Published online 7 October 2004
Epiphytism and pollinator specialization:
drivers for orchid diversity?
Barbara Gravendeel1 , Ann Smithson2, Ferry J. W. Slik1 and
Andre Schuiteman1
1
Nationaal Herbarium Nederland–Universiteit Leiden, P.O. Box 9514, Leiden, The Netherlands
University of Exeter, Department of Biological Sciences, Hatherly Laboratories, Prince of Wales Road, Exeter EX4 4PS, UK
(a.smithson@exeter.ac.uk)
2
Epiphytes are a characteristic component of tropical rainforests. Out of the 25 000 orchid species currently
known to science, more than 70% live in tree canopies. Understanding when and how these orchids diversified is vital to understanding the history of epiphytic biomes. We investigated whether orchids managed to
radiate so explosively owing to their predominantly epiphytic habit and/or their specialized pollinator systems by testing these hypotheses from a statistical and phylogenetic standpoint. For the first approach, species numbers of 100 randomly chosen epiphytic and terrestrial genera were compared. Furthermore, the
mean number of pollinators per orchid species within the five subfamilies was calculated and correlated with
their time of diversification and species richness. In the second approach, molecular epiphytic orchid phylogenies were screened for clades with specific suites of epiphytic adaptations. Epiphytic genera were found to
be significantly richer in species than terrestrial genera both for orchids and non-orchids. No evidence was
found for a positive association between pollinator specialization and orchid species richness. Repeated
associations between a small body size, short life cycle and specialized clinging roots of twig epiphytes in
Bulbophyllinae and Oncidiinae were discovered. The development of twig epiphytism in the first group
seems repeatedly correlated with speciation bursts.
Keywords: Bulbophyllinae; epiphytism; Oncidiinae; orchids; pollinator specialization; twig epiphytes
1. INTRODUCTION
Epiphytes are a characteristic component of modern tropical rain and cloud forests, both in terms of species diversity
and biomass. Understanding when and how this particular
life form diversified is vital to understanding the history of
epiphytic biomes. Approximately 7.5% of all vascular plant
species are epiphytes (Gentry & Dodson 1987; Bramwell
2002). Although many epiphytic species exist (more than
23 000), most of them are accounted for (Benzing 1990)
by only a few higher taxa (876 genera in 84 angiosperm
families). Apparently, relatively few lineages have been able
to enter the epiphytic niche successfully. And out of those
taxa that have evolved an epiphytic habit, only few have
radiated into species-rich groups.
Benzing (1990) offered a possible explanation for the few
large epiphytic vascular plant radiations. He postulated
that a complex suite of adaptations is needed for an epiphytic habit. Canopy habitats are indeed difficult to colonize
for four reasons. First, substrate stability is low (Nieder
2004). Structures such as multiple minute climbing roots
that increase adhesion to the host are essential for survival.
Second, nutrient and water supplies are limited owing to
Author for correspondence (gravendeel@nhn.leidenuniv.nl).
One contribution of 16 to a Discussion Meeting Issue ‘Plant phylogeny
and the origin of major biomes’.
Phil. Trans. R. Soc. Lond. B (2004) 359, 1523–1535
doi:10.1098/rstb.2004.1529
the frequently thin substrate cover with low water-carrying
capacity (Chase 1987). Adaptations such as succulence,
crassulacean acid metabolism sequential production of
individual shoots operating as independent physiological
units, and special absorptive tissues prolonging contact
with transitory fluids such as velamentous roots, are
required to overcome severe drought stress (Benzing
1990). Third, canopy habitats are not the most accessible
for colonizing seeds owing to patchiness in the epiphytic
biotope. Arrays of suitable branches within individual hosting crowns are usually scattered and sometimes far apart
(Ibisch et al. 1996). Dust-like seeds, which are easily dispersed by wind and have a high germination success due to
fungal intervention, significantly enhance successful propagation. Fourth, the population density of epiphytes is often
low (Wolf & Flamenco 2003). Highly specialized pollination systems may be required for effective pollen transfer
between such scattered populations.
The morphological and physiological adaptations
mentioned above are most strikingly developed in the
Orchidaceae. This is not surprising, given the fact that out
of the almost 25 000 orchid species currently described, ca.
18 000 are epiphytes (Royal Botanic Gardens, Kew 2003).
In addition, almost half of the 47 largest epiphytic genera
are orchids (Benzing 1990). The question arises as to
whether orchids, in contrast to other vascular plant
1523
# 2004 The Royal Society
1524
B. Gravendeel and others
Epiphytism and pollinator specialization in orchids
6
6
5
4
3
epiphytic microhabitats
5
4
3
2
2
1
1
0
50
number of species
Figure 1. Vertical distribution of orchid species (open bars)
and individuals (black bars) in four of the six different
epiphytic microhabitats as first described by Johansson
(1974). Picture modified from Ek et al. (1997) and Nieder
(2004).
families, managed to radiate so explosively owing to their
predominantly epiphytic habit.
Several hypotheses for the large species richness of orchids have been described. The first theory was postulated by
Gentry & Dodson (1987), who proposed that the high species diversity of orchids might be correlated with their
exceptionally fine niche partitioning. Floristic inventories
by Pittendrigh (1948), Johansson (1974), Ter Steege &
Cornelissen (1989), Ek et al. (1997), Wolf & Flamenco
(2003) and Nieder (2004) indeed demonstrated microhabitat specialization in the tropical epiphytic environment
and show that the total bark and branch surface area available for occupation by epiphytic species greatly exceeds
that of the ground area. These studies also show that many
more orchid species and individuals are crowded into the
tree crown and branches as compared with a similar
ground area. Four main epiphytic microhabitats exist (figure 1). The first consists of the shaded and humid tree
base, where species growing directly on the bark survive.
The second microhabitat encompasses the upper trunk,
where epiphytes grow only when suitable germination sites
are present. The third microhabitat comprises the inner
canopy, which is a heterogeneous assemblage of the
environment of the upper tree trunk and outer canopy.
Here, shade-adapted species can survive in the inner forks
and branches next to hemi-epiphytes growing in packages
of moss and humus and species that can endure sites in
direct sun. The observation that the largest numbers of epiphytic species and crown-foraging pollinators are usually
found here might be correlated with the large diversity of
this environment. The fourth microhabitat is the outer canopy, with high levels of disturbance, prolonged periods of
drought and large fluctuations in temperature, where
largely xeromorphic species are present rooting directly on
the outer well-illuminated twigs. Speciation may be
increased since specialized morphological adaptations
allow a survival advantage in each of these four microhabitats. In addition, the high diversity in tropical tree
Phil. Trans. R. Soc. Lond. B (2004)
species might stimulate further niche differentiation owing
to host specificity.
A second hypothesis explaining the large species richness
of orchids is given by Benzing (1990). He describes the
highly fragmented nature of the epiphytic substratum,
especially in mid-montane rainforest, as an ideal speciation
condition since it should promote allopatric speciation.
According to Benzing (1990), this would explain why not
only orchids, but also epiphytic aroids and bromeliads are
so species-rich despite their different morphological adaptations to the arboreal habitat. This argument is contradicted by Ibisch et al. (1996), who mention that, in plant
families that have evolved epiphytism other than the Orchidaceae, the terrestrial species have higher rates of speciation. They use this observation as support for their
hypothesis that the fragmented nature of the epiphytic substratum cannot be the main driver for diversity in the
orchids.
A third hypothesis explaining the large species richness
of orchids is pollinator specialization. Orchids are widely
believed to have the highest degree of pollinator specialization when compared with other families of flowering plants
(Tremblay 1992; Ibisch et al. 1996). Clearly, a high degree
of prezygotic (‘ethological’) isolation can be imposed if
each group of plants receives visits from only one, unique
pollinator species, and this can easily lead to reproductive
isolation and thus potentially speciation with, or without,
physical barriers to gene flow. The orchids are well known
for certain pollination strategies that are argued to cause
such highly specialized relationships between plant and
pollinator. For example, flowers of species of Ophrys in
Europe and several Australian genera are thought to produce highly specific suites of olfactory and visual stimuli
that each attract a unique species of pollinator, usually a
male bee or wasp (Dafni & Bernhardt 1990). The insect
confuses these stimuli with the conspecific female and
pollinates through repeated mistake copulations. It is suggested that high speciation rates within these genera are
because different species of bees use slightly different olfactory stimuli to attract mates, so only a small number of
mutations in orchid olfactory stimuli genes should be sufficient to cause adaptation to a new species of pollinator
and thus reproductive isolation (Schiestl & Ayasse 2002).
Even in moderately specialized ‘pollinator syndromes’—
such as those associated with bumble-bee and hummingbird pollination, where a plant species is pollinated by a
small number of bumble-bee or hummingbird species—
one major gene mutation may be all that is required to
cause an adaptive switch between the two syndromes
(Bradshaw & Schemske 2003). Thus chance mutations
within plant species that are only moderately specialized on
pollinators could also potentially drive speciation. Gentry
& Dodson (1987) consider bee, fly, hummingbird, hawkmoth, bat and small mammal pollination syndromes all to
be moderately specialized in terms of the number of pollinator species attracted. Epiphytic taxa with these types of
pollination systems are present in Apocynaceae (Asclepiadaceae), Bignoniaceae, Bromeliaceae, Cactaceae, Ericaceae, Gesneriaceae, Marcgraviaceae and Rubiaceae. Apart
from Bromeliaceae, these families have much lower species
diversities compared with the Orchidaceae, and therefore
this does not necessarily support the hypothesis that pollinator specialization per se has driven speciation. However,
Epiphytism and pollinator specialization in orchids
for many authors the specialized pollination mechanisms
displayed in some orchid genera are sufficient to warrant
this argument. Tremblay (1992) attempted to test the pollinator specialization hypothesis by comparing information
on numbers of pollinator species per orchid available from
the literature. Out of the 456 orchid species for which pollinators were known in 1992, he argued that 67% were pollinated by a single pollinator. A much lower proportion of
orchid species were pollinated by two (14%) or more
(17%) species. Only a few orchids (< 5%) exploit pollinators indiscriminately, suggesting high pollinator specialization. Further, Tremblay (1992) averaged the numbers of
pollinator species across the orchid subfamilies recognized
by phylogeneticists at the time, and argued that more
recently derived subfamilies have fewer pollinators, suggesting that pollinator specialization could drive speciation.
However, the information from both orchid phylogenetics
and on pollinating species has increased since 1992. It is
also recognized that time spent studying a single plant
species often increases the number of pollinator species
recognized, and that including casual observations can
erroneously categorize species as specialized (Ollerton &
Cranmer 2002).
A fourth hypothesis for the large species richness of orchids is given by Vasquez et al. (2003). They use longdistance dispersal modes as an explanation for the high
species diversity of orchids by using the argument that their
dust seeds enable the establishment of innumerable small
and dispersed founder populations, and thus high rates of
allopatric speciation. Benzing (1990) also uses this argument as the explanation for high species richness amongst
epiphytic ferns. Kessler (2002) however, shows that
bromeliad taxa with adaptations to long-distance dispersal
(dust seeds) have lower species numbers as compared with
taxa with short-distance dispersal modes (winged seeds
and berries). He attributed this to the hypothesis that longdistance dispersal is much more efficient in colonizing canopy habitats and preventing population differentiation
owing to a high gene flow.
A fifth hypothesis for the large species richness of orchids
can be deduced from Wikström & Kenrick (2000, 2001).
They correlate development of a closed-canopy vegetation
of broad-leafed angiosperm forests during the Late
Cretaceous and Early Tertiary with an increase in the ranges of light and humidity conditions. According to these
authors, this had a positive effect on epiphytic diversification and a negative effect on terrestrial species diversity
because the quality and intensity of light reaching the forest
floor was greatly reduced. The Orchidaceae, with their
highest percentage of epiphytism, might have had the best
chances of diversification in this new environment with
light in the canopy and shade at the forest floor. That epiphytic diversification in the Early Tertiary is not a general
phenomenon for vascular plants, however, is pointed
out by Benzing (1990). He states that many speciesrich, canopy-based orchid and non-orchid genera such as
Anthurium, Peperomia, Rhododendron and Tillandsia are
concentrated in geologically young montane habitats,
which suggests that much of the current epiphyte diversity
dates from the Pliocene/Pleistocene only. Molecular dating
studies on the diversification of the epiphytic orchid genus
Coelogyne from terrestrial allies in the Himalayan and
Phil. Trans. R. Soc. Lond. B (2004)
B. Gravendeel and others 1525
southeast Asian region performed by Gravendeel (2004)
support this hypothesis.
2. GENERAL AIMS
We have highlighted that there is a debate as to whether the
epiphytic habit promotes speciation rates in the Orchidaceae. Pollinator specialization has been advanced as a main
alternative to this hypothesis. We test these contrasting
hypotheses from both a statistical and a phylogenetic
standpoint.
(i) To test statistically whether epiphytism could have
driven speciation, species numbers of 100 randomly chosen epiphytic and terrestrial genera were compared. This
was carried out for both orchids and non-orchids to test
whether epiphytic species have a significantly different
number of species compared with terrestrial species. (ii) To
test statistically whether pollinator specialization could
have driven speciation, pollinator numbers of all orchid
species described in Van der Cingel (1995, 2001) of the
(predominantly terrestrial) subfamilies Apostasioideae,
Cypripedioideae, Orchidoideae and (climbing) Vanilloideae were compared with the numbers of pollinators per
species of the (predominantly epiphytic) subfamily Epidendroideae.
For the second approach, two different phylogenies of
tropical epiphytic orchids were used to investigate whether
(iii) special suites of epiphytic adaptations characterize specific clades; (iv) how many times these combination of
characters evolved; and (v) whether these characters are
repeatedly associated with main speciation bursts.
3. MATERIAL AND METHODS
(a) Concept of epiphytism used
The designation ‘epiphyte’ is here reserved for rain-fed ‘atmospheric’ vascular plants, germinating on their host (which can
either be trees or rocks) to which they are anchored by a few roots
only and which never become host vasculature- or ground-connected so that they can allocate resources for growth and reproduction that soil-based terrestrials have to allocate to self-support
(Benzing 1987, 1990). For practical reasons, we included epilithic
species in our survey as well, although some epilithic orchids are
restricted to rocks only. The term ‘obligate twig epiphytes’ is used
for those epiphytes that occur predominantly on less than 2-yearold ultimate branchlets and twigs (Chase 1987).
(b) Statistical analyses
To determine whether epiphytes have significantly more species
compared with terrestrial orchids, the epiphytic genera listed in
table 1.1 of Benzing (1990) were numbered. Subsequently, 100
numbers were generated (using the random number generation
option as implemented in Microsoft EXCEL v. 2002) for both
orchid and non-orchid genera. Out of the 100 genera in each dataset, the next terrestrial genus according to the alphabet was looked
up in Mabberley (1998). Out of all genera, the number of species
was recorded (table 1). Genera found were sorted from small to
large species numbers and subsequently subdivided into 10
classes of 10 observations each.
A likelihood-ratio test for goodness of fit (or G-test), as
described by Sokal & Rohlf (1995), was subsequently performed
to compare the distributions of species over genera between epiphytes and terrestrials, both for orchids and non-orchids.
To determine whether a high degree of pollinator specialization
is correlated with a large orchid species diversity, the number of
1526
B. Gravendeel and others
Epiphytism and pollinator specialization in orchids
Table 1. Species numbers of 100 randomly chosen (a) orchid genera and (b) non-orchid genera from the list of epiphytes as given
in Benzing (1990) and the next terrestrial genus according to the alphabet in Mabberley (1998).
(a) orchid genera
epiphytic genus
species number
terrestrial genus
species number
Acineta
Aeranthes
Ancistrochilus
Ancistrorhynchus
Angraecopsis
Ansellia
Armodorum
Basiphyllaea
Benzingia
Bogoria
Bollea
Bolusiella
Brachypeza
Bulbophyllum
Caucaea
Ceratochilus
Chamaeangis
Chamaeanthus
Chitonanthera
Chondrorhyncha
Chroniochilus
Chysis
Cleisocentrum
Cleisostoma
Clowesia
Coryanthes
Cryptopus
Cycnoches
Cymbidiella
Cyrtorchis
Dendrobium
Dilomilis
Dimerandra
Diothonaea
Dracula
Dressleria
Dryadorchis
Drymoda
Encyclia
Eparmatostigma
Epidendrum
Eria
Gongora
Hagsatera
Hofmeisterella
Hygrochilus
Hymenorchis
Ischnogyne
Jacquiniella
Laelia
Leptotes
Lockhartia
Lopharis
Macroclinium
Mediocalcar
Mendoncella
Mesospinidium
Neofinetia
Neogyna
Notylia
10
30
2
13
14
2
2
3
2
4
7
10
7
1000
1
2
15
10
7
16
5
6
1
95
5
20
3
17
3
18
900
4
2
7
93
4
2
2
130
1
500
500
40
2
1
1
9
1
11
69
5
29
25
25
20
11
7
1
1
75
Acrolophia
Altensteinia
Androcorys
Androchilus
Anoectochilus
Anthogonium
Arnottia
Baskervilla
Beloglottis
Bonatea
Brachystele
Brachionidium
Brownleea
Burnettia
Centrostigma
Cheirostylis
Chiloglottis
Chloraea
Chusua
Chrysoglossum
Coeloglossum
Collabium
Claderia
Cleistes
Codonorchis
Corybas
Cryptostylis
Cynorkis
Cyrtosia
Cyrtostylis
Diceratostele
Diphylax
Diplomerus
Diplolabellum
Drakaea
Duckeella
Eleorchis
Ephippianthus
Epiblema
Epipactis
Epipogium
Eriaxis
Goniochilus
Hancockia
Holothrix
Hylophila
Isotria
Ipsea
Kreodanthus
Kuhlhasseltia
Ligeophila
Ludisia
Lyperanthus
Malaxis
Megalorchis
Mexipedium
Microtis
Neottia
Neotinea
Odontorrhynchus
9
9
4
1
35
1
2
7
1
20
18
35
7
1
5
15
18
47
17
6
1
10
2
55
3
100
20
125
5
5
1
1
2
1
4
3
1
1
1
22
3
3
1
1
55
6
2
2
6
6
8
1
5
300
1
1
11
9
2
5
(Continued.)
Phil. Trans. R. Soc. Lond. B (2004)
Epiphytism and pollinator specialization in orchids
B. Gravendeel and others 1527
Table 1. (Continued.)
430
8
25
1
15
1
7
3
1
6
1
9
3
1
1
4
6
14
4
2
5
2
2
1
3
1
1
4
2
25
5
187
11
10
4
2
8
21
84
5
Oncidium
Otoglossum
Pachyphyllum
Papperitzia
Pedilochilus
Peristeranthus
Phloeophila
Pinelia
Platyrhiza
Plectrophora
Polyotidium
Ponera
Porphyrodesme
Porphyroglottis
Pseudacoridium
Quisqueya
Rangaeris
Renanthera
Rhipidoglossum
Rhyncholaelia
Rossioglossum
Rudolfiella
Sanderella
Sepalosiphon
Smitinandia
Sphyrarhynchus
Stenia
Stolzia
Summerhayesia
Sunipia
Systeloglossum
Taeniophyllum
Tetramicra
Thelasis
Trevoria
Trias
Trichoceros
Trichopilia
Trichosalpinx
Tuberolabium
25
3
2
1
73
70
14
4
8
2
53
2
60
21
6
1
1
2
1
1
7
40
13
6
1
4
1
3
1
14
1
1
46
1
19
35
5
2
2
3
Ophrys
Otostylis
Pachyplectron
Papuaea
Pelexia
Peristylus
Phragmipedium
Piperia
Platythelys
Pogonia
Ponthieva
Porphyrostachys
Prasophyllum
Prescottia
Pseudocentrum
Raycadenco
Rhamphorhynchus
Rhizanthella
Rimacola
Risleya
Sarcanthopsis
Sarcoglottis
Serapias
Sertifera
Spiculaea
Stenoglottis
Satyridium
Solenocentrum
Symphyosepalum
Tainia
Thulinia
Thaia
Thelymitra
Thelyschista
Triphora
Tropidia
Trudelia
Tsaiorchis
Tubilabium
Tylostigma
(b) non-orchid species
family
Melastomataceae
Gesneriaceae
Gesneriaceae
Gesneriaceae
Araceae
Polypodiaceae
Caryophyllaceae
Polypodiaceae
Aspleniaceae
Cyclanthaceae
Begoniaceae
Burmanniaceae
Campanulaceae
Ericaceae
Melastomataceae
Bromeliaceae
Asclepiadaceae
Campanulaceae
Melastomataceae
Liliaceae
epiphytic genus
species number
family
terrestrial genus
species number
Adelobotrys
Agalmyla
Alloplectus
Alsobia
Amydrium
Anathropteris
Arenaria
Arthromeris
Asplenium
Asplundia
Begonia
Burmannia
Burmeistera
Calopteryx
Calvoa
Catopsis
Ceropegia
Clermontia
Clidemia
Clivia
21
15
25
2
4
1
1
6
400
60
30
2
5
1
4
20
3
10
11
1
Lamiaceae
Apocynaceae
Asteraceae
Icacinaceae
Euphorbiaceae
Brassicaceae
Palmae
Poaceae
Asteraceae
Brasssiceae
Sapindaceae
Campanulaceae
Alismataceae
Restionaceae
Asteraceae
Bombacaceae
Palmae
Clethraceae
Rosaceae
Myrtaceae
Adelosa
Aganonerion
Allopterigeron
Alsodeiopsis
Amyrea
Anastatica
Arenga
Arthroostachys
Asplundianthus
Asta
Beguea
Burmeistera
Burnatia
Calorophus
Calycadenia
Catostemma
Ceroxylon
Clethra
Cliffortia
Cloezia
1
1
1
11
2
1
20
1
17
2
1
80
1
1
11
11
15
64
115
8
(Continued.)
Phil. Trans. R. Soc. Lond. B (2004)
1528
B. Gravendeel and others
Epiphytism and pollinator specialization in orchids
Table 1. (Continued.)
Ericaceae
Moraceae
Polypodiaceae
Asclepiadaceae
Cyperaceae
Gesneriaceae
Melastomataceae
Davalliaceae
Polypodiaceae
Polypodiaceae
Ericaceae
Dioscoreaceae
Ericaceae
Solanaceae
Rapateaceae
Asteraceae
Cyclanthaceae
Onagraceae
Gnetaceae
Liliaceae
Bromeliaceae
Solanaceae
Ericaceae
Gentianaceae
Bromeliaceae
Gesneriaceae
Gentianaceae
Rubiaceae
Polypodiaceae
Melastomataceae
Myrtaceae
Polypodiaceae
Vittariaceae
Myrsinaceae
Bromeliaceae
Polypodiaceae
Melastomataceae
Davalliaceae
Polypodiaceae
Liliaceae
Pandanaceae
Gesneriaceae
Melastomataceae
Polypodiaceae
Urticaceae
Bromeliaceae
Polypodiaceae
Ericaceae
Polypodiaceae
Epacridaceae
Urticaceae
Bromeliaceae
Solanaceae
Rubiaceae
Clusiaceae
Araceae
Liliaceae
Ericaceae
Marcgraviaceae
Gesneriaceae
Bignoniaceae
Rubiaceae
Araceae
Polypodiaceae
Davalliaceae
Costera
Coussapoa
Crypsinus
Cynanchum
Cyperus
Cyrtandra
Dalenia
Davallia
Dendroconche
Diblemma
Didonica
Dioscorea
Disterigma
Ectozoma
Epidryos
Eupatorium
Evodianthus
Fuchsia
Gnetum
Hippeastrum
Hohenbergia
Juanulloa
Lateropora
Leiphaimos
Lymania
Lysionotus
Macrocarpaea
Malanea
Marginariopsis
Medinilla
Metrosideros
Microgramma
Monogramma
Myrsine
Navia
Neocheiropteris
Neodissochaeta
Nephrolepis
Oleandropsis
Pamianthe
Pandanus
Paradrymonia
Phainantha
Photinopteris
Pilea
Pitcairnia
Pleopeltis
Plutarchia
Polypodiopteris
Prionotes
Procris
Pseudaechmea
Rahowardiana
Relbunium
Renggeria
Rhaphidophora
Rhodocodon
Rusbya
Ruyschia
Sarmienta
Schlegelia
Schradera
Scindapsus
Scleroglossum
Scyphularia
8
20
40
2
1
10
2
40
2
1
2
1
15
1
3
7
1
15
3
2
20
10
2
1
4
2
2
2
1
300
3
13
2
12
2
10
10
15
1
1
4
8
4
1
20
75
10
6
3
1
10
1
1
2
1
100
1
1
7
1
18
12
20
6
8
Cyperaceae
Rubiaceae
Poaceae
Apiaceae
Acanthaceae
Amaryllidaceae
Fabaceae
Asteraceae
Annonaceae
Rubiaceae
Rubiaceae
Menispermaceae
Juncaceae
Poaceae
Orobanchaceae
Sapindaceae
Rutaceae
Apiaceae
Thymelaeaceae
Asteraceae
Malvaceae
Palmae
Scrophulariaceae
Aizoaceae
Ericaceae
Epacridaceae
Melastomataceae
Olacaceae
Apiaceae
Sarcolaenaceae
Palmae
Asteraceae
Chenopodiaceae
Myrtaceae
Malvaceae
Fabaceae
Urticaceae
Menyanthaceae
Asteraceae
Asteraceae
Chenopodiaceae
Euphorbiaceae
Orchidaceae
Loranthaceae
Myrtaceae
Fabaceae
Rutaceae
Poaceae
Poaceae
Brassicaceae
Amaryllidaceae
Euphorbiaceae
Asteraceae
Poaceae
Rubiaceae
Rubiaceae
Bignoniaceae
Aizoaceae
Flacourtiaceae
Sarraceniaceae
Sapindaceae
Oleaceae
Apiaceae
Chenopodiaceae
Scytopetalaceae
Costularia
Coussarea
Crypsis
Cynapium
Cyphacanthus
Cyrthanthus
Dalhousiea
Daveaua
Dendrokingstonia
Dibrachyionostylus
Didymaea
Dioscoreophyllum
Distichia
Ectrosia
Epifagus
Euphorianthus
Evodiella
Fuernrohria
Gnidia
Hippia
Hoheria
Jubaea
Lathraea
Leipoldtia
Lyonia
Lissanthe
Macrocentrum
Malania
Margotia
Mediusella
Metroxylon
Microgynella
Monolepis
Myrtastrum
Nayariophyton
Neochevalierodendron
Neodistemon
Nephrolephyllidium
Olearia
Pamphalea
Panderia
Paradrypetes
Phaius
Phragmanthera
Pileanthus
Pithecellobium
Plethadenia
Poa
Polypogon
Prionotrichon
Proiphys
Pseudagrostistachys
Raillardella
Relchela
Rennellia
Rhaphidura
Rhodocolea
Ruschia
Ryania
Sarracenia
Schleichera
Schrebera
Sciothamnus
Sclerolaena
Scytopetalum
20
100
8
1
1
47
3
1
1
1
5
3
3
11
1
1
2
1
140
8
5
1
7
10
35
6
15
1
1
1
5
1
6
1
1
1
1
1
75
6
1
2
45
6
3
37
2
200
10
4
3
2
3
1
4
1
6
360
8
8
1
10
4
64
3
(Continued.)
Phil. Trans. R. Soc. Lond. B (2004)
Epiphytism and pollinator specialization in orchids
B. Gravendeel and others 1529
Table 1. (Continued.)
Solanaceae
Cyclanthaceae
Araceae
Gesneriaceae
Vitaceae
Polypodiaceae
Cyclanthaceae
Bromeliaceae
Melastomataceae
Ericaceae
Vittariaceae
Bromeliaceae
Cactaceae
Bromeliaceae
Agavaceae
15
2
30
10
2
1
1
400
2
95
6
200
2
4
2
Solanum
Stelestylis
Stenospermation
Streptocarpus
Tetrastigma
Thayeria
Thoracocarpus
Tillandsia
Triolena
Vaccinium
Vaginularia
Vriesea
Werckleocereus
Wittrockia
Yucca
Table 2. Species numbers of taxa in Bulbophyllinae investigated, their taxonomic rank and estimated number of species
(based on Schlechter 1912, 1925; Vermeulen 1987).
genus
Bulbophyllum
section
species number
Altisceptrum
Aphanobulbon
Careyana
Cirrhopetalum
Desmosanthes
Globiceps
Hybochilus
Hymenobractea
Intervallatae
Leptopus
Macrobulbon
Macrouris
Oxysepalum
Pelma
Polyblepharon
Sestochilus
9
110a
33
78
50
48
33a
5
74
45
7
27a
19a
22a
90a
90
3
36a
27
15
Drymoda
Pedilochilus
Sunipia
Trias
a
Groups containing obligate twig epiphytes.
pollinators of orchid species belonging to the (predominantly terrestrial) subfamilies Apostasioideae, Cypripedioideae, Orchidoideae, (climbing) Vanilloideae, and (predominantly epiphytic)
subfamily Epidendroideae was recorded from Van der Cingel
(1995, 2001). We excluded casual observations regardless of plant
species, focusing on quantitative pollinator studies. For comparison, the mean number of pollinators and standard error were calculated per subfamily. Overall mean number of pollinator species
per orchid, and relationships between more recently derived subfamilies and mean pollinator numbers, were also compared using
a G-test. For comparison, the mean number of pollinators and
standard deviation were calculated per subfamily.
(c) Selection of phylogenetic datasets
Phylogenies of the largely epiphytic Bulbophyllinae
(B. Gravendeel, unpublished data) and Oncidiinae (Williams et al.
2001) were selected because of the availability of floristic,
molecular phylogenetic, and macromorphological data.
Phil. Trans. R. Soc. Lond. B (2004)
Alliaceae
Caryophyllaceae
Acanthaceae
Poaceae
Olaceae
Euphorbiaceae
Ericaceae
Myrtaceae
Burseraceae
Amaryllidaceae
Apocynaceae
Orchidaceae
Brassicaceae
Alseuosmiaceae
Rubiaceae
Solaria
Stellaria
Stenostephanus
Streptochaeta
Tetrastylidium
Thecacoris
Thoracosperma
Tillospermum
Triomma
Vagaria
Vahadenia
Vrydagzynea
Werdermannia
Wittsteinia
Yutajea
2
200
6
3
3
20
10
36
1
4
2
40
4
3
1
(d) Phylogenetic analysis
The character states of morphological characters at internal
nodes were reconstructed with Mesquite v. 1.0 (Maddison &
Maddison 2003) using the ‘likelihood ancestral states’ option.
This likelihood reconstruction finds, for each node, the state
assignment that maximizes the probability of arriving at the
observed states in the terminal taxa under a Mk1 (gains and losses
are equally likely) model of evolution, and allows the states at all
other nodes to vary. The relative likelihoods found are indicated as
pie diagrams in the cladograms analysed.
4. RESULTS
(a) Statistical analyses
The results of the G-tests show that epiphytic genera are
significantly richer in species than terrestrial genera
( p < 0:01) both for orchids and non-orchids (figures 2
and 3). The magnitude of the epiphyte–terrestrial imbalance is greater for orchids than for non-orchids, especially
in classes 9 and 10.
Overall, the mean number of pollinators amongst the
orchids was 3:98^0:97 per species, with 46% of species
having one pollinator (n ¼ 424). Excluding those genera
proposed to have specialized pollination, such as pseudocopulation, there were 5:4^1:8 pollinators per species,
with only 38% of species having just one pollinator
(n ¼ 232). The mean number of species pollinating each
subfamily, along with species richness per subfamily (number of species currently described per subfamily), are
shown in figure 4. There was a trend for more recently
derived subfamilies to have a larger mean number of
pollinators per species, but this correlation was not significant (r p ¼ 0:59, p > 0:05). Further, there was a trend for
the mean number of pollinators per species within subfamilies to increase with subfamilial species richness,
although this correlation was again not significant
(r p ¼ 0:50, p > 0:05). The predominantly epiphytic Epidendroideae tended to have more pollinators per species
when compared with the other families, which are predominantly terrestrial (Apostasioideae, Cypripedioideae,
Orchidoideae), or climbing (Vanilloideae), but again this
was not significant (Epidendroideae: 4:61^1:80 pollinators per species; others: 3:27^0:25 pollinators per species;
t 422 ¼ 0:69, p > 0:05).
Epiphytism and pollinator specialization in orchids
7
mean number of pollinators
500
1
2
3
4
5
6
size class
7
8
9
10
Figure 2. Species numbers of randomly sampled epiphytic
(filled bars) and terrestrial (open bars) orchid genera sorted
from small to large and subsequently divided into 10 classes of
10 observations each. The two distributions differed
significantly from each other according to the G-test
(d:f: ¼ 9; v2 ¼ 1015; p < 0:01).
175
4
15 000
3
2
17
1
0
orchid subfamily
Figure 4. Mean number of pollinators calculated from
observations listed in Van der Cingel (1995, 2001) for orchid
species belonging to subfamilies Apostasioideae (n ¼ 2),
Cypripedioideae (n ¼ 8), Vanilloideae (n ¼ 4), Orchidoideae
(n ¼ 184) and Epidendroideae (n ¼ 227) with corresponding
standard errors. Subfamilial species richness (numbers of
species currently described per subfamily; Royal Botanic
Gardens, Kew 2003) is indicated above the corresponding
bars.
4000
number of species
5
Epidendroideae
1000
4600
Orchidoideae
1500
235
6
Apostasioideae
number of species
2000
Vanilloideae
B. Gravendeel and others
Cypripedioideae
1530
3000
2000
1000
1
2
3
4
5
6
size class
7
8
9
10
Figure 3. Species numbers of randomly sampled epiphytic
(filled bars) and terrestrial (open bars) non-orchid genera
sorted from small to large and subsequently divided into 10
classes of 10 observations each. The two distributions differed
significantly from each other according to the G-test
(d:f: ¼ 9; v2 ¼ 99; p < 0:01).
(b) Phylogenetic analyses
When optimized on the molecular phylogenies presented
in figures 5 and 6, several clades are characterized by a
small body size, short life cycle and climbing roots with
multiple adhesion points. This suite of epiphytic adaptations seems to be present in obligate twig epiphytes only
and evolved multiple times in the Bulbophyllinae and
Oncidiinae.
In Bulbophyllinae, Bulbophyllum sections Aphanobulbon,
Coelochilus, Fruticicola, Hybochilus, Lepanthanthe, Macrouris,
Monilibulbus, Nematorhizis, Oxysepalum, Pelma, Pedilochilus,
Peltopus and Polyblepharon contain obligate twig epiphytes.
In particular, sections Aphanobulbon, Coelochilus and
Polyblepharon are species-rich as compared with the other
sections and genera in Bulbophyllinae (table 2) and show
independent associations between twig epiphytism and
speciation bursts.
The twig epiphytic groups in Oncidiinae are confined to
31 genera, of which Comparettia, Erycina, Ionopsis, Macroclinium, Notylia, Rodriguezia, and Tolumnia were included
in the molecular phylogeny of Williams et al. (2001). It is
Phil. Trans. R. Soc. Lond. B (2004)
impossible to say whether the number of species of these
twig epiphytic genera is high or low because many of the
traditionally recognized genera of Oncidiinae are polyphyletic and in need of revision (Chase & Palmer 1992;
Williams et al. 2001).
5. DISCUSSION
(a) Higher diversity of epiphytic genera
Our results show that epiphytic genera are significantly
richer in species than terrestrial genera, both in orchids and
non-orchids. It must be stressed, however, that a random
sample of only 100 genera might be too small to represent
trends in the more than 13 000 genera of vascular plants
currently described. We think, however, that a random
sample of 100 of the 438 epiphytic orchid genera currently
described should be sufficient to discover trends in species
richness.
The larger taxonomic diversification found for epiphytes
contradict Ibisch et al. (1996), who state that in plant families which have evolved epiphytism other than the Orchidaceae, the terrestrials have more elevated levels of
speciation. The correlation found here between high species diversity and epiphytism supports the hypotheses of
Gentry & Dodson (1987) and Benzing (1990) who postulate that the epiphytic habitat offers more possibilities for
speciation owing to its larger number of niches and more
fragmented nature compared with the forest floor.
Another factor that could explain this difference in
species richness is that, within a forest, total bark surface
area greatly exceeds ground area and can be more densely packed with plants. One square metre of ground
area can be equivalent to more than 10 m2 of canopy
area immediately above it. The larger the area, the more
Epiphytism and pollinator specialization in orchids
B. Gravendeel and others 1531
Bulbophyllum tenuifolium
B. trirhopalon
B. ochroleucum
B. specnov
B. specnov
B. lineolatum
B. formosum
B. nitidum
B. maxillare
B. maxillare
B. coloratum
B. stabile
B. ramulicola
B. maquilingense
B. spec
B. alticola
B. dolichoglottis
B. spec
B. tortuosum
B. gadgarrense
B. nummularioides
B. teysmannii
B. affine
B. tentaculiferum
B. specnov
B. caudatisepalum
B. farinulentum
B. mutabile
B. grudense
B. furcillatum
B. membranaceum
B. specnov
B. stipulaceum
B. levatii
B. macrourum
Pedilochilus guttulatum
B. agastor
B. cruentum
B. phalaenopsis
B. reptans
B. carunculatum
B. papulosum
B. vanvuurenii
B. oobulbum
B. fractiflexum
B. orbiculare
B. hymenobracteum
B. bariense
B. echinolabium
Drymoda siamensis
Sunipia andersonii
Sunipia anamensis
B. coniferum
B. lobbii
B. trigonobulbum
B. modestum
B. skeatinium
Trias oblonga
Trias intermedia
Drymoda gymnopus
B. kutbuense
B. specnov
B. patens
B. tolleniferum
B. emiliorum
B. pileatum
B. digoelense
B. specnov
B. brienianum
B. lepidum
B. auratum
B. pulchellum
B. medusae
B. vaginatum
B. acuminatum
B. plumatum
B. mirum
B. obtusum
B. rubroguttatum
B. maculatum
B. sp.
B. hainanense
B. spMad
B. sarcophylloides
B. annandalei
B. lasiochilum
B. taeter
B. adelphidium
B. refractum
B. spMad
B. spMad
B. baronii
B. leandrianum
B. spMad
B. spMad
B. pleurothallopsis
B. nutans
B. spMad
B. spMad
B. spMad
B. spMad
B. spMad
B. spMad
B. hamelinii
B. spMad
B. erectum
B. sp.
B. sp.
B. cirrhosum
B. bracteolatum
B. micropetalum
B. sp.
B. wedelii
B. wederbauerianum
B. oxychilum
B. pumilum
B. barbigerum
B. sp.
Dendrobium aphyllum
D. salaccense
D. crystallinum
D. kingianum
Figure 5. One of more than 10 000 maximum parsimonious trees (randomly chosen) based upon matK sequences of
Bulbophyllinae (B. Gravendeel, unpublished data). Trunk epiphytes, black circles; twig epiphytes, grey circles. Pie diagrams
depict the relative likelihoods found.
1532
B. Gravendeel and others
Epiphytism and pollinator specialization in orchids
Rodriguezia delcastilloi
R. satipoana
R. lanceolata
Notylia barkerii
Macroclinium bicolor
Ionopsis utricularioides
I. satyrioides
I. minutiflora
Comparettia macroplectron
Zelenkoa onusta
Capanemia superflua
Oncidium dasystlye
O. flexuosum
Rodrigueziella gomezoides
Gomesa planifolia
Erycina hyalinobulbon
E. echinata
E. pusilla
E. pumilio
E. cristagalli
Rhynchostele bicfonensis
Amparoa costaricense
Rhynchostele londesboroughiana
Tolumnia calochila
T. variegata
T. tuerckheimiii
T. henekenii
Ada aurantiaca
A. sp.
Mesospinidium panamense
Ada allenii
Brassia caudata
B. gireoudiana
B. arcuigera
Brachtia andina
Aspasia epidendroides
Cischweinfia dasyandra
Systeloglossum acuminatum
Miltonia flavescens
M.candida
Cyrtochilum edwardii
C. annulare
C. rhodoneurum
C. cimiciferum
Miltoniopsis warscewiczii
Otoglossum chinquense
Miltoniodes reichenheimii
Mexicoa ghiesbreghtiana
Oncidium sphacelatum
O. leucochilum
O. ornithorrhynchum
O. cheirophorum
Symphyglossum sanguineum
Odontoglossum hanyanum
Cochlioda noezliana
Odontoglossum multistellare
Sigmatostalix picta
Trichopilia subulata
T. maculata
T. sanguinolenta
T. brevis
Psychopsis papilio
Pescaforea lehmaniii
Dichaea municata
Trichocentrum pfavii
T. lanceanum
T. splendidum
T. cebolleta
T. jonesianum
Lockhartia oerstedii
L. amoena
Stellilabium pogonostalix
Telipogon parvulus
Hofmeisterella eumicroscopica
Trichoceros parviflorus
Ornithocephalus inflexus
Sphyrastylis escobanana
Dipteranthus grandiflorus
Fernandezia sp.
Figure 6. Single maximum parsimonious tree based upon nuclear (nrITS1–5.8S–ITS2) sequences of Oncidiinae (Williams et al.
2001). Trunk epiphytes, black circles; twig epiphytes, grey circles. Pie diagrams depict the relative likelihoods found.
different species can coexist in this area. A hypothetical
species–area curve should therefore become saturated at
a higher maximum number of species of epiphytes compared with terrestrial ones (figure 7).
Epiphytic orchids are overwhelmingly tropical, whereas
terrestrial orchids are tropical and temperate. The general
increase in species diversity in the tropics could be another
explanation for the significant difference in species diversity found between epiphytes and terrestrials. This hypothesis is not supported, though, by the species diversity of
subfamily Orchidoideae, which is largely terrestrial in the
tropics as well and evenly diverse in both temperate and
tropical regions.
Phil. Trans. R. Soc. Lond. B (2004)
(b) Correlation between pollinator specialization
and species richness
Our analysis has found no evidence that pollinator specialization has driven speciation in the Orchidaceae.
Instead, we have found that more recently derived subfamilies tended to show decreased, rather than increased,
pollinator specialization (cf. Tremblay 1992). Similarly,
species richness tended to decrease with increased pollinator specialization. The predominantly epiphytic subfamily Epidendroideae is the most species-rich, but does
not have a significantly different level of pollinator specialization from other subfamilies (indeed, if anything there is a
trend for decreased rather than increased pollinator
Epiphytism and pollinator specialization in orchids
B. Gravendeel and others 1533
Strunk epiphytes max
number of species
number of species
Sepiphytes max
Sterrestrials max
0
100
percentage of ground area
Figure 7. Hypothetical species–area curve for epiphytes
(dotted line) and terrestrials (solid line).
specialization). Orchids generally are less pollinatorspecialized than is generally assumed—most species have
more than one pollinator.
It must be stressed that certain groups of orchids may be
underrepresented within our dataset, and that the results
presented here might be influenced by the fact tha
pollination has been most extensively studied in European
orchids. The sample size within Apostasioideae is particularly small, and we encourage further studies within this
subfamily. Further, we cannot rule out the hypothesis that
specialization has been important for speciation in some
orchid clades, nor that some pollinator-associated mechanisms more complex than mere specialization may have
driven speciation. We feel, however, from this analysis that
pollinator specialization per se is unlikely to have driven
orchid speciation.
(c) Twig epiphytism
Twig epiphytes appeared to have several unique features
that make them highly specialized for growth and survival
on outermost branchlets. The first is a small body size to
make efficient use of space in the canopy. According to
Nieder (2004), a small body size allows some orchids to
grow on extremely tiny twigs.
The second feature is related to the fact that twigs are
short-lived habitats: they either break off or develop into
larger branches with a more textured surface and thicker
substrate layer. During these two processes, they become
unsuitable habitats for twig epiphytes and are colonized
instead by ‘trunk epiphytes’ with long-lived life cycles
(Dungeon 1923; Benzing 1990). A short life cycle forces
twig epiphytes to ensure that they colonize, mature and
reproduce before the twig they are growing on is either
abscised or develops into a large branch with different substrate characteristics. Many of the twig epiphytes in the
Oncidiinae and Bulbophyllinae mature within the course
of a single season (Chase 1986; G. Fischer, personal communication).
Phil. Trans. R. Soc. Lond. B (2004)
Stwig epiphytes max
0
100
percentage of canopy area
Figure 8. Hypothetical species–area curve for twig epiphytes
(solid line) and trunk epiphytes (dotted line).
A third factor is the large amount of vegetative reduction
that most twig epiphytes display. Growth forms can be
laterally flattened and small in size. Leaves are unifacial or
completely absent (Chase 1987). In addition, shoot development is limited, whereas root production is increased.
Bloom et al. (1985) postulated that this vegetative
reduction is meant to mitigate specific resource scarcities.
According to Chase (1987), a leafless habit is a more
efficient way for small plants to deal with water/surface area
relations compared with standard habits.
A fourth factor is also related to the fact that twigs are
short-lived habitats. Many twig epiphytes have clinging
roots with one or more secondary points of attachment,
which have the effect of increasing the number of growing
seasons before they are shed by their hosts. These host trees
can often be found with several twig epiphytic plants
loosely dangling by their clinging roots (Chase 1986).
A fifth factor is owing to the fact that tiny twigs only have
a very thin or completely absent substrate cover with very
limited absorbent capacities. Chase (1986) and Benzing
(1990) therefore postulate that the restriction of species to
twigs might reflect tolerances for certain moisture levels.
On young bare twigs, humidity is low, whereas humidity
increases on older branches with thicker layers of substrate
(Winter et al. 1985).
(d) The number of times twig epiphytism
has evolved
Different orchid groups have independently developed a
surprising number of parallel adaptations related to twig
epiphytism throughout the tropics. In South American
orchids, obligate twig epiphytism developed significantly in
the Oncidiinae and Pleurothallidinae (Chase 1986). In
southeast Asia, it developed in some species groups of
Bulbophyllum (Bulbophyllinae) and Taeniophyllum (Aeridinae). In Africa, this habit also occurs in Microcoelia
(Aerangidinae).
1534
B. Gravendeel and others
Epiphytism and pollinator specialization in orchids
(e) Species-rich clades
Epiphytism evolved multiple times in the Orchidaceae.
Rather than a key innovation, we are inclined to define epiphytism as the result of a suite of key innovations among
which are the ability to cope with nutrient-poor and
temporarily very dry conditions. The development of twig
epiphytism seems repeatedly associated with several speciation bursts in Bulbopyllinae. In contrast with general
epiphytic radiations, however, the total number of twig epiphytes in both Bulbophyllinae and Oncidiinae is lower than
the number of trunk epiphytes. This is not surprising as the
larger axes in the canopy have a much higher total surface
area as compared with the twigs ( Johansson 1975). A
hypothetical species–area curve therefore should become
saturated at a higher maximum number of species of trunk
epiphytes as compared with twig epiphytes (figure 8).
6. CONCLUSIONS
Our results show that epiphytic genera are significantly
more species-rich as compared with terrestrial genera, both
for orchids and non-orchids. Species diversity could not be
explained by a high degree of pollinator specialization. A
small body size, short life cycle and highly specialized clinging roots evolved multiple times and independently in
unrelated orchid clades. Furthermore, clades with these
suites of traits seem to have undergone extensive speciation
in one the orchid groups investigated on more than one
occasion. It seems therefore that epiphytism stimulated the
development of a high taxon diversity in the Orchidaceae.
Evidence for an adaptive value of twig epiphytism was not
provided, as phylogenetic techniques are not suitable for
this. Correlations between twig epiphytism and repeated
bursts of speciation in the orchid groups investigated, however, do indicate that several groups of tropical orchids are so
species-rich thanks to, and not despite, their predominantly
arboreal habit. Only future studies that also include other
plant families containing epiphytes, such as bromeliads, can
show whether repeatedly occurring correlations between
twig epiphytism and lineage diversifications as presented
here are a general phenomenon within epiphytic floras.
Bill Baker, Freek Bakker, Mark Chase, John Dransfield,
Marcel Eurlings, Gunter Fischer, Peter Hovenkamp, Peter
Linder, and Niels Raes are gratefully acknowledged for stimulating discussions, literature suggestions, help with the
molecular and statistical analyses, and critically reading drafts
of this manuscript.
REFERENCES
Benzing, D. H. 1987 Vascular epiphytism: taxonomic participation and adaptive diversity. Ann. Miss. Bot. Gard. 74,
183–204.
Benzing, D. H. 1990 Epiphytism: a preliminary overview. In
Vascular epiphytes. General biology and related biota (ed. P. S.
Ashton, S. P. Hubbell, D. H. Janzen, A. G. Marshall, P. H.
Raven & P. B. Tomlinson), pp. 1–42. Cambridge University
Press.
Bloom, A. J., Chapin, F. S. & Mooney, H. A. 1985 Resource
limitation in plants—an economic analogy. A. Rev. Ecol.
Syst. 16, 363–392.
Bradshaw, H. D. & Schemske, D. W. 2003 Allele substitution
at a flower colour locus produces a pollinator shift in
monkeyflowers. Nature 426, 176–178.
Phil. Trans. R. Soc. Lond. B (2004)
Bramwell, D. 2002 How many plant species are there? Pl. Talk
28, 32–33.
Chase, M. W. 1986 A monograph of Leochilus (Orchidaceae).
Syst. Bot. Monogr. 14, 1–97.
Chase, M. W. 1987 Obligate twig epiphytism in the Oncidiinae and other neotropical orchids. Selbyana 10, 24–30.
Chase, M. W. & Palmer, J. D. 1992 Floral morphology and
chromosome number in subtribe Oncidiinae (Orchidaceae): evolutionary insights from a phylogenetic analysis of
chloroplast DNA restriction site variation. In Molecular
systematics of plants (ed. P. S. Soltis, D. E. Soltis & J. D.
Doyle), pp. 324–339. London: Chapman & Hall.
Dafni, A. & Bernhardt, B. 1990 Pollination of terrestrial orchids of southern Australia and the Mediterranean region.
Evol. Biol. 24, 193–252.
Dungeon, W. 1923 Succession of epiphytes in the Quercus
incana forest at Landour, Western Himalayas. J. Ind. Bot.
Soc. 3, 270–272.
Ek, R. C., Ter Steege, H. & Biesmeijer, K. C. 1997 Vertical
distribution and association of vascular epiphytes in four
different forest types in the Guianas. In Botanical diversity in
the tropical rain forest of Guyana (ed. R. C. Ek), pp. 65–89.
PhD thesis, Utrecht University, The Netherlands.
Gentry, A. H. & Dodson, C. H. 1987 Diversity and biogeography of neotropical vascular epiphytes. Ann. Miss. Bot.
Gard. 74, 205–233.
Gravendeel, B. 2004 Coelogyne—matching molecules with
morphology and distribution patterns. In Proc. Eur. Orchid
Conf. and Show, March 2003 (ed. J. Hermans & P. Cribb),
pp. 143–159. London: The British Orchid Council and the
Royal Horticultural Society.
Ibisch, P. L., Boegner, A., Nieder, J. & Barthlott, W. 1996
How diverse are neotropical epiphytes? An analysis based
on the catalogue of the flowering plants and gymnosperms
of Peru. Ecotropica 2, 13–28.
Johansson, D. R. 1974 Ecology of vascular epiphytes in West
African rain forests. Acta Phytogeogr. Suec. 59, 1–136.
Johansson, D. R. 1975 Ecology of epiphytic orchids in West
African rain forests. Am. Orchid Soc. Bull. 44, 125–136.
Kessler, M. 2002 Environmental patterns and ecological correlates of range size among bromeliad communities of
Andean forests in Bolivia. Bot. Rev. 68, 100–127.
Mabberley, D. J. 1998 The plant book. A portable dictionary
of the vascular plants, 2nd edn. Cambridge University Press.
Maddison, W. P. & Maddison, D. R. 2003 Mesquite: a modular system for evolutionary analysis, v. 1.0. Available at
http://mesquiteproject.org.
Nieder, J. 2004 Distribution patterns of epiphytic orchids—
present research, past causes and future consequences.
In Proc. Eur. Orchid Conf. and Show, March 2003 (ed. J.
Hermans & P. Cribb), pp. 241–258. London: The British
Orchid Council and the Royal Horticultural Society.
Ollerton, J. & Cranmer, L. 2002 Latitudinal trends in plant–
pollinator interactions: are tropical plants more specialised?
Oikos 98, 340–350.
Pittendrigh, C. S. 1948 The bromeliad–Anopheles–malaria
complex in Trinidad. I. The bromeliad flora. Evolution 2,
58–89.
Royal Botanic Gardens, Kew 2003 Monocot checklist. Published on the Internet at http://www.rbgkew.org.uk/data/
monocots (accessed 2 February 2004; 9:30 GMT).
Schiestl, F. P. & Ayasse, M. 2002 Do changes in floral odor
cause speciation in sexually deceptive orchids? Pl. Syst. Evol.
234, 111–119.
Schlechter, R. 1912 Die Orchidaceen von Deutsch-NeuGuinea. Rep. Spec. Nov. Regni Veg. Beih. 1, 693–888.
Schlechter, R. 1925 Orchidaceae Perrierianae. Rep. Spec. Nov.
Regni Veg. Beih. 33, 1–391.
Epiphytism and pollinator specialization in orchids
Sokal, R. R. & Rohlf, F. J. 1995 Biometry: the principles and
practice of statistics in biological research, 3rd edn. New York:
Freeman.
Ter Steege, H. & Cornelissen, H. J. 1989 Distribution and
ecology of vascular epiphytes in lowland rain forest of
Guyana. Biotropica 21, 331–339.
Tremblay, R. L. 1992 Trends in the pollination ecology of the
Orchidaceae: evolution and systematics. Can. J. Bot. 70,
642–650.
Van der Cingel, N. A. 1995 An atlas of orchid pollination:
European orchids. Rotterdam, The Netherlands: Balkema.
Van Der Cingel, N. A. 2001 An atlas of orchid pollination: orchids of America, Africa, Asia & Australia. Rotterdam, The
Netherlands: Balkema.
Vermeulen, J. J. 1987 A taxonomic revision of the continental
African Bulbophyllinae. Orchid Monogr. 2, 1–300.
Vasquez, R., Ibisch, P. L. & Gerkmann, B. 2003 Diversity
of Bolivian Orchidaceae—a challenge for taxonomic, floristic and conservation research. Org. Divers. Ecol. 3,
93–102.
Phil. Trans. R. Soc. Lond. B (2004)
B. Gravendeel and others 1535
Wikström, N. & Kenrick, P. 2000 Phylogeny of epiphytic Huperzia (Lycopodiaceae): paleotropical and neotropical clades corroborated by rbcL sequences. Nordic J. Bot. 20, 165–171.
Wikström, N. & Kenrick, P. 2001 Evolution of Lycopodiaceae
(Lycopsida): estimating divergence times from rbcL gene
sequences by use of nonparametric rate smoothing. Mol.
Phylogenet. Evol. 19, 177–186.
Williams, N. H., Chase, M. W., Fulcher, T. & Whitten, M. W.
2001 Molecular systematics of the Oncidiinae based on evidence from four DNA sequence regions: expanded circumscriptions of Cyrtochilum, Erycina, Otoglossum, and
Trichocentrum and a new genus (Orchidaceae). Lindleyana
16, 113–139.
Winter, K., Medina, E., Garcia, V., Mayoral, M. L. & Muniz,
R. 1985 Crassulacean acid metabolism in roots of a leafless
orchid, Campylocentrum tyrridion Garay & Dunsterv. J. Pl.
Physiol. 118, 73–78.
Wolf, J. H. D. & Flamenco, A. 2003 Patterns in species
richness and distribution of vascular epiphytes in
Chiapas, Mexico. J. Biogeogr. 30, 1689–1707.