#+popula3on+
G.speciosa#
Knuckles#
122
Figure 4.10: The PCA (A) and the PCoA (B) using the genotypes of the G. speciosa population (green dots) in the Adams Peak region and the Gorodnia population in the Knuckles region (blue dots).
B
A
123
Figure 4.11: The neighbor joining tree showing the clustering of the populations from Knuckles (blue) and G. speciosa (green) from Adams Peak region as two different groups.
124
Figure 4.12: The UPGMA tree showing the clustering of the populations from Knuckles (blue) and G. speciosa (green) from Adams Peak region as two different groups.
125
Chapter 5
General conclusions
A review of the published literature indicate that the flora of Sri Lanka has
affinities mainly with the Indian and the Southeast Asian flora, and is the result of
dispersal. A signature of Gondwanan vicariance was absent within Sri Lankan flora,
contrary to a common assumption in the floristic and taxonomic literature. Even though a
few plant taxa show relationships to former Gondwanan landmasses such as Africa and
Madagascar, the divergence times of these groups are too young for them to be
considered a result of Gondwanan vicariance. The species of Polyspora (= Gordonia)
endemic to Sri Lanka are consistent with these results. The genus Polyspora is restricted
to tropical and warm regions of Asia. The Sri Lankan species are thus expected to
closely related to taxa from other parts of Asia, and thus a result of dispersal.
The comparative analysis of seed coat micromorphology in Theaceae shows
variation that consistently distinguishes the three tribes of the family. Seeds of tribe
Theeae are characterized by smooth seed coats with small isodiametric or elongated cells,
while Gordonieae are characterized by seed coats with irregular protrusions. Seeds of
Stewartieae have seed coats made up of small isodiametric cells and some groups have
intricate sculpting visible. These morphological features can be used to assign species of
uncertain relationships, including members of Gordonia s.l. and fossil taxa. Results
support the conclusion that G. lasianthus and G. brandegeei are the only species of
Gordonia s.l. in Gordonieae. Seed coat micromorphology of all other species of
Gordonia s.l. sampled, including all Sri Lankan taxa, is consistent with placement in
Theeae, in either Polyspora or Laplacea.
126
The phylogenetic analyses of Polyspora using chloroplast and mitochondrial
sequence data indicate that the Sri Lankan species are nested within Theeae and closely
related to Indian and Chinese species of Polyspora. Polyspora and Theaceae as a whole
are characterized by low sequence variation, and thus relationships among taxa within Sri
Lanka were unresolved. Population genetic analyses using three microsatellite loci show
some genetic structuring within the country for the three main populations identified
(Adams Peak, Nuwara Eliaya and Knuckles) based on geographical distance. However,
these results are not supported by a STRUCTURE analysis, which indicate that the entire
population of Polyspora within Sri Lanka behaves as one panmictic group. Before any
taxonomic changes are made, data from additional loci are needed to confirm these
results.
Ecological Niche Models developed for populations of Polyspora in the Western
Ghats region of India and Sri Lanka show reciprocal areas of suitable habitats for
endemic species in each region. Models for the last glacial maximum show that these
populations in both Sri Lanka and the Western Ghats were at lower elevations and more
extensive. However, the models suggest that these populations were still isolated from
each other, despite the greater extent and the land connection between Sri Lanka and
India. The distribution models generated for the Sri Lankan species for 2080AD using the
projected changes in the climate show drastic reductions in suitable habitat for these
endangered species within the central highlands of Sri Lanka.
The results of this dissertation stress upon the uniqueness of the Sri Lankan
endemic flora. Also it provides more credibility to the hypothesis of a close affinity
between the Sri Lankan flora and the South and Southeast Asian flora and thereby builds
127
a foundation for future research in plant biogeography in the region. In addition the
results also sends a strong message about the re-evaluating the conservation priorities of
this understudied biodiversity hotspot especially in the light of the projected future
changes in the climate.
128
REFERENCES
Abeywickrama BA (1956). The origin and affinities of the flora of Ceylon. Proceedings of the Annual Sessions of Ceylon Association for the Advancement of Science, 11th, Part 2. 99-121.
Abeywickrama BA (1958). The evolution of the flora of Ceylon. Proceedings of the
Annual Sessions of Ceylon Association for the Advancement of Science, 14th (2) 217-219.
Adams JM & Faure H (1997). Preliminary vegetation maps of the world since the last
glacial maximum: an aid to archaeological understanding. Journal of Archaeological Science 24: 623-647.
Airy-Shaw HK (1936). Notes on the genus Schima and on the classification of the
Theaceae - Camellioïdeae. Bulletin of Miscellaneous Information (Royal Gardens, Kew): 496-499.
Antonelli A (2008). Higer level phylogeny and evolutionary trends in Campanulaceae
subfam. Lobelioideae: molecular signal overshadows morphology. Molecular Phylogenetics and Evolution 46: 1-18.
Ashton MS, Gunatilleke S, Zoysa de N, Dassanayake MD, Gunatilleke N, & Wijesundara
S (1997). A field guide to the common trees and shrubs in Sri Lanka. Colombo, Sri Lanka: Wildlife Heritage Trust of Sri Lanka. vii + 432.
Ashton PS (1988). Dipterocarp biology as a window to the understanding of tropical
forest structure. Annual Review of Ecology & Systematics 19: 347-370.
Ashton PS & Gunatilleke CVS (1987). New light on the plant geography of Ceylon I: historical plant geography. Journal of Biogeography 14: 249-285.
Axelrod DI (1974). Plate tectonics in relation to the history of the angiosperm vegetation
in India. Lucknow, India: Birbal Sahini institute of Paleobotany. Bahulikar RA, Lagu MD, Kulkarni BG, Pandit SS, Suresh HS, et al. (2004). Genetic
diversity among spatially isolated populations of Eurya nitida Korth. (Theaceae) based on inter - simple sequence repeats. Current Science 86: 824-831.
Balasubramanium S (1985). Tree flora of Sri Lanka. International Conference on Timber Technology, Moratuwa - Sri Lanka. 58-67.
Baldwin MF (1991). Natural resources of Sri Lanka. Conditions and trends. A report
prepared for the Natural Resources, Energy and Science Authority of Sri Lanka. Colombo, Sri Lanka: Keels Business Systems Ltd. 280p.
129
Barker WR (1980). Taxonomic revisions in Theaceae in Papuasia I. Gordonia, Ternstroemia, Adinandra and Archboldiodendron. Brunonia 3: 1-60
Bartholomew B & Tienlu M (2005). New combinations in Chinese Polyspora (Theaceae). Novon 15: 264-266.
Bartholomew B & Tienlu M (2007). Theaceae. Pp 366-478 in: Flora of China - eFloras.
Published on the Internet http://www.efloras.org [accessed on February, 10, 2015]. Missouri Botanical Garden, St. Louis, MO & Harvard University Herbaria, Cambridge, MA.
Bellard C, Bertelsmeier C, Leadley P, Thuiller W & Courchamp F (2012). Impacts of climate change on the future of biodiversity. Ecology Letters 15: 365-377.
Benjamin SP (2011). Collaborative studies on the fauna of the Western Ghats-Sri Lanka
biodiversity hotspot. Current Science 100: 280-280. Berry EW (1929). Gordonia from the miocene of Idaho and Washington. American
Journal of Sceince 18: 429-432. Bonebrake TC & Mastrandrea MD (2010). Tolerance adaptation and precipitation
changes complicate latitudinal patterns of climate change impacts. PNAS 107: 12581-12586.
Bossuyt F, Meegaskumbura M, Beenaertd N, Gower DJ, Pethiyagoda R, et al. (2004).
Local endemism within the Western Ghats- Sri Lanka biodiversity hotspot. Science 306: 479-481.
Briggs JC (2003). The biogeographic and tectonic history of India. Journal of
Biogeography 30: 381-388. Brown JH & Gibson AC (1983). Biogeography. St. Louis: Mosby. 643 p. Burt TP & Weerasinghe KDN (2014). Rainfall distribution in Sri Lanka in time and
space: an analysis based on daily rainfall data. Climate 2 (4): 242-263. Cavieres LA, Brooker RW, Butterfield BJ, Cook BJ, Kikvidze Z, et. al. (2013).
Facilitative plant interactions and climate simultaneously drive alpine plant diversity. Ecology Letters 17: 193-202
Chanderbali AS, van der Werff H & Renner SS (2001). Phylogeny and historical
Biogeography of Lauraceae: evidence from the chloroplast and the nuclear genomes. Annals of the Missouri Botanical Garden 88: 104-134.
Chandran MDS (1997). On the ecological history of the Western Ghats. Current Science
73: 146-155.
130
Chapman VJ (1974). The application of aerial photography to ecology as exemplified by the natural vegetation of Ceylon. Indian Forester 73: 287-314
Chapuis M & Estoup A (2007). Microsatellite null alleles and estimation of population differentiation. Molecular Biology Evolution 24: 612-631.
Chen L, Grimm GW, Wang Q & Renner SS (2015). A phylogeny and biogeographic analysis for the Cape- Pondweed family Aponogetonaceae (Alismatales). Molecular Phylogenetics and Evolution 82: 111-117.
Cincotta RP, Wisnewski J & Engelman R (2000). Human population in the biodiversity
hotspots. Nature 404: 990-992. Conti E, Eriksson T, Schönenberger J, Systma KJ & Baum DA (2002). Early tertiary out-
of-India dispersal of Crypteroniaceae: evidence from phylogeny and molecular dating. Evolution 56: 1931-1942
Cook EK (1953). Ceylon; its geography, its resources and its people. London: Macmillan
and Company Ltd. ix + 360. Cooray PG (1967). The geology of Ceylon. Colombo, Sri Lanka: National Museums of
Ceylon. xxvii + 324. Cruz H (1973). Nature conservation in Sri Lanka (Ceylon). Biological Conservation 5:
199-208 Damayanthi MKN & Gamage D (2011). Transformation of smallholder agriculture
Sector in Sri Lanka: an annotated compendium of statistics. Colombo, Sri Lanka: Hector Kobbekaduwa Agrarian Research and Training Institute. xii + 161.
Darriba D, Taboada GL, Doallo R & Posada D (2012). jModelTest 2: more models, new heuristics and parallel computing. Nature Methods 9: 772.
Das A, Krishnaswamy J, Bawa KS, Kiran MC, Sirinivas V, et al. (2006). Proritisation of conservation areas in the Western Ghats, India. Biological Conservation 133: 16-31.
de Queiroz A (2005). The resurrection of oceanic dispersal in historical biogeography.
Trends in Ecology and Evolution 20: 68-73. Dittus WPJ (1977). The ecology of s semi - evergreen forest community in Sri Lanka.
Biotropica 9: 268-286 Doyle JA, Sauquet H, Scharaschkin T & Thomas AL (2004). Phylogeny, molecular and
fossil dating, and biogeographic history of Annonaceae and Myristicaceae (Magnoliales). International Journal of Plant Sciences 165: S55-S67
131
Draper D, Roselló-Graell A, Garcia C, Gomes CT & Sérgio C (2003). Application of GIS
in plant conservation programmes in Portugal. Biological Conservation 113: 337-349.
Dufresne J-L, Foujols M-A, Denvil S, Caubel A, Marti O et al. (2013). Climate change
projectons using the IPSL - CM5 earth system model: from CMIP3 to CMIP5. Climate Dynamics 40: 2123-2165.
Edirisooriya G & Dharmagunawardhane HA (2013). New species of Bennettitales from
the jurassic rocks of Sri Lanka. International Journal of Scientific and Research Publications 3: 1-7.
Edwards WN (1935). The flora of the London clay. In: EM Reid, MEJ Chandler, editors.
The London Clay Flora. London: British Museum of Natural History. xxxiii + 561.
Elith J, Graham CH, Anderson RP, Dudík M, Ferrier S. et. al (2006). Novel methods
improve prediction of species' distribution from occurence data. Ecography 29: 129-151.
Erdelen W (1984). The genus Calotes (Sauria, agamidae) in Sri Lanka: distributon
patterns. Journal of Biogeography 11: 515-525. Erdelen W, Preu C (1990). Quaternary coastal vegetation dynamics in the Palk strait
region, south Asia - the evidence and hypotheses. In: Thornes JB, editor. Vegetation and Erosion. London: John Wiley & Sons Ltd. pp. 491-520.
Erdelen W (1996). Tropical rain forests of Sri Lanka: characteristics, history of human
impact, and the protected area system. In Edwards D.S., Booth W.E., Choy S.C. eds. Tropical rainforest research - Current Issues. Netherlands: Kluwer Academic Publishers. 503-511.
Evanno G, Regnaut S, Goudet J ( 2005). Detecting the number of clusters of individuals
using the software STRUCTURE: a simulation study. Molecular Ecology 14: 2611-2620.
Excoffier L, Smouse PE, & Quattro JM (1992). Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial - DNA restriction data. Genetics 131: 479-491.
Fernando CH (1984). Preface: ecology and biogeography in Sri Lanka. In: Fernando C.H. ed. Ecology and biogeography in Sri Lanka. Boston: Dr. W. Junk Publishers. V-VII.
132
Fernando P, Pfrender ME, Encalada SE, Lande R (2000). Mitochondrial DNA variation, phylogeography and population structure of the Asian Elephant. Heredity 84: 362-372.
Flenley JR (1998). Tropical forests under the climates of the last 30,000 years. Climate
Change 39: 177-197.
Freeman S, West J, James C, Lea V & Mayess S (2004). Isolation and characterization of highly polymorphic microsatellites in tea (Camellia sinensis). Molecular Ecology Notes 4: 324-326.
Gimaret-Carpentier C, Dray S, Pascal J-P (2003). Broad - scale biodiversity pattern of the endemic tree flora of the Western Ghats (India) using canonical correlation analysis of herbarium records. Ecography 26: 429-444.
Givinish TJ & Renner SS (2004). Tropical intercontinental disjunctions: Gondwana
breakup, immigration from the boreotropics and transoceanic dispersal. International Journal of Plant Sciences 165: S1-S6.
Gower DJ, Dharne M, Bhatta G, Giri V, Vyas R, et al. (2007). Remarkable genetic
homogeneity in unstriped, long tailed Ichthyophis along 1500km of the Western Ghats, India. Journal of Zoology 272: 266-275.
Groeninckx I, Dessein S, Ochoterena H, Persson C, Motley TJ, et al. (2009). Phylogeny
of the herbaceous tribe Spermacoceae (Rubiaceae) based on plastid DNA data. Annals of the Missouri Botanical Garden 96: 109-132.
Grote, PJ & Dilcher DL (1989). Investigations of angiosperms from the Eocene of North
America: a new genus of Theaceae based on fruit and seed remains. Botanical Gazette 150: 190-206.
Grote PJ & Dilcher DL (1992). Fruits and seeds of tribe Gordonieae (Theaceae) from the Eocene of North America. American Journal of Botany 79: 744-753.
Guindon S & Gascuel O (2003). A simple, fast and accurate method to estimate large
phylogenies by maximum-likelihood. Systematic Biology 52: 696-704. Gunathilake LAAH, Prince JS & Whitlock BA (2015). Seed coat micromorphology of
Gordonia sensu lato (including Polyspora and Laplacea; Theaceae). Brittonia 67 (3): 68-78
Gunathilleke CVS & Ashton PS (1987). New light on the plant geography of Ceylon II. The ecological biogeography of the lowland endemic tree flora. Journal of Biogeography 14; 295-327.
Gunatilleke IAUN & Gunatilleke CVS (1990). Distribution of floristic richness and its
conservation in Sri Lanka. Conservation Biology 4: 21-31
133
Gunatilleke IAUN & Gunatilleke CVS (1991). Threatened woody endemics of the wet lowlands of Sri Lanka and their conservation. Biological Conservation 55: 17-36.
Gunatilleke IAUN, Gunatilleke CVS & Dilhan MAAB (2005). Plant biogeography and
conservation of the south - western hill forests of Sri Lanka. The Raffles Bulletin of Zoology Supplement 12: 9-12.
Gunawardene NR, Daniela AE, Gunatilleke IAUN, Gunatilleke CVS, Karunaran PV, et
al. (2007). A breif overview of Western Ghats - Sri Lanka biodiversity hotspot. Current Science 93: 1567-1572.
Hammer O, Harper DAT & Ryan PD (2001). PAST: Paleontological statistics software package for education and data analysis. Palaeontologica Electronica 4: 9pp.
Hawkins BA, Field R, Cornell HV, Currie DJ, Guegan J, et al. (2003). Energy, water and broad-scale geographic patterns of species richness. Ecology 84: 3105-3117.
Heart TR & Theobold WL (1977). Comparative studies of vegetative anatomy in the
Theaceae in Sri Lanka. Botanical Journal of the Linnaean. Society. 75: 375-386. Hedrick PW (2005). A standardized genetic differentiation measure. Evolution 59: 1633-
1638. Heinrichs J, Bombosch A, Feldberg K, Kreier H, Hentschel J. et al. (2012). A phylogeny
of the northern temperate leafy liverwort genus Scapania (Scapaniaceae, Jungermanniales). Molecular Phylogenetics and Evolution 62: 973-985.
Hentschel J, Zhu R, Long DG, Davison PG, Schneider H., et al. (2007). A phylogeny of
Porella (Porellaceae, Jungermanniopsida) based on nuclear and chloroplast DNA sequences. Molecular Phylogenetics and Evolution 45: 693-705.
Hewitt G (2000). The genetic legacy of the quaternary ice ages. Nature 405: 907-913. Hijmans RJ, Cameron SE, Parra JL, Jones PG & Jarvis A (2005). Very high resolution
interpolated climate surfaces for global land areas. International Journal of Climatology 25: 1965-1978.
Holmes CH (1956). The broad patter of climate and vegetational distribution in Ceylon.
Study of tropical vegetation - proceedings of the Kandy symposium. Kandy, Sri Lanka: 99-113
Huelsenbeck JP & Ronquist F (2001). MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17: 754-755.
134
IUCN Sri Lanka and the Ministry of Environment and Natural Resources. (2007). The 2007 red list of threatened flora and fauna of Sri Lanka. Colombo, Sri Lanka: IUCN Sri Lanka and the Ministry of Environment and Natural Resources-Sri Lanka. xiii+148.
Kale VS, Shejwalker N (2008). Uplift along the western margin of the Deccan Basalt
Province: is there any geomorphometric evidence? Journal of Earth System Science 117: 959-971.
Kandu SR (2005). A synopsis of Theaceae in Indian subcontinent: its distribution and
endemism. Acta Botanica Hungarica 47: 369-377. Kårehed J, Groeninckx I, Dessein S, Motely TJ & Bremer B (2008). The phylogenetic
utility of chloroplst and nuclear DNA markers and the phylogeny of the Rubiaceae tribe Spermacoceae. Molecular Phylogenetics and Evolution 49: 843-866.
Katz KB (2000). Sri Lanka - India interplate tectonics - preambrian to present. Gondwana
Research 3: 3-5. Keng H (1962). Comparative morphological studies in Theaceae. University of
California Publications in Botany 33: 269-384. Keng H (1980). On the unification of Laplacea and Gordonia (Theaceae). Gardens’
Bulletin Singapore 33: 303-311. Keng H (1984). Florae Malesianae precursors - LVIII, part two. The genus Gordonia
(Theaceae) in Malesia. Gardens’ Bulletin Singapore 37: 1-47. Kissling WD, Eiserhardt WL, Baker WJ, Borchsenius F, Couvreur TLP, et al. (2012).
Cenozoic imprints on the phylogenetic structure of palm species assemblages worldwide. PNAS 109: 7379-7384.
Kita Y & Kato M (2001). Infrafamilial phylogeny of the aquatic angiosperm
Podostemaceae inferred from the nucleotide sequences of the matK gene. Plant Biology 3: 156-163.
Kita Y & Kato M (2004). Molecular phylogeny of Cladopus and Hydrobryum
(Podostemaceae, Podostemoideae) with implications for their biogeography in East Asia. Sysematic Botany 29: 921-932.
Kobuski CE (1950). Studies in the Theaceae, XX. Notes on the South and Central
American species of Laplacea. Journal of the Arnold Arboretum 31: 405-429.
135
Kress WJ, Liu A, Newmann M & Li Q (2005). The molecular phylogeny of Alpinia (Zingiberaceae): a complex and polyphyletic genus of gingers. American Journal of Botany 92: 167-178.
Kundu SR (2005). A synopsis of Theaceae in Indian subcontinent: its distribution and
endemism. Acta Botanica Hungarica 47: 369-377. Kuttapetty M, Padmesh PP, Varghese RJ, Seeni S (2014). Genetic diversity analysis in
disjunct populations of Rhododendron arboreum from the temperate and tropical forests of Indian subcontinent corroborate Satpura hypothesis of species migration. Biologica 69: 311-322.
Li Y, Dressler S, Zhang D & Renner SS (2009). More miocene dispersal between Africa
and Asia - the case of Bridelia (Phyllanthaceae). Systematic Botany 34: 521-529. Mabberley DJ (2008). Mabberley's plant-book: a portable dictionary of plants, their
classification and uses. Cambridge University Press, Cambridge. Macey JR, Schulte II JA, Larson A, Ananjeva NB, Wang Y, et al. (2000) Evaluating
trans-Tethys migration: an example using acrodont lizard phylogenetics. Systematic Biology 49: 233-256.
Maddison WP & Maddison DR (2015) Mesquite: a modular system for evolutionary analysis. 3.02 ed. (http://mesquiteproject.org)
Malcomber ST (2002). Phylogeny of Gaertnera Lam. (Rubiaceae) based on multiple DNA markers: evidence of a rapid radiation in a widespread morphologically diverse genus. Evolution 56: 42-57.
Mantel N (1967) The detection of disease clustering and a generalized regression approach. Cancer Research 27: 209-220.
Mathavan V, Parme WKBN & Cooray PG (1999). Geology of the high grade Proterozoic terrains of Sri Lanka and the assembly of Gondwana: an update on recent developments. Gondwana Research 2: 237-250.
McLoughlin S (2001). The breakup history of Gondwana and its impact on pre -
Cenozoic floristic provincialism. Australian Journal of Botany 49: 271-300. Meegaskumbura M, Bossuyt F, Pethiyagoda R, Manamendra - Arachchi K, Bahir M et al.
(2002) Sri Lanka: an amphibian hot spot. Science 298: 379. Meiers ST, Proctor VW & Chapman RL (1999). Phylogeny and biogeography of Chara
(Charophyta) inferred from 18S rDNA sequences. Australian Journal of Botany 47: 347-360.
136
Meimberg H, Wistuba A, Dittrich P & Heubl G (2001). Molecular phylogeny of Nepenthaceae based on cladistic analysis of plastid trnK intro sequence data. Plant Biology 3: 164-175.
Meirmans PG (2006). Using the AMOVA framework to estimate a standardized genetic differnetiation measure. Evolution 60: 2399-2402.
Menon S & Bawa KS (1997). Applications of geographical information systems (GIS), remote - sensing, and a lanscape ecology approach to biodiversity conservation in the Western Ghats. Current Science 73: 134-145.
Ming TL & Bartholomew B (2007). Theaceae. Pp. 366-478. In Z. Y. Wu & P. H. Raven
(eds.), Flora of China. Missouri Botanical Garden Press, St. Louis. Miryeganeh M, Takayama K, Tateishi Y & Kajita T (2014). Long-distance dispersal by
sea-drifted seeds has maintained the global distributio of Ipomea pes-caprae subsp. brasiliensis (Convolvulaceae). PLoS ONE 9: e91836.
Miththapala S, Seidensticker J & O'Brien SJ (1996) Phylogeographic subspecies
recognition in leopards (Panthera pardus): molecular genetic variation. Conservation Biology 10: 1115-1132.
Moore JC (1960) Squirrel geography of the Indian sub-region. Systematic Biology 9: 1-
17. Morley RJ (2000). Origin and evolution of tropical rain forests. Chichester, England:
John Wiley & Sons. Morley RJ & Dick CW (2003). Missing fossils, molecular clocks and the origin of the
Melastomaraceae. American Journal of Botany 90: 1638-1644 Muller-Dombois D (1968). Ecogeograhic analysis of a climate map of Ceylon with
particular reference to vegetation The Ceylon Forester 8: 39-58. Myers N, Mittermeier RA, Mittermeier CG, Gustavo ABdaF & Kent J (2000).
Biodiversity hotspots for conservation priorities. Nature 403: 853-858.
Nei M (1972) Genetic distance between populations. The American Naturalist 106: 283-292.
Nianthi KWGR & Jayakumara MAS (2010). Progress of research on cascade irrigation systems in the dry zones of Sri Lanka. In: Shaw R & Thaitakoo D. eds. Water Communities. Bingley, UK: Emerald Group Publishing Ltd. 109-137.
O'Shea BJ (2003). Bryogeographical relationships of the mosses of Sri Lanka. J Hattori
Bot Lab 93: 293-304.
137
Oguri E, Yamaguchi T, Tsubota H, Deguchi H & Murakami N (2013). Geographical origin of Leucobryum boninense Sull. & Lesq. (Leucobryaceae, Musci) endemic to the Bonin Islands, Japan. Ecology and Evolution 3: 753-762.
Oosterhout CV, Hutchinson WF, Wills DPM & Shipley P (2004). MICRO-CHECKER: software for identifying and correcting genotyping errors in microsatellite data Molecular Ecology Notes 4: 535-538.
Orel G, Wilson PG, Curry AS & Luu HT (2012). Polyspora huongiana sp. nov.
(Theaceae) from Vietnam and notes on related species. Nordic Journal of Botany 30: 47-52.
Panabokke CR (1996). Soils and agro-ecological environments of Sri Lanka. Colombo, Sri Lanka: Natural Resources Energy and Science Authority. xvi + 220p
Peakall R & Smouse PE (2006) GENALEX 6: genetic analysis in Excel. Population gentic software for teaching and research. Molecular Ecology Notes 6.
Peakall R & Smouse PE (2012). GenAlEx 6.5: genetic analysis in Excel. Population
gentic software for teaching and research - an update. Bioinformatics 28: 2537-2539.
Perera NP (1975). A Physiognomic vegetation map of Sri Lanka. Journal of Biogeography 2: 185-203.
Philips SJ, Anderson RP & Schapire RE (2006). Maximum entropy modeling of species
geographic distributions. Ecological Modelling 190: 231-259. Philips SJ, Dudik M & Schapire RE (2004). A maximum entropy approch to species
distribution modeling. proceedings of the 21st International Conference on Machine Learning. New York: ACM Press. pp. 655-662.
Phillips S (2005). A brief tutorial on Maxent. AT & T research. Available:
http://www.cs.princeton.edu/~schapire/maxent/tutorial/tutorial.doc. Accessed February 2, 2014.
Phillips SJ & Dudik M (2008). Modeling of species distributions with Maxent: new
extensions and a comprehensive evaluation. Ecography 31: 161-175. Prince LM & Parks CR (2001). Phylogenetic relationships of Theaceae inferred from
chloroplast DNA sequence data. American Journal of Botany 88: 2309-2320. Prince LM (2009). Theaceae. Pp. 322-328. In: Flora of North America Editorial
Committee (eds.), Flora of North America north of Mexico, volume 8. Oxford University Press, Oxford and New York.
138
Prince LM (2002). Circumscription and biogeographic patterns in the Eastern North American-East Asian genus Stewartia (Theaceae: Stewartieae): insight from chloroplast and nuclear DNA sequence data. Castanea 67: 290-301.
Prince LM (2007). A brief nomenclatural review of genera and tribes in Theaceae. Aliso
24: 105-121. Pritchard JK, Stephens M & Donnelly P (2000). Inference of population structure using
multilocus genotype data. Genetics 155:945-959. Pritchard JK, Wen W & Falush D (2010). Documentation for STRUCTURE Software:
Version 2. 3. Available from http://pritchardlab.stanford.edu/structure_software Qian H & Ricklefs RE (2000). Large-scale processes and the Asian bias in species
diversity of temperate plants. Nature 407: 180-182. Ramen TRS (2006). Effect of habitat structure and adjacent habitats on birds in tropical
rainforest fragments and shaded palntations in the Western Ghats, India. Biodiversity and Conservation 15: 1577-1607.
Ramesh BR & Pascal JP (1997). Atlas of endemics of the Western Ghats (India):
distribution of tree species in the evergreen and semi-evergreen forests. Pondicherry, India: Institut français de Pondichéry. iv +403p.
Ramette A (2007). Multivariate analyses in microbial ecology. FEMS Microbial Ecology
62: 142-160. Ravindranath NH & Sukumar R (1998). Climate change and tropical forests in India.
Climatic Change 39: 563-581. Reeves C (2009). Re-examining the evidence from plate tectonics for the initiation of
Africa’s passive margins. Conference of Petroleum Exploration Society of Great Britain. London
Renner SS, Strijk JS, Strasberg D & Thébaud C (2010). Biogeography of the
Monimiaceae (Laurales): a role for East Gondwana and long-distance dispersal, but not West Gondwana. Journal of Biogeography 37: 1227-1238.
Robin VV, Sinha A & Ramakrishnan U (2010). Ancient geographical gaps and paleo-
climate shape the phylogeography of an endemic bird in the sky islands of southern India. PLoS ONE 5: e13321.
Ronquist F & Huelsenbeck JP (2003). MrBayes 3: Bayesian phylogentic inference under mixed models. Bioinformatics 19: 1572-1574.
Rosayro RA de (1950). Ecological conceptions and vegetational types with special reference to Ceylon. The Tropical Agriculturist 151: 108-121.
139
Rutschmann F, Eriksson T, Schönenberger J & Conti E (2004). Did Crypteroniaceae really disperse out of India? Molecular dating evidence from rbcl, ndhF and rpl16 intron sequences. International Journal of Plant Sciences 165: S69-S83.
Schaefer H & Renner SS (2010). A three-genome phylogeny of Momordica
(Cucurbitaceae) suggests seven returns from dioecy to monoecy and recent long-distance dispersal to Asia. Molecular Phylogenetics and Evolution 54: 553-560.
Schatz GE (1996). Malagasy/Indo-Australia-Malesian phytogeographic connections. In
Lourenco W.R. ed. Biogéographie de Madagascar. Paris: ORSTOM. 73-84. Schmitt T (2009). Biogeographical and evolutionary importance of the European high
mountain systems. Frontiers in Zoology 6: DOI 10.1186/1742-9994-6-9 Schulte II JA, Macey JR, Pethiyagoda R & Larson A (2002). Rostral horn evolution
among agamid lizards of the genus Ceratophora endemic to Sri Lanka. Molecular Phylogenetics and Evolution 22: 111-117.
Sealy JR (1958). A revision of the genus Camellia. The Royal Horticultural Society.
London. Senaratna LS (2001). A checklist of the flowering plants of Sri Lanka. Colombo, Sri
Lanka: National Science Foundation of Sri Lanka. ix+ 451 Sepkoksi Jr. JJ (1998) Rates of speciation in the fossil record. Phil Trans R Soc Lond B
353: 315-326 Seward AC & Holttum RE (1922). Jurassic plants from Ceylon. Quarterly Journal of the
Geological Society 78: 271-277. Sitholey RV (1944). Jurassic Plants from the Tabbowa series in Ceylon. Spolia Zeylanica
42: 3-17. Somaratne S, Dhanapala AH (1996). Potential impact of global climate change on forest
distribution in Sri Lanka. In: Erda L, Bolhofer W,Haq C, Lenhart S, Mukherjee SK, Smith JB, Wisniewski J, editors. Climate change vulnerability and adaptation in Asia and the Pacific. Netherlands: Kluwer Academic Publishers. pp. 129-135.
Stevens PF, Dressler S & Weitzman AL (2004). Theaceae. Pp. 463–471. In: K. Kubitzki
(ed.), The families and genera of flowering plants, volume 6. Flowering plants: dicotyledons (Celastrales, Oxalidales, Rosales, Cornales, Ericales). Springer-Verlag, Berlin.
140
Subramanyam K & Nayar MP (1974). Vegetation and phytogeography of the Western Ghats. In: Mani MS, editors. Ecology and Biogeography of India. Monographiae
Suppiah R & Yoshino MM (1984). Rainfall variations of Sri Lanka part 1: spatial and
temporal patterns. Archives for Meteorology, Geophysics, and Bioclimatology Ser. B 34: 329-340.
Thulin M, Lavin M, Pasquet R & Delgado-Salinas A (2004). Phylogeny and
biogeography of Wajira (Leguminosae): A monophyletic segregate of Vigna centered in the Horn of Africa region. Sysematic Botany 29: 903-920.
Tribsch A & Schönswetter P (2003). Patterns of endemism and compartive
phylogeography confirm palaeoenvironmental evidence for pleistocene refugia in the eastern Alps. Taxon 52: 477-497.
Tsou C-H (1997). Embryology of the Theaceae–anther and ovule development of
Camellia, Franklinia, and Schima. American Journal of Botany 84: 369-381. Tsou C-H (1998). Early floral development of Camellioideae (Theaceae). American
Journal of Botany 85: 1531-1547. Vaz GG (2000). Age of relict coral reef from the continental shelf off of Karaikal, Bay of
Bengal: Evidence of last glacial maximum. Current Science 79: 228-230. Vidya TNC, Fernando P, Melnick DJ & Sukumar R (2005). Population differentiation
within and among Asian elephant (Elephas maximus) populations in southern India. Heredity 94: 71-80.
Vithanage PW (1972). Post-Precambrian uplifts and regional neotectonic movements in Ceylon. 29th International Geological Conference . Montreal. 642-654.
Wadhwa BM (1996). Theacae. Pp. 386-395 in: A revised handbook to the Flora of Ceylon. Vol 10 (Eds. M.D. Dassanayake and W.D. Clayton). Oxford and IBH Publishing Co. Pvt. Ltd., New Delhi. pp. 386-395.
Wang YH, He H, Min TL, Zhou LH & Fritsch PW (2006). The phylogenetic position of
Apterosperma (Theaceae) based on morphological and karyotype characters. Plant Systematics and Evolution 260: 39-52.
Webb Jr. JLA (2002). Tropical Pioneers: human agency and ecological change in the
highlands of Sri Lanka, 1800-1900; In Webb Jr. JLA, series editor. Ohio University Press series in ecology and history. Athens: Ohio University Press. xviii+243.
141
Weerakoon D (2012). A breif overview of the biodiversity of Sri Lanka. In: Weerakoon D, Wijesundara S, editors. The national red list of the 2012 of Sri Lanka; conservation status of the fauna and flora. Colombo, Sri Lanka: Ministry of Environment. pp. xvii-xix
Weerasooriya A (1998). Gordonia - A genus with fascinating endemics in Sri Lanka. Sri
Lanka Nature 1: 23-24. Wen J (1999). Evolution of Eastern Asian and Eastern North American disjunct
distributions of flowering plants. Ann. Rev. Ecol. Syst. 30: 421-455.
Wen Q, Xu L, Gu Y, Huang M & Xu L (2012). Development of polymorphic microsatellite markers in Camellia chekiangoleosa (Theaceae) using 454-ESTs. American Journal of Botany: e203-e205.
Whitlock BA, Silver J & Prince JS (2010). Seed coat morphology in Gentianopsis (Gentianaceae). Rhodora 112: 58-79.
Widdowson M (1997). Tertiary paleosurfaces of the SW Deccan, Western India:
implications for passive margin uplift. In: Widdowson M, editor. Paleosurfaces: Recognition, reconstruction and paleoenvironmental interpretation. Bath, UK: The Geological Society of London. pp. 221- 248 (vi + 330).
Wijesinghe MR, De Silva VR (2012). Conservation value of forest plantations: a study of
four timber species in Sri Lanka. Journal of Tropical Forestry and Environment 2: 36-47.
Wikström N, Neupane S, Kårehed J, Motley TJ & Bremer B (2013). Phylogeny of
Hedyotis L. (Rubiaceae: Spremacoceae): redefining a complex Asian-Pacific assemblage. Taxon 62: 357-374.
Wold S, Esbensen K & Geladi P (1987). Principal component analysis. Chemometrics and Intelligent Laboratory Systems 2: 37-52.
Wright S (1943). Isolation by distance. Genetics 28: 114-138.
Yakandawala DMD & Gunathilake LAAH (2008). Phylogenetic relationships of Gordonia species endemic to Sri Lanka. Journal of National Science Foundation, Sri Lanka 36: 51-58.
Yang S-X, Yang J-B, Lei L-G, Li D-Z, Yoshina H & Ikeda T (2004). Reassessing the
relationships between Gordonia and Polyspora (Theaceae) based on the combined analyses of molecular data from the nuclear, plastid and mitochondrial genomes. Plant Systematics and Evolution 248: 45-55.
142
Yuan Y, Song Y, Geuten K, Rahelivololona E, Wohlhauser S et al. (2004). Phylogeny and biogeography of Balsaminaceae inferred from ITS sequences. Taxon 53: 391-403.
Yuan Y, Wohlhauser S, Möller M, Klackenberg J, Callmander MW, et al. (2005).
Phylogeny and biogeography of Exacum (Gentianaceae): a disjunctive distribution in the Indian Ocean basin resulted from long distance dispersal and extensive radiation. Systematic Botany 54: 21-34.
Zona S, Davis P, Gunathilake LAAH, Prince J & Horn JW (2012). Seeds of
Eriocaulaceae of the United States and Canada. Castanea 77: 37-45.