PHYLOGENY AND BIOGEOGRAPHY
OF ELATOSTEMA (URTICACEAE)
FROM MOUNT KINABALU, SABAH, MALAYSIA
REED S. BEAMAN
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Reed S. Beaman
For Ethan and Sydney
I thank everyone who provided research material, assistance, helpful advice, or
encouragement during the tenure of this work. Most notable is Walter S. Judd, my
committee chair, who has been my primary mentor for many years. I am also very
grateful to the others who have served on my committee, John Atwood, Dana Griffin, Jon
Reiskind, Scot Smith, and Norris Williams. Stephen Mulkey, although not a member of
my committee, provided food for thought. Alexis Thomas furnished the rectified digital
satellite imagery that is featured prominently throughout this document. The original raw
imagery was kindly donated to him by the EOSAT corporation.
The staff of the University of Florida Herbarium (FLAS), particularly Kent
Perkins and Trudy Lindler, provided essential assistance, without which this project
would have been impossible. The staff at A, BM, GH, K, MSC, SAN, UC, and US have
been extremely helpful, providing access to specimens and some work space. I also thank
those in the Botany Department of the University of Texas, particularly Robert Jansen,
Beryl Simpson, and Todd Barkman, for a semester of training in molecular systematics.
Mark Whitten and Norris Williams have also been extremely helpful and supportive in
my attempts to obtain nucleotide sequence data.
Many individuals have generously provided assistance, transportation, and
lodging during trips to Sabah. Datuk Lamri Ali, Francis Liew, Jamili Nais, Remi Repin,
Ludi Apin, Alim Biun, and many others from Sabah Parks made my work more
productive and enjoyable through their help, hospitality, and companionship.
Special thanks to my wife, Nicoletta Cellinese, for her encouragement and
tolerance, especially in these last few months. I also wish to thank Peter Alcorn, who has
always been ready to help in a pinch. Natasha Akopian provided some much-needed last-
minute help in preparing the manuscript. My employer, Andre Clewell, has been
extremely supportive throughout this entire endeavor, as has my co-worker Marion
Lasley. I also thank my parents, John and Teofila Beaman, for their life-long support and
Finally, I thank the National Science Foundation (Grant DEB-9400888), the
MacArthur foundation, and the American Society of Plant Taxonomists for their generous
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ........................................................................................... iv
ABSTRACT ............ ............................. ............... viii
1 M O U N T K IN A B A LU ................................................... .............................. 1
Introduction ..................................................................... . . 1
T he F lora.................................................................. . 5
Geography ........................................ 7
Geomorphology .............. ................... ........ .................... 9
G e o lo g y ................................................................................................................ 1 0
C lim ate ........................................................ ......... ..... 16
Elevational Zonation .. ..... ................................................... .............. 17
2 GEOGRAPHICAL INFORMATION SYSTEM ............................................... 29
S satellite Im ag ery ...................................................................... 3 0
Predicting Species Ranges using Digital Elevation Models ............................. 32
D diversity A nalyses through G IS ........................................ ........................ 33
3 PHYLOGENY OF THE URTICACEAE...................... ..... ......... ..... 42
In tro d u ctio n ................................. ........................................ 4 2
The U rticales .................... ................. ............................... 42
The Urticaceae ................................................... ...................... 45
Methods ............................................. 46
Results ............ ............................... ............... 47
Discussion ............................................ 50
4 PHYLOGENY OF KINABALU ELATOSTEMA ............................................... 89
In tro d u ctio n .......................................................................................................... 8 9
Elatostem a in M alesia .......................................................... 89
Generic Delimitation and Monophyly of Elatostema ................................. 89
M methods ......................................... ... ... ..... ................ .......... 90
Character and Character State Categories and Clarification ...................... 91
R results ................................... ........ ............ .................. ........... 95
Interpretation of Selected Clades. ......................................... .............. 96
Interpretation of Selected Characters ............................................................ 98
D iscu ssion .. ................................................... ............... . 10 1
Conclusions ............ ............................ ............... 104
5 BIOGEOGRAPHY OF KINABALU ELATOSTEMA .................................. 147
In tro d u ctio n ................................ ........................................ 14 7
D elim itation of A reas ................................................................................. 150
Phylogenetic and Cladistic Biogeography ................................. .............. 151
M materials and M methods ................................ .. ......................................... 153
Delimitation of Classes ..................... .............. ...... ......... 153
Im age Classification ...................................... ............. 154
Results and D discussion ............................... ......... .......... ............. 156
Im age Classification ...................................... ............. 156
Cladistic Biogeography ... .................................................................. 160
Speciation ... ....... ............................... ........... 162
6 TAXONOMIC TREATMENT ................................. 176
Introduction ............................................ .......... 176
Key to the Species of Elatostema ..................................... ............... 179
REFEREN CES ......... ............................. ...... .......... 255
BIOGRAPH ICAL SKETCH ....................... ......................................................... 263
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
PHYLOGENY AND BIOGEOGRAPHY
OF ELATOSTEMA (URTICACEAE)
FROM MOUNT KINABALU, SABAH, MALAYSIA
Reed S. Beaman
Chairman: Walter S. Judd
Major Department: Botany
Methods from phylogenetic systematics and cladistic biogeography are combined
with geographical information system (GIS) and remote sensing technology to develop a
novel approach to biogeography at a landscape scale, bridging the gap between traditional
perceptions of ecological and historical biogeography. This study focused on determining
phylogenetic and biogeographic relationships between 23 species of Elatostema
(Urticaceae), eight of which are new, from Mount Kinabalu, the highest mountain in
Borneo at 4094 m. Kinabalu is a biodiversity and endemicity hot spot of global
significance with ca. 6000 vascular plants species. Factors contributing to biodiversity
and endemicity include the extensive elevation and concomitant climate variation, diverse
habitats including those on ultramafic (serpentine) outcrops of ophiolitic origin, the
Miocene and late Pliocene uplift of the Kinabalu granitic pluton, Pleistocene glaciation,
and Borneo's history of tectonic accretion.
Phylogenetic relationships between the taxa were applied to a cladistic
biogeography analysis. The approach involved defining the areas in the analysis as
satellite image classes using a maximum likelihood supervised image classification and
class probability analysis using Landsat TM imagery of the area. Image signatures for
each class represented 12 habitats at varying elevations and geology. The assumption was
made that the image classes represent ecological entities with a basis in the historical
geology of Mount Kinabalu, Borneo, and its neighbors. Speciation, extinction, dispersal,
and failure to vicariate were accommodated by the cladistic biogeography analysis.
Vicariant speciation is hypothesized to have resulted from the geologically recent
emplacement and uplift of the Kinabalu pluton, habitat displacement during Pleistocene
climatic changes, and population fragmentation due to the cleaving action of glacial
melting to form deep river valleys. Dispersal may have taken the form of Bornean
tectonic accretion of ophiolitic fragments originating in New Guinea, rafting not just into
present day Sulawesi and the Philippines, but also into Borneo. Results indicate that
recent speciation has occurred primarily in taxa occurring at mid and high elevation
habitats. Species ofElatostema on low elevation habitats, both ultramafic and otherwise,
are phylogenetically basal, suggesting that adaptation to ultramafics may have occurred
early in the phylogenetic history of Elatostema, long before Kinabalu arose.
Kinabalu is a scenic wonder, a test for mountaineers from the amateur to
the skilled rock-climber, a holiday from the hot lowlands, and botanically
a paradise. The people of Sabah possess on this famous mountain what I
believe is the richest and most remarkable assemblage ofplants in the
E. J. H. Corner (1978, page 113)
The purpose at hand is to examine the phylogeny and biogeography of the 23
species ofElatostema Forst. (Urticaceae) known from Mount Kinabalu (Figure 1-1) in
Sabah, Malaysia. Kinabalu is located near the northern tip of Borneo, and is the tallest
mountain in Southeast Asia, between the Himalaya and New Guinea (Figure 1-2). With
an elevation of 4,093.7 m, this isolated massif stands 1000 m taller than any other in
Borneo. Kinabalu is, in a sense, an isolated alpine island. Barkman et al. (1997)
developed this analogy and the concept that the theories of island biogeography may be
applied to isolated montane floras such as that on Kinabalu. Earlier workers, including
Stapf(1894), Corner (1964), and van Steenis (1964, 1967) recognized the significance of
the Kinabalu flora. Van Steenis pioneered biogeographical work in the region by
assessing relationships between Kinabalu and neighboring mountains based on their
distance and number of shared species and genera.
Elatostema species are abundant on Kinabalu in herbaceous undergrowth,
especially along rivers and streams and near waterfalls. Some species occur on exposed
ridges and slopes. Friis (1993) estimates there are 300 species of Elatostema, but notes
that the genus is in need of study. Few species of Elatostema have been studied beyond
their original descriptions or listings in floristic treatments. The total number is probably
higher because many areas in the montane paleotropics, particularly in Borneo and New
Guinea, remain poorly explored. Shroter and Winkler (1936) monographed 106 species
of Elatostema, including six Kinabalu taxa. Their monograph included none of the 14
species of Elatostema known only from Kinabalu. The other Kinabalu species appear to
be restricted to Borneo, with the exception of E. acuminatum Brong., which occurs in
Java, Sumatra, and southern India. Descriptions of eight new species of Elatostema from
Kinabalu are included in this dissertation.
This study was conceived as a means to elucidate and quantify evolutionary
mechanisms pertaining to the high floristic diversity on Kinabalu. With as many as 6000
species of vascular plants on its slopes (Beaman and Beaman 1998), Kinabalu ranks
among the global hot-spots for biological diversity. Beaman and Beaman (1990)
hypothesized that the high diversity on Kinabalu is the collective result of several factors.
Geological, geomorphological, and ecological factors addressed herein include (1)
diversity of habitats due to elevational variation from lowland rainforest to alpine scrub,
(2) precipitous topography, often opening up new habitat owing to landslides, (3)
geological diversity and patchiness (particularly of ultramafic outcrops) resulting in
highly localized edaphic conditions, (4) historical geology and tectonic history of
Kinabalu, the rest of Borneo, and the Malesian floristic region, and (5) climatic
oscillations such as those during Pleistocene glacial and inter-glacial cycles, as well as
more recent oscillations influenced by EINiho events (Beaman et al. 1986).
An underlying assumption is that these factors have charted present day species
distributions on a geologically very recent mountain. How and to what extent do they
contribute to speciation and high diversity? While that question may never be answered
explicitly, it is possible to discern patterns in species relationships that bear on the issue
of speciation. The technical goal of this study is to determine whether patterns in the
phylogeny and distribution of species are congruent with discernible spatial attributes and
patterns in the landscape. Of special interest are the phylogenetic and biogeographic
relationships between very localized species and their habitats. The Kinabalu Elatostema
make an appropriate choice as a research taxon primarily because of the occurrence of
endemic species at mid and high elevations and on ultramafic substrates. Table 1-1 lists
the species studied and their general distributions.
Using geographical information system (GIS) and remote sensing tools, analysis
and visualization of phylogenetic and biogeographical data are greatly enhanced.
Figures 1-3 and 1-4, showing a Landsat TM image of the Kinabalu area and a three-
dimensional terrain model combined with an image drape, while visually striking, are
only a glimpse at the digital data contained therein. Remote sensing and three-
dimensional terrain modeling provide an opportunity to make localized analytical spatial
queries on ecological and physical characteristics of the places where species and clades
occur. In Chapter 2, a more detailed description of the Kinabalu GIS that forms the basis
for the spatial data set is provided.
On Kinabalu, endemic and sometimes closely related sympatric species occur in
very local areas. Spatial patterns or correlation between remotely sensed, terrain defined
data and phylogeny are explored for the Kinabalu Elatostema. Barkman et al. (1997)
addressed similar questions for the genus Dendrochilum (Orchidaceae) using a molecular
based phylogeny. An approach based on developing hypotheses of phylogenetic
relationships and cladistic biogeographic relationships between areas on Kinabalu is
developed to address issues of the origin, distribution, and diversity of Kinabalu
Elatostema, and of the mountain's flora in general. Specific questions formulated to
address the issues stated above include the following:
* Did higher elevation endemic species on Kinabalu evolve from lower elevation
species as uplift and Pleistocene glacial/interglacial cycles drastically and dynamically
altered habitat and increased natural selection pressure? Elatostema bulbothrix Stapf
is a Kinabalu endemic found only on some of the highest elevation ultramafic
outcrops. What is the relationship of this rare species to other Kinabalu Elatostema?
* Does precipitous topography hinder gene exchange between isolated populations,
driving rapid adaptive radiation and speciation? If so, one would expect to see closely
related species with micro-disjunct area relationships.
* What role does the occurrence of ultramafic geology play in the large number of
To answer these questions, a phylogeny of Kinabalu Elatostema based on
morphological characters was constructed. Elatostema is especially rich in vegetative
characters, particularly in the form and distribution of cystoliths and trichomes on both
vegetative and reproductive structures. Floral and inflorescence characters also proved to
be quite informative. The phylogenetic analysis is presented in Chapter 4, with the
higher-level analysis in Chapter 3 providing the appropriate outgroups.
In Chapter 5, a biogeographical analysis to determine relationships between taxa
and areas is presented. Novel techniques for visualizing phylogenetic patterns, levels of
endemicity, and species distributions are employed. A cladistic biogeography approach is
also employed. Congruence between evolution of Elatostema and the geological history
of Kinabalu is evaluated. Modes of speciation via vicariance and long distance dispersal
are considered, as well as the roles that differing geological substrates, changing climatic
conditions and mountain orogeny play in speciation. In addition, the possibility exists that
high biological diversity in the Kinabalu region may be due to the intermingling of floras
resulting from continental accretion. This may be even more significant than long
distance dispersal. Finally, revision and taxonomic treatment of Kinabalu Elatostema is
presented in Chapter 6, including nomenclature, descriptions, diagnoses of new species,
and specimens cited.
General observations on the flora, geography, geology, geomorphology, climate,
and elevational zonation are described in the balance of this first chapter. For further
information on the natural history of Kinabalu the reader is directed to Wong and Phillips
(1996) who provide a recently revised compendium by authors in numerous disciplines.
In a summary of global plant diversity, Barthlott et al. (1996) ranked northwest
Borneo, inclusive of Kinabalu, among the top ten biological diversity centers in the
world. They indicated that the area has more than 5,000 vascular plant species per
10,000 km2. Kinabalu actually may harbor 5,000-6,000 species in an area of only 1150
km2 (Beaman and Beaman 1998). More than 1,000 genera representing over 200 families
are known in the Kinabalu vascular plant flora. Roos (1997) asserted that the Malesian
vascular plant flora consists of about 42,000 species in an area of 3.1 million km2. Thus
the relatively small area (0.04% of Malesia) encompassed by Kinabalu may include as
much as 14% of the Malesian flora and 2% of the global flora of an estimated 300,000
vascular plant species. To place this in perspective, the flora of Florida probably
comprises about 1.3% of the global flora, and Europe about 3.7%.
Exceptionally high diversity provides a compelling argument for recognizing
Kinabalu as a biodiversity hot-spot. However, numbers alone do not make a botanical
paradise. Kinabalu serves as home to a number of non-urticaceous taxa of special
botanical significance. Brief mention of certain taxa is warranted here.
The largest family of flowering plants on Kinabalu is the Orchidaceae with over
643 species. Many are well known for their horticultural value and rarity (Cribb 1998,
Wood et al. 1993, Wood and Cribb 1994). Notable taxa include several species of the
well known genera Paphiopedilum and Dendrobium. Bulbophyllum, the largest orchid
genus on Kinabalu, includes over 87 species.
No less engaging is Rafflesia keithii Meijer (Rafflesiaceae), an achlorophyllous
root parasite, having flowers measuring over one meter in diameter. The flowers look and
smell like rotting meat which attracts carrion flies as pollinators (Beaman et al. 1988).
The genus Rafflesia is known for having the largest flowers in the world.
Npe,,i/we,\ rajah (Nepenthaceae) is a carnivorous plant whose pitchers are formed
from extensions of the leaf midrib, and the pitchers of this species may hold over a liter of
water laden with digestive enzymes, variously digested insects and even small mammals.
Numerous other pitcher plants species occur on Kinabalu, some of them endemic, and
many on ultramafic substrates.
The genus Rhododendron (Ericaceae) deserves special mention, primarily because
80% of the Kinabalu species are endemic (Argent et al. 1988). Other Ericaceae, such as
Diplocosia, are also abundant, with numerous endemic species on Kinabalu.
Other large families on Kinabalu include the Rubiaceae, Euphorbiaceae,
Melastomataceae, Lauraceae, Myrtaceae, and Araceae. The single largest genus on
Kinabalu is Ficus (Moraceae), with ca. 100 species. Floristic enumerations have been
completed for the ferns and fern allies (Parris et al. 1992) and gymnosperms (Beaman and
Beaman 1993, 1998) as well as for the monocotyledons (Beaman and Beaman 1998).
Sipman (1993) provided a checklist of lichens, listing 286 known species, and estimated
the actual number to be about double this value.
Low's Peak, the highest on Kinabalu, is a mere 40 km from the South China Sea.
The massif rises impressively from the surrounding foothills and the coastal plane (Figure
1-3). The oft-quoted 4,101 m (13,455 ft) elevation for this peak was determined by the
East India Shipping Company in 1910 (Jenkins 1996), but several of the peaks on
Kinabalu have been recently surveyed. In 1997, the Sabah Department of Lands and
Surveys, using carrier phase global positioning system (GPS) equipment, measured the
height of Low's peak at 4093.7 m.
Kinabalu Park, one of six parks in the Sabah Parks system, encompasses the
Kinabalu massif in an area of about 750 km2. The park also includes Mount Madeleng
(the name Tambuyukon is often misapplied to this peak), the third-highest (ca. 2540 m)
mountain in Sabah. Kinabalu forms the northern terminus of the Crocker Range (also
under protection by Sabah Parks), which parallels the west coast of Sabah.
A Kinabalu location map, reproduced here at reduced scale (Figure 1-5) and
gazetteer (Beaman et al., in prep.) are near completion. Botanical collection sites, as well
as locations of general geographical interest have been mapped. Much of this
geographical knowledge was obtained through interviews with the local Dusun people. A
large number of sites, especially historical collection localities, were known only to some
of the oldest individuals in their communities. Without much of this detailed geographical
knowledge, the micro-geographic specimen mapping employed in this study would not
have been possible.
Kinabalu is the only mountain above 2649 m in Sabah and the only one in Borneo
above 3200 m. Sarawak, Sabah's neighboring East Malaysian state to the south, has a
number of mountains and mountain ranges including Mount Murud, the Dulit Range, and
the Hose Mountains, but none of these approach Kinabalu's stature. Likewise, the interior
of Kalimantan (Indonesian Borneo) is rugged and mountainous (Figure 1-2) but not well
explored botanically. Numerous other mountains between 2500 and 3000 m elevation
occur in the neighboring Philippine islands, including Palawan and Mindanao, and in
Sulawesi (the Celebes), which is close to Borneo but on the other side of Wallace's line.
Steep slopes abound on Kinabalu; razorback ridges and deep valleys make for
some very rugged terrain. This combination may foster effective geographic and
reproductive isolation of species over short distances.
The summit area is bisected by Low's Gully. Low's Gully formed along a thrust
fault (Jacobson 1970) that was sculpted at its upper end by Pleistocene glaciers. The
gully decends into what becomes the Penataran River. Together, they drop more than
3500 m over a four kilometer distance. Low's Gully exemplifies the precipitousness of
Kinabalu and it remains a serious challenge to explorers. In 1993, ten British Army
soldiers set out on a ten-day descent of Kinabalu through Low's Gully. It was concluded
tardily after a 12-day search and rescue operation used a helicopter to save five soldiers
trapped between waterfalls. The other five made it out on their own.
The highest peaks of the Mountain are west of Low's Gully. King George Peak is
the highest peak east of the gully and forms a pivot point between Eastern Ridge and
Northern Ridge. The northern slopes of Kinabalu are not well explored. Most botanical
exploration has been concentrated on and around the southwest slopes of Kinabalu
(Figure 1-5). This is exemplified by the higher concentration of named places in this area.
The southwest slopes have historically afforded the easiest ascent. Hugh Low was the first
European to reach the summit, in 1852. Granite crags line both of these ridges at their
upper elevations. The Eastern Ridge gradually descends to 600 m at Poring Hot Springs.
The Northern Ridge runs nearly level towards the north.
Five majors river systems originate near the summit of Kinabalu. The Northern
Ridge terminates where the headwaters of the Mekedeu River run to the east, and the
Wariau river runs to the northwest past Sayap and into the western coastal plain. The
Liwagu River originates on the southern slopes and ultimately flows out the eastern coast
of Sabah. The Kadamaian River cascades off the southwest slopes in a 1000 m high
waterfall. Numerous rivers, including the Penataran, join with the Kadamaian as it flows
north through the coastal plain. The Kadamaian becomes the Tampassuk River near Kota
Belud, the city from which the earliest expeditions up Kinabalu commenced.
Mount Kinabalu is one of earth's most recent major granitic plutons (primarily
composed of ademellite, with porphyritic margins). It was emplaced diapirically about 9
ma, cooling until 4.9 ma and uplifted in the last 1.5 million years (Jacobson 1970). It may
still be rising at a rate of between 3 and 5 mm per year. This granite core is exposed in
most areas above 3000 m. The summit area was much affected by the sculpting action of
glaciers during the Pleistocene, during which a 5 km2 ice cap formed. Deglaciation,
ending about 9,200 years ago, further eroded deep river valleys and deposited alluvial
sediments such as the Pinosuk gravels. Habitats above 3000 m were colonized after the
glacier retreated. Below the summit, a number of other substrates, including highly folded
and faulted Tertiary Crocker and Trusmadi sedimentary formations and Quaternary
Pinosuk gravels support a range of tropical forest communities.
Of particular interest are plant species and communities growing on the ultramafic
(also referred to as ultrabasic or serpentine) outcrops that occur at various elevations on
Mount Kinabalu. A higher proportion of endemic species are found on ultramafic
substrates than on non-ultramafics (Beaman and Beaman 1993, Parris et al. 1992, Wood
et al. 1993). Elatostema exemplifies this trend (see Table 1-1). More than half of the 15
Kinabalu plant communities recognized by Kitayama (1991) occur on ultramafics.
Plants growing on ultramafic substrates must adapt to low availability of nitrogen,
phosphorous, potassium, and calcium and high levels of iron, magnesium, nickel,
manganese, chromium, and cobalt. This is a toxic mixture to many plants. While there is
no evidence that any ultramafic adapted plants require high levels of heavy metals, some
species occur only on ultramafic outcrops. Competitive exclusion may be invoked to
explain why some of these ultramafic endemics do not occur on less toxic soils.
The stature of trees and shrubs on ultramafic habitats is often stunted, and is one
of the factors that makes these ultramafic patches quite recognizable with remote sensing
software. Serpentine (ultramafic) vegetation was noted by Kitayama (1991) to be
strikingly different from that of surrounding forests on non-ultramafic substrates. Not all
trees on ultramafics are stunted, however. The commercial value of some enormous trees
in the family Dipterocarpaceae and the genus Agathis (Podocarpaceae) endangers this
community at lower elevations.
Kitayama (1991) also indicated that there are at least three altitudinal subdivisions
of woody serpentine vegetation. The first subdivision is that dominated by Tristania
(Tristaniopsis) elliptica, the second by Leptospermum javanicum and Tristania elliptica,
and the third by Leptospermum recurvum and Dacrydium gibbsiae. Another ultramafic
vegetation type on which Kitayama did not comment are the forests dominated by
Casuarina sumatranum or C. terminal. These are particularly distinctive markers of
lower and mid elevation ultramafic soils. In addition, there are very interesting graminoid
ultramafic communities at Marai Parai and on the summit ofMt. Madeleng. These
graminoid communities often support the highest levels of local endemism.
The occurrence of ultramafics around Kinabalu is not anomalous for the region.
Ultramafic outcrops occur in Palawan, Sulawesi, New Guinea, as well as around the
volcanic belt known as the "Pacific Ring of Fire." These are active or previously active
zones of subduction of the spreading oceanic Pacific Plate beneath the Asian, Australian
and American continental plates.
Van Steenis (1964) concluded that there must have been a former mountain
connection between Asia and Australia, and that Kinabalu was a part of a "transtropical
bridge of temperate genera." Stapf (1894) held that Kinabalu must have shared an
"immediate connection with the highlands of New Guinea, or what was then its
equivalent," for the presence of Australian elements in the Kinabalu flora to be
In light of recent tectonic evidence (Hall 1996, 1999, Metcalfe 1999, Moss and
Wilson 1999), Stapf and van Steenis may have been more prophetic than we realized even
recently (Beaman 1996). Malesia is a tectonically complex region and the biogeography
is as convoluted and complex as that of the Caribbean.
Borneo, the Philippine islands, Sulawesi, New Guinea and other Indonesian
islands are situated in an area of striking tectonic activity at the juncture of the Eurasian
and Indian/Australian cratons, and the Pacific, Caroline and Philippine Sea plates.
Metcalfe (1999, page 25) described the area as "giant 'jigsaw puzzle' of continental
fragments bounded by major geological discontinuities... now huge strike-slip faults,
whereas others are actual suture zones that include remnants of oceanic crust (ophiolites),
oceanic and continental-margin sedimentary rocks, accretionary complexes, melange, and
sometimes volcanic arcs."
Ophiolites are typically ocean floor material thrust up onto and sometimes folded
into volcanic and sedimentary material. Brooks (1987) suggests that the ultramafics on
Kinabalu may be part of the Palawan Ophiolitic Belt. Palawan, a southern Philippine
island just to the north of Borneo is rich in ultramafics. It is a marginal fragment of
Eurasia that rafted eastward as the South China sea opened in the Mid Oligocene through
the Late Miocene, an event that coincided with the final uplift of Kinabalu. It was
submerged until the early Pliocene. Alternatively, the ultramafics, or at least the
ultramafic elements of the flora, on Kinabalu may have more closely shared a history with
those in Mindanao and Western Sulawesi, and as a result with New Guinea and Australia.
Ophiolites in Sabah crop out in a discontinuous S-shaped curve, with the Darvel Bay
Ophiolitic complex at the southeastern end and those on Kinabalu at the northwestern
end. The Darvel Bay ophiolites were already eroding by the Eocene (Omang and Barber
1996) and Brooks (1987) indicates these as related to the Mindanao ophiolites.
Borneo appears to be the result of dispersion of Gondwana marginal slivers and
later accretion of a number of these fragments from Paleozoic to Cenozoic times
(Metcalfe 1999). Continental blocks and fragments believed to contribute to Borneo
include South West Borneo, Semitau, Luconia, and Kelabit/Long Bowan fragments
derived from Cathaysialand, which was formed by the amalgamation of South China,
Indochina, and East Malaya in the Late Devonian to Early Carboniferous. These
fragments and parts of the Philippines basement separated from Cathaysialand opening up
the Proto-South China Sea in the Late Cretaceous. The Mangkalihat and Paternoster
fragments found to the east of the Adio and Meratus sutures in Borneo were derived from
Gondwanaland in the Late Triassic to Late Jurassic, along with West Sulawesi. These
fragments rafted from the northern margin of Australia in the Late Jurassic through the
Late Cretaceous. East Sulawesi, once contiguous with New Guinea/Australia (Metcalfe
1999), is shown by Hall (1999) as accreting to West Sulawesi in the Mid to Late Miocene,
a time when there may have also been a discontinuous volcanic arc connection to Borneo
via the eastern Philippine islands and Mindanao. Moss and Wilson (1999) indicate that
microcontinental blocks from the New Guinea-Australian margin that accreted onto
eastern Sulawesi in the Miocene or Pliocene may have remained emergent as they rafted.
Moss and Wilson (1999) infer a mountainous land connection between southern
Borneo and mainland SE Asia in the Eocene into the Pliocene, possibly dating as far back
as the Jurassic. They also suggest that dispersal may have occurred along volcanic arcs
between Borneo, Sulawesi, and the Philippine islands. Hall (1999) postulated that an
Eocene collision of India into Asia about 45 ma resulted in major mountain building and
resulting changes in climate and drainage systems in SE Asia. This event also allowed
dispersal of the Gondwana flora into Asia. The northward movement of the Australian
craton brought about contact between north Australia, New Guinea, Sulawesi, Borneo,
and the Philippines ca. 25 ma. The uplift of the central mountains of Borneo also occurred
around this time, and Borneo and the Malay Peninsula maintained connection along the
exposed Sunda Shelf through the Pliocene.
While much of the geological activity described above occurred before the final
uplift of Kinabalu, it was important in setting the stage for high levels of biological
diversity. The active orogeny of Kinabalu only 1.5-9 ma opened new niches, including
what is now the only high elevation habitat between the Himalaya and New Guinea. Also
important is the recognition that while the ultramafic flora on Kinabalu is represented by a
large proportion of endemic, possibly neo-endemic species, organisms have been adapting
to ultramafic toxicity since at least the Eocene. A cross-section diagram by Jacobson
(1970) shows the intrusion of the Kinabalu pluton occurring through both ultramafic and
sedimentary formations. Mount Madeleng, 24 km northeast of the Kinabalu summit, is a
predominantly ultramafic mountain. Madeleng may provide an indication of what the
Kinabalu area might be like had the final emplacement and uplift of the Kinabalu pluton
in the Pliocene never occurred.
Wallace's line (Wallace 1898) passes just to the east of Borneo. The position of
Kinabalu west of Wallace's line would suggest a Laurasian origin of the flora. However,
there are many representatives of the Australasian flora on Kinabalu. These include many
of the gymnosperms (i.e., Dacrydium, Agathis, Podocarpus) and many Myrtaceae (i.e.,
Leptospermum, Tristania and Tristaniopsis). Recent knowledge of the timing of tectonic
events along with application of cladistic biogeography in the Malesian region are
beginning to provide answers to the question of why some organisms may have crossed
Wallace's line and why others have not.
Dispersal from Australasia of Dacrycarpus imbricatus (Podocarpaceae) into
Sundaland is cited by Morley (1999) as a classic example of westward migration across
Wallace's line. Pollen of this species was common in the Lower Pliocene from offshore of
Sabah and Sarawak. The genus is typically restricted to SE Asian moist montane habitat,
and is represented by two species on Kinabalu, D. imbricatus and the endemic D.
kinabaluensis (Beaman and Beaman 1998).
Kinabalu supports extreme climatic ranges from tropical rain forests near sea level
to freezing alpine conditions at the summit. Stapf (1894), in the first floristic account of
the Kinabalu flora, made special note of temperate taxa occurring in the flora such as
Ranuculus lowii (Ranunculaceae) and Potentilla s. (Rosaceae). The closest relatives to
either of these are in the Himalaya and New Guinea.
Pleistocene climatic oscillations may have caused upwards and downwards
migrations of species during glacial and interglacial cycles. Glaciers extended as low as
3000-3100 m (Choi 1996, Jacobson 1978), about 1000 below the summit. A 1000 m
difference in elevation may be correlated with a mean temperature drop of 5.50 C on
Kinabalu (Kitayama 1992, 1994). Below the physical extent of the glaciers, this degree of
change in mean temperature would strongly select against many tropical cloud forest
plants. In addition, downward migrations may have isolating effects on some plant
populations. Besides having elevational restrictions, most plants on Kinabalu tend to be
restricted to ridges, stream valleys or slopes. In the case of a species occurring only on
ridges, when populations are forced downwards, those on adjacent ridges would become
more isolated, since ridges radiate as elevations decrease. The opposite effect would occur
with stream species, since streams conjoin as elevations decrease.
Recent El Niho events have lead to extreme short-term climatic instability
throughout Borneo. Beaman et al. (1986) documented El Niho related forest fires
occurring in 1983 in Sabah. Almost 10 % of the forested land of Sabah burned during this
event. Two other ElNiho events have occurred since, one in 1990 and another in 1998.
Both were associated with fires burning through primary forest within Kinabalu Park.
Many of these fires burned vegetation on ultramafic substrates. This vegetation type
seems particularly susceptible to drought. Many of the plants on ultramafics exhibit
characteristics of xerophytes (thick leathery leaves with waxy cuticles), but their ability to
tolerate fire is dubious. One of the characteristics of the El Niho droughts is that they are
often followed by unusually heavy rains. Numerous landslides have occurred on Kinabalu
following the El Niho events. While the wildfires may be only recently occurring on
Kinabalu because of human disturbance (and arson), ElNiho is not new, nor are
landslides. Many very interesting ultramafic species occur only on landslides.
A vegetation map of Mount Kinabalu Park was published by Kitayama (1991) in
which 21 vegetation map units were recognized. Kinabalu Park does not include all the
area covered by this study, but enough is included to make the map useful for
understanding the vegetation of the entire Kinabalu massif. Kitayama indicated that
diagnostic canopy tree species could be used to distinguish vegetation zones. He
considered these species to be mutually exclusive in distribution and their occurrence
correlated with elevation. The upper boundary of lowland rain forest, where the majority
of emergent trees (mostly Dipterocarpaceae) disappear from the canopy, is at about 1200
m. The upper limit of lower montane forest is at between 2000 and 2350 m, and that of
upper montane forest at between 2800 and 3000 m, the latter particularly marked by the
upper limit of Lithocarpus havilandii. Above this level occurs a lower-subalpine
coniferous forest dominated by Dacrycarpus kinabaluensis, Dacrydium gibbsiae,
Podocarpus brevifolius, and angiosperms Photinia davidiana, Styphelia malayana, and
Leptospermum recurvum. The upper limit of this forest is at about 3400 m and
corresponds to the closed forest line. A fragmented upper-subalpine forest extends above
this level to the tree line at about 3700 m. Above this level is a summit area zone of alpine
rock-desert with scattered communities of alpine scrub. Kitayama suggested that the tree
line may coincide with the lowest elevation where nocturnal ground frost is frequent. He
noted that great variations in dominance type, species composition, and forest structure
occur within each zone. These were attributed to altitudinal temperature effects, soil
nutrient status in relation to topography (particularly ridge and valley differences) and
Elevational zonation is so strong on Kinabalu that it may be used as the primary
basis for recognition of vegetation types. Below 500-600 m is what Kitayama refers to as
"substituted vegetation" and is meant to refer to lowlands that have long been subject to
human management and disturbance. "Hill forest" is a much used term for the upper part
of the dipterocarp-dominated lowland forest, and we have applied this to habitats between
the levels of about 600 and 1400 m. Kitayama considered the lowland forest to be six-
layered with emergent trees characteristic and undergrowth sparse. Lower montane forest
is in the elevational range of about 1200-2200 m, is five-layered, and does not have
emergents; upper montane forest has a dense herb layer. The lower subalpine forest is
sparser in undergrowth and lower in height. Upper montane forest is used here to refer to
all forest vegetation above the lower montane forest. The area above the continuous upper
montane forest is referred to as the summit area.
Figure 1-1. Mount Kinabalu. Top: View of the upper south face from the West
Mesilau River, above the Pinosuk Plateau. Bottom: View of the southwestern
slopes showing ephemeral waterfalls after a rain storm.
U C~ ~ -
cE -65 -b
- c -E
o -, ctrj
CA .2 j
F4 mrt CA
Figure 1-3. Satellite image map of Mount Kinabalu, based on Landsat 5
Thematic Mapper data from 14 June 1991. Bands 3, 2, and 1 are shown applied
to a red- green-blue (RGB) color model. Raw image data was rectified using
digital elevation model (DEM) data and global positioning system (GPS) ground-
control points. Original pixel values were maintained by this method, thus the
image may be used in further analysis. The Kinabalu Park boundary is shown in
\ ..un kin I l lu I i' l l I l l i -
..... . I\II
Figure 1-4. Drape of the Landsat TM image of Mount Kinabalu over the DEM, as seen
from the northwest. This figure was produced using image bands 7, 2, 1. The Kinabalu
Park boundary shown in red is an additional coverage. Black holes in the draped image are
areas where surface visibility was miscalculated owing to three-dimensional surface-
resolution limitations. The surface resolution is 50 x 50 m while the image resolution is
28.5 x 28.5 m.
Figure 1-5. Reduced scale (1:250,000) reproduction of Mount Kinabalu location
map showing place names of kampungs, peaks, rivers and botanical collecting
localities. The shaded relief map was produced by hill shading and color coding
by elevation the digital elevation model (DEM).
NIMI I KIN %\ \I I
I O( 1O N N1 1 ,1'
t .. "
S "* "r . "'*
: ~ ~ "' "4.
p s l i ,,
, V _,
,. "tt, ; r"
i i i : )
Table 1-1. Species of Elatostema included in this study, their general
distributions, and occurrence on ultramafic substrates. 'Species occurs
exclusively on ultramafic substrates. Undescribed species indicated by single
quotes around the specific epithet.
E. 1th,/i i it i k'
Kinabalu, Borneo, Java, India
Kinabalu, Sabah, Sarawak
GEOGRAPHICAL INFORMATION SYSTEM
Development of a geographical information system (GIS) for Mount Kinabalu
was originally conceived to augment the botanical inventory of Mount Kinabalu, with a
location map and gazetteer, and as a basis for mapping species distributions. As the
technology has improved, GIS has become much more than just a mapping tool; it forms
a spatial framework for analysis of biological data. Though beyond the scope of this
dissertation, we are developing GIS tools to (1) predict where species occur in unexplored
areas based on occurrence in known localities, (2) develop diversity indices based on
surface areas and elevation, and (3) identify Itramafic habitat islands as part of an effort to
understand plant speciation patterns on Kinabalu. Identification of ultramafic outcrops is
of particular interest because of the correlation between these outcrops and distribution of
endemic species, particularly those hypothesized to have evolved on Kinabalu.
Coverages included in the Kinabalu GIS to date are topography, hydrography
(rivers and streams), the Kinabalu Park boundary, ground-control points, a vegetation
map published by Kitayama (1991), place names toponymss) relating to specimen
collections made since 1851 as well as by local collectors for the Projek Etnobotani
Kinabalu (PEK), the Landsat TM image, and a digital elevation model (DEM) was
developed from topographic coverages.
Predictions of where a taxon occurs in unexplored areas are useful in planning
expeditions into areas such as the still unexplored north side of Mount Kinabalu.
Likewise, park management may use this type of data to target certain areas for
protection. Predicting the occurrence of a particular taxon is based on knowing the
characteristics of its habitat. Sheila Collenette (pers. comm.) probably was the first person
to use remote sensing in this manner in her rediscovery of Paphiopedilum
1 i/l, hil li, i 1unl on Mount Kinabalu. Using black and white aerial photographs of
Kinabalu she was able to recognize the Casuarina s.l. forests that are characteristic of
many ultramafic areas. Site visits proved her technique worthwhile, because the rare
slipper orchid, P. nIti',hiihl inlnu, had not been known from the wild in almost 70 years.
The type collection had been (mischievously) attributed to New Guinea.
Remote sensing and GIS software (ERDAS Imagine and ESRI Arc/Info,
respectively) provide image enhancement and classification techniques that make these
sorts of predictions possible in a systematic way. Digital elevation models (DEMs) allow
further refinement of predictive models by the inclusion of elevation, slope and aspect
data. Diversity indices may also be enhanced by developing models to incorporate DEM
data into the diversity index. How species diversity relates to habitat diversity and the
spatial extent of certain plant communities is of particular interest on Mount Kinabalu,
because of the high endemicity of species, particularly in ultramafic areas.
The satellite image used extensively in this study is a Landsat 5 Thematic Mapper
(TM) product, from an orbit ca. 700 km above the earth's surface. The image shown in
Figure 1-3 is a quarter of a quarter scene, i.e., one-sixteenth of the total scene that covers
185 x 185 km. This image, recorded 14 June 1991, is one of only two cloud-free images
available for the area, which is characterized by frequent and heavy cloud cover. The
nominal ground resolution is 28.5 x 28.5 m for each pixel. The data are provided from
seven bands of the electromagnetic spectrum, as designated by the EOSAT corporation:
Band 1, visible blue (0.45-0.52 [tm); band 2, visible green (0.52-0.60 [tm); band 3, visible
red (0.63-0.69 ipm); band 4, near infrared (0.76-0.90 ipm); band 5, mid infrared (1.55-1.75
[tm); band 6, thermal infrared (10.4-12.5 [tm); and band 7, a second mid infrared band
(2.08-2.35 [tm). Applying the visible bands 3, 2, and 1 to a red, green and blue (RGB)
color model creates a true-color image (Figure 1-3). Several other band combinations are
shown in Figures 2-1 through 2-4 for comparative purposes. These are pseudo-color
images in which differentiation between vegetation types is often enhanced. As the term
implies, pseudo-color images are less faithful at approximating colors seen by the human
eye. The human eye is not sensitive to infra-red wavelengths. Most of the Landsat TM
bands provide useful information on vegetation (Lillesand and Kiefer 1987). For example,
band 1 is useful in differentiating between soil and vegetation types. Band 2 measures
green reflectance of healthy vegetation. Band 5 is sensitive to the amount of water in the
leaves of plants and penetrates through smoke and haze better than bands 3, 2 or 1.
Some particularly conspicuous elements of the landscape are apparent in all
combinations, including primary forest, secondary forest, and shifting agriculture.
Furthermore, the bare granitic outcrop of the summit area of Kinabalu is readily apparent,
as is the Mamut Copper Mine, another feature without vegetation cover. Other less
conspicuous features with which we are working, such as the occurrence of ultramafic
outcrops, require special techniques to enhance. Enhancement techniques we have
employed so far include tasseled cap analysis (enhances brightness, wetness and
greenness), basic mineral composition, and ferro-magnesium indices (e. g., occurrence of
ferrous minerals can be enhanced by ratioing band 5 to band 4). In terms of the vegetation
on Kinabalu, the latter method has proved more useful than the former. Ultramafic
outcrops stand out particularly well in Figure 2-4.
Finally, satellite imagery made the production of the location map for Mount
Kinabalu (Figure 1-5), on which accurate and up-to-date placement of roads and
settlements (kampungs) are important, easier and more accurate. Roads vs. rivers, for
instance, are more conspicuous in different band combinations (Figures 2-2 and 2-3).
Predicting Species Ranges using Digital Elevation Models
It is possible to convert the elevational data from contours on topographic maps
into a three-dimensional digital elevation model (DEM) of a landscape. In a grid-based
DEM, elevation values are stored within cells, each of which represents a discrete unit of
space on earth. The cells are organized into rows and columns forming a Cartesian matrix
in the same way that satellite data are stored. The Kinabalu DEM uses cell sizes of 50 x
50 m. When visualized, the process is like draping an enormous fishnet over the
mountain. Different elevations can be represented by different colors, thus enhancing the
three-dimensional aspects of the image (Figure 2-5). Views from any direction can be
produced. Elevation, slope, aspect, and insolation (intensity of sunlight falling on a slope)
attributes can be easily extracted from a DEM. In an analytical environment, these
attributes can be used to predict where a particular rare species may occur in unexplored
areas, based on elevation, slope, and aspect data for known occurrences of the taxon.
A particularly exciting aspect of the DEM is that it provides a mechanism for
draping other types of coverages over basic details such as elevation. For example, river
systems can be shown three-dimensionally draped over the DEM. An even more dramatic
example is in the ability to drape satellite imagery. Thus, a realistic three-dimensional
model of Mount Kinabalu can be produced as a drape over the DEM, and additional
coverages, for example the Kinabalu Park boundary, can be added as shown in
Diversity Analyses through GIS
Plant diversity and distribution patterns on Mount Kinabalu were discussed by
Beaman and Beaman (1990), and the hypotheses explaining the high diversity were
outlined there and somewhat elaborated in Beaman et al. (1997). Significant
environmental variables involved in this hypothesis that are amenable to GIS analysis
include extreme elevational variation from just above sea level to 4,094 m, precipitous
topography, diverse geological substrates, relatively open habitats (e.g., landslides), slope,
aspect, and, as dependable data about species distributions on the mountain become
available, relative isolation of individual populations.
In Parris et al. (1992) and Wood et al. (1993), histograms showing generic and
species diversity for the pteridophytes and Orchidaceae indicate that the highest number
of taxa occur at about 1500 m. This type of diversity index is independent of the actual
surface area at any particular elevation range. Among the analytical uses of the digital
elevation model is that it allows us to calculate planimetric surface area for each of the
elevation ranges on Kinabalu. To compare diversity for areas of unlike size it is necessary
to calculate a diversity index. There is of course less overall surface area on any mountain
at progressively higher elevations. The simplest type of diversity index is derived by
dividing the number of taxa by the surface area. In this instance, elevation ranges in 500
m bands around Mount Kinabalu are the areas for which diversity is compared. The
results for pteridophytes are shown in Figure 2-6. Although the highest number of
pteridophyte taxa occurs between 1001 and 1500 meters, the highest number of
pteridophyte taxa per unit area is between 2001 and 3000 meters. This area-based
diversity index is more comparable to diversity indices based on plot sampling.
Figure 2-1. Landsat TM pseudocolor image of Mount Kinabalu, produced by
applying bands 7, 6, and 4 to the RGB color model. Low elevation ultramafic areas are
quite evident with this band combination.
Figure 2-2. Landsat TM pseudocolor image of Mount Kinabalu, produced by
applying bands 7, 5, and 2 to the RGB color model.
Figure 2-3. Landsat TM pseudocolor image of Mount Kinabalu, produced by
applying bands 5, 2, and 1 to the RGB color model. This band combination
penetrates haze better than bands 3, 2, and 1. Water features are very clear.
Figure 2-4. Landsat TM pseudocolor image of Mount Kinabalu produced by
applying bands 4, 3, and 2 to the RGB color model. This band combination
produces an image very similar to that of color infrared film.
8- a ~
*number of species *species per10 sq km
500 1000 1500 2000 2500 3000 3500
Upper elevation range (m)
Figure 2-6. Pteridophyte diversity on Mount Kinabalu. Note that although the greatest
number of species occurs at 1500 m, the greatest diversity of species per unit area is
between 2500 and 3000 m.
PHYLOGENY OF THE URTICACEAE
Choice of outgroup for the phylogenetic analysis of Elatostema from Mount
Kinabalu in Chapter 4 is based on an analysis of intergeneric relationships presented in
this chapter. Outgroup choice is often critical to the topology of the tree, and tree
topology embodies the direction of character evolution. Patterns of character evolution
cannot be reconstructed from an unrooted tree. Outgroup analysis is an almost universally
recognized method for determining character polarity (i.e., rooting a cladogram). It has
become more or less automatic using most phylogenetic software packages. The selection
of an outgroup, however, is not automatic. When an appropriate phylogeny is not
available on which to base outgroup choice, an ad hoc approach is often used. The
purpose herein is to provide a phylogenetic basis for outgroup choice upon which to base
the analysis of Elatostema in Chapter 4, thus avoiding the ad hoc approach. A brief
review of pertinent higher level classifications places the analysis in context.
Circumscription of the Urticales has been relatively consistent for over 150 years.
Weddell (1856) included the Ulmaceae, Cannabineae, Artocarpeae, Moreae, and
Urticaceae. Takhtajan (1997) and Thorne (1992) used the same circumscription. Dahlgren
(1989) and Cronquist (1981) included the Barbeyaceae, a monotypic family occurring in
north-east tropical Africa and the Arabian Peninsula.
Placement of the Urticales has been less certain. Cronquist placed the Urticales
within the Hamemelidae, a group that also includes the Fagales, Casuarinales, Myricales,
Juglandales, etc. Berg (1977, 1989) suggested affinities with the Malvales. Dahlgren,
Takhtaj an, and Thorne followed suit. Recently, relationships within the Urticales and with
other orders have been evaluated by Chase et al. (1993), Sytsma et al. (1996) and Qiu et
al. (1998) based on phylogenetic analysis ofrbcL nucleotide sequences. Based in part
upon these analyses, Judd et al. (1999) and the Angiosperm Phylogeny Group (1998)
considered the Urticales (suborder Urticineae sensu Judd) to be nested within the Rosales.
Monophyly of the Rosales is weakly supported by a reduction in endosperm and presence
of a hypanthium. Qiu et al. (1998) also reported that in all Urticales, Elaeagnaceae, and
Rhamnaceae, the nad] mitochondrial gene carries a trans-spliced intron, whereas in most
other dicots, including the Rosaceae, the intron is cis-spliced. This is perhaps the
strongest molecular evidence that the Urticales is nested within the Rosales as sister to the
Rhamnaceae/Elaeagnaceae clade. The tree of Qiu et al. (1998) also shows Barbeya as
basal to the Urticales.
Monophyly of the Urticales is supported by occurrence of cystoliths, reduction in
flower size, reduction in the number of stamens to five or fewer, a two-carpellate
unilocular ovary with a single apical ovule (a basal ovule in the Urticaceae), presence of
urticoid teeth, and at least one prominent prophyllar bud (Judd et al. 1999). The
hypanthium is lost in the Urticales with exception of the Ulmaceae, possibly because of
reduction in flower size. Laticifers are a synapomorphy for the (Moraceae(Cecropiaceae/
Urticaceae)) clade, but are lacking or reduced to occurring in the bark in the Cecropiaceae
and Urticaeae (in Urera), or lost or perhaps modified into mucilage cells and ducts
(Renner 1907) in the rest of the Urticaceae.
In the description of the Cecropiaceae as a new family, Berg (1978) included
Poikilospermum, but noted the genus had a number of characters in common with the
Urticaceae, including presence of elongated cystoliths, stamens that are inflexed in bud,
lack of milky latex (laticifers still present but reduced in size), and production of
mucilaginous sap. The mucilage is distinct from that of of Urticaceae, however, in that it
turns black upon exposure to air. Berg (1989) reiterated his doubt that Poikilospermum
belonged to the Cecropiaceae, noting that anatomical features discussed by Bonsen and
ter Welle (1983) suggested a relationship closer to the Urticaceae than other
Humphries and Blackmore (1989) carried out a phylogenetic analysis at the tribal
level for the Moraceae (Ficeae, Dorstenieae, Castilleae, Moreae, Artocapeae), and
included the genera Poikilospermum, Conocephalus (= Poikilospermum), and
Sparattosyce (Moraceae) and families Urticaceae, Ulmaceae, Cannabaceae, Cecropiaceae,
and Barbeyaceae, with the latter as outgroup. Their tree showed a polytomy between the
Urticaceae, Poikilospermum, and Conocephalus, with the Cecropiaceae as the sister
lineage, and below that, the Moraceae.
Morphological, anatomical and molecular data now support the hypothesis that
Poikilospermum is more closely related to the Urticaceae than the Cecropiaceae. In the
morphology-based analysis of Judd et al. (1994) Poikilospermum and Pilea were placed
as sister taxa (no other Urticaceae s.s. were analyzed), with Cecropia subtending the pair.
Phylogenetic analysis of 25 urticalian taxa (Sytsma et al. 1996) based on rbcL sequence
data hypothesized a slightly different relationship, suggesting that the Cecropiaceae is
polyphyletic and that Poikilospermum is actually embedded deeply within the Urticaceae.
Judd et. al. (1994) circumscribed the Urticaceae to include the Cecropiaceae,
Moraceae, and Cannabaceae because their analysis showed the Moraceae to be
metaphyletic. Molecular evidence, based on phylogenetic analysis of rbcL nucleotide
sequences by Sytsma et al. (1996) and Qiu et al. (1998) suggests that the Moraceae are
monophyletic. Judd et al. (1999) adopted a narrower circumscription of the Urticaceae
based on synapomorphies of elongate cystoliths, incurved stamens in bud, wood
anatomical characters, and molecular data. Phylogenetic analysis of molecular data from
rbcL sequences (Chase et al. 1993 and Qiu et al. 1998) also supports the monophyly of
the Urticaceae, but neither of these analyses included Poikilospermum.
The Urticaceae sensu Friis (1993) consists of 45 genera and about 1000 species.
He did not include Poikilospermum. Weddell (1854, 1856) recognized five tribes as did
the more recent work of Friis (1989, 1993). The names adopted here follow the tribal
names used by Friis (1993). They are the Urticeae (10 genera), Lecantheae (17 genera,
including Elatostema), Boehmerieae (19 genera), Parietarieae (5 genera), Forsskaoleae (4
genera). Although Friis provided a rich discussion of characters, no rigorous phylogenetic
analysis was conducted. Friis (1989) presented two ad hoc cladograms. These showed
confidence in the circumscription of the tribes, but did not suggest what the relationship
might be between the Urticeae, Lecantheae, and Boehmerieae.
The purpose of this analysis is not to determine phylogenetic relationships for the
entire family, but to find an appropriate outgroup for the analysis of Elatostema in the
next chapter. The disposition of Pellionia is one of the more interesting taxonomic
problems within the Lecantheae. Species of Pellionia are often placed in Elatostema.
Robinson (1910) considered them distinct. Schroter and Winkler (1935) treated Pellionia
as a subgenus ofElatostema. Friis (1989) considered them as two entities, but later he
(Friis 1993) combined Pellionia into Elatostema. In the analysis below they are treated as
separate taxa. However, an analysis at the species level will better address this issue in the
next chapter. In this chapter, generic relationships within the Lecantheae and between the
five tribes are evaluated.
A subset of 36 genera of Urticaceae was selected, representing each of the five tribes. In
addition, four other genera, Cannabis, Ficus, Morus, and Cecropia, were included as
outgroup taxa. Poikilospermum and Conocephalus were included as part of the ingroup
owing to the evidence supporting their placement in the Urticaceae. A total of 28
characters was scored. Friis (1989, 1993) was consulted for much of the data for
Urticaceae. Humphries and Blackmore (1989), Berg (1989), and Judd et al. (1994, 1999)
were consulted for outgroup characters. Characters (Table 3-1) were selected that were
most consistent within genera, but where they were known not to be, they were scored as
variable. Data (Table 3-2) was analyzed with PAUP* 4.0d65 (Swofford 1998) utilizing
the heuristic search algorithm (500 replicates) and using tree bisection reconnection
(TBR) and random sequence addition options.
The heuristic search generated 219 equally parsimonious trees, each of 117 steps,
a consistency index (CI) of 0.752, and a retention index (RI) of 0.800. A strict consensus
tree is shown in Figure 3-1 and a 50% majority rule consensus tree in Figure 3-2. The
tribal classification of Friis (1993) is mapped onto the strict consensus tree. In all trees,
the Urticaceae + Cecropiaceae form a monophyletic group. The Urticaceae s.s is
monophyletic in 92% of the 219 trees. The Conocephalus/Poikilospermum clade is sister
to the Urticaceae s.s. in 92% of trees, in those cases suggesting that the Cecropiaceae is
paraphyletic. Based on the strict consensus tree, relationships between the Cecropiaceae
and basal Urticaceae appear unclear. The situation may be clarified by comparing two
alternative tree topologies from the pool of 219 equally parsimonious trees.
The first of the 219 most parsimonious trees is shown in Figure 3-3, with locations
of unambiguous character state changes mapped. In this tree, the Conocephalus/
Poikilospermum clade is sister to the Urticaceae, and Cecropia basal. With respect to the
base of the tree, a competing, equally parsimonious basal topology (shown by the tree in
Figure 3-4) is found in 8% of trees. In this topology, the Cecropiaceae is monophyletic
and nested within part of the Boehmerieae. In this second topology, moving Cecropia to a
position sister to the Urticaceae s.s., while leaving Poikilospermum and Conocephalus
embedded, increases the length of the tree by only two steps. Poikilospermum embedded
in the Urticaceae s.s. was the hypothesis supported by the analyses of Sytsma et al. (1996)
based on rbcL sequences.
Cecropiaceae embedded within the Urticaceae s.s. also requires the independent
evolution of laticifers in two lineages (see char. 6, Figure 3-5), as opposed to a single loss
(see char. 6, Figure 3-6), reduction, or modification of laticifers in the Urticaceae.
Laticifers are present in the bark only in the Cecropiaceae and Urera in the Urticaceae.
(Urera was not included in this analysis.) Neraudia often has milky sap, so was scored as
having laticifers. Renner (1907) argued that the mucilage cells and ducts of the
Cecropiaceae and Urticaceae were derived from laticifers. That line of reasoning supports
a tree topology with a paraphyletic Cecropiaceae basal to the the Urticaceae s.s.
For these reasons, the topology of the tree shown in Figure 3-3 is preferred and
was selected to illustrate other character distributions. Five characters support the
monophyly of the Cecropiaceae/Urticaceae in this tree. Mucilage cells or ducts occur in
the Cecropiaceae and throughout the Urticaceae, although some homoplasy is evident
(char. 5, Figure 3-7). There is also a reduction in the number of stigmas (char. 25, Figure
3-8), a change to basal placentation and an orthotropus ovule (char. 27), and reduction to
a pseudomonomerous gynoecium (char. 28). Arachnoid hairs (char. 11, Figure 3-9) are
present in the Cecropiaceae and basal Urticaceae, but are lacking in the Urticeae,
Lecantheae, Parietarieae, and Boehmerieae (p.p.).
Three anatomical wood characters occur in Poikilospermum and Conocephalus
and in the Urticaceae, but not in Cecropia. These are presence of unlignified vessel
elements (char. 1, Figure 3-10), wood fiber dimorphism (char. 2, Figure 3-11), and
presence of tangential and radial fiber pits (char. 3, Figure 3-12). In the Urticaceae,
unlignified vessel elements occur in members of the Urticeae and Lecantheae, and in
Touchardia and Nothocnide in the Boehmerieae.
The relationships between the Boehmerieae, Forsskaoleae, and Parietarieae remain
unresolved. In all trees, however, the Lecantheae together with the Urticeae form a
monophyletic group (Figure 3-13). Two floral characters support this clade. All members
of this clade have a glabrous male pistillode (char 19, Figure 3-14), while the pistillode is
either lanate or absent in other taxa. Free or lobed pistillate perianth parts (char. 24,
Figure 3-15) are also consistent for this clade. The perianth in pistillate flowers is either
lacking (as in the Forsskaolaep.p., Phenax, and Leucoscyse), or strongly fused into a tube
(as in the Cecropiaceae, Moraceaep.p., and Boehmerieae p.p.).
The Lecantheae, as circumscribed by Friis (1989, 1993), is also monophyletic.
The Lecantheae, itself, are supported by presence of staminodia (char. 20, Figure 3-16) in
the pistillate flowers. These staminodia function in explosive achene dispersal. Pellionia
and Elatostema are sister taxa and are sister to the remaining genera of Lecantheae. The
Elatostema and Pellionia clade is supported by anisophyllous phyllotaxy as opposed to
opposite or spiral phyllotaxy (char. 7). Alternate leaves in Elatostema (including
Pellionia) are considered to be derived via anisophyllous reduction (see Chapter 4).
Taxa belonging to the Urticeae sensu Friis form clade with the exception of
Gyrotaenia. Gyrotaenia lacks the stinging hairs (char. 9, Figure 3-17) characteristic of the
Urticeae, and is sister to the Lecantheae in the tree shown, but lacks unambiguous branch
The Forsskaoleae are always monophyletic but may be nested within part of the
Boehmerieae or may alternatively be a basal clade of the Urticaceae. Monophyly of the
this tribe is supported by a reduction in the number of stamens to one.
The Boehmerieae appear to consist of a number of small clades that tend to be
basal within the family. In the majority rule tree, two small genera of the Boehmerieae are
basal within the Urticaceae. Neraudia consists of five species found on the Hawaiian
islands (Cowan 1949) and Debregeasia with four species occurring in Northeast Africa
and the Arabian peninsula, Pakistan, India, in the Himalaya, Japan, and Southeast Asia
throughout much of Malesia, but not in New Guinea (Wilmot-Dear 1989). Debregeasia
velutina Gaud. and D. squamata Kina ex Hook. f occur on Mount Kinabalu.
Delimitation of tribes by Weddell (1854, 1856) and Friis (1989, 1993) are in many
instances well supported by this analysis, with the exception of taxa in the Boehmerieae.
In presenting character distributions on a phenetic map of genera, Friis (1989) placed the
Boehmerieae in a central location relative to the four other tribes, visually suggesting that
members of the other tribes may have been derived from elements within the
Boehmerieae. The Boehmerieae is the largest tribe in terms of number of genera, and is
defined at least partially on pleisiomorphic characters for the Urticaceae, such as a
tubular pistillate perianth and filiform stigmas (see Weddell 1854), both character states
found also in the Cecropiaceae. In addition, many character states are polymorphic at the
tribal level for the Boehmerieae, including phyllotaxy, cystolith and indumentum
morphology, and pollen grain characteristics. Thus, it is not surprising that inter- and
intra-tribal relationships concerning the Boehmerieae are still not well resolved. The tribe
is surely non-monophyletic.
The monophyly of the Urticaceae s.s. is tentatively supported by the single loss of
laticifers (or conversion into mucilage cells/ducts) in almost all genera of the Urticaceae
(except Urera and possibly Neraudia, where they are present and functional). This is
preferable to the equally parsimonious alternative that laticifers evolved independently in
the Cecropiaceae (in the bark only) and the Moraceae (throughout the plant). The
transformation proposed by Renner (1907) that the mucilage cells and ducts found in
many taxa in the Urticaceae are derived from laticifers seems plausible.
The suggestion by Sytsma et al. (1996) that Poikilospermum but not Cecropia
may be embedded in the Urticaceae still requires either the independent evolution of
laticifers, or the less parsimonious solution of laticifer loss in multiple lineages.
Constraining the analysis to force Cecropia into a position as the sister group to the
Urticaceae, and leaving Poikilospermum and Conocephalus embedded in the Urticaceae,
produced a tree only two steps longer. Thus, morphology does not strongly reject this
hypothesis. The present analysis suggests that the Cecropiaceae may indeed be
paraphyletic (as currently circumscribed), but probably not polyphyletic. The doubt
expressed by Berg (1989) that Poikilospermum belongs in the Cecropiaceae is reinforced
by the analysis of Sytsma et al. (1996) and by the results of the present analysis, which
place Poikilospermum as sister to the Urticaceae s.s. Additional sampling of taxa for both
molecular and morphological characters may place Poikilospermum slightly or even
deeply nested within the Urticaceae. More study is needed.
The task of choosing an outgroup for the Elatostema analysis becomes more
objective based on the foregoing hypothesized phylogenetic relationships. In a
preliminary analysis, representatives of Pilea and Boehmeria were used based on ad hoc
reasoning. Relationships within the Lecantheae/Urticeae clade are now better resolved
and the Elatostema/Pellionia clade is shown to be sister to the remaining Lecantheae,
with Pilea a basal member of the sister group. Thus Pilea continues to be an appropriate
outgroup taxon. On the other hand, Boehmeria appears to be relatively distantly related.
Therefore, in the ensuing analysis, Pilea and Gyrotaenia are used to root the resulting
trees. Gyrotaenia is either basal to the rest of the Lecantheae, or basal in the Urticeae.
Representatives of Pellionia would not make an appropriate outgroup, since it
may actually be part of the ingroup. A number of species of Elastostema have at some
point or another been described as or transferred to Pellionia. Pellionia was included as
distinct in the generic analysis only because some authors have recognized it (Weddell
1856, Friis 1989) at the generic level. However, others have not (Schroter and Winkler
1935, Friis 1993). Fortunately, these two genera consistently show up as sister taxa. The
next chapter treats taxa at the species level, and the monophyly of Pellionia and
Elatostema is addressed.
Figure 3-1. Strict consensus of 219 equally parsimonious trees, showing generic
relationships between 30 genera of the Urticaceae and six genera in the Cannabinaceae,
Cecropiaceae, and Moraceae. Cannabis, Ficus, and Morus were designated as outgroup
taxa. Tribes, sensu Friis (1993), are mapped for the Urticaceae.
Figure 3-2. Majority rule (50%) consensus tree of 219 equally parsimonious trees
showing generic relationships between 30 genera of the Urticaceae and six genera in
the Cannabaceae, Cecropiaceae, and Moraceae. Cannabis, Ficus, and Morus were
designated as outgroup taxa.
100 E Morus
5g 0 Lecanthus
92 10 Meniscogyne
92 84 Phenax
51 10 Drogeuetia
Figure 3-3. Preferred cladogram showing generic relationships between 30 genera of
the Urticaceae and six genera in the Cannabinaceae, Cecropiaceae, and Moraceae.
Cannabis, Ficus, and Morus were designated as outgroup taxa. Positions where
unambiguous character state changes occur are mapped onto the tree. The
Cecropiaceae is paraphyletic and subtends the Urticaceae in this tree.
7c 15 u*1 2 2
713 15 8 1 2 24 1
15 3 16 11
..22 7 26
15 21 22
9 10 10
Figure 3-4. Representative cladogram showing generic relationships between 30
genera of the Urticaceae and six genera in the Cannabinaceae, Cecropiaceae, and
Moraceae. Cannabis, Ficus, and Morus were designated as outgroup taxa. Positions
where unambiguous character state changes occur are mapped onto the tree. The
Cecropiaceae is embedded within the Urticaceae in this tree.
Figure 3-5. Representative most parsimonious tree. Branch tracing indicates
occurrence of laticifers. This topology requires independent evolution of laticifers or
multiple loss. Laticifers absent (0), Laticifers present (1).
B o0 5 g.
5 Cz 5 0
70) 0 U -j 0 0. 75,
:) ~ ~ ~ m 4 &P. P w P4U 4 ):
Figure 3-6. Preferred most parsimonious tree. Branch tracing indicates occurrence of
laticifers. This topology involves a single loss of laticifers. Laticifers absent (0),
Laticifers present (1).
C) z 0 B 0
-a U El 0
Figure 3-7. Preferred most parsimonious tree. Branch tracing indicates occurrence of
mucilage cells or ducts. Mucilage cells or ducts absent (0), mucilage cells or ducts
Figure 3-8. Preferred most parsimonious tree. Branch tracing shows number of
stigmas. Two (0), one (1).
Figure 3-9. Preferred most parsimonious tree. Branch tracing indicates occurrence of
arachnoid hairs. Arachnoid hairs absent (0), arachnoid hairs present (1).
0. S 0
~ o S T'
Figure 3-10. Preferred most parsimonious tree. Branch tracing indicates occurrence
ofunlignified vessel elements. Vessel elements lignified (0), vessel elements
C) ) U C)
z0 C) z
Figure 3-11. Preferred most parsimonious tree. Branch tracing indicates occurrence
of dimorphic wood fibers. Wood fibers monomorphic (0), wood fibers dimorphic (1).
33aIj & --0
~~~r 3FZ 6 a, v EC
Figure 3-12. Preferred most parsimonious tree. Branch tracing indicates occurrence
of tangential and radial fiber pits. Fiber pits radial only (0), fiber pits radial and
Figure 3-13. Preferred most parsimonious tree, showing relationships within the
Lecantheae/Urticeae clade. Characters and character state changes are mapped.
Ct CA CA
C.) 15 I 0 0 0 I
*I ~~C I ^ 1 I^ s
g~ c ^ dQ^g^
Figure 3-14. Preferred most parsimonious tree. Branch tracing indicates occurrence
of the male pistillode and vestiture. Pistillode glabrous (0), pistillode lanate or pilose
(1) pistillode absent (2).
3 3 I . S 11 I
s g-ii 9~i isi
* a 8
*^ S 23 2
Figure 3-15. Preferred most parsimonious tree. Branch tracing indicates degree of
fusion of female perianth. Perianth absent (0), perianth free or lobed (1), perianth
Figure 3-16. Preferred most parsimonious tree. Branch tracing indicates occurrence
of staminodia in pistillate flowers. Staminodia absent (0), staminodia present (1).
C) 00 5
o o~ 8 :~a~
0 ~ F 00~) ~0 -
Figure 3-17. Preferred most parsimonious tree. Branch tracing indicates occurrence
of stinging hairs. Stinging hairs absent (0), stinging hairs present (1).
5 z 0 t
0z -z -'
vl 8 'l ECn
Table 3.1. Characters scored for phylogenetic analysis of the Urticales. All considered
1) Vessel elements:
2) Wood fibers:
3) Fiber pits:
4) Storied fibers:
5) Mucilage cells:
9) Stinging hairs:
10) Hooked hairs:
11) Arachnoid hairs:
14) Female receptacles:
17) Number of stamens:
18) Stamens filaments:
19) Male pistilode:
21) Pollen shape:
22) Pollen spinules:
24) Female perianth:
25) Stigma number:
lignified (0), unlignified (1)
monomorphic (0), dimorphic (1)
radial (0), tangential and radial (1)
absent (0), present (1)
absent (0), present (1)
absent (0), present (1)
spiral (0), opposite (1), anisophyllus (2)
absent (0), punctate (1), elongate (2)
absent (0), present (1)
absent (0), present (1)
absent (0), present (1)
absent (0), present (1)
intrapetiolar (0), interpetiolar (1), lateral and free (2)
not fleshy (0), fleshy (1)
no involucrate disk (0), involucrate disk present (1)
not glomerulate (0), glomerulate (1)
one (1), two (2), four (4), five (5)
+ straight (0), incurved (1)
glabrous (0), lanate or pilose (1), absent (2)
absent (0), present (1)
sphaeroidal (0), subprolate (1), intermediate (2),
small, evenly dispered, and dense (0),
larger, arranged in groups (1),
larger, cone-shaped, in groups (2)
curved (0), straight (1)
absent (0), free or lobed (1), tubular (2)
two (0), one (1)
capitate-penicillate (0), oblong (1), linear (2),
apical and anatropus (0), basal and orthotropus (1)
2 carpellate (0), pseudomonomerous (1)
0 0 c0
(N (N 0
0 0 0
0 0 0
0 0 0
0 0 O
0 0 0
O O (
0 0 0
i- C -
C ^ I i -
0 0 0
0 0 0
(N (N (N
m C- (N
0 0 0
't 7t 7
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
(N (N (N
0 0 0
0 0 0
CD D C
CD CD CD
0 o -S a
c l *4 | | I |
0 t3 Q 2: 0e Q O S. <; u c;M 3
c - c -
00 - -
-(N N - - 'N - -
- - - - - - - -
-) -o - -- - -
SO 0 O N 0 N OO O O (N
MO O O 0 0 00 O
0 O - - -
N0 0 C 0 0 0 0 0 0 C O 0 0 0 O
- 0 0 0 - 0 0 -0 -
0 0 i
0- - -i 0 0 - -
Pg- P- 4 4 r-)
33 3 33 3 3 NN
U 0 0 0 0 0 0 0 0 0 0
'*6b b b b b
^3 c i f f f f f ^ f - - - -
V M 3
Q) ^ M^ ^ i:^ Q (S^ Q <
PHYLOGENY OF KINABALU ELA TOSTEMA
This chapter focuses on phylogenetic relationships among the Elatostema s.l.
from Mount Kinabalu. In Chapter 5, the phylogenetic reconstruction is applied to a
biogeographical analysis with special detail given to micro-geographical areas and
ecological associations within the Kinabalu floristic community.
Elatostema in Malesia
Based on a review of Index Kewensis entries, the genus Elatostema in the
Malesian floristic region includes ca. 250 species. About 120 of these were originally
described from the Philippines, about 75 from New Guinea and approximately 40 from
Borneo. There are 19 species of Elatostema on Mount Kinabalu already published, and an
additional seven species that are newly described here (see Chapter 6). Twelve species are
known only from Mount Kinabalu.
Generic Delimitation and Monophyly of Elatostema
Schr6ter and Winkler (1935, 1936) recognized four subgenera of Elatostema s.l.,
i.e., E. subg. Pellionia (Gaudich.) Hall. f., E. subg. Elatostematoides (C. B. Rob.) H.
Schroter, E. subg. Weddellia H. Schroter, and E. subg. Elatostema (as subg.
Euelatostema H. Schr6ter). They monographed 106 species in the first three of these
subgenera. Their published work never treated E. subg. Elatostema. However, they list
taxa from Borneo
that they recognized within this subgenus, including a number of Kinabalu taxa. There
are no taxa listed from Kinabalu in E. subg. Weddellia. Of the four subgenera, E. subg.
Pellionia is most frequently recognized at the generic level, and it still is in the
horticultural trade. Pellionia sensu Robinson (1910) has cymose or fasciculate staminate
and pistillate inflorescences (sometimes very condensed, but never involucrate), while the
pistillate inflorescences ofElatostema s.s. are fused into involucral heads.
In a revision of Philippine Urticaceae, Robinson (1910) described the genus
Elatostematoides as a segregate of the Elatostema/Pellionia alliance. He treated
Elatostema s.s. as distinct based on two characters, the presence of the involucrate
inflorescences mentioned above, and what he referred to as a cup-shaped perianth on the
pistillate flowers. Robinson doubted that this cup-shaped perianth was really a whorl of
staminodia as suggested by Stapf (1894). This issue, as well as the monophyly of
Elatostema s.s., Pellionia, and Elatostematoides, is considered further in the discussion
section of this chapter. Elatostema s.l. is distinct among the Urticaceae on the basis of
anisophyllous leaves and alternate leaves through reduction via anisophylly.
The phylogenetic analysis included 26 taxa and was based on the observation of
76 morphological characters. Outgroup taxa included two species ofPilea, P. leptocardia
and P. leptograma, and Gyrotaenia myriocarpa. The choice of these outgroups was based
on a preliminary phylogeny of the Urticaceae presented in Chapter 3.
Characters and character states used are listed in Table 4-1. While some of the
characters are self explanatory, others require clarification. These are discussed below.
Some suites of characters are used to describe the overall pattern of distribution of hair or
cystolith types over the plant surface. For example, cystoliths sometimes occur on just
one leaf surface or they may line the veins but not the interstices of a leaf surface. By
coding these as separate characters, the overall pattern is defined in binary format. Care
was taken to avoid unintentional weighting due to character correlation in these suites of
characters. Characters 44, 46, 52, and 66 were deleted for this reason. Characters 6, 23,
41, and 62 were found to be parsimony uninformative, and were also deleted.
Data were analyzed with PAUP* 4.0d65 (Swofford 1998) utilizing the heuristic
search algorithm (500 replicates) and using tree bisection reconnection (TBR) and
random sequence addition options.
Character and Character State Categories and Clarification
Sexual dimorphism. Both monoecious and dioecious plants (char. 1) are
common in Elatostema. Only three species were recorded as polymorphic for this
Leaf venation. Venation types (char. 2) are as defined by Schroter and Winkler
(1935). Leaf venation in the Urticaceae is likely to be derived from palmately veined
ancestors. Related families (i.e., Cecropiaceae and Moraceae) often have palmately
veined leaves. In Elatostema, leaf venation appears to be derived from camptodromus
ancestors (common in Pilea). A transformation series occurs to brochidodromus and
craspedromus states in which the two main arching secondaries appear to be compressed
down toward the base of the leaf to varying degrees. This occurs to different degrees even
on either side of the midvein, making the venation pattern asymmetrical across the
midvein. Venation is then camptodromus on one side of the midvein and brochidodromus
on the other.
Type 1 venation is more or less symmetrical, and camptodromus. Type 2 is more
or less symmetrical, camptodromus, but apically brochidodromus. Type 3 and 4a-b