| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
2 Florida Museum of Natural History, University of Florida, Gainesville, Florida 32611-7800 USA; and 3 Royal Botanic Gardens, Kew, Richmond, Surrey, TW9 3DS, UK
Received for publication September 30, 1999. Accepted for publication March 2, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
35% of currently recognized genera) and four outgroup taxa produced resolved and highly supported cladograms. Based on the cladograms, we recognize six subtribes: Eriopsidinae, Oncidiinae (including Pachyphyllinae, Ornithocephalinae, and Telipogoninae), Stanhopeinae, Coeliopsidinae, Maxillariinae (including Lycastinae and Bifrenariinae), and Zygopetalinae (including Cryptarrheninae, Dichaeinae, Huntleyinae, and Warreinae). Stanhopeinae were sampled most intensively; their generic relationships were highly resolved in the analysis and largely agree with currently accepted generic concepts based on morphology. Coeliopsidinae (Coeliopsis, Lycomormium, Peristeria) are sister to Stanhopeinae. Correlations are drawn among phylogeny, pollination mechanisms, and life history traits.
Key Words: Coeliopsidinae • ITS • matK • Maxillarieae • Maxillariinae • Orchidaceae • Stanhopeinae • trnL-F
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
consist of 2600 species in 165 genera (10% of Orchidaceae) and contain many of the showy epiphytic orchids of the Neotropics including horticulturally important genera such as Lycaste Lindl., Oncidium Sw., Odontoglossum Kunth, Stanhopea Frost ex Hook., and Zygopetalum Lindl. The tribe includes most Neotropical orchids possessing a complex pollinarium with a viscidium and stipe, pseudobulbs usually of a single internode, and the Maxillaria seed type (Barthlott and Ziegler, 1981
; Chase and Pippin, 1987
). They have highly diverse vegetative habits, floral morphology, and pollination mechanisms. Within Maxillarieae, subtribe Stanhopeinae contain some of the most exotic pollination mechanisms found in orchids (Darwin, 1877
; van der Pijl and Dodson, 1966
). They are pollinated exclusively by male euglossine bees collecting fragrant chemicals from the flowers, and their complex floral shapes manipulate the bees to place pollinaria on precise locations on the bee's body. We now know a considerable amount about the pollination biology of Stanhopeinae from the fieldwork of Dodson, Dressler, and others, but understanding the evolution of pollination-related traits requires a reliable phylogeny.
In this study, we present analyses of Maxillarieae based on combined analyses of nuclear and plastid regions. Early in the course of our analyses of Stanhopeinae, we concluded that the choice of outgroups was problematic because an explicitly phylogenetic classification of the tribe Maxillarieae was lacking. Consequently, we broadened our analysis to include representatives of all major groups within the tribe confirmed by the broad rbcL analyses of Cameron et al. (1999)
. This paper examines subtribal circumscription within Maxillarieae with emphasis on generic relationships within Stanhopeinae. Subsequent papers will focus on the individual subtribes and character evolution in greater detail.
Few modern workers have produced classifications of Maxillarieae that attempt to recognize phylogenetic relationships. Those of Schlechter (1926), Dressler and Dodson (1960)
, and Garay (1972)
were clearly artificial. Dressler (1981)
originally placed Stanhopeinae in Cymbidieae (among subtribes such as Catasetinae, Cyrtopodiinae, and Oncidiinae) based on the occurrence of two pollinia; those subtribes with four pollinia were placed in Maxillarieae. Chase (1987)
and Dressler (1989)
hypothesized that Vandeae (sensu lato) are a polyphyletic grade and that reduction in pollinium number occurred repeatedly. Dressler (1993)
placed Stanhopeinae in Maxillarieae together with subtribes such as the Zygopetalinae and Lycastinae, reflecting a change of opinion about the importance of pollinium number and increased emphasis on vegetative characters such as the number of nodes in the pseudobulbs. By contrast, Burns-Balogh and Funk (1986)
separated Maxillarieae and Vandeae solely on the basis of pollinium fusion; they included Stanhopeinae in their broad Vandeae together with such taxa as Sarcanthinae, Catasetinae, and Oncidiinae (all with two pollinia). In their classification, Maxillarieae are footnoted as being "not monophyletic" (presumably polyphyletic).
None of the most recent classifications of Maxillarieae (Dressler, 1993
; Brieger, Maatsch, and Senghas, 1994–2000
; Szlachetko, 1995
) are explicitly cladistic, but all provide modern hypotheses of relationships. The three classifications disagree on circumscriptions within the tribe; Dressler favors a broad Maxillarieae with eight subtribes, whereas Szlachetko splits these same taxa among seven tribes and 26 subtribes, and Brieger splits them into three tribes and at least 19 subtribes. In spite of different ranking criteria, these classifications are largely congruent on the delimitation and composition of many taxa. They disagree on the placement of several anomalous genera, such as Eriopsis, Dichaea, Vargasiella, Scuticaria, and Cryptarrhena. The classifications differ most in their treatment of oncidioid orchids; Dressler recognized a broad Oncidiinae, whereas Brieger and Szlachetko divided these taxa among numerous subtribes. The composition of oncidioid clades also varied greatly among systems. Günter Gerlach recently prepared the treatment of Stanhopeinae and Coeliopsidinae for the Brieger, Maatsch, and Senghas classification (Brieger, Maatsch, and Senghas, 2000); his treatment incorporated the molecular evidence from our study.
Freudenstein and Rasmussen (1999)
produced a more rigorous morphology-based cladistic analysis of Orchidaceae, but the resulting cladograms yield little resolution at the subtribal level. Recent molecular studies at the family (Cameron et al., 1999
) and lower levels (Cox et al., 1997
; Douzery et al., 1999
; Kores et al., 1997
) are helping to define higher level relationships within the orchids, but none of these sampled extensively within Maxillarieae. The Cameron et al. (1999)
rbcL analysis of Orchidaceae included 20 representatives of Maxillarieae (sensu Dressler, 1993
), which form a monophyletic clade sister to Catasetinae and Cyrtopodiinae in the shortest trees, although without high bootstrap support.
We attempted to clarify these generic and subtribal relationships by performing parsimony analyses of sequence data from three regions: matK (plastid), trnL intron and trnL-F spacer (hereafter treated as a single matrix designated trnL-F; plastid), and ITS 1, 5.8S, and ITS 2 (nuclear ribosomal DNA; hereafter referred to as ITS nrDNA). The matK gene codes for a maturase that is 1550 bp in length and several times more variable than rbcL in most angiosperms (Soltis and Soltis, 1998
). The trnL-F region (Taberlet et al., 1991
) is largely noncoding and consists of an intron in the trnL (UAA) gene and the trnL-trnF (GAA) intergene spacer. The widely used ITS nrDNA region (Baldwin et al., 1995
) consists of two noncoding spacer regions flanking the 5.8S gene. Many recent studies have indicated that combined molecular data sets using regions with different levels of variation provide resolution at different areas of the cladogram, and phylogenetic resolution and levels of support are improved by directly combining independent molecular data sets (Chase and Cox, 1998
; Soltis et al., 1998
).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
) and matK analyses (Freudenstein, Kores, Goldman, Chase, and Whitten, unpublished data). We included Dressleria (Catasetinae), Eulophia (Eulophiinae), and Cyrtopodium and Grammatophyllum (Cyrtopodiinae) as outgroups based on the results of these broader analyses.
|
) according to the methods of Doyle and Doyle (1987)
, scaled down to 1.0-mL extraction volumes. DNA was precipitated overnight at -20°C with 0.65 volumes of isopropanol, centrifuged, washed twice with 70% ethanol, and dried. The pellet was resuspended in 75 µL of Tris-EDTA buffer (TE) and stored at -20°C. Amplification of DNA was generally performed using 50-µL reactions with 35 cycles, 2.5 mmol/L MgCl2, and a hot start, using Promega (Promega Corp., Madison, Wisconsin, USA) or Epicentre (Epicentre Technologies, Madison, Wisconsin, USA) buffers and Taq polymerase. Annealing temperatures for amplification were 51°C for matK and 55–58°C for trnL-F. For ITS, a touchdown thermal cycling program was used; the initial annealing temperature was 76°C, decreasing 1°C per cycle for 15 cycles, followed by 15 cycles at 61°C. DNA samples that failed to amplify were cleaned to remove inhibitors using QIAquick columns (Qiagen, Inc., Santa Clarita, California, USA). Some samples that failed to amplify for ITS using standard conditions were amplified successfully by adding betaine (1.0 mol/L final conc.) to the PCR (polymerase chain reaction) mix. Amplification and sequencing primers were those of Sun et al. (1994)
for ITS 1 and 2 and Taberlet et al. (1991)
for trnL-F. Some amplifications for trnL-F using primers c and f produced multiple bands; for these taxa, the region was amplified in two separate reactions using primers c and d and e and f, which yielded single products. Primers used for matK are (5' to 3'; locations approximate): 56F—ACTTCCTCTATCCGCTACTCCTT; 749F—TTGAGCGAACACATTTTTCTATGGAA; 832R—ACATAATGTATGAAAGTATMTTTGA; 1520R—CGGATAATGTCCAAATACCAAATA. Some taxa were amplified using the matK primers trnK-3914F and trnK-2R of Johnson and Soltis (1995)
and then sequenced using the above primers.
PCR products were purified using Wizard PCR Preps system (Promega, Inc., Madison, Wisconsin, USA) or QIAquick columns (Qiagen, Inc., Santa Clarita, California, USA) and directly sequenced on an PE Biosystems, Inc. (ABI; Foster City, California) 373 or 377 automated sequencer using standard dye-terminator chemistry according to manufacturer's protocols, except that cycle sequencing reactions were scaled down to 5 µL. Both strands were sequenced to assure accuracy in base calling. The ABI software packages "Sequence NavigatorTM" and "AutoassemblerTM" were used to edit and assemble complementary and overlapping sequences, and each individual base position was examined for agreement of the two strands. DNA sequences were aligned manually, and gaps were coded as missing values. The ends of matrices were trimmed to exclude sequencing artifacts. One region of ambiguous alignment was excluded in the ITS nrDNA matrix, and five regions (totaling 149 bp) were excluded from the trnL-F matrix. Sequences are deposited in GenBank (matK AF239415–AF239510; trnL-F AF239511–AF239606; ITS AF239319–AF239510). The aligned data matrix is available from the authors and is archived at http://www.botany.org/bsa/ajbsupp/v87/whitten.html. All cladistic analyses were performed using PAUP* version 4.0b2 (Swofford, 1999
). MacClade version 3.08a (Maddison and Maddison, 1997
) was used to plot the number of steps per site.
Search strategies
Each matrix (three separate and the combined ITS nrDNA/trnL-F/matK) was subjected to 1000 replicates of random taxon entry additions, MULTREES on, using subtree pruning and regrafting (SPR) swapping, but saving only ten trees per replicate to minimize time spent swapping on suboptimal islands. The shortest trees from this search were used as starting trees and up to 10 000 trees were swapped to completion using SPR; the tree limit was set to 10 000 for each matrix due to computer memory limitations. For the combined analysis, up to 10 000 Fitch trees were used as starting trees for successive weighting (SW; Farris, 1969
). SW was used to downweight sites that are highly homoplasious. Lledó et al. (1998)
provided a lucid discussion of this successive weighting strategy. The characters were reweighted on the rescaled consistency index (RC) based on the best performance on trees using the menu command in PAUP*. Each round of analysis consisted of ten replicates of random taxon entry, MULTREES on, SPR swapping, and holding ten trees per replicate. The shortest trees collected in each of these ten replicates were used as starting trees to collect all shortest SW trees, which were swapped to completion. Rounds of search followed by reweighting were repeated until tree length remained the same in two successive rounds. Trees were evaluated on the basis of tree length, consistency index (CI), and retention index (RI) as calculated in PAUP*. Confidence limits for trees were assessed by performing 1000 replicates of bootstrapping (Felsenstein, 1985
) using equal weighting, SPR swapping, MULTREES on, and holding only ten trees per replicate. For the combined analysis, bootstrapping was performed using equal weights and also using the final weight set from successive weighting.
Patterns of sequence evolution were estimated using MacClade (Maddison and Maddison, 1997
) with matrices stripped to include only the positions included in the cladistic analyses. Because we believe that the trees from the combined (SW) analyses are the most accurate estimates of phylogeny (due to higher overall bootstrap support), we assessed the evolution of each region on one of these trees, rather than trees produced from analyses of separate data sets. To calculate the number of transitions and transversions [and their consistency indices (CI) and retention indices (RI)] observed on one of the shortest SW trees, we used a stepmatrix to calculate the number of transversions at each base position by weighting the transitions to zero. After invoking the "Typesets" command in PAUP* and loading one of the shortest SW trees, the "Tree score" command was used to calculate the number of transversions and their collective CI and RI (ACCTRAN optimization). From these, we calculated those of transitions.
We assessed congruence of the separate data sets by visual inspection of the individual bootstrap consensus trees. We considered the bootstrap trees to be incongruent only if they displayed "hard" (i.e., highly bootstrap supported) incongruence, rather than "soft" (poorly bootstrap supported) incongruence (Seelanan, Schnabel, and Wendel, 1997
; Wiens, 1998
). We use the following descriptions for categories of bootstrap support: unsupported, <50%; weak, 50–74%; moderate, 75–84%; strong 85–100%. We consider percentages <50% to be unsupported because such groups often are not present in all shortest Fitch trees. All trees illustrated with branch lengths include autapomorphies; CIs reported also include autapomorphies (even without autapomorphies the CI is a less meaningful statistic and so we emphasize RI as the more meaningful measure of performance; see Davis et al., 1998
, for a discussion of the problem of using even a standard calculation of CI).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
|
|
; Xiang, Soltis, and Soltis, 1998
). Third-codon positions contributed the most steps (38.8% based on the combined tree), slightly more steps than first (32.2%) or second (28.5%) positions, but all three sites displayed approximately equal CI and RI values (Table 3). Transitions were less numerous and had both a higher CI and RI than transversions (Table 4). Heuristic search (Fitch criterion) yielded more than 10 000 equally parsimonious trees of 1140 steps (CI 0.56, RI 0.69).
|
are unsupported and divided into a Coeliopsis/Lycomormium/Peristeria clade (Coeliopsidinae), Braemia, and a moderately supported remainder, hereafter referred to as core Stanhopeinae. Many clades within core Stanhopeinae are highly supported. The position of Eriopsis is unresolved within Maxillarieae in the strict consensus of all shortest trees and the bootstrap consensus.
|
957 bases of intron, the 3' exon (50 bases), and 528 bases of the intergenic spacer (Fig. 1). Numerous large indels occur in the intergenic spacer and especially the intron; some taxa (e.g., Lycomormium, Coeliopsis, Telipogon, Stellilabium) have deletions of 300 bp or more. Of the 1282 bases included, 547 (43%) were variable, of which only 285 (22%) were potentially informative. Transversions are more numerous than transitions in all three regions, and the ts/tv ratio is slightly higher in the exon than in the two noncoding regions (Table 4). Heuristic search (Fitch criterion) yielded more than 10 000 equally parsimonious trees of 1247 steps (CI 0.58, RI 0.64). The bootstrap consensus tree (Fig. 3) is congruent with that of matK (Fig. 2), although less resolved at the subtribal level. Zygopetalinae are not monophyletic; only the Huntleyinae clade (Dichaea to Pescatorea) and the Zygopetalum clade (Zygopetalum to Batemannia) receive weak bootstrap support. Oncidiinae are unsupported, again with only the core Oncidiinae receiving weak support. Stanhopeinae show the same division into three clades as in the matK bootstrap consensus tree (Braemia; Coeliopsis clade; core Stanhopeinae). Core Stanhopeinae are highly resolved at the generic level, with most nodes displaying moderate to high bootstrap support. The placement of Eriopsis is again unresolved.
|
The ITS bootstrap consensus tree (Fig. 4) is highly congruent with the two plastid data sets (Figs. 2, 3). Eriopsis is weakly supported as sister to all other genera of Maxillarieae. Zygopetalinae (including Cryptarrhena) are weakly supported, with high support for several clades within the subtribe. Stanhopeinae are weakly monophyletic, again consisting of a Coeliopsis clade and a core Stanhopeinae including Braemia. Many clades within core Stanhopeinae are highly supported. Maxillariinae (sensu lato) are unsupported, but Maxillariinae (sensu stricto), Bifrenariinae (minus Rudolfiella), Lycastinae, and Xylobium show weak to high levels of support. Oncidiinae are not monophyletic; a weakly supported clade unites the core Oncidiinae with Telipogon, Stellilabium, Ornithocephalus, and Fernandezia. However, all of these deeper nodes are only weakly supported (none higher than 63% bootstrap support).
|
The combined matrix under the Fitch criterion yielded 1024 trees of 5096 steps (CI 0.48, RI 0.62). Successive weighting produced three trees of 1575 steps (CI 0.79, RI 0.83), corresponding to a Fitch length of 5097 (one step longer than the equally weighted trees). One of these three trees is shown in Figs. 5 and 6; the number of Fitch steps is shown above each branch, and the Fitch and SW bootstrap values are shown below. An arrowhead in Fig. 6 indicates a single node that collapses in the strict consensus of the three shortest trees. The bootstrap analysis of the combined data set yielded higher levels of support (54 clades with Fitch bootstrap support >85%; Table 2) than did any of the single data sets.
|
|
The remaining subtribes form a weakly/moderately supported clade (54% Fitch; SW 87% bootstrap) that is sister to Oncidiinae/Eriopsidinae. Zygopetalinae are sister to a broad Maxillariinae plus (Coeliopsidinae plus Stanhopeinae). The highly supported (100%) Zygopetalinae lack support at several deeper nodes within the clade, but Dichaea is clearly sister to the Pescatorea/Chaubardia clade. Cryptarrhena is embedded within Zygopetalinae, but its relationships within the subtribe are poorly supported. Maxillariinae (as defined here) are strongly supported in the SW bootstrap analysis and contain several strongly supported clades that have been recognized as subtribes by various authors: Lycastinae (Lycaste, Anguloa, Neomoorea); Maxillariinae (sensu stricto) (Maxillaria, Trigonidium, Cryptocentrum); Bifrenariinae (Bifrenaria, Rudolfiella, Scuticaria); and Xylobium. Three genera (Coeliopsis, Lycomormium, Peristeria) form a highly supported clade sister to the remaining genera of Stanhopeinae. Clades within Stanhopeinae are highly resolved, with high bootstrap support for most nodes.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Johnson and Soltis, 1994
; Xiang, Soltis, and Soltis, 1998
). In the large angiosperm data sets for rbcL and atpB (Savolainen et al., in press), the ts/tv ratios are 1.6 and 1.8, respectively, and 70% of substitutions are at third positions. These statistics contrast with the low ts/tv ratio and even numbers of substitutions at all codon sites found in Maxillarieae. Although the matK indels occur in triplets consistent with a coding region, this might indicate either a recent loss of function or that patterns of indel activity continue for some time after function has been lost. The trnL-F region has more variable sites than matK, but variable sites in both regions are evolving at similar rates (2.2 steps/site vs. 2.3 steps/site, respectively). Numbers of phylogenetically informative sites are similar, although matK is longer than trnL-F in each species. Performance, as measured by RI, is almost identical. Frequency of ts and tv is not related to performance (RI) in any of these regions. All three codon positions perform similarly, but third positions are not better as reported for rbcL and atpB (Savolainen et al., in press). The ts/tv ratios show no clear correlations of frequency with RI; therefore, there seems to be no basis for whole-category weights. Thus, we favor SW that weights each position individually based upon RC.
Classification of Maxillarieae based upon molecular data
The high level of congruence among the three data sets, representing coding and noncoding as well as plastid and nuclear regions, and the high bootstrap values in the combined analysis support the resulting tree as a good hypothesis of phylogenetic relationships among the taxa sampled. To translate this tree into a subtribal classification, we used several criteria (Backlund and Bremer, 1998
): (1) subtribes and clades must be monophyletic and highly supported; (2) nodes defining subtribes preferably should include morphological synapomorphies that permit recognition of members; and (3) the classification should be as consistent as possible with previous systems. We did not consider levels of sequence divergence to be a primary criterion (although it is clear that patterns of bootstrap support and branch lengths are highly correlated), because these and other data indicate that some clades within the Oncidiinae (e.g., Ornithocephalus, Telipogon, and Stellilabium) have accelerated rates of sequence divergence, perhaps due to rapid life cycles and altered life history strategies. Based upon these criteria, we recognize the following subtribes within Maxillarieae:
Eriopsidinae
Dressler (1993)
left this small enigmatic genus as incertae sedis and suggested that it might warrant placement in its own subtribe. He also noted that it has Maxillaria-type seeds, so a placement within Maxillarieae seemed likely. Szlachetko (1995)
created the monogeneric subtribe Eriopsidinae within his Maxillarieae. In a morphology-based cladogram of relationships of Maxillarieae and Zygopetalinae, Szlachetko (1995)
placed Eriopsidinae sister to Maxillariinae based on synapomorphies of duplicate leaves and entire tegulae. Our observations of living plants indicate that leaves of Eriopsis are revolute, not duplicate. In the rbcL trees (Cameron et al., 1999
), Eriopsis is nested within Maxillarieae but without strong bootstrap support. Our molecular data indicate that Eriopsis is isolated and sister to all other Maxillarieae in the shortest trees, although with weak bootstrap support. Its isolated position and lack of clear affinities support subtribal status. Further study is needed to find morphological characters that define this subtribe. The flowers, pollinaria, and seeds of Eriopsis possess no unique characters that separate them from the diversity within other subtribes, but the combination of characters exhibited fits well with Maxillarieae. To a field botanist, Eriopsis is immediately recognizable by a peculiar warty texture of the pseudobulbs and distinctive leathery leaf texture, but these traits are difficult to define as character states. This genus is rare in cultivation, and we were unable to obtain living material of other species to include in the analysis. We have no reason to suspect that the genus is not monophyletic, and inclusion of more species in the analysis is unlikely to change its subtribal placement.
Oncidiinae sensu Whitten et al
In previous classifications, the subtribes Ornithocephalinae and Telipogoninae have sometimes been regarded as allied to Oncidiinae but separated on the basis of pollinium number (two in Oncidiinae vs. four in the others). Earlier classifications of Orchidaceae have placed great weight upon pollinium number, with the assumption that reduction in pollinium number is largely irreversible. Our molecular data indicate a major reevaluation of Oncidiinae and related subtribes. Pachyphyllinae (represented by Fernandezia), Ornithocephalinae, and Telipogoninae (Telipogon, Stellilabium) form a highly supported clade that is embedded deeply within Oncidiinae. Maintaining these three clades as separate subtribes would necessitate recognition of several other segregates of the Oncidiinae at subtribal level (e.g., Trichopilia clade, Cuitlauzina/Palumbina clade, Lophiaris clade, Lockhartia). A greatly expanded combined molecular analysis of the Oncidiinae (Williams, Chase, and Whitten, unpublished data) supports the inclusion of these three subtribes within a broad Oncidiinae and indicates that increases in pollinium number are possible. In all other aspects, these three subtribes are highly compatible with Oncidiinae and only pollinium number has been used to exclude them. For example, the leafy stems composed of several nodes found in Pachyphyllinae are present also in Lockhartia. The distinctive columns with elongate rostellar beaks found in Ornithocephalinae and Telipogoninae are also found in Erycina, Sigmatostalix, and some members of Oncidium section Rostrata Rolfe. The oil-secreting lip calli of Ornithocephalinae are also found in Oncidium sect. Waluewa Pabst, and the pseudocopulatory flowers of Telipogoninae are observed in Tolumnia henekenii (R. H. Schomb. ex Lindl.) Nir and Oncidium sect. Crispa Rchb.f. ex Pfitzer. Although many of these have to be viewed as parallelisms as a result of the phylogenetic patterns observed here, these tendencies are rare outside Oncidiinae and the flower structure of these groups is generally typical of Oncidiinae.
Zygopetalinae
The highly supported Zygopetalinae include two anomalous genera: Dichaea and Cryptarrhena. Dichaea flower and pollinarium structure are consistent with other Zygopetalinae, but its unusual vegetative habit (long, pseudobulbless, monopodial stems and warty/spiny capsules) make it appear out of place in Zygopetalinae. The molecular analysis places Dichaea sister to the mostly pseudobulbless Huntleyinae clade (Chaubardia, Chondrorhyncha, and Pescatorea in Fig. 5). Cryptarrhena is a small genus of perhaps only four species, and the presence of pseudobulbs varies among species. The numerous, small flowers are borne on an arching raceme; the lip is clawed and anchor-shaped, and the column has a hooded, toothed clinandrium. The pollinarium lacks a conspicuous stipe and has a pair of long, hyaline caudicles and four flattened pollinia. This combination of characters led Dressler to place it in a monotypic subtribe (Dressler, 1971
) and then tribe (Dressler, 1980
). Our results show it firmly embedded within Zygopetalinae, and thus it does not warrant subtribal status. This placement was also supported by rbcL data (Cameron et al., 1999
). Although the hyaline caudicles are apparently autapomorphic for the genus, the other character states might represent synapomorphies shared with other Zygopetalinae, e.g., the toothed clinandrium present in Huntleya and the anchor-shaped lip present in Dichaea.
Previous classifications divide Zygopetalinae s. l. into several groups, recognized either formally as separate subtribes (Huntleyinae, Zygopetalinae, Warreinae, Dichaeinae; Szlachetko, 1995
) or as informal clades (Dressler, 1993
). Although our generic sampling in the combined data set is far from complete, several of these clades are recognizable: Huntleyinae (mostly pseudobulbless, leaves duplicate); Zygopetalinae (pseudobulbed, leaves usually convolute); and the Warrea clade (pseudobulbs of several internodes, leaves plicate), plus Dichaea and Cryptarrhena. Based on the low sequence divergence among these clades and the high bootstrap support for the larger clade, we favor recognition of a broad Zygopetalinae (similar to Dressler's). Unfortunately, the vegetative and floral diversity makes identification of synapomorphies defining Zygopetalinae s. l. difficult. Perhaps the most obvious synapomorphies for this subtribe are the usual combination of four flattened, superposed pollinia and a transverse slit-like stigma.
Dressler (1993)
included Vargasiella in Zygopetalinae, but stated that it might be placed in its own subtribe; Romero and Carnevali (1993)
validated the subtribe Vargasiellinae, which was also recognized by Szlachetko (1995)
. We were unable to obtain extractable material of this genus for inclusion in this study.
Maxillariinae
A bootstrap supported clade (Fitch 67%; SW 88%) includes Maxillariinae, Lycastinae, Bifrenariinae, and Xylobium. Most of the genera within this clade possess a distinct column foot and mentum, four rounded or ovoid pollinia, and a broad, open stigma, but their habits vary greatly. Maxillariinae s. s. are distinguished by conduplicate leaves and usually a crescent-shaped viscidium, whereas Lycastinae have plicate leaves and strap-like viscidia. Bifrenariinae have plicate or conduplicate leaves and often have a forked stipe. Although these four clades are individually all highly supported, there is no strong support for the position of Xylobium relative to Maxillariinae s. s. and Lycastinae. Recognition of three subtribes in this group would therefore necessitate creation of a separate subtribe or leave Xylobium as incertae sedis. We favor lumping these clades into a single more broadly defined Maxillariinae to reflect the close relationship among these clades and to avoid creation of a monogeneric subtribe.
Dressler (1993)
included Scuticaria in Zygopetalinae whereas Brieger, Maatsch, and Senghas (1993) and Szlachetko (1995)
placed it in Maxillariinae s. s., but our molecular data place this genus within the Bifrenaria clade (100% bootstrap support). Although its terete, whip-like leaves are anomalous, the flowers and pollinaria of some Scuticaria species are similar to those of Rudolfiella, and the few-flowered inflorescence is a synapomorphy shared with Bifrenaria. Although more intensive sampling of this subtribe might improve resolution, this clade is marked by low levels of sequence divergence (relative to other subtribes, e.g., Oncidiinae). Sequencing of additional regions will be necessary to resolve relationships within these clades.
This subtribe includes the large and vegetatively diverse genus Maxillaria for which estimates range from 450 to 600 species (J. Atwood and E. Christenson, personal communication). Our sampling of only two placeholder species indicates that the genus is polyphyletic, and more extensive sampling with ITS nrDNA (Whitten et al., unpublished data) indicates that generic boundaries need to be reevaluated within this species-rich group.
Coeliopsidinae sensu Whitten et al
These genera (Coeliopsis/Lycomormium/Peristeria) usually have been placed in Stanhopeinae, but they are distinguished from Stanhopeinae by: (1) smooth, unribbed, ovoid pseudobulbs bearing 3–4 large, thin, plicate leaves; (2) thick inflorescence rachis bearing globose flowers with thick, fleshy sepals and petals; (3) presence of a column foot and mentum; (4) roots with prominent root hairs; and (5) most distinctively, viscidia that are button-like and sclerified with short stipes. All three genera possess elongated but typical Maxillaria-type dust seeds (Whitten, unpublished data), not Stanhopea-type balloon seeds. Like Stanhopeinae, members of this subtribe are all pollinated by fragrance-collecting male euglossine bees. The button or tab-like viscidia of this clade are adapted to attachment on the smooth surface of the scutum of male bees (in Peristeria elata, the vertex of the bee's head; in Coeliopsis, on the frons of the bee's head; Williams, 1982
). The viscidia are reminiscent of those of some Oncidiinae, and it is unclear whether this represents a symplesiomorphy or (more likely) is a convergent adaptation for attachment to smooth surfaces of the pollinator. In all species of this clade observed in the greenhouse, the unpollinated flowers do not abscise from the inflorescence as they senesce. The flowers wither on the rachis, and the entire dried inflorescence remains attached to the plant for weeks or months in the greenhouse. To our knowledge, this trait is unique within Maxillarieae.
Szlachetko (1995)
split Stanhopeinae into two subtribes, creating Coeliopsidinae to accommodate Coeliopsis, Lacaena, Lueddemannia, Lycomormium, and Peristeria. In the errata accompanying his text, he excludes Lacaena from his concept of Coeliopsidinae; however, he does not explicitly place it within Stanhopeinae. Lacaena and Lueddemannia have pollinaria typical of other Stanhopeinae, lack a distinct column foot, and have a floral abscission layer; on the basis of morphology and molecular data, they clearly belong in Stanhopeinae and not in Coeliopsidinae.
Stanhopeinae sensu stricto
In the remaining genera of Stanhopeinae s. s., viscidia and stipes are thin and strap-like and adapted for attachment to the edge of the bee's scutellum or to a leg. The pseudobulbs are usually either ribbed/four-angled or flattened, and leaf texture is often thicker than in Coeliopsidinae. Roots are smooth, without prominent root hairs, and a column foot is lacking or indistinct. Unpollinated flowers quickly abscise and fall from the inflorescence. Although most members of Stanhopeinae possess highly distinctive balloon seeds (see Dressler, 1993
, for example), this character apparently shows several reversals to the typical Maxillaria dust-type seeds (in the Acineta clade and in several clades within Gongora; Whitten, unpublished data). Because of these differences, we favor recognition of separate Coeliopsidinae and Stanhopeinae. Although both clades are highly supported sister taxa, lumping them into a broader Stanhopeinae creates a subtribe that is difficult to define by morphological synapomorphies. Splitting them into separate subtribes allows Stanhopeinae (s. s.) to be characterized by pollinaria with ligulate or triangular viscidia, distinct stipes, two flattened pollinia, no column foot, a floral abscission layer, and (usually) balloon seeds.
Clades within Stanhopeinae (Fig. 6)
Braemia
Jenny (1985)
created the monotypic genus Braemia to accommodate the anomalous Polycycnis vittata (Lindl.) Rchb.f., citing differences in several vegetative and floral characters (e.g., lack of trichomes on rachis and lip) that distinguish it from members of Polycycnis. Our results support this transfer and affirm its generic uniqueness. We sequenced two different accessions of Braemia to verify its position in the trees; sequences were identical. The morphology of its column and pollinaria are typical of most Stanhopeinae, and its lip shape is somewhat similar to that of Polycycnis and Kegeliella, but the molecular results indicate that it is quite isolated within the subtribe and has no clear affinities with any other clade. Like most genera of Stanhopeinae, it possesses balloon seeds, although they are unique in being pointed at both ends (Whitten, unpublished data). No pollination data exist for Braemia, but floral/pollinarium morphology indicates that the pollinarium is probably deposited on the bee's scutellum.
Gongora clade (Cirrhaea/Gongora)
These taxa produce moderately to strongly ribbed pseudobulbs with one or two thin leaves. The inflorescence is pendent, many-flowered, thin, and wiry. The lip is fleshy and complexly three-parted. The lip is uppermost in both genera, but they have different pollination mechanisms. In Cirrhaea, the column curves upward, and the viscidium is hook-shaped; the pollinaria are attached to the base of the bee's leg. In Gongora, the bee hangs upside down from the lip and slides past the end of the column when exiting, guided by the adnate petals; pollinaria are attached to the scutellum. Jenny (1993)
recognized several subgenera/sections of Gongora, and these subgeneric taxa are recognizable in the combined cladogram (Fig. 6). Relationships within Gongora will be examined in detail in a subsequent paper. Cirrhaea and Gongora subgenus Portentosa (the sister clade of the rest of Gongora) share a character not found in any other Stanhopeinae; the ovary and pedicel are complexly twisted so that the flowers face outwards from the rachis of the inflorescence. In the remaining species of Gongora, the twist is absent and the flowers face inwards towards the rachis.
Acineta clade (Acineta/Lacaena/Lueddemannia/Vasqueziella)
These taxa are distinguished by flattened, slightly ribbed pseudobulbs bearing 3–4 leaves, pendent inflorescences with many fleshy flowers, and complex lips with large lateral lobes. Pollinarium structure varies within Acineta; in most species, the viscidium is rectangular with a 45° crease that allows it to fit onto the underside of the bee's scutellum where it joins the posterior thorax. In at least one species (Acineta densa Lindl.), the viscidium and stipe are narrow and typical of most Stanhopeinae, indicating that it might attach to the edge of the scutellum or perhaps back of the head. The monotypic Bolivian Vasqueziella is strongly supported as sister to Acineta and perhaps should be included within Acineta.
Polycycnis clade (Kegeliella/Polycycnis/Soterosanthus)
Plants of this clade vary in size from <10 cm to nearly 1 m and produce one or two thin leaves per pseudobulb. Some species produce a reddish anthocyanin pigmentation on the underside of the leaves, a character not found in other Stanhopeinae. Most distinctively, the many-flowered inflorescence has persistent black or brown trichomes. The column apex has broad wings, and the pollinaria are small and delicate. Jenny (1986)
created the monotypic Soterosanthus to accommodate Sievekingia shepheardii Rolfe; it does not agree with the vegetative and floral characters of Sievekingia. Our analyses support his conclusions and place Soterosanthus sister to Kegeliella. The lips of Kegeliella and Polycycnis are three-parted, with broad hypochilar wings and a deltoid to ligulate epichile, and their flowers are resupinate. Pollinaria are deposited on the bee's scutellum in Polycycnis and on the back of the head in Kegeliella. Flowers of Soterosanthus are nonresupinate, the lip is simple and entire, and the viscidium is curved; these are adaptations to pollinarium placement on the leg of the pollinator. Gerlach (1999)
examined a Euglossa crassipunctata Moure bearing pollinaria of both Soterosanthus and an unidentified Sievekingia. Pollinaria of both species were attached to the trochanter of the middle or hind legs, but the pollinaria of Soterosanthus are borne on a long, stiff stipe that projects outward from the bee's body, whereas the stipes of Sievekingia are shorter and twist inward under the bee's body. These differences indicate that subtle mechanical isolating mechanisms might exist even among orchids that utilize the same pollinator and site of pollinarium attachment.
The molecular data also indicate paraphyly of Polycycnis; the two species with a ligulate epichile and two leaves per pseudobulb (P. ornata and P. aurita) are sister to Kegeliella/Soterosanthus, not to the other species of Polycycnis with a deltoid epichile and mostly one leaf per pseudobulb (P. gratiosa). More sampling within the genus (especially P. annectans Dressler, P. tortuosa Dressler, and P. breviloba Summerhayes) is needed to clarify generic relationships. Recently, Jenny (1999)
transferred P. breviloba to a new monotypic genus, Luekelia Jenny, which has priority over Brasilocycnis Gerlach and Whitten (Gerlach and Whitten, 1999
). Living material of this species was not available for inclusion in the combined analysis, although preliminary ITS data affirm its distinctiveness from the rest of Polycycnis.
Stanhopea clade (Coryanthes/Embreea/Stanhopea/Sievekingia)
This clade includes the greatest diversity of floral morphologies and pollination mechanisms within the subtribe. Coryanthes flowers are complex in form and function; the large, fleshy lip forms a bucket trap that is filled with a watery fluid produced by a pair of "faucet" glands on the base of the column. Bees are lured to osmophores on the hypochile and fall into the bucket during fragrance collection; the pollinarium is deposited between the thorax and abdomen as the bee crawls out through an exit formed by the tip of the lip and the column. In contrast, the flowers of Sievekingia are nonresupinate and simple; the thin, cupped lip bears only a toothed callus, and the lip margin may be entire or highly fimbriate. The pollinarium has a curved viscidium and is attached to the trochanter of the bee's leg (Gerlach, 1999
). In spite of the great morphological differences, these two genera are sister taxa with high bootstrap support in our analyses. If this relationship is correct, we suggest that the intermediate forms leading to the evolution of Coryanthes are either extinct or never existed (i.e., Coryanthes is the result of a swift and massive reorganization of floral morphology), leaving us with few clues as to how the complex floral mechanism of Coryanthes evolved. The flowers of Stanhopea are also large, fleshy, and have a usually three-parted lip with a deeply saccate hypochile that encloses the osmophore tissue. One might hypothesize that the bucket of Coryanthes evolved by enlargement of the saccate hypochile of something like Stanhopea. However, artificial hybrids between Stanhopea and Coryanthes clearly indicate that the bucket of Coryanthes is entirely derived from the epichile, not the hypochile (Whitten, personal observation).
Stanhopea can be divided into several monophyletic groups on the basis of lip shape and number of flowers per inflorescence. The "two-flowered" species of Stanhopea (e.g., S. annulata, S. pulla) possess more or less entire lips and operate as simple gullet flowers; the pollinarium is deposited on the bee's scutellum as the bee backs out of the flower. The multiflowered species of Stanhopea (e.g., S. wardii) have fall-through flowers with three-parted lips; the bee slips while backing out of the waxy hypochile and falls through a channel formed by the mesochile, epichile, and the column. A third clade (not represented in this analysis) is formed by the Amazonian species such as S. candida Barb. Rodr. and S. grandiflora (Bonpl.) Rchb.f. The molecular data indicate that the genus is monophyletic in spite of these differences in morphology and pollination systems. The monotypic Embreea was created by Dodson (1980)
to accommodate Stanhopea rodigasiana Claes ex Cogn., which has numerous autapomorphies that separate it from the rest of Stanhopea species. The molecular data confirm its distinctiveness; it is unsupported as sister to any other genus in this clade. Although we have no pollination data on Embreea, its morphological similarity to Stanhopea indicates that it also has a fall-through flower.
Houlletia clade (Horichia/Houlletia/Paphinia/Schlimmia/Trevoria) clade
Although well supported, this clade lacks obvious morphological synapomorphies. Most members have a lip clearly divided into several parts, with a broad, triangular epichile and a hypochile with acute, curved lateral projections. Resupination of flowers varies among and within genera. The flowers of Schlimmia and Trevoria are nonresupinate and the pollinaria are deposited on the base of the bee's legs. The fleshy lateral sepals are partly (Trevoria) or completely (Schlimmia) fused to form a saccate cavity that encloses the lip and guides the pollinator. The lip is greatly reduced and vestigial in Schlimmia. Trevoria is distinguished by an asymmetric twist of the column (G. Gerlach, personal communication), similar to that of Mormodes (Catasetinae); the functional significance of this twist is unknown. Most species of Paphinia have resupinate flowers and deposit pollinaria on the rear edge of the scutellum, except for the nonresupinate P. subclausa Dressler that attaches pollinaria to the bee's leg (Dressler, 1968
). The monotypic Horichia is clearly distinct from the other genera of this clade on the basis of both morphology and molecular data. Its lip consists of a small, hemispherical hypochile bearing a pair of acute horns plus a narrow, acute epichile.
In the combined tree (Fig. 6), Houlletia is not monophyletic; the H. sanderi clade is sister to Schlimmia/Trevoria (which also have nonresupinate flowers and leg pollination). Houlletia tigrina is sister to all other members of this clade. Sampling of additional species is needed to evaluate the apparent paraphyly of Houlletia. As presently defined, Houlletia consists of two morphologically distinct groups. The group containing H. brockelhurstiana (the type species), H. tigrina, H. odoratissima Linden ex Lindl. and H. juruenensis Hoehne have open, resupinate flowers that are heavily spotted in red-brown. The epichile is triangular and the hypochile bears a pair of curved, acute projections; the lip shares many features of the lip of Paphinia. The viscidium is narrow, approximately the same width of the long stipe, and the pollinaria are deposited on the bee's scutellum. In contrast, the group containing H. sanderi, H. wallisii, H. clarae Schltr., and H. lowiana Rchb.f. has globose, nonresupinate flowers that are white to yellow, mostly unspotted, and borne on an erect inflorescence. The epichile is rectangular or ovate (not triangular), and the lateral projections on the hypochile are broad instead of acute. The pollinarium has a broad, concave viscidium. Pollination has not been observed, but the nonresupinate flowers and viscidium shape indicate that the pollinaria might be deposited on the bee's leg.
Recently, Lueckel, and Fessel (1999)
created the genus Jennyella to accommodate the globose, white flowered taxa (H. sanderi, H. kalbreyeriana Kraenzl., H. clarae Schltr.), but excluding H. wallisii and H. roraimensis Rolfe. Additional sampling and careful morphological analyses are needed to evaluate monophyly and generic boundaries within and among Houlletia, Trevoria, and Schlimmia.
Archivea kewensis Christenson & Jenny
The type and only specimen of this monotypic genus is a watercolor by T. Duncanson in the herbarium archives of the Royal Botanic Gardens, Kew; no pressed or living material is known (Christenson and Jenny, 1996
). Consequently, we were unable to include it in our analyses and its relationships remain uncertain.
General conclusions
As noted by previous workers (Chase and Cox, 1998
; Graybeal, 1998
), the greater taxon sampling appears to increase levels of bootstrap support and resolution. In the bootstrap consensus trees of the three separate analyses, Stanhopeinae (densely sampled) were consistently well resolved and supported, whereas the other subtribes (sparsely sampled) were less resolved. Likewise, the combined analysis yielded fewer shortest trees and higher levels of support than any of the separate analyses.
Generic limits within Stanhopeinae generally are well supported by the DNA data, in contrast to Oncidiinae (Chase and Palmer, 1988, 1997
) and several other subtribes (Zygopetalinae, Maxillariinae: Williams, Whitten, and Chase, unpublished data). Floral morphology of Stanhopeinae and Coeliopsidinae is complex and adapted for relatively precise placement of pollinaria on the body of male euglossine bees. The diversity of sites on the bee's body used by different genera is a form of mechanical isolation mechanism that reduces competition for pollination. This diversity of different pollination mechanisms and floral morphologies provided characters that earlier workers used to define genera. The molecular data generally support these generic concepts based on morphology.
Stanhopeinae and Coeliopsidinae provide a legitimate reward to pollinators in the form of floral fragrances. The majority of species in other subtribes appear to be deceit-pollinated; many Oncidiinae are mimics of oil-secreting Malphigiaceae or are nectar-deceit mimics (e.g., Ada, Aspasia, Cochlioda, Trichocentrum). The specificity and set of reward-offering morphologies make floral morphology a generally good indicator of phylogenetic relationships among taxa of reward flowers (e.g., Stanhopeinae), whereas in other groups with nonspecific deceit flowers or a mixture of syndromes, floral morphology is highly misleading of phylogenetic relationships. This hypothesis is supported by recent molecular analyses of Catasetinae (Pridgeon and Chase, 1998
), which are all pollinated by fragrance-collecting male euglossine bees. As in Stanhopeinae, there is excellent congruence between generic relationships based on floral morphology (Romero, 1990
) and the molecular data. Zygopetalinae exhibit a mixture of pollination syndromes, with some species providing a true fragrance reward to male euglossines and other species that are nectar deceit flowers (Williams, 1982
; Ackerman, 1983, 1986
). There is considerable disagreement on generic concepts within the Huntleyinae clade, and preliminary molecular analyses (Whitten et al., unpublished data) indicate that most of the confusion in generic boundaries is centered around taxa that are nectar deceit flowers.
The real importance of DNA patterns is not just in clarifying generic limits, but rather as the foundation for examination of evolutionary effects of life history traits (e.g., pollination biology, ecology, vegetative habit), as well as specific morphological characters (seed and pollinarium morphology, anatomy). A robust phylogeny is critical to an understanding of the diversity of form observed in this family that has been studied extensively since the time of Darwin (1877)
. Previous attempts to understand the evolution of orchid morphology and pollination systems used phylogenies constructed with characters largely based upon pollination-related traits, with attendant hazards of circularity. With the aid of DNA information, this area of research can be placed in a more robust evolutionary framework than was previously possible.
|
|
FOOTNOTES |
|---|
4 Author for reprint requests (e-mail: whitten{at}FLMNH.UFL.EDU
).
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
———. 1986 Mechanisms and evolution of food-deceptive pollination systems in orchids. Lindleyana 1: 08–113
Backlund, A., and K. Bremer. 1998 To be or not to be: Principles of classification and monotypic plant families. Taxon 47: 391–400[CrossRef][ISI]
Baldwin, B. G., M. J. Sanderson, J. M. Porter, M. F. Wojciechowski, D. S. Campbell, and M. J. Donoghue. 1995 The ITS region of nuclear ribosomal DNA: a valuable source of evidence on angiosperm phylogeny. Annals of the Missouri Botanical Garden 82: 247–277[CrossRef][ISI]
Barthlott, W., and B. Ziegler. 1981 Mikromorphologie der Samenschalen als systematisches Merkmal bei Orchideen. Berichte der Deutschen Botanischen Gesellschaft 94: 267–273
Brieger, F. G., R. Maatsch, and K. Senghas [eds.]. 1994, 1995, 1996, 2000 In Schlechter Die Orchideen. 1713–2056, 2309–2436 Blackwell Wissenschafts-Verlag, Berlin-Wein, Germany
Burns-Balogh, P., and V. A. Funk. 1986 A phylogenetic analysis of the Orchidaceae. Smithsonian Contributions to Botany 61: 1–79
Cameron, K. M., M. W. Chase, W. M. Whitten, P. J. Kores, D. C. Jarrell, V. A. Albert, T. Yukawa, H. G. Hills, and D. H. Goldman. 1999 A phylogenetic analysis of the Orchidaceae: evidence from rbcL nucleotide sequences. American Journal of Botany 86: 208–224
Chase, M. W. 1987 Systematic implications of pollinarium morphology in Oncidium Sw., Odontoglossum Kunth, and allied genera (Orchidaceae). Lindleyana 2: 8–28
———, and A. V. Cox. 1998 Gene sequences, collaboration and analysis of large data sets. Australian Systematic Botany 11: 215–229[CrossRef][ISI]
———, and H. H. Hills. 1991 Silica gel: An ideal material for field preservation of leaf samples for DNA studies. Taxon 40: 215–220[CrossRef][ISI]
———, and J. D. Palmer. 1988 Chloroplast DNA variation, geographical distribution, and morphological parallelism in subtribe Oncidiinae (Orchidaceae). American Journal of Botany 75(6II): 163–164 (Abstract)
———, and ———. 1997 Leapfrog radiation in floral and vegetative traits among twig epiphytes in the orchid subtribe Oncidiinae. In T. J. Givnish and K. J. Sytsma [eds.], Molecular evolution and adaptive radiation, 331–352. Cambridge University Press, Cambridge, UK
———, and J. S. Pippin. 1987 Seed morphology in the Oncidiinae and related subtribes (Orchidaceae). Systematic Botany 13: 313–323[CrossRef][ISI]
Christenson, E. A., and R. Jenny. 1996 Archivea. Orchids 65: 496–499
Cox, A. V., A. M. Pridgeon, V. A. Albert, and M. W. Chase. 1997 Phylogenetics of the slipper orchids (Cypripediodeae, Orchidaceae): nuclear rDNA ITS sequences. Plant Systematics and Evolution 208: 197–223[CrossRef][ISI]
Darwin, C. 1877 The various contrivances by which orchids are fertilised by insects, 2nd. ed. D. Appleton, New York, New York, USA
Davis, J. I., M. P. Simmons, D. W. Stevenson, and J. F. Wendel. 1998 Data decisiveness, data quality, and incongruence in phylogenetic analysis: an example from the monocotyledons using mitochondrial atpA sequences. Systematic Biology 47: 282–310[CrossRef][ISI][Medline]
Dodson, C. H. 1980 Embreea—new genus in the Orchidaceae. Phytologia 46: 389–390
Douzery, E. J. P., A. M. Pridgeon, P. Kores, H. P. Linder, H. Kurzweil, and M. W. Chase. 1999 Molecular phylogenetics of Diseae (Orchidaceae): a contribution from nuclear ribosomal ITS sequences. American Journal of Botany 86: 887–899
Doyle, J. J., and J. S. Doyle. 1987 A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11–15
Dressler, R. L. 1968 Observations on orchids and euglossine bees in Panama and Costa Rica. Revista Biologia Tropical 15: 143–183
