| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics and Phytogeography |
2Florida Museum of Natural History, University of Florida, Gainesville, Florida 32611-7800 USA; 3Department of Botany, University of Florida, Gainesville, Florida 32611-8526 USA; 4Department of Biochemistry, University of Stellenbosch, Private Bag x1, 7602 Matieland, South Africa
Received for publication September 2, 2005. Accepted for publication February 21, 2006.
ABSTRACT
Members of tribe Vandeae (Orchidaceae) form a large, pantropical clade of horticulturally important epiphytes. Monopodial leafless members of Vandeae have undergone extreme reduction in habit and represent a novel adaptation to the canopy environment in tropical Africa, Asia, and America. To study the evolution of monopodial leaflessness, molecular and structural evidence was used to generate phylogenetic hypotheses for Vandeae. Molecular analyses used sequence data from ITS nrDNA, trnL-F plastid DNA, and matK plastid DNA. Maximum parsimony analyses of these three DNA regions each supported two subtribes within monopodial Vandeae: Aeridinae and a combined Angraecinae + Aerangidinae. Adding structural characters to sequence data resulted in trees with more homoplasy, but gave fewer trees each with more well-supported clades than either data set alone. Two techniques for examining character evolution were compared: (1) mapping vegetative characters onto a molecular topology and (2) tracing vegetative characters onto a combined structural and molecular topology. In both cases, structural synapomorphies supporting monopodial Vandeae were nearly identical. A change in leaf morphology (usually reduced to a nonphotosynthetic scale), monopodial growth habit, and aeration complexes for gas exchange in photosynthetic roots seem to be the most important characters in making the evolutionary transition to leaflessness.
Key Words: ITS • matK • monopodial leaflessness • Orchidaceae • phylogeny • trnL-F • Vandeae
Orchidaceae are one of the largest, most diverse plant families with more than 24000 species (Board of Trustees of the Royal Botanic Gardens Kew, 2004
). With the recent spate of molecular data, evolutionary relationships among orchids are being reestablished and recircumscribed (Chase et al., 2003
). Surprisingly, Vandeae, a large horticulturally important group with almost 2000 species, has been essentially untouched by molecular taxonomists. Traditional classification of taxa within Vandeae is based primarily on floral morphology, especially for leafless taxa with few vegetative characteristics. However, floral characteristics are highly homoplasious within the family and have proven to be treacherous in making hypotheses of phylogenetic relationships (Chase and Palmer, 1989
; Bateman et al., 1997
).
Vandeae were first circumscribed by Lindley during the early 1800s in his treatment of the orchid family (1830–1840). Members of Vandeae sensu Lindley were grouped based on the presence of distinct caudicles of the pollinarium. In addition to Vandeae sensu Dressler (1993; e.g., Vanda and Angraecum), a significant part (40%) of Lindley's group was of tropical American taxa that are now placed in Maxillarieae. However, after describing Vandeae as a new tribe, Lindley admitted his circumscription of Vandeae could probably be even further subdivided into well-defined groupings, given further study.
More than a century later, Dressler and Dodson (1960)
placed subtribe Sarcanthinae within a large tribe Epidendreae. They noted that although this monopodial subtribe may be difficult to distinguish from the tropical American Oncidiinae using floral morphology, they are probably not closely related. However, Garay (1960
, 1972a
) reinstated tribe Vandeae based on the presence of an incumbent anther, porrect rostellum, and well-developed stipes. Vandeae sensu Garay included tropical American subtribes Cyrtopodiinae, Zygopetalinae, and Oncidiinae as well as Vandinae (Garay, 1972a
).
Dressler (1981)
elevated units of the original tribe Epidendreae delimited by Dressler and Dodson (1960)
to form two subfamilies: Epidendroideae and Vandoideae. Although Vandoideae had long been recognized (at varying taxonomic levels) on the basis of floral morphology, there were few consistent characters to delimit Vandoideae from Epidendroideae. Dressler admitted that the only clear character delimiting these two subfamilies was anther development and several years later (1989) reinstated members of Vandoideae into a broadly defined Epidendroideae, much like his original systematic treatment with Dodson (Dressler and Dodson, 1960
). In either case, Vandeae formed a well-defined group of orchids split into three subtribes: Sarcanthinae, Angraecinae, and Aerangidinae.
In Dressler's most recent classification (1993), Vandeae were essentially unchanged from his previous work (1981) with two exceptions: Vandeae were allied with Dendrobieae and Podochileae in a dendrobioid subclade of Epidendroideae, and the illegitimate subtribal name Sarcanthinae (Bentham, 1881
) was replaced with Aeridinae.
Unlike any previous workers, Szlachetko (1995)
provided a much more subdivided classification in which he split orchids into three families and subdivided Orchidaceae sensu stricto (s.s.) into eight subfamilies, including Vandoideae. Szlachetko's system was primarily based on overall similarity in floral morphology, not on shared derived characters. In his organization of Vandeae (based primarily on rostellum morphology), Szlachetko created 10 new subtribes to complement the three existing subtribes at the time (Aeridinae, Angraecinae, and Aerangidinae). Current molecular phylogenetic evidence often conflicts with Szlachetko's extremely subdivided classification system, and it has been essentially disregarded by most taxonomists today (Pridgeon et al., 1999
).
The most comprehensive cladistic morphological study of orchid systematics was carried out by Freudenstein and Rasmussen (1999)
. In their analyses, Vandeae formed a monophyletic tribe with several synapomorphies: isodiametric exodermal cell shape, monopodial growth habit, spherical stegmata, and seeds with laterally compressed walls. Aerangis and Angraecum were united by the reduction of four pollinia to two, and they were sister to a paraphyletic grade of Aeridinae (Acampe and Phalaenopsisv). Their study also indicated a sister relationship with Polystachyinae (Polystachya and Neobenthamia).
Existing molecular evidence for relationships among members of tribe Vandeae is scarce. This tribe has often been referred to as the "black box" of Orchidaceae (R. Dressler, Florida Museum of Natural History, personal communication). Molecular analyses of Vandeae included sequence data from several plastid-encoded DNA regions: Jarrell and Clegg (1995)
using matK; Neyland and Urbatsch (1996)
using ndhF; Cameron et al. (1999)
using rbcL; and Cameron (2001) using atpB and psaB. The only study to find a nonmonophyletic Vandeae sensu Dressler (1993)
was the one performed by Neyland and Urbatsch (1996)
in which Angraecum philippinense Ames (= Amesiella philippinense (Ames) Garay, Aeridinae) was erroneously used as a representative of Angraecinae.
In the most recent classification of Orchidaceae (Chase et al., 2003
), Vandeae form a monophyletic tribe within a large polytomy of advanced epidendroid groups (Cymbidieae and Agrostophyllinae) and include the sympodial subtribe Polystachyinae. The sister relationship of Vandeae sensu Dressler (1993)
and Polystachyinae is well-supported by analyses of Cameron (2001)
as well as van den Berg (2005).
Vandeae sensu Chase et al. (2003)
represent the most complete classification based on a modern phylogenetic approach. However, we chose to focus on the morphologically diagnosable clade of monopodial Vandeae (Vandeae s.s.) for our study, which includes approximately 158 genera with 1968 species of monopodial epiphytes. Distinguishing characteristics of Vandeae s.s. include velamen of the Vanda type (Porembski and Barthlott, 1988
); monopodial growth habit; tegular stipe (Rasmussen, 1986
); pollinia reduced from eight to four or two (Rasmussen, 1986
); pollinia superposed (Rasmussen, 1986
); and Vanda-type seeds (Ziegler, 1981
).
Subtribal relationships
Lower-level phylogenetic studies of Vandeae are even more uncommon than higher-level studies of the tribe. Topik et al. (2005)
studied the subtribe Aeridinae, using African members of Vandeae sensu Dressler (1993)
as outgroups. The horticulturally important genus Phalaenopsis was also examined by several workers from Taiwan (Tsai et al., 2003
), and Carlsward et al. (2003)
performed detailed studies of the New World genusDendrophylax (including Harrisella and Polyradicion).
Within Vandeae s.s., three monopodial subtribes are defined by floral morphology and biogeography. Aeridinae form a monophyletic group (Topik et al., 2005
) and make up the largest subtribe, with 103 genera (1253 species) primarily distributed throughout Asia, Australia, and a few species in Africa. They are distinguished from members of the other two subtribes by having an entire rostellum, a relatively small spur formed by the lip, and four (or two) pollinia. Several genera, such as Acampe and Renanthera, may also possess a prominent column foot.
Aerangidinae contain 36 genera (307 species) from tropical Africa and Madagascar. Members of Aerangidinae typically possess an elongate rostellum, elongate spur, and two pollinia. Their appearance closely resembles that of members of subtribe Angraecinae, which contain 19 genera (408 species). Angraecinae are distributed throughout Madagascar, the Mascarene and Comoros Islands, and tropical Africa, with two genera in tropical America. Angraecinae possess an apron-like rostellum, elongate spur, and two pollinia. Other than the shape of the rostellum, Aerangidinae and Angraecinae are similar in appearance and have collectively been referred to as the "angraecoids."
Leaflessness
Although leaflessness within angiosperms has arisen several times, the type of leaflessness found in monopodial Vandeae is unique to this specialized orchid tribe (Fig. 1). Most leafless vascular plants are typically either succulent and xeric-adapted (e.g., Cactaceae, Euphorbiaceae) or in some way parasitic on other plants or fungi (e.g., Orobanchaceae, Orchidaceae). In almost all cases, the shoot system of these leafless plants is well developed and forms the main body (as it does in leafless Vanilla species in which the stem has taken over the role of photosynthesis). However, the shoot system of leafless monopodial Vandeae is extremely reduced, contributing no net carbon gain to the plant (Benzing and Ott, 1981
). Instead, the roots form the main body of the plant and have assumed the role of food assimilation using a Crassulacean Acid Metabolism (CAM)-like system of recycling CO2 (Cockburn et al., 1985
). The degree of reduction is so extreme that these leafless orchids have even been referred to as "shootless" and "semishootless" by some workers (Benzing and Ott, 1981
; Benzing et al., 1983
). Benzing and Ott (1981)
hypothesized that these extreme epiphytes have reduced their carbon investment in vegetative organs by limiting shoot formation, while increasing their potential fitness by allocating more resources for flower and fruit production.
|
Three genera within Aerangidinae contain leafless species. Microcoelia is an exclusively leafless genus with approximately 26 species found throughout tropical Africa and Madagascar. Chauliodon is a monotypic African genus that resembles Microcoelia (Jonsson, 1979
). Solenangis is mainly a leafy genus with two leafless members: S. aphylla (Thouars) Summerh. and S. cornuta (Rchb.f.) Summerh. Both Microcoelia and Solenangis have leaves reduced to small, nonphotosynthetic scales. The main vegetative difference between these genera is the length of their stems. Species of Microcoelia all possess abbreviated stems (<3 cm long), whereas S. aphylla and S. cornuta possess elongate stems (>4 cm long).
The primarily paleotropical Angraecinae contain two leafless genera restricted to the neotropics (Carlsward et al., 2003
). Dendrophylax is an exclusively leafless genus with 13 species bearing brown scales along an abbreviated stem axis. The larger genus, Campylocentrum Benth. (35 species), is primarily leafy with approximately 12 leafless species, ranging from plants with abbreviated stems bearing nonphotosynthetic scales to one species [C. poeppigii (Rchb.f.) Rolfe] with elongate stems bearing minute, falcate, caducous leaves.
MATERIALS AND METHODS
Molecular techniques
Specimens were obtained from cultivated material, herbarium specimens, and wild-collected plants. Taxon names and authorities of sampled taxa (Appendix) follow Brummitt and Powell (1992)
and the Royal Botanic Gardens, Kew (2004)
World Checklist of Monocotyledons. When possible, the type species for each genus was sampled (Appendix S1, see Supplemental Data with the online version of this article). The internal transcribed spacer (ITS) nrDNA region for representatives of all subtribes was sequenced to obtain an overall hypothesis of relationships within the monopodial Vandeae. Polystachyinae were used as outgroup taxa, based on results of previous analyses (Neyland and Urbatsch, 1996
; Cameron et al., 1999
; Freudenstein and Rasmussen, 1999
; Cameron, 2001
; Chase et al., 2003
; van den Berg et al., 2005
). Two additional plastid regions were sequenced for all available Aerangidinae and Angraecinae and a few representative Aeridinae: matK (part of the exon encoding maturase K) and trnL-F (trnL intron + trnL 3' exon + trnL-trnF intergenic spacer).
DNA from fresh, silica-dried, and herbarium material was extracted using a modified cetyl trimethylammonium bromide (CTAB) technique from Doyle and Doyle (1987)
, scaled down to a 1-mL volume. For DNA extracted from fresh or silica-dried specimens, the entire internal transcribed spacer region (ITS 1 + 5.8S + ITS 2) was amplified using primers of Sun et al. (1994)
designed for Sorghum (17SE and 26SE). Polymerase chain reactions (PCRs) were carried out in 50-µL volumes using an initial denaturation of 98°C for 6 min (for more complete PCR thermocycler protocols, see Appendix S2 in Supplemental Data with the online version of this article). For degraded DNA extracted from herbarium specimens, ITS 1 and ITS 2 were amplified separately using primers of Blattner (1999)
designed for general angiosperms (ITS 1: ITS-A + ITS-C and ITS 2: ITS-D + ITS-B). These PCRs were carried out in a 50-µL volume using a hot start of 80°C and an initial denaturation of 94°C for 2 min.
In some Vandeae, apparent paralogy in ITS was detected as a large deletion (approximately 40 bp long) at the beginning of the ITS 2 region (plus numerous base substitutions). In our study, paralogy was restricted to genera of Angraecinae. These paralogous sequences were checked using the Basic Local Alignment Search Tool (BLAST) from the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov) to eliminate the possibility of fungal contamination; results of the BLAST search indicated all sequences were from vandaceous orchids. Apparent paralogy has also been detected in Aeridinae by other workers (J. Schulman, University of Texas at Austin and A. Kocyan, Ludwig-Maximilians-University Munich, personal communications). The PCR products of Angraecum calceolus were initially cloned to separate orthologous and paralogous DNA regions, but only one copy (presumably a paralogous one) was ever detected. Taxa with putatively paralogous sequences were therefore excluded from all analyses and are indicated with an asterisk in the Appendix.
For DNA extracted from fresh or silica-dried plant material, the plastid-encoded matK and trnL-F regions were amplified using a hot start of 80°C and an initial denaturation at 94°C for 3 min. Primers for part of matK were those of Whitten et al. (2000)
, and primers for trnL-F were those of Taberlet et al. (1991)
. Cycle sequence products were directly sequenced, and all sequences (including paralogous sequences) were submitted to GenBank (see Appendix for accession numbers).
Phylogenetic analyses
The ITS region was sequenced for all three monopodial subtribes of Vandeae. Multiple individuals were sequenced for several species, as material was available. This large ITS matrix was used to examine relationships among all monopodial Vandeae. To more closely examine relationships among the angraecoids, we collected additional sequence data from the plastid regions trnL-F and matK. Therefore, two sets of analyses were performed: (1) ITS analyses of all monopodial Vandeae and (2) three-region analyses (ITS, matK, and trnL-F) of mainly angraecoids and a reduced subset of Aeridinae.
Sequence data for each region were aligned manually within Se-Al version 2.0a11 (Rambout, 1996
) and imported into PAUP* version 4.0b10 (Swofford, 2003
) for phylogenetic analysis. All characters were weighted equally, and indels were coded as missing data. No sequence data were excluded from ITS and matK regions. However, due to difficulties with alignment and multiple repeat regions within the trnL-F matrix, all analyses excluded 356 aligned base pairs from the central region (preceding the trnL exon). Maximum parsimony (MP) analyses were run using a heuristic search strategy of branch swapping by subtree pruning and regrafting (SPR), stepwise addition with 1000 random-addition replicates holding 10 trees at each step, saving multiple trees (MULTREES on). The maximum number of trees saved for all analyses was limited to 20000. The resulting shortest trees from this initial analysis were then swapped to completion. Levels of support were estimated with 1000 bootstrap replicates, using the SPR algorithm of branch swapping for 10 random-addition replicates per bootstrap replicate.
Phylogenetic analyses were run separately for each region (ITS, matK, and trnL-F). Bootstrap trees generated from each region were then manually compared for congruence, as described by Whitten et al. (2000)
. When there were no conflicting, well-supported clades (bootstrap percentage [BP] > 74) between regions, data were combined. Data congruence was also tested using the partition homogeneity test (HTF) in PAUP* version 4.0b10 (Swofford, 2003
) as described by Johnson and Soltis (1998)
. Heuristic searches for the HTF tests were performed using 100 replicates and an SPR algorithm. Five random addition replicates were performed per HTF replicate. Probability values (p) greater than 0.05 were used to identify data sets that were not significantly different from one another and could therefore be combined. Combined analyses were performed with the same heuristic search strategies as described earlier.
Character mapping
To compare consistent data sets and perform combined analyses of Angraecinae and Aerangidinae, taxon sampling from the anatomical data set of Carlsward et al. (in press) and from our molecular data set were pruned so that each contained the same species (Appendix S3, see Supplemental Data with the online version of this article). However, corresponding species from each data set were not necessarily from the same specimen, and anatomical data from several specimens of the same species were often combined. A total of 112 ingroup (monopodial Vandeae) and four outgroup species (Polystachyinae) were used in all analyses.
Twenty-four structural characters (Appendix S4, see Supplemental Data with the online version of this article) from Carlsward et al. (in press) were used to make a character matrix for this subset of Vandeae (Appendix S5, see Supplemental Data with the online version of article). For combined structural and molecular analyses, sequence data from ITS, matK, and trnL-F were used. All MP analyses were run with the same search strategies described.
Evolution of structural characters was examined in two ways using MacClade version 4.06 for Mac OS X (Maddison and Maddison, 2000
): (1) molecular data were used to create a topology onto which anatomical characters were mapped and (2) anatomical characters were traced onto a combined structural and molecular topology.
RESULTS
In all of the analyses, monopodial Vandeae formed a strongly supported clade (>90 BP). Within this clade, Aeridinae also formed a well-supported clade with >90 BP in all analyses.
ITS analyses of Vandeae
The large ITS matrix of all Vandeae included 207 individuals representing 193 species. Of the 777 aligned sequence positions, 394 were potentially parsimony informative. Heuristic analysis produced 20000+ trees (Figs. 2, 3) with length (L) = 3205 steps, consistency index (CI1) = 0.29, consistency index excluding uninformative characters (CI2) = 0.26, retention index (RI) = 0.76, and rescaled consistency index (RC) = 0.22.
|
|
ITS and plastid analyses of Angraecinae and Aerangidinae
The ITS submatrix included 116 ingroup species. Most genera of Angraecinae possessed paralogous ITS sequences and were therefore excluded from phylogenetic analyses. Results of the Angraecinae + Aerangidinae ITS submatrix (Fig. 4A, Table 1) were similar to those of the larger Vandeae matrix (Figs. 2, 3).
|
|
The trnL-F submatrix included 138 ingroup species (Ypsilopus viridiflorus was excluded due to amplification difficulties). The data from the trnL-F region gave similar results to those of the matK region; the primary difference was that trnL-F generally gave more resolution among taxa (for the trnL-F bootstrap tree see Appendix S7 in Supplemental Data with the online article).
The combined plastid submatrix included 139 ingroup species. There were no conflicting, well-supported clades between the matK and trnL-F topologies, which was supported by the partition homogeneity test (p = 0.18). The topologies of the combined plastid trees (Fig. 4B) were similar to both the matK and trnL-F topologies, and in most cases were more resolved with better support than either region separately (Table 1).
There were only three minor instances in which the ITS data conflicted with the plastid data (indicated with arrowheads in Fig. 4). To eliminate the possibility of taxon sampling incongruence, reduced plastid analyses were performed in which taxa with paralogous ITS sequences were removed from the matrix. Topologies and BP for the conflicting taxon groupings were the same in the reduced and complete plastid analyses.
Probability values from the partition homogeneity tests between trnL-F/ITS (p = 0.01) and matK/ITS (p = 0.01) had significant heterogeneity between both plastid and ITS data sets. However, this statistical test has been shown to be overly sensitive (Graham et al., 1998
; Reeves et al., 2001
), and given the infrequency of topology conflicts between the plastid and ITS data sets and the minor position of these topologies within the overall tree, this incongruence is more likely due to technical issues such as taxon sampling or phylogenetic signal and homoplasy (Wendel and Doyle, 1998
) than they are to real genealogy differences. Therefore, the data sets were combined to give a more robust phylogenetic tree of all Aerangidinae and Angraecinae than would otherwise be attainable with individual data sets (Soltis et al., 1998
).
The combined three-region matrix included 139 ingroup species. Topologies of the resulting trees were more similar to the plastid trees than the ITS trees (Figs. 4, 5). This combined three-region matrix resulted in fewer trees with better resolution and support than any of the three data sets alone (Table 1).
|
|
Mapping the 24 structural characters onto a representative molecular tree (showing all most parsimonious states at each node) revealed few patterns of anatomical character evolution because of the extreme homoplasy in the structural data set (for mapping of each character see Appendix S9 in Supplemental Data with the online version of this article). The synapomorphies of Vandeae s.s. supported by character mapping (Appendix S10, see Supplemental Data with the online version of this article) were character 2, loss of tilosomes; character 6, presence of aeration complexes; character 14, loss of a hypodermis; character 18, presence of stegmata; character 21, loss of mucilage; character 22, presence of a monopodial stem.
Structural characters traced on combined topologies
Bootstrap consensus topologies of the structural and molecular data sets were manually compared for homogeneity assessment. Because there were no hard incongruencies among the data sets, the structural data were combined with the molecular data for phylogenetic analysis. Heuristic searches produced 690 most parsimonious trees with only slightly lower CI and RI values than molecular data alone (Table 2). However, there were more clades with BP > 85 and many fewer trees (Table 2) when the combined data set was used than when molecular data alone was used.
Results of tracing 24 anatomical and morphological characters along one of the 690 most parsimonious trees of the combined analysis were almost identical to that of mapping the same structural characters onto the molecular tree (Fig. 6 and Appendix S10, see Supplemental Data with the online version of this article). The synapomorphies supporting Vandeae s.s. were almost the same in both instances, with the exception of hypodermal composition (character 15). However, when ACCTRAN optimization was used to map hypodermal composition onto a molecular topology, it also formed a synapomorphy of Vandeae s.s. Most differences in the patterns of character evolution between mapping and tracing were due to the slightly different topologies of the combined and molecular trees.
|
Taxonomy of Vandeae
The two African and Malagasy subtribes of monopodial Vandeae, Angraecinae and Aerangidinae, were originally circumscribed by Summerhayes (1966)
based on rostellum shape and chromosome number. These differences, however, seem to be phylogenetically misleading. Chromosome counts from several sources (Charard, 1963
; Jones, 1967
; Arends et al., 1980
; Jonsson, 1981
; Arends and Van der Laan, 1983
, 1986
) support a general trend of x = 25 for Aerangidinae and x = 19 for Angraecinae, as indicated by Summerhayes (1966)
, but within each subtribe there is tremendous variation. This chromosomal variation is most prominent in large genera such as Angraecum (x = 19, 20, 21, 23, 24, 25) and Aerangis (x = 21, 23, 25, 27).
Morphologically, members of these two subtribes have similar vegetative and floral features and are often referred to collectively as "angraecoids." Angraecum distichum, which is sister to the Angraecopsis amaniensis/Diaphananthe millarii clade (traditionally Aerangidinae) in the combined three-region analysis of our study, has a chromosome number of x = 25. Several other African species of Angraecum not included in our study (A. bancoense Burg, A. podochiloides Schltr., and A. subulatum Lindl.) also have been reported as having x = 25 (Arends et al., 1980
). The chromosome number common to many Aerangidinae is x = 25, suggesting affinities between Aerangidinae (which is primarily African) and the African species of Angraecum.
Individually, Aerangidinae and Angraecinae are polyphyletic, but together they form a well-supported monophyletic group in all molecular analyses. Therefore, we propose recognizing a broadly circumscribed subtribe Angraecinae (Summerh., Kew Bulletin 20: 188, 1966) that includes Aerangidinae (Summerh., Kew Bulletin 20: 188, 1966).
Unlike Angraecinae or Aerangidinae, Aeridinae are a well-supported subtribe in all analyses with a consistent chromosome number of x = 19 (Woodard, 1951
; Storey, 1952
; Sagawa, 1962
; Charard, 1963
; Shindo and Kamemoto, 1963
; Storey et al., 1963
; Kamemoto, 1965
; Jones, 1967
; Tara and Kamemoto, 1970
; Arends and Van der Laan, 1983
). Therefore, our current circumscription of monopodial Vandeae consists of only two subtribes: (1) the primarily Asian Aeridinae and (2) the African, Malagasy, and American Angraecinae.
For the majority of Aeridinae in our study, only data from the ITS region were gathered, and few species were collected for each genus. Therefore, many of our taxonomic conclusions are speculative. Based on our sampling, Acampe, Chiloschista, Pelatantheria, Phalaenopsis, Taeniophyllum (including Microtatorchis), and Trichoglottisv are monophyletic (Fig. 3). Cleisostoma, Robiquetia, and Vanda (as currently circumscribed) are not monophyletic. In the strict consensus, Chiloschista and Taeniophyllum (both leafless) are part of one large clade. Taeniophyllum is sister to the leafy genus Sarcochilus, and Chiloschista is sister to a leafy clade of Amesiella, Dyakia, and Tuberolabium. This would suggest that leaflessness probably evolved at least three times within Aeridinae (once in each of the leafy ancestors of Taeniophyllum, Chiloschista, and Phalaenopsis).
Within Phalaenopsis (including Doritis, Kingidium, and Kingiella), two main clades exist (subgeneric classification following Christenson, 2001
): (1) subgenus Phalaenopsis including sections Aphyllae, Deliciosae, Esmeralda, Parishianae, and Proboscidioides and (2) subgenus Phalaenopsis including sections Phalaenopsis and Stauroglottis as well as subgenus Polychilos including sections Amboinensis, Polychilos, and Zebrinae. Section Aphyllae is monophyletic (72 BP) if P. lowii (section Proboscidioides) is included. Phalaenopsis lowii is morphologically similar to section Aphyllae (Christenson, 2001
), which would support its inclusion within section Aphyllae. The monophyly of section Deliciosae is questionable because P. deliciosa is more closely related to section Aphyllae (65 BP) than to P. chibae. Subgenus Polychilos is weakly supported as monophyletic (78 BP), excluding P. fuscata, which is unresolved in a clade with P. deliciosa/P. pulcherrima and P. amabilis/P. cornucervi.
Gussonea was originally circumscribed by Richard (1828)
for Angraecum aphyllum Thouars. The genus was adopted by Schlechter (1918)
and Perrier de la Bâthie (1941)
to include all leafless African and Malagasy angraecoids. Schlechter (1918)
divided Gussonea into subgenus Eugussonea with elongate stems and loosely spaced roots (i.e., Solenangis aphylla (Thouars) Summerh.) and subgenus Taeniophylloides with shortened stems and densely tufted roots (i.e., Microcoelia spp.). As noted by Summerhayes (1943)
, Gussonea A.Rich. is actually a later homonym of the Euphorbiaceae genus Gussonia (now a synonym of Sebastiana) previously described by Sprengel. In his treatment of the leafless angraecoids, Summerhayes (1943)
split Gussonea into two genera, which are comparable to Schlechter's subgenera (1918), based on floral morphology and overall habit: the long-stemmed Solenangis and the short-stemmed Microcoelia. Currently, Solenangis includes two leafless species [S. aphylla and S. cornuta (Rchb.f.) Summerh.], one intermediate species with reduced leaves [S. conica (Schltr.) L. Jonss.], and three species with well-developed leaves [S. clavata (Rolfe) Schltr., S. scandens (Schltr.) Schltr., and S. wakefieldii (Rolfe) P.J. Cribb & J. Stewart]. Based on sequence data (Fig. 5), the leafless S. aphylla and S. cornuta are more closely related to other species of Microcoelia than to the leafy species of Solenangis (S. clavata and S. wakefieldii). In the strict consensus of the combined analyses, S. clavata + S. wakefieldii are sister to Microcoelia + leafless Solenangis (although there is no support for this relationship). Solenangis aphylla was originally transferred from Gussonea into the genus Microcoelia by Summerhayes (1936)
, and the results of our molecular phylogenetic analyses support the transfer of S. aphylla back to M. aphylla (Thouars) Summerh.
Microcoelia aphylla (Thouars) Summerh., Bulletin of Miscellaneous Information, Royal Botanic Gardens, Kew: 233, 1936.
Basionym
Angraecum aphyllum Thouars, Histoire Particulière des Plantes Orchidées: 72, 1822.
Synonyms
Gussonea aphylla A. Rich., Mémoires de la Société d'Histoire Naturelle de Paris 4: 68, 1828. Saccolabium aphyllum Lindl., Genera and Species of Orchidaceous Plants: 223, 1833. Mystacidium aphyllum T. Durand & Schinz, Conspectus Florae Africae 5: 51, 1894. Rhaphidorhynchus aphyllus Finet, Mémoires de la Société Botanique de France, Paris 9: 35, 1907. Angraecum defoliatum Schltr., Annales du Musee d'Histoire Naturelle de Marseille 1: 191, 1913. Gussonea defoliata Schltr., Beihefte zum Botanischen Centralblatt. Kassel 33: 425, 1915. Solenangis aphylla (Thouars) Summerh., Botanical Museum Leaflets 11: 159, 1943.
Solenangis cornuta was a new combination by Summerhayes (1942)
, transferred from Gussonea cornuta Ridl. The results of our molecular phylogenetic analyses support the transfer of S. cornuta to M. cornuta (Fig. 5). The type species of Solenangis, S. scandens, possesses well-developed photosynthetic leaves. Further sampling of S. scandens (the type species) and S. conica (an intermediate species with small, deciduous, conical leaves) would be needed before conclusions about the monophyly of the leafy species of Solenangis can be made.
Microcoelia cornuta (Ridl.) Carlsward, comb. nov.
Basionym
Gussonea cornuta Ridl., Journal of Botany, London 23: 310, 1885.
Synonyms
Angraecum cornutum Rchb.f, Flora 68: 538, 1885. Angraecum cyclochilum Schltr., Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie 38: 160, 1906. Rhaphidorhynchus cornutus Finet, Mémoires de la Société Botanique de France, Paris 9: 34, 1907. Solenangis cornuta (Ridl.) Summerh., Botanical Museum Leaflets 11: 160.
Two other African leafless genera that have been recognized as segregates of Microcoelia are Chauliodon Summerh. and Encheiridion Summerh. Summerhayes transferred Microcoelia macrorrhynchia (Schltr.) Summerh. to Encheiridion macrorrhynchium (Schltr.) Summerh. based on the trilobed lip and horizontally elongate rostellum (Summerhayes, 1943
). Jonsson (1981)
in his revision of Microcoelia found several intermediate species [namely M. leptostele (Summerh.) L. Jonss., M. nyungwensis L. Jonss., and M. sanfordii L. Jonss.] that formed a continuum in lip morphology with Encheiridion macrorrhynchium. Given this floral evidence, Jonsson logically transferred Encheiridion back into Microcoelia (Jonsson, 1981
).
In the same publication on leafless angraecoids in which Summerhayes (1943)
introduced Encheiridion, he also created a new genus and species, Chauliodon buntingii Summerh., based on callus morphology in the lip. Summerhayes mentioned the resemblance of this species to Microcoelia, especially in vegetative habit and pollinarium structure. During his monographic work on Microcoelia, Jonsson (1981)
transferred Summerhayes' Chauliodon buntingii (Summerhayes, 1943
) to Chauliodon deflexicalcaratum (DeWild.) L. Jonss. Chauliodon deflexicalcaratum was originally described by De Wildeman (1916)
as Angraecum deflexicalcaratum and was later transferred to the short-stemmed section of Gussonea by Schlechter (1918)
. Summerhayes (1943)
transferred Gussonea deflexicalcarata (DeWild.) Schltr. to Microcoelia deflexicalcarata (DeWild.) Summerh. in the same publication in which he described the new genus Chauliodon. Jonsson (1979)
attributed Summerhayes' confusion to not having access to De Wildeman's original material of Angraecum deflexicalcaratum. In examining type material for both M. deflexicalcarata and Chauliodon buntingii, Jonsson felt that these species were conspecific and should be lumped together under the older name of C. deflexicalcarata. Although Chauliodon and Encheiridion were not sampled in our study, their vegetative and floral similarity to Microcoelia is a strong indication that they should be included in Microcoelia.
The only remaining leafless Old World member of Aerangidinae is Taeniorrhiza Summerh., a monotypic genus native to Africa. In his description of the genus, Summerhayes (1942)
separated Taeniorrhiza from the remaining leafless Aerangidinae by its flattened roots, single-flowered inflorescences, winged column, and Eulophia-like lip. Unfortunately, Taeniorrhiza was not sampled in our study, and because it is so morphologically distinct, conclusions of its phylogenetic position cannot be made.
Rhipidoglossum has traditionally been segregated from Diaphananthe by two key features: a column foot and a distinct conical projection or "tooth-like" callus at the opening of the spur (Summerhayes, 1960
; Garay, 1972b
; Senghas, 1986
). However, Summerhayes (1960)
and Cribb (1989) observed a complete continuum in these floral character states within Diaphananthe and Rhipidoglossum. Sequence data support the inclusion of Rhipidoglossum with Cribbia and Diaphananthe millarii, whereas most species of Diaphananthe in our study are most closely related to Chamaeangis (Fig. 5). It is clear from these preliminary analyses that Rhipidoglossum and Diaphananthe do not form a monophyletic group. However, given the relatively small percentage of Diaphananthe and Rhipidoglossum species sampled in our study (Appendix S1, see Supplemental Data with the online version of this article), no nomenclatural changes are made here.
In all molecular analyses, species of Aerangis form a monophyletic group with the inclusion of Microterangis hariotiana (the type species) and M. hildebrandtii. Before Senghas erected the genus Microterangis (1985), most of its members were included in Chamaeangis section Microterangis (Schlechter, 1918
). Morphologically, flowers of Microterangis look more similar to Chamaeangis than to Aerangis, but molecular data clearly indicate this to be convergence. The type species (M. hariotiana) and M. hildebrandtii should both be transferred to Aerangis, but doing this would leave the five species not sampled for our study in a state of nomenclatural limbo (with no genus name). The remaining five species of Microterangis may form part of the large Aerangis clade (along with M. hariotiana and M. hildebrandtii); they may form a clade that will need a new generic name; or they may be closely related to other genera of Angraecinae. We feel the genus needs further study before name changes can be made.
Paralogous ITS sequences were never found in New World Angraecinae or their African Angraecum relatives, but paralogy was common in the Malagasy Angraecinae (especially Aeranthes, Angraecum, Oeoniella, Sobennikoffia, Jumellea, and Lemurorchis). Cryptopus, Oeonia, Neobathiea, and Lemurella (also Malagasy Angraecinae containing orthologous ITS sequences) are most closely related to Beclardia (Malagasy Aerangidinae), and this group is embedded in a large, poorly resolved angraecoid clade (Fig. 5). The topology of the combined tree suggests that duplication events creating paralogy probably originated in Madagascar.
Angraecum is clearly polyphyletic in all analyses. With over 200 species (only 15 of which were included in our study) and distribution in America (Campylocentrum and Dendrophylax) as well as Madagascar and Africa (Aeranthes, Jumellea, Lemurorchis, and Aerangidinae), its polyphyletic nature is not surprising. Bonniera, Oeoniella, and Sobennikoffia are embedded within a clade composed primarily of Angraecum (Fig. 5). Jumellea and Aeranthes each seem to be monophyletic. However, without more extensive taxon sampling, taxonomic decisions about most traditional Angraecinae are premature.
The clade of Cyrtorchis, Listrostachys, Podangis, Rangaeris, Tridactyle, and Ypsilopus (Fig. 5) show that generic circumscriptions for many of these genera may need to be reevaluated with denser taxon sampling. Cyrtorchis is clearly monophyletic, whereas Listrostachys and Podangis may also be distinct genera (only one of the two species of Listrostachys was sampled and Podangis is monotypic). Some species of Rangaeris are more closely related to other genera (Podangis and Tridactyle) than they are to one another, but no clear pattern of relationships was found. The two species of Ypsilopus sampled are more closely related to Tridactyle than they are to one another, so Ypsilopus should probably be transferred to Tridactyle. However, more taxon sampling within Tridactyle (only six of the 43 species were sampled) would be needed to make any taxonomic conclusions about the generic concept of Tridactyle.
Mystacidium forms a monophyletic group in the combined bootstrap analysis, but it is embedded in a larger clade that includes several species of Angraecopsis and one species of Sphyrarhynchus (Fig. 5). However, more extensive taxon sampling of Mystacidium and Angraecopsis is needed for generic recircumscription.
Although there are problems with the monophyly of several of the genera mentioned earlier, many angraecoid genera are monophyletic in the combined bootstrap consensus (Fig. 5). The New World Campylocentrum and Dendrophylax, as redefined by Carlsward et al. (2003)
, form well-supported clades. Old World angraecoid genera that also form monophyletic groups are Cryptopus, Aeranthes, Jumellea, Eurychone, Ancistrorhynchus, Bolusiella, and Cyrtorchis. Oeonia, Neobathiea, Lemurella, Beclardia, and Lemurorchis are all small genera with only one species sampled from each, and they are likely monophyletic as well.
Evolution of leaflessness
Sequence data derived from the ITS region for all monopodial subtribes (Aeridinae, Aerangidinae, and Angraecinae) supported a monophyletic Vandeae s.s. (Figs. 2, 3). With this overall tree, it appears that leaflessness has arisen six to seven times within Vandeae, depending on the occurrence of reversals to a leafy condition: three to four times in Asia with Taeniophyllum, Chiloschista, and Phalaenopsis; once in Africa and Madagascar with Microcoelia (including leafless Solenangis); and at least twice in the New World with Dendrophylax and Campylocentrum. From examining the pattern of variation in anatomical and morphological characters among leafy and leafless Vandeae, a monopodial growth habit, a reduction to nonphotosynthetic leaves, and the presence of aeration complexes in photosynthetic roots (the cortical stomatal complexes of Carlsward et al., in press) may be important structural precursors to the leafless habit. This extremely reduced epiphytic habit probably represents an adaptation for carbon conservation (Benzing and Ott, 1981
) and/or water conservation in a canopy environment.
Our molecular phylogenetic work not only represents a foundation for examining evolutionary questions within a horticulturally important plant group, it also represents an integrated approach to answering a general question of parallelism that has occurred systematically and geographically in different lineages.
Taxon; Voucher (location); Source; ITS, trnL-F, matK.
Tribe Vandeae, Subtribe Aerangidinae
Aerangis biloba (Lindl.) Schltr.; Whitten 1938 (FLAS); FLMNH hort.; DQ091594; DQ091455; DQ091331. A. citrata (Thouars) Schltr.; Whitten 1788 (FLAS); Countryside Orchids; DQ091600; DQ091461; DQ091337. A. confusa J.Stewart; Bytebier s.n. (EA); PCP hort.; DQ091595; DQ091456; DQ091332. A. coriacea Summerh.; No voucher; Hoosier Orchid Co.; DQ091610; —; —. A. coriacea; Bytebier 562 (EA); PCP hort.; DQ091609; DQ091469; DQ091345. A. ellisii var. grandiflora J.Stewart; Carlsward 401 (FLAS); Hoosier Orchid Co.; DQ091602; DQ091463; DQ091339. A. fastuosa (Rchb.f.) Schltr.; Carlsward 402 (FLAS); Countryside Orchids; DQ091604; DQ091464; DQ091340. A. kirkii (Rchb.f.) Schltr.; Bytebier 637 (EA); PCP hort.; DQ091596; DQ091457; DQ091333. A. kotschyana (Rchb.f.) Schltr.; Bytebier 671 (EA); PCP hort.; DQ091598; DQ091459; DQ091335. A. luteoalba var. rhodosticta (Kraenzl.) J.Stewart; Bytebier 691 (EA); PCP hort.; DQ091611; DQ091471; DQ091347. A. macrocentra (Schltr.) Schltr.; Kew 779 (K); J. Hermans hort.; DQ091601; DQ091462; DQ091338. A. modesta (Hook.f.) Schltr.; Carlsward 242 (FLAS); ABG hort.; DQ091603; DQ091465; DQ091341. A. punctata J.Stewart; Carlsward 328 (FLAS); Hoosier Orchid Co.; DQ091605; DQ091466; DQ091342. A. somalensis (Schltr.) Schltr.; Bytebier 1549 (EA); PCP hort.; DQ091612; DQ091470; DQ091346. A. thomsonii (Rolfe) Schltr.; Kirika 968 (EA); PCP hort.; DQ091599; DQ091460; DQ091336. A. ugandensis Summerh.; Bytebier 681 (EA); PCP hort.; DQ091597; DQ091458; DQ091334. A. verdickii (DeWild.) Schltr.; No voucher; Countryside Orchids; DQ091613; DQ091472; DQ091348.
Ancistrorhynchus capitatus (Lindl.) Summerh.; Carlsward 276 (FLAS); Uzumara Orchids; DQ091643; DQ091475; DQ091351. A. metteniae (Kraenzl.) Summerh.; Carlsward 398 (FLAS); Uzumara Orchids; DQ091644; DQ091476; DQ091352. A. recurvus Finet; No voucher, ver. La Croix; Uzumara Orchids; DQ091646; DQ091478; DQ091354. A. straussii (Schltr.) Schltr.; Carlsward 285 (FLAS); Uzumara Orchids; DQ091645; DQ091477; DQ091353.
Angraecopsis amaniensis Summerh.; No voucher, ver. La Croix; Uzumara Orchids; DQ091568; DQ091482; DQ091357. A. breviloba Summerh.; Bytebier 307 (EA); PCP hort.; DQ091569; DQ091483; DQ091358. A. parviflora (Thouars) Schltr.; Carlsward 291 (FLAS); Uzumara Orchids; DQ091575; DQ091489; DQ091364. A. parviflora; Kew 4363 (K); J. Hermans hort.; DQ091576; —; —.
Beclardia macrostachya (Thouars) A.Rich.; Carlsward 289 (FLAS); Uzumara Orchids; DQ091592; DQ091497; DQ091372. B. macrostachya; Kew 3536 (K); J. Hermans hort.; DQ091593; —; —.
Bolusiella batesii (Rolfe) Schltr.; Carlsward 152 (FLAS), Nkongmeneck 2087 (SEL); SEL hort. (1997–0173A); DQ091663; DQ091479; —. B. iridifolia (Rolfe) Schltr.; Bytebier 1113 (EA); PCP hort.; DQ091665; DQ091481; DQ091356. B. maudiae (Bolus) Schltr.; Bytebier 485 (EA); PCP hort.; DQ091664; DQ091480; DQ091355.
Chamaeangis ichneumonea (Lindl.) Schltr.; La Croix 1101 (?), Carlsward 286 (FLAS); Uzumara Orchids; DQ091616; DQ091498; DQ091373. C.odoratissima (Rchb.f.) Schltr.; No voucher, ver. La Croix; Uzumara Orchids; DQ091617; DQ091499; DQ091374. C. sarcophylla Schltr.; Bytebier 339 (EA); PCP hort.; DQ091621; DQ091503; DQ091378. C.vesicata (Lindl.) Schltr.; Bytebier 796 (EA); PCP hort.; DQ091622; DQ091504; DQ091379. C.vesicata; Kew 399 (K); J. Hermans hort.; DQ091623; —; —.
Cribbia brachyceras (Summerh.) Senghas; Bytebier 361 (EA); PCP hort.; DQ091577; DQ091490; DQ091365. C.confusa P.J.Cribb; Kew 3936 (K); J. Hermans hort.; DQ091578; DQ091491; DQ091366.
Cyrtorchis arcuata (Lindl.) Schltr.; Bytebier 676 (EA); PCP hort.; DQ091624; DQ091505; DQ091380. C.chailluana (Hook.f.) Schltr.; Carlsward 156 (SEL); SEL hort. (1996–0294); DQ091625; DQ091506; DQ091381. C. praetermissa Summerh.; No voucher; Countryside Orchids; DQ091627; DQ091507; DQ091382. C.ringens (Rchb.f.) Summerh.; Carlsward 226 (FLAS), Nkongmeneck 1388 (SEL); SEL hort. (1997–0154A); DQ091626; DQ091508; DQ091383.
Diaphananthe fragrantissima (Rchb.f.) Schltr.; Kirika 536 (EA); PCP hort.; DQ091618; DQ091500; DQ091375. D. lorifolia Summerh.; Bytebier 346 (EA); PCP hort.; DQ091619; DQ091501; DQ091376. D. millarii (Bolus) H.P. Linder; Carlsward 346 (FLAS); Countryside Orchids; DQ091579; DQ091493; DQ091368. D. pellucida (Lindl.) Schltr.; Carlsward 241 (FLAS); ABG hort.; DQ091620; DQ091502; DQ091377.
Eurychone galeandrae (Rchb.f.) Schltr.; Carlsward 293 (FLAS); Uzumara Orchids; DQ091614; DQ091473; DQ091349. E. rothschildiana (O'Brien) Schltr.; Carlsward 407 (FLAS); Sunset Orchids; DQ091615; DQ091474; DQ091350.
Listrostachys pertusa (Lindl.) Rchb.f.; Carlsward 399 (FLAS); Uzumara Orchids; DQ091637; DQ091509; DQ091384.
Microcoelia bulbocalcarata L.Jonss.; No voucher, ver. La Croix; Uzumara Orchids; DQ091647; DQ091521; DQ091396. M. corallina Summerh.; No voucher, ver. La Croix; Uzumara Orchids; DQ091653; DQ091527; DQ091402. M. exilis Lindl.; Whitten 1937 (FLAS); Sunset Orchids; DQ091658; DQ091531; DQ091406. M. exilis; Bytebier 1255 (EA); PCP hort.; DQ091659; —; —. M. gilpinae (Rchb.f. & S. Moore) Summerh.; Carlsward 290 (FLAS); Uzumara Orchids; DQ091649; DQ091522; DQ091397. M. globulosa (Hochst.) L.Jonss.; Carlsward 259 (FLAS); Uzumara Orchids; DQ091655; DQ091529; DQ091404. M. globulosa; PCP 488 (EA); PCP hort.; DQ091656; —; —. M. macrantha (H.Perrier) Summerh.; Kew 5391 (K); J. Hermans hort.; DQ091650; DQ091524; DQ091399. M. megalorrhiza (Rchb.f.) Summerh.; Bytebier 1250 (EA); PCP hort.; DQ091648; DQ091523; DQ091398. M. obovata Summerh.; Bytebier 1256 (EA); PCP hort.; DQ091654; DQ091528; DQ091403. M. physophora (Rchb.f.) Summerh.; Bytebier 629 (EA); PCP hort.; DQ091661; DQ091533; DQ091408. M. smithii (Rolfe) Summerh.; Bytebier 1248 (EA); PCP hort.; DQ091660; DQ091532; DQ091407. M. stolzii (Schltr.) Summerh.; Carlsward 287 (FLAS); Uzumara Orchids; DQ091657; DQ091530; DQ091405.
Microterangis hariotiana (Kraenzl.) Senghas; Carlsward 227 (FLAS); Andy's Orchids; DQ091606; DQ091467; DQ091343. M.s hariotiana; Carlsward 292 (FLAS); Uzumara Orchids; DQ091607; —; —. M. hildebrandtii (Rchb.f.) Senghas; Kew 2616 (K); J. Hermans hort.; DQ091608; DQ091468; DQ091344.
Mystacidium aliceae Bolus; Whitten 1787 (FLAS); Countryside Orchids; DQ091571; DQ091485; DQ091360. M. braybonae Summerh.; Carlsward 179 (FLAS); Andy's Orchids; DQ091572; DQ091486; DQ091361. Mystacidium capense (L.f.) Schltr.; Whitten 1781 (FLAS); Countryside Orchids; DQ091573; DQ091487; DQ091362. M. flanaganii (Bolus) Bolus; Kew 5084 (K); J. Hermans hort.; DQ091574; DQ091488; DQ091363.
Podangis dactyloceras (Rchb.f.) Schltr.; Kew 4999 (K); J. Hermans hort.; DQ091628; DQ091510; DQ091385.
Rangaeris amaniensis (Kraenzl.) Summerh.; Bytebier & Kirika 26 (EA); PCP hort.; DQ091634; DQ091512; DQ091386. R. muscicola (Rchb.f.) Summerh.; Carlsward 169 (SEL); SEL hort. (1997–0177A); DQ091630; DQ091513; DQ091387. R. muscicola; Carlsward 400 (FLAS); Countryside Orchids; DQ091631; —; —. R. rhipsalisocia (Rchb.f.) Summerh.; Carlsward 235 (SEL); S
