| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
2Laboratoire de Paléontologie, Paléobiologie et Phylogénie - CC064, Institut des Sciences de l'Evolution UMR5554-CNRS, Université Montpellier II; Place E. Bataillon, 34 095 Montpellier Cedex 05, France; 3Molecular Systematics Section, Jodrell Laboratory, Royal Botanic Gardens, Kew Richmond, Surrey TW9 3DS, United Kingdom; 4Bolus Herbarium, University of Cape Town, Rondebosch 7700, South Africa; and 5Compton Herbarium, National Botanical Institute, Private Bag 7, Claremont 7735, South Africa
Received for publication April 16, 1998. Accepted for publication November 6, 1998.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Brownleeinae • Coryciinae • Diseae • Disinae • molecular phylogeny • Orchidaceae • ribosomal ITS • Satyriinae
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
20;th000 species in 800 genera), great variety of floral morphology and pollinator relationships, and diversity of ecological preferences (subterranean and epiphytic species, terrestrial herbs, and lianas). Among the several major Orchidaceae lineages that have been identified, the sister group to the rest of the orchids is probably the three- or two- anthered apostasioids, as indicated by the distribution of primitive features in the family (Dressler, 1993
, p.62), and evidence from plastid rbcL sequences (Chase et al., 1994
; Kores et al., 1997
; Cameron et al., 1999
). Similarly, the two-anthered cypripedioids (slipper orchids) with their saccate labellum could represent the second diverging group and are likely the sister group of the monandrous orchids (Dressler, 1993
), but the molecular data cited above provide weak support for this hypothesis. Analysis of rbcL sequences shows that apostasioids are sister to an unresolved clade with four major components: vanilloids, cypripedioids, epidendroids, and orchidoids (Kores et al., 1997
; Cameron et al., 1999
). Subfamily Orchidoideae comprises three clades: (1) core Diurideae, predominantly from Australasia; (2) core spiranthoids, mostly tropical but also well represented in the north temperate zone, and perhaps including Chloraeinae from South America and New Caledonia; and (3) the sister group to the Diurideae-spiranthoid lineage, the core orchidoids, which are both nearly cosmopolitan.
We will here focus on core orchidoids, which are considered to form a monophyletic assemblage well defined by morphological, anatomical (Dressler, 1993
), and molecular data (Kores et al., 1997
; Cameron et al., 1999
). As circumscribed here, the core orchidoids include only those genera with basitonic anthers. These genera are assigned to two tribes: Orchideae with an erect anther and generally spurred lip (Ophrys being one exception) and Diseae with a reflexed anther. The latter have either no spur or a spur formed by the dorsal sepal or two spurs formed by the lip (see Linder and Kurzweil, 1994
, and references therein). Orchideae include subtribes Orchidinae (mostly northern hemisphere but ranging into Africa and tropical Asia) and Habenariinae (largely pantropical, but ranging into temperate Eurasia and North America; Dressler, 1993
). Diseae comprise five subtribes of mainly South African provenance (Linder and Kurzweil, 1994
): Disinae (168 species in two currently recognized genera, Disa and Schizodium), Coryciinae (134 species in five genera, Ceratandra, Corycium, Disperis, Evotella, and Pterygodium), and Satyriinae (103 species in two genera, Satyrium and Pachites), and the monogeneric subtribes Brownleeinae (seven species) and Huttonaeinae (five species).
Each of these five subtribes is defined by a variety of morphological and anatomical synapomorphies, but their interrelationships remain speculative. The single putative synapomorphy, a reflexed or bent anther, assumes a variety of forms that would appear to share little in addition to being bent back dorsally. Dressler (1986)
suggested that Disinae and Satyriinae are closer to each other than either was to Coryciinae. Conversely, Dressler (1993)
later favored a Disinae/Coryciinae association based on an apomorphic galeate sepal. Kurzweil et al. (1995)
subjected 36 anatomical characters to cladistic analysis and found a weakly supported sister-group relationship between Coryciinae/Huttonaeinae on the one hand (both have the stigma derived only from the median carpel apex) and Satyriinae/Disinae on the other (both have sepals brightly colored), with uncertain affinities for Brownleeinae. No attempt has previously been made to evaluate monophyly of tribe Diseae, and the apomorphies mentioned above for the orchidoid tribes require further evaluation.
These phylogenetic ambiguities both at the generic and subtribe levels call for collection of molecular data. The nuclear ribosomal RNA genes (rDNAs) of higher plants are organized in long tandem repeats, each consisting of a single transcribed region spanning the 18S, 5.8S, 26S rDNAs, two small internal transcribed spacers (ITS1 and ITS2), and a large external nontranscribed intergenic spacer. The ITS1 and ITS2 spacers have been shown to constitute a valuable source of molecular characters to reconstruct plant phylogeny, from the generic to the familial levels (reviewed in Baldwin et al., 1995
). At deeper phylogenetic levels, the 5.8S -ITS2 sequences may provide some information as well (Hershkovitz and Lewis, 1996
). The ITS1-5.8S rDNA-ITS2 locus has been previously used to study the relationships of other groups of Orchidaceae, including cypripedioids (Cox et al., 1997
), Orchideae (Bateman, Pridgeon, and Chase, 1997
; Pridgeon et al., 1997
), and the genus Platanthera (Hapeman and Inoue, 1997
). At present, the only other DNA sequence studies of orchids have involved plastid genes: rbcL for the whole family (Chase et al., 1994
; Kores et al., 1997
; Cameron et al., 1999
), Cypripedioideae (Albert, 1994
), Dendrobiinae (Yukawa, Cameron, and Chase, 1996
); ndhF for mostly Epidendroideae (Neyland and Urbatsch, 1996
).
This molecular study of Diseae was undertaken to address the following topics: (1) the pattern of molecular evolution of ITS among Diseae; (2) the evolutionary affinities of Diseae relative to other orchidoids; (3) the monophyly of subtribes Disinae, Satyriineae, Coryciineae, and their relation to Brownleeinae; (4) the monophyly of Disa, in particular to address whether Monadenia and Herschelia(nthe) are imbedded in Disa, as recent morphological analyses have indicated; and (5) the congruence of morphological and anatomical data with the molecular data (although the former are not formally analyzed here).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
) and purified by ultracentrifugation on a CsCl-ethidium-bromide gradient (1.55 g/mL). Amplification of the ITS1, 5.8S, and ITS2 region was conducted in a polymerase chain reaction (PCR) reagent volume of 100 µL, using 2.5 units of the Taq polymerase (Promega, Madison, WI), and 100 ng of the two primers designed by G. Sheridan (University of Bath, personal communication): a forward primer (AB101) annealing in the 18S gene, 5'-ACGAATTCATGGTCCGGTGAAGTGTTCG, and a reverse primer (AB102) annealing in the 26S gene, 5'-TAGAATTCCCCGGTTCGCTCGCCGTTAC. In some cases we used the ITS5 (forward) and ITS4 (reverse) primers of White et al. (1990)
or the combination AB101F/ITS4. The PCR protocol included: 25 cycles of 1 min denaturation (94°C), 1 min annealing (54°C), and 2 min, 30 s elongation (72°C), with two additional seconds elongation per cycle. PCR products were purified using the Wizard (Promega, Madison, WI) or QIAquick (QIAGEN, Amsterdam, The Netherlands) purification columns following the manufacturers' protocols. Cycle sequencing directly on the amplified product was conducted with the ABI Prism Dye Terminator Cycle Sequencing Ready Reaction kit (Perkin Elmer, Norwalk, CT), using 2.5 ng of primer in a 5-µL reaction volume. Sequencing conditions were the following: 26 cycles of 15 s denaturation (96°C), 1 s annealing (50°C), and 4 min elongation (60°C) in a Perkin-Elmer 9600 thermocycler. Sequencing reactions were purified by ethanol precipitation and run on a ABI Prism 377 automated sequencer. Electropherograms were analyzed with Sequencher 3.0 software (Gene Codes, Ann Arbor, MI). Complementary strands of the entire ITS1-5.8S rDNA-ITS2 region were both sequenced in single reactions each, providing better than 90% overlap.
Phylogenetic analyses
The end of the 18S rDNA, the beginning of the 26S rDNA, and the whole ITS1 - 5.8S rDNA - ITS2 region were sequenced for all the taxa, and these were compared to sequences produced for other orchidoids (12 taxa were selected from Pridgeon et al., 1997
). Sequences were manually aligned and analyzed using the MUST package (Philippe, 1993
). Congruence of the ITS1, 5.8S rDNA, and ITS2 data sets was evaluated by the XARN program developed by Farris et al. (1994). Phylogenetic reconstructions were obtained by the maximum parsimony (MP) method (PAUP 3.1.1; Swofford, 1993
), and by the maximum likelihood (ML) quartet puzzling method (PUZZLE 2.5.1; Strimmer and von Haeseler, 1996
), using the Tamura and Nei (1993)
model of sequence evolution.
Cranichis fertilis (Spirantheae), Chloraea sp. (Chloraeinae), and Elythranthera emarginata and Diuris sulphurea (all Diurideae) were used as outgroup for all core orchidoids based on the results in Kores et al. (1997)
. All molecular characters were assessed as independent, unordered, and equally weighted (Fitch parsimony; Fitch, 1971
). Indels were coded in two ways, either as a fifth character or as missing data. The MP analyses were conducted with heuristic searches using 200 replicates of random taxon addition, tree bisection-reconnection swapping (TBR) with MULPARS on (i.e., saving all shortest trees). ACCTRAN optimization was used for all illustrated MP trees.
Robustness of the clades was assessed by three different approaches: (1) the bootstrap (Felsenstein, 1985
) yielding bootstrap percentages (BP) for each node, computed after resamplings followed by a MP reconstruction (bootstrap option in PAUP 3.1.1., with 100 replicates of heuristic search, one random sequence addition per replicate, and MAXTREES = 100); (2) Bremer support (BS), i.e., the number of extra steps required to collapse the corresponding node (Bremer, 1988
) using enforced topological constraints with PAUP 3.1.1.; and (3) the reliability percentages (RP), i.e., the number of times the group appears after 1000 ML puzzling steps (PUZZLE 2.5.1; Strimmer and von Haeseler, 1996
). The tests of alternative topologies were conducted by the ML method of Kishino and Hasegawa (1989)
as implemented in the DNAML program (version 3.572c) of the PHYLIP package (Felsenstein, 1993
). The ML trees were drawn by the TREEVIEW program (Page, 1996
). Character-state changes were mapped on the trees using MacClade 3.04 (Maddison and Maddison, 1992
).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
. The ITS1 and ITS2 regions exhibited the same base composition: 25–27% A, 19% C, 24–26% G, and 30% T. By contrast, the 5.8S region showed a lower T content (21%) and a higher C content (25%). The overall G + C content (43–45%) was rather low for Diseae by comparison to other angiosperms (see representative values in Baldwin et al., 1995
). Diseae had an extensive ITS1 length polymorphism, representing 13% of the maximum length (Corycium nigrescens and C. dracomontanum). No ITS polymorphism was detected in two individuals of Satyrium membranaceum, which produced the same sequence throughout the ITS region. The two individuals of Disa (Monadenia) bracteata likewise exhibited the same ITS sequence, one from Australia (naturalized) and the other from South Africa. The final matrix alignment has a total length of 711 sites (300, 163, and 248 sites for ITS1, 5.8S, and ITS2, respectively), of which 566 (283, 51, and 232, respectively) are variable and 506 (261, 35, and 210, respectively) phylogenetically informative. The mean divergence between all pairs of taxa is 14.4 ± 4.9% for Disinae (66 pairs), 21.2 ± 9.6% for Coryciinae (10 pairs), and 21.6 ± 5.5% for Satyriinae (45 pairs). Similarly, the mean pairwise divergence is 17.6 ± 4.9% for Orchidinae (36 pairs) and 19.9 ± 6.1% for Habenariinae (55 pairs). The two Disperis species were more distant relative to other Diseae and Orchideae: 42.2 ± 1.9% (96 pairs). Comparisons between all Diseae plus Orchideae and the four cranichid and diurid outgroup taxa (Cranichis, Diuris, Elythranthera, and Chloraea) exhibit 39.6 ± 3.1% divergence (200 pairs).
Molecular evolution of ITS sequences
The pattern of length polymorphisms (indels) among Satyriinae involves single-nucleotide (nt), 3-nt, or 6-nt long motifs of two classes. Firstly, length polymorphism occurs in stretches of the same base: deletion of one or two Ts in Satyrium acuminatum or S. humile after a (T)3-5 repeat (positions 50 in the alignment) or insertion of two As in S. rhynchantum after a (A)2 repeat (end of ITS2). Secondly, all the Satyrium species except S. odorum and S. humile possess four tandemly repeated 3-nt long nearly identical motifs at the end of the ITS2 (positions 647–665: ATC-[C]TC-ATC-ATC). Satyrium humile exhibits a deletion of one motif, whereas S. odorum shows the insertion of a fifth ATC block. Thirdly, Satyrium acuminatum is the only orchid species so far studied that exhibits a 6-nt long insertion close to the beginning of ITS1 (position 70). This GCTATT motif likely represents the result of a duplication because it occurs directly downstream of two nearly identical tandem copies (GCTATT-GC[A]ATT). At least two Satyrium species exhibit a large deletion in the ITS1 region: 19 bp for S. carneum and 24 bp for S. membranaceum, located in approximately the same 43 nt-long region (positions 236–278). Since these two species are not directly related (Fig. 1), the question of how often large indels occur in the ITS1 region is relevant.
|
T over A G transitions. Among transversions, the A T substitutions predominate over G T by a 1.29 factor and are twice as abundant as A C and G C. Using a stepmatrix in which all transitions were set to zero, we separately calculated the consistency index (CI) and retention index (RI) of transitions and transversions to see if the more numerous transitions produce weaker patterns of relationships (frequency is often used as the basis for reweighting). These were calculated for the matrix in which gaps were coded as missing. For transversions and transitions, respectively, the CIs were 0.34 and 0.41, and the RIs were 0.57 and 0.50, as based on the shortest trees in the MP analysis. Thus, although more frequent, transitions were not substantially different in performance from transversions.
Phylogenetic trees reconstructed by maximum parsimony
The ITS1, 5.8S, and ITS2 matrices are not significantly incongruent as shown by the XARN test (
= 100%, after 1000 counts). To keep the ratio between the number of informative sites and the number of taxa at the highest level, we directly combined the ITS1 with ITS2 and 5.8S regions.
When indels were coded as a fifth character, the MP analysis yielded two most parsimonious trees (length = 3471; CI = 0.40; RI = 0.56; CI excluding uninformative sites = 0.39). The strict consensus topology and the corresponding branch supports are indicated in Fig. 1. The main phylogenetic results deduced from the comparison of the ITS1-5.8S-ITS2 region for Diseae, Orchideae, and outgroups (Cranichis, Chloraea, Elythranthera, and Diuris) are as follows. Seven major core-orchidoid groups are depicted in Fig. 1. Five lineages of Diseae are evident: Satyriinae, Coryciinae s.s. (i.e., all Coryciinae excluding Disperis: Kurzweil, Linder, and Chesselet, 1991
), Disinae, Brownleea coerulea, and the two Disperis species. Two Orchideae clades are depicted: Orchidinae and Habenariinae. The relative robustness for these groups varies from high support (BP = 97 to 100 and BS = +10 to +17 for Satyriinae, Orchidinae, and Disinae) to low support (BS = +1 for Habenariinae). The branching order among these seven groups indicates that Diseae are paraphyletic, with successive emergence of Disperis, Brownleeinae, Disinae, Coryciinae s.s., Satyriinae, and then Orchideae (Fig. 1). All the internodal segments are, however, either weakly or not supported (BP at most only 53 and BS of +1). Constraining Diseae to be monophyletic involves 17 extra steps (one tree; length = 3488; CI = 0.40; RI = 0.56). The relationships between these lineages indicate that Satyriinae are closely related to Orchidinae + Habenariinae and Disperis is the sister group of the rest of the core orchidoids. The other results are now described for each subtribe.
Disinae
The 12 Disinae species here considered form a strongly supported monophyletic group (Fig. 1). Four multifurcating clades were recovered: (1) the section Disa s.s.: D. cardinalis, Disa tripetaloides, D. racemosa, D. uniflora, and D. pillansii; (2) the section Micranthae sensu lato (s.l.): Disa chrysostachya, Disa (Herschelia) spathulata, and Disa (Monadenia) bracteata; (3) the Disa glandulosa-D. longicornis clade; and (4) the Disa rosea-D. sagittalis clade. It is clear that these clades do not conform to the sections Disa (Disa rosea falls outside of the five other members of the section Disa) and section Coryphaea (Disa glandulosa and D. sagittalis cluster separately) as defined by Linder (1981a)
. By comparison to other Disinae species, members of Disa section Disa s.s. share some synapomorphic features: a 9-bp (Disa pillansii), 10-bp (Disa tripetaloides and D. cardinalis), and 16-bp (D. racemosa and D. uniflora) deletion (positions 126–142) in ITS1, a T or A to C change in the 5.8S gene (position 408), and a 1-bp deletion (position 513) and two A replacements (positions 530 and 670) in ITS2. Within section Disa, D. cardinalis and D. tripetaloides are the closest relatives, with D. racemosa plus D. uniflora as a sister group, and finally D. pillansii as an outlier.
Satyriinae
The ten Satyrium species considered here belong to five sections according to Schlechter (1901)
and form a strongly supported monophyletic group (Fig. 1). All the bifurcations but one in the Satyriinae subtree are supported by high BP and BS. Given our current taxonomic sampling, the first divergence within the Satyrium genus separates S. bicallosum and Satyrium rhynchantum from the other species (Fig. 1). The second deepest split separates Satyrium nepalense and S. odorum from a cluster of six species. In this latter group, Satyrium carneum and S. ligulatum are sister, with S. acuminatum outside, then followed by S. stenopetalum and S. humile. The sister species of these five is S. membranaceum.
Coryciinae
Subtribe Coryciinae is diphyletic, with two distinct lineages branching at different locations in the MP trees. Coryciinae s.s., excluding Disperis, are sister to the Satyriinae/Orchideae clade. Moreover, Corycium is paraphyletic due to the nesting of Pterygodium within Corycium.
The two Disperis species represent the second unrelated Coryciinae lineage. The ITS sequences of Disperis capensis and D. lindleyana are highly divergent compared to other Diseae and Orchideae, and these two species are at present the only orchids to have a single nucleotide insertion in the 5.8S gene.
Orchideae
The sampling in this tribe is too restricted to draw any firm conclusions, but see Pridgeon et al. (1997)
and Bateman, Pridgeon, and Chase (1997)
for a better sampled study of this tribe. Most members of Orchidinae sensu Dressler (1993)
form a well-supported monophyletic group. However, Brachycorythis (tropical Africa) falls outside Orchidinae, which could be interpreted to mean that Orchidinae are strictly limited to Eurasia and North America. Sampling here is too sparse to be very useful in identifying subclades within the tribe, but Orchis clearly falls into two well-supported clades, which is compatible with the ITS results of Pridgeon et al. (1997)
and differences in chromosome number and floral pigments (Pridgeon et al., 1997
; Bateman, Pridgeon, and Chase, 1997
). Habenariinae are also supported as monophyletic, but weakly so (BS of only +1), and Habenaria would appear to be well supported as grossly polyphyletic. This requires much more thorough taxon sampling before firm conclusions are drawn. Orchidinae and Habenariinae are moderately supported as sister groups.
If indels are coded as missing data, the MP analysis yielded three most parsimonious trees (length = 2903; CI = 0.37; RI = 0.53; CI excluding uninformative sites = 0.35). A comparison of the topologies obtained with indels coded either as missing data or as a fifth character state indicates switches of 1-3 taxa. The species involved are Orchis morio (either sister to Barlia, Ophrys, and Serapias, or to Barlia, respectively), and Disa uniflora and D. pillansi (D. uniflora sister to D. racemosa, or sister to D. racemosa, D. pillansii, D. cardinalis, and D. tripetaloides). The method of indel coding does not have an influence on any of the major phylogenetic conclusions that are summarized by Fig. 1.
Pattern of accumulation of ITS-rDNA substitutions
The number of steps between each pair of taxa was computed on one of the two most parsimonious trees (length: 3471 steps; indels coded as fifth character state). The total number of differences observed between pairwise sequences, i.e., transitions and transversions and indels, was plotted against the number of substitutions inferred by maximum parsimony (Fig. 2) as indicated by Philippe et al. (1994)
. For species differing by up to 150 differences in their ITS-rDNA region, the actual number of substitutions inferred increases regularly, up to 150 steps (Fig. 2: empty circles). For species exhibiting between 150 and 250 differences, the profile remains linear though more dispersed, indicating that multiple changes began to occur at the same sites for comparisons involving distant genera of the same tribe and genera of different tribes. For more distant comparisons, namely those involving the cranichid, diurid, and Disperis taxa relative to the others, all the substitutions are multiple (Fig. 2: black triangles). At these high levels of divergence, many indels occurred, and the ITS alignment becomes difficult: local misalignment can therefore create hypothesized multiple substitutions at sites that are likely to be nonhomologous. This analysis of the accumulation pattern of substitutions, however, confirms the isolated status of Disperis among core orchidoids; it acts more like a member of the outgroup.
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Disinae contain the genera Disa, Herschelia, Monadenia, and Schizodium. The type of the subtribe is the large genus Disa (128 species). The delimitation of Disa has been problematic due to: (1) the status of Monadenia, which had been considered a member of Disa by some authors, but with no clear affinities to any section (Kurzweil et al., 1995
); (2) recognition of the genus Herschelia (often known under the name Herschelianthe), despite its close relationships to D. section Stenocarpa (Kurzweil et al., 1995
); and (3) possible distinction of D. section Micranthae as a separate genus (Chesselet, 1989
). The molecular phylogeny supports a circumscription of the genus Disa that includes Monadenia and Herschelia (as Disa bracteata and Disa spathulata, respectively; Figs. 1, 3), therefore confirming previous observations (Linder and Kurzweil, 1994
) and supporting the circumscription of the genus advocated by Bolus (1884,
1885)
and Schlechter (1901)
. Additionally, Disa (Herschelia) spathulata strongly clusters with Disa chrysostachya (section Micranthae), and this clade is weakly supported as sister to Disa (Monadenium) bracteata. Among the 14 Disa sections recognized by Linder (1981a,
b)
, our study included only four sections: Disa (six representatives), Coryphaea (two), Micranthae (one), and Phlebidia (one; Table 1). According to Linder (1981a,
b)
, plants possess either erect anthers (rarely, section Micranthae) or reflexed anthers. In the latter case, petals are reflexed next to the anther, either blue (section Phlebidia) or another color (section Disa), or plants have erect petals, free from the rostellum, and having lateral lobes squared (section Coryphaea). Sections Disa and Coryphaea are paraphyletic because Disa rosea (section Disa) clusters with D. sagittalis (section Coryphaea) in a clade distinct from other representatives of sections Disa and Coryphaea. This result agrees with Linder (1981a
, p.272) who noted that D. rosea is morphologically isolated within the Corymbosae series. By contrast, a well-defined clade including D. cardinalis, D. tripetaloides, D. racemosa, D. uniflora, and D. pillansii is always recovered. This section Disa s.s. is characterized by chromosome numbers ranging from 2n = 36 to 38, and is distinct from the 2n = 40 found in D. sagittalis (Pienaar et al., 1989
). Within D. section Disa s.s., the maximum parsimony analysis of ITS data with indels coded as a fifth character (Fig. 1) supports the monophyly of series Racemosae (D. racemosa, D. cardinalis, D. tripetaloides, D. uniflora).
The results within Disa must be considered in light of the relatively thin sampling of sections of the genus for ITS sequences (12 of the 128 Disa species). In morphological analyses (Johnson, Linder, and Steiner, 1998
; Linder and Kurweil, unpublished data), D. sect. Stenocarpa was sister to the Disa (Herschelia) spathulata group and the Disa (Monadenia) bracteata group clustered near the base of Disa. Sampling of more species diversity within the large genera (Disa, Satyrium, Disperis, and Corycium) would be required to more accurately assess infrageneric boundaries.
The naturalness and content of subtribe Satyriinae
Satyriinae include the genera Satyrium and Pachites (not available for this study), defined by two morphological synapomorphies: an elongated column and lack of differentiation between sepals and petals (Linder and Kurzweil, 1994
). Five of the seven sections recognized in the large genus Satyrium (100 species, Schlechter, 1901
; 88 species in Kurzweil and Linder, in press
) were included in our molecular study: Eusatyrium, Leptocentrum, Chlorocorys, Brachysaccium, and Satyridium. The ITS results do not correspond to some of these taxonomic sections, but sampling is too limited (ten of the 100 Satyrium species) to draw conclusions. These sections, with the exception of Brachysaccium and Chlorocorys (one species of each sampled here), are also not supported by morphological, anatomical, and pollen/seed ultrastructural characters (Kurzweil and Linder, in press
). Representatives of sections Eusatyrium (Satyrium membranaceum, S. humile, S. acuminatum, and S. carneum) and Leptocentrum (S. stenopetalum and S. ligulatum) are reciprocally embedded, thus indicating paraphyly of both groups. The apomorphy of two basal leaves appressed to the ground for S. section Eusatyrium thus represents a homoplastic condition (also found in Kurweil and Linder, in press
). For the same reason, the labellum possessing two threadlike calli (perhaps spurs) in sections Leptocentrum (pinkish-redish, whitish, or yellowish leaves; labellum oblong) and Chlorocorys (greenish leaves; suborbicular labellum) likely represents a convergent character state (identical to the findings of Kurzweil and Linder [in press]
, but they found that the latter also has other apomorphic characters and was nonetheless monophyletic).
The monotypic section Satyridium (Satyridium rostratum = Satyrium rhynchantum) has been often treated as a monotypic genus. Our molecular data do not support this view: Satyrium rhynchantum is always strongly associated with S. bicallosum, indicating that Satyridium should be synonymized with Satyrium. In the morphological/anatomical analysis (Kurzweil and Linder, in press
), it was also well nested within Satyrium, but the association with S. bicallosum was not predicted. Both species are morphologically unusual and are highly reduced or modified in several characteristics, leaving the impression that they are each isolated within the genus. As noted by Schlechter (1901)
, the labellum has short "little sacs" ("Säckchen") in sections Brachysaccium (here represented by S. bicallosum). In S. rhynchantum, these are fleshy and elongate-conical, and so this character may represent a true synapomorphy defining the S. bicallosum/S. rhynchantum clade. To evaluate this hypothesis, ITS should be sequenced for representatives of sections Aviceps and Leucocomus because the "little sacs" of the labellum are also encountered in these two groups.
Satyriinae also include Pachites, but its status is unclear (no material was available for this study). This genus comprises two rare species (P. appressa and P. bodkinii) and may be paraphyletic. Pachites bodkinii may actually be misplaced within Satyriinae because of the presence of monostelic (i.e., an undissected siphonostele) tubers, a character potentially of great assistance in resolving phylogenetic questions among Diseae (Kurzweil et al., 1995
).
Coryciinae: relationships between the Pterygodium/Corycium complex and Disperis
Coryciinae originally included the genera Corycium, Pterygodium, Evotella, Ceratandra, and Disperis, all with flowers possessing a lip with an appendage (Kurzweil et al., 1991
, hypothesized that the absence of the appendage in three species of Ceratandra was a secondary loss). Kurzweil et al. (1995)
suggested that the two clades Corycium/Pterygodium and Ceratandra/Evotella cluster together with Disperis as their sister group. Morphological support for Coryciinae s. s. is clear, with five apomorphic characters, whereas only two characters support a relationship of these with Disperis (Linder and Kurzweil, 1994
).
The MP and ML trees indicated paraphyly of Coryciinae with two distinct clades (Figs. 1, 3): (1) Coryciinae s.s. including Corycium and Pterygodium as sister group to Orchideae and Satyriinae; and (2) Disperis as sister to Orchideae plus all other Diseae. However, the test of significance in likelihood differences does not reject the monophyly of the subtribe (Table 2), and it should be noted that we have sampled only two members of large and widespread Disperis. If monophyly of Coryciinae is constrained with a requirement of six extra steps, Disperis pairs with Pterygodium, conflicting with the conclusions of Kurzweil et al. (1995)
.
The amount of ITS divergence is high within the Corycium/Pterygodium complex (e.g., Fig. 3). Corycium is paraphyletic due to the internal position of Pterygodium with C. carnosum sister to the whole clade. More sampling is needed to evaluate better the position of Pterygodium, and it is possible that our result is an effect of undersampling (5 of the 32 Corycium/Pterygodium species were sampled) and the higher levels of divergence encountered. The isolated position of Corycium carnosum was emphasized on the basis of its morphology and anatomy (Kurzweil et al., 1991
). The ITS data confirm its early diverging position among Coryciinae s.s. Corycium nigrescens and C. dracomontanum are close and sister to C. flanaganii, in full agreement with the morphological and anatomical study of Kurzweil et al. (1991)
. This clade is defined by a unique synapomorphy: the fusion of the lateral sepals.
Generic delimitation of Disperis has never been problematic, and the genus is isolated among Coryciinae (Kurzweil et al., 1991
). The 5.8S rDNA sequence of Disperis capensis and D. lindleyana confirm their uniqueness because these two species are the only orchids possessing a single nucleotide insertion in this gene (i.e., cytosine in position 128), in a region known to be of variable length over a broad spectrum of plants (Hershkovitz and Lewis, 1996
). A caveat to this statement must include as well an acknowledgment that <500 ITS sequences exist for a family with >20 000 species. The ITS1 and ITS2 sequences of the two Disperis are also highly divergent relative to other Diseae and Orchideae orthologues, as illustrated by the mean overall divergence of 51.0 ± 3.0% (ITS1) and 53.7 ± 3.0% (ITS2) for 96 pairwise comparisons. We cannot exclude the possibility that we have amplified an ITS pseudogene because divergent paralogues have been described in the nuclear genome of angiosperms (Buckler, Ippolito, and Holtsford, 1997
). However, the amplification and sequencing of the ITS region were completed twice independently for Disperis lindleyana, without any differences detected between the two derived sequences. The two Disperis sequences also exhibited similarities (BS = +65; Fig. 1), including three consecutive 16-bp, 5-bp, and 9-bp deletions and a unique [AATTGC] insertion in the ITS1 region.
The divergent position of Disperis (Figs. 1, 3) could be explained by its apparent accelerated substitution rate leading to a long-branch attraction (Felsenstein, 1978
) by the cranichid and diurid outgroups. Additional ITS sequencing among Disperis may help to clarify the occurrence of the molecular rate acceleration and to break up the long branch leading to Disperis capensis and D. lindleyana. Moreover, it is hypothesized that all phylogenetic reconstruction methods provide estimates of tree shape that are more unbalanced than the true tree, especially when rates of evolution are high (Huelsenbeck and Kirkpatrick, 1996
). Such a trend is clear with the ITS data judging by the asymmetry of our MP and ML trees and the long branches for Diuris, Elythranthera, and Disperis (Figs. 1, 3). Paraphyly of Diseae may simply reflect the general behavior of phylogenetic algorithms to produce unbalanced trees rather than the true evolutionary history of the main clades of Diseae and Orchideae.
In support of the patterns obtained here with ITS, studies of three plastid regions, trnL-F, rbcL, and matK, demonstrate the same pattern of relationships for Disperis (Kores, Molvray and Chase, unpublished data), and we suggest that the basal position of Disperis is neither the result of a long-branch problem, nor the consequence of a tree-shape artifact. Likewise, Diseae s.l. are not monophyletic in these plastid trees: Satyriinae are sister to genera in Orchidinae/Habenariinae, but Disinae and Coryciinae are together sister to Satyriinae/Orchidinae/Habenariinae (a relationship supported here by the ML tests) Overall patterns of relationships as estimated by both plastid and nuclear regions are highly congruent, and we anticipate eventually being able to combine these and the morphogical/anatomical data in a single analysis once the sampling has been made more comparable and the remaining critical taxa are added to the molecular matrices.
The isolated position of Brownleeinae
Brownleeinae are monogeneric and have been hypothesized to have originated as a hybrid between species of Disinae and Coryciinae (Linder and Kurzweil, 1994
). Linder and Kurzweil (1994)
could not support the grouping of Brownleea with any confidence. Recognition of the tribe Brownleeinae is rather recent (Linder and Kurzweil, 1994
), and we here included a single representative, Brownleea coerulea. The ITS sequence of the B. coerulea was without any indication of heterogeneity, but a long-established hybrid line may have its initial heterogeneity eliminated by concerted evolution. The constraint of Diseae monophyly (cladogram not shown) and the ML evaluation of alternative hypotheses (Table 2) both indicate that Brownleeinae may be the sister group of Disinae, despite the early divergence of Brownleea coerulea among Diseae/Orchideae in the MP (Fig. 1) and ML trees (Fig. 3). Linder and Kurzweil (1996)
demonstrated the morphological heterogeneity with Brownleea, and ITS sequences from the six remaining Brownleeinae species, at least B. parviflora and B. mulanjiensis, are required to assign better the phylogenetic position of this isolated tribe. Brownleea has not yet been sampled for any plastid regions.
Phylogenetic relationships within Orchideae
Monophyly of Orchidinae is strongly supported by the molecular data (Figs. 1, 3). The phylogeny of the subset of Orchidinae genera included in the present study generally confirms the detailed molecular ITS analyses by Pridgeon et al. (1997)
and Bateman, Pridgeon, and Chase (1997)
. Barlia, Ophrys, Serapias, and Orchis morio cluster in a clade characterized by globose tubers and chromosome numbers 2n = 32 or 36. Gymnadenia, Dactylorhiza, Pseudorchis, Platanthera, and Orchis italica fall into a more weakly supported clade, whereas they form a paraphyletic assemblage in the study by Pridgeon et al. (1997)
. This discrepancy likely arises because of the difference of level of sampling. Pridgeon et al. (1997)
included 88 species with 79 Orchidinae and nine outgroups, whereas the present study included 54 species with nine Orchidinae and 45 outgroups.
Monophyly of Habenariinae appears in our molecular trees but is not well supported (Figs. 1, 3). The position of Brachycorythis (Orchidinae sensu Dressler, 1993
) is different in the MP and ML trees but has no internal support in either position. This result for Brachycorythis was not unanticipated; although the flowers look superficially like those of Orchidinae, it is becoming clear that Orchidinae is a north temperate subtribe with no representation likely in Africa (Pridgeon et al., 1997
). The relationships between taxa of Habenariinae are actually poorly resolved by the ITS data, except for the clustering of Herminium lanceum with Habenaria sagittifera (cf. Pridgeon et al., 1997
) and Habenaria procera with H. arenaria/Bonatea speciosa. The sequences of Holothrix scopularia and Habenaria repens are virtually identical, requiring additional sequencing of other Habenaria repens individuals and other Holothrix species to check for possible problems in sequence determination or species identification (they were not extracted at the same time, so switching of the samples is not likely). This pattern of relationships would be difficult to justify with morphological data.
Many of the other genera now assigned to Habenariinae have at one time or other been considered members of Habenaria. For example, Gennaria, Herminium, and Bonatea have combinations already made in Habenaria. We expected the circumscription of Habenaria to be problematic and so these results do not surprise us. Habenaria is a large cosmopolitan genus and at one time even included species of the distantly related genus Platanthera (Orchidinae). We expect that some species groups may have switched pollinator relationships and thus been segregated by authors relying solely upon floral characters for generic delimitation.
The position of Satyriinae as sister to Orchidinae/Habenariinae has its parallel in the studies of Linder and Kurzweil (in press), in which the placement of the former was ambiguous, with two of three possible arrangements present in their set of minimal length trees. The flowers of this subtribe are nonresupinate, and the column is highly modified, bent so that the base is uppermost to permit the entry of the pollinator from the opposite side of the flower. The presence of two spurs produced by the lip is not like other Diseae and is more similar to spurs in Orchideae, in which a single spur is produced, also by the lip. The column in the flowers of most Habenariinae (Orchideae) is so short that its exact orientation is not possible to determine. The main reason for placement of Satyriinae in Diseae is the presence of the bent anther, but an erect anther would be expected in flowers such as those of Orchideae in which the lip is the lowermost part of the perianth.
The congruence between morphological and ITS data
Numerous studies and cladistic analyses are available for Diseae (e.g., Kurzweil, Linder, and Chesselet, 1991
; Kurzweil et al., 1995
; Johnson, Linder, and Steiner, 1998
). The superimposition of morphological, anatomical, and palynological characterstates on molecular cladograms should allow us to distinguish among characters incongruent with the rest of the data set, characters that support small subsets of taxa, and those that support a large set of taxa (Kurzweil et al., 1995
). The vegetative and reproductive anatomical characters of Kurzweil et al. (1995)
were mapped onto our MP cladograms to reveal potential synapomorphies defining the clades evidenced by the ITS data (Fig. 4).
|
This review of diagnostic characters across Diseae (Fig. 4) indicates broad and general agreement between morphology/anatomy and the ITS data sets. However, the arrangement of vascular strands in tubers was thought to be of great assistance in resolving the phylogenetic questions in Diseae (Kurzweil et al., 1995
). These molecular data validate the scenario of a monostelic (i.e., undissected tuber siphonostele) ancestral condition in Diseae and Orchideae. The "polystelic" condition (i.e., dissected siphonostele) is convergently derived in Disperis lindleyana, Brownleea, Herschelia, and in the Satyrium/Brachycorythis/Holothrix clade. Further studies to elucidate orchidoid phylogeny may focus on dissemination structures, either pollen or seed, following the example of Molvray and Kores (1995)
in determining qualitative and quantitative characters of the diurid seed coat.
The phylogenetic affinities of Diseae relative to other core orchidoids
The tribe Diseae includes 400 species distributed in South Africa, Arabia, Madagascar, Mascarenes, India, China, and Indonesia. The greatest diversity is found in southern Africa, with several endemic genera (e.g., Schizodium). The tribe has been recognized as monophyletic, with all Diseae being characterized by the reflexed anther, thus placing pollinia on the ventral surfaces of pollinators (e.g., Dressler, 1993
; this is actually a single character related to the orientation in which pollen vectors approach these flowers). The major discrepancy between morphology and the ITS tree thus centers on the question of whether or not Diseae are monophyletic. To investigate whether the weakly supported paraphyly of Diseae in the ITS tree could be taken as a rejection of monophyly, we constrained the ITS data to this topology (cladogram not shown). Monophyly of Diseae involves 19 extra steps compared to the MP trees (Fig. 1), which is a significantly less likely topology when compared to the highest-likelihood tree (Fig. 3, Table 2). In the context of Diseae monophyly, Satyriinae emerges first, and then Coryciinae (sistergroup of Disinae) plus Brownleea. The distinctness of Satyriinae relative to other Diseae is thus confirmed, and future studies should explore the evolutionary affinities between Satyriinae, Orchidinae, and Habenariinae. Dressler (1986)
suggested that Disinae and Satyriinae are closer to each other than to Coryciinae because they share brightly colored sepals and a subterminal stigma. These characters may, however, be homoplastic because they are part of the pollinator-attracting/orienting syndrome. The ITS data reject this hypothesis (Table 2) but favor the association between Disinae and Coryciinae. This view agrees with Dressler (1993)
and Kurzweil et al. (1995)
who stressed the horizontally reflexed anther and the galeate dorsal sepal as synapomorphies for Disinae plus Coryciinae. This relationship is also found in our current plastid tree (Kores, Molvray and Chase, unpublished data).
Constraining Diseae monophyly actually involves the rooting of the MP tree (Fig. 1) nearly on the Orchideae internode (cladogram not shown). This alternative rooting of the core orchidoids subtree would neither explain the position of Disperis nor the relationships between Habenariinae. Thus, rooting of the ITS tree with diurids and spiranthoids as outgroups may be inadequate because of the great divergence between these outgroups and the core orchidoids (Fig. 2). Diseae and Orchideae represent a derived and well-characterized clade defined by a long ancestral segment (Kores et al., 1997
), precluding the use of closer outgroups. This problem will hold for phylogenetic studies of core orchidoids whatever the kind of characters used (morphological, anatomical, palynological, chromosomal, and molecular). In the present case, ITS characters appear suitable to reconstruct the phylogeny within the major core orchidoid lineages but reach their limits for distant comparisons in subfamily Orchidoideae.
Conversely, this molecular study supports paraphyly of Diseae (Figs. 1, 3; Table 2), therefore conflicting with all previous interpretations of floral characters. However, analysis of morphological/anatomical data revealed no apomorphic characters in the vegetative anatomy for Diseae (Kurzweil et al., 1995
). Our ITS data could be interpreted to mean that a reflexed anther represents: (1) the ancestral state among core orchidoids and the erect anther a derived state; or (2) a labile character state controlled by pollinator pressures that has arisen multiple times. It is more likely that this scenario is too simple. The column and lip structure in each of the seven clades (Disperis, Brownleeinae, Disinae, Corycium/Pterogodium, Satyriinae, Orchidinae, and Habenariinae) identified here as well as in the morphological cladistic studies as monophyletic are so different from each other that simply describing the column as either bent or erect is exceedingly unsatisfactory and almost certainly inaccurate. When taken in conjuction with the finding of Kurzweil et al. (1995)
that there are no apomorphic vegetative characters for these plants, the best assessment is that there has never been an accurate evaluation of phylogeny in Orchideae/Diseae because there are as yet no synapomorphies identified that could apply to any pair of subtribes other than Disinae/Corycium/Pterogodium, which all share the saccate cavity/spur on their dorsal sepals. Perhaps Orchideae/Diseae are best considered a single tribe.
Perspectives
This overview of molecular phylogeny of Diseae emphasizes the need for additional taxonomic sampling. The genera Schizodium and Pachites will complete the generic representation of Disinae and Satyriinae. Inclusion of the genera Ceratandra and Evotella will help us to reconstruct the phylogeny of Coryciinae, which appears as the keystone subtribe in understanding the evolution of Diseae and Orchideae. Molecular studies should also include the rare Huttonaea, characterized by flowers with spathulate sepals and previously placed in Orchideae (Dressler, 1981
). The monogeneric tribe Huttonaeinae was recently recognized and included within Diseae (Linder and Kurzweil, 1994
) and may represent an essential taxon to establish the exact relationship between Diseae and Orchideae (Dressler, 1993
).
Further studies of core orchidoids using the ITS1 and ITS2 markers could explore their molecular evolution. Evaluation of ITS secondary structures as well may help to align divergent sequences and to characterize the dynamics of indel occurrence. Sequencing a phylogenetic marker distinct from noncoding nuclear DNA for the same Diseae, Orchideae, Diurideae, and spiranthoid taxa would also help to resolve remaining ambiguities. Promising alternative markers may be found in the plastid genome, for example the protein-coding matK, rps4, and rbcL genes, or the noncoding rps4/trnS, trnL-F and atpB/rbcL introns and spacers.
This study of ITS sequences confirms the findings of a large body of previous work on morphology of Diseae (cited in the text) and to a large degree identifies the same groups that were already clear from these studies. Support is high for most of the same groups supported by morphological apomorphies, and the points at which descrepancies exist could be the result of insufficient taxon sampling for ITS. Two major new and discordant patterns appear in the ITS trees: (1) the removal of Disperis from Coryciinae, and (2) the rooting of the basitonic orchids within Diseae rather than between Diseae and Orchideae, thus making Diseae paraphyletic. These two results are similar in that they are the weakest points in our understanding of these plants derived from previous works. The exact position of Disperis has been a problematic issue for morphology to address, and it clearly lacks the pattern of abundant synapomorphies that link Pterygodium and Corycium. The identification of an alternate position for Disperis is intriguing and will be followed up in future studies (some of which, at least the molecular ones, are already underway). The second pattern, the rooting, has never been subject to previous analysis with any form of data, and so it is perhaps not so surprising, especially if the bent anther of these groups, the sole potential synapomorphy of Diseae, is viewed against the diversity of column/lip structures observed. Although the ITS topology itself is not strongly supported along its spine, the alternative is much less parsimonious and also less likely from a ML perspective. The significance of this study then resides, first, in the confirmation by DNA sequence analysis of the results from meticulous and lengthy studies of morphology/anatomy and, second, in the discovery of new patterns that appear to be reasonable alternatives to those previously held and upon which future work can now be focused.
|
FOOTNOTES |
|---|
6 Current address: Oklahoma Natural Heritage Inventory, Oklahoma Biological Survey, 111 E. Chesapeake Street, Norman, OK 73019-0575 USA
7 Author for correspondence (e-mail: douzery@isem.univ-montp2.fr).
This work is dedicated to the memory of botanist Alain Lecointe (1943–1998). We wish to thank James Richardson for supplying material of some taxa, Anette de Bruijn and Jeffrey Joseph for providing technical assistance in the laboratory, and John V. Freudenstein and Ken Cameron for providing helpful comments and criticisms. The first author thanks the EMBO for a short-term fellowship, which was carried out at the Royal Botanic Gardens, Kew, during October - December 1996. ED would also like to thank Serge Aubert, Fred Berger, Philippe Choler, and Jean-Yves Dubuisson for their constant botanical encouragement. This is the contribution number 98-076 of the Institut des Sciences de l'Evolution de Montpellier (UMR 5554 - CNRS).
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Baldwin, B. G., M. J. Sanderson, J. M. Porter, M. F. Wojciechowski, C. 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]
Bateman, R. M., A. M. Pridgeon, and M. W. Chase. 1997 Phylogenetics of subtribe Orchidinae (Orchidoideae, Orchidaceae) based on nuclear ITS sequences. 2. Infrageneric relationships and reclassification to achieve monophyly of Orchis sensu stricto. Lindleyana 12: 113–141.
Bolus, H. 1884 Contributions to South African Botany. Botanical Journal of the Linnean Society 20: 467–488.
———. 1885 Contributions to South African Botany. Botanical Journal of the Linnean Society 22: 65–80.
Bremer, K. 1988 The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795–803.[CrossRef][ISI]
Buckler, E. S., IV, A. Ippolito, and T. P. Holtsford. 1997 The evolution of ribosomal DNA: divergent paralogues and phylogenetic implications. Genetics 145: 821–832.[Abstract]
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., K. G. Cameron, H. G. Hills, and D. Jarrell. 1994 Molecular systematics of the Orchidaceae and other lilioid monocots. In A. Pridgeon [ed.], Proceedings of the 14th World Orchid Conference, 61–73. HMSO, London.
Chesselet, P. C. 1989 Systematic implications of leaf anatomy and palynology in the Disinae and Coryciinae (Orchidaceae). Master's Thesis, University of Cape Town, South Africa.
Cox, A. V., A. M. Pridgeon, V. A. Albert, and M. W. Chase. 1997 Phylogenetics of the slipper orchids (Cypripedioideae: Orchidaceae): nuclear rDNA ITS sequences. Plant Systematics and Evolution 208: 197–223.[CrossRef][ISI]
Doyle, J. J., and J. L. Doyle. 1987 A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin of the Botanical Society of America 19: 11–15.
Dressler, R. L. 1981 The orchids: natural history and classification. Harvard University Press, Cambridge.
———. 1986 Recent advances in orchid phylogeny. Lindleyana 1: 5–20.
———. 1993 Phylogeny and classification of the orchid family. Dioscorides Press, Portland, OR.
Farris, J. S., M. Källersjö, A. G. Kluge, and C. Bult. 1994 Testing significance of incongruence. Cladistics 10: 315–319.[CrossRef][ISI]
Felsenstein, J. 1978 Cases in which parsimony or compatibility methods will be positively misleading. Systematic Zoology 27: 401–410.[CrossRef][ISI]
——— 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791.[CrossRef][ISI]
———. 1993 PHYLIP (Phylogeny Inference Package) version 3.5c. Department of Genetics, University of Washington, Seattle, WA.
