Am. J. Bot. Join the BSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (57)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Asmussen, C. B.
Right arrow Articles by Chase, M. W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Asmussen, C. B.
Right arrow Articles by Chase, M. W.
Agricola
Right arrow Articles by Asmussen, C. B.
Right arrow Articles by Chase, M. W.
(American Journal of Botany. 2001;88:1103-1117.)
© 2001 Botanical Society of America, Inc.

Systematics

Coding and noncoding plastid DNA in palm systematics1

Conny B. Asmussen2 and Mark W. Chase3

Botanical Section, Department of Ecology, Royal Veterinary and Agricultural University, Rolighedsvej 21, DK-1958 Frederiksberg C, Denmark

Received for publication November 12, 1999. Accepted for publication August 15, 2000.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Plastid DNA sequences evolve slowly in palms but show that the family is monophyletic and highly divergent relative to other major monocot clades. It is therefore difficult to place the root within the palms because faster evolving, length-variable sequences cannot be aligned with outgroup monocots, and length-conserved regions have been thought to give too few characters to resolve basal nodes. To solve this problem, we combined 94 ingroup and 24 outgroup sequences from the length-conserved rbcL gene with ingroup and alignable outgroup sequences from noncoding rps16 intron and trnL-trnF regions. The separate rps16 intron and trnL-trnF region contained about the same number of variable sites (autapomorphies not included) as rbcL, but gave higher retention indices and more clades with bootstrap support. In general, the strict consensus tree based on combined rbcL, rps16 intron, and trnL-trnF data showed more resolution towards the base of the palm family than previous hypotheses of relationships of the Arecaceae. An important result was the position of subfamily Calamoideae as sister to the rest of the palms, but this received <50% bootstrap support. Another result of systematic significance was the indication that subfamily Phytelephantoideae is related to two tribes from subfamily Ceroxyloideae, Cyclospatheae and Ceroxyleae.

Key Words: Arecaceae • coding and noncoding plastid DNA • molecular systematics • Palmae • phylogeny • rbcL • rps16 intron • trnL-trnF.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Phylogenetic analyses of molecular data have shown that the palm family is monophyletic and a member of the derived commelinoid monocot clade (Duvall et al., 1993a, b ; Chase et al., 1995a, b, 2000 ). The palms are resolved on a long branch relative to other major clades in the monocots, but within the family itself plastid DNA sequences, both coding and noncoding, evolve slowly compared to other monocot groups (Wilson, Gaut, and Clegg, 1990 ; Gaut et al., 1996 ; Asmussen, 1999a, b ; Baker, Hedderson, and Dransfield, 2000 ).

The shortage of plastid DNA substitutions has been the main argument for choosing the supposedly faster evolving noncoding regions, rps16 intron and trnL-trnF (the intron of the transfer RNA leucine-UAA gene, the trnL exon II, the intergenic spacer between trnL exon II, and the transfer RNA phenylalanine-GAA gene), as molecular markers for resolving relationships within the Arecaceae (Asmussen, Baker, and Dransfield, 2000 ; Baker et al., 2000) . Cladistic analyses of separate rps16 intron and trnL-trnF data sets resulted in trees that were partially resolved, and a combined analysis improved both resolution and support (Asmussen, Baker, and Dransfield, 2000) . These noncoding plastid DNA regions are length-variable, and numerous gaps must be introduced to align the sequences. Within the palms, alignment is straightforward and unambiguous, but if trnL-trnF and to some extent also rps16 intron sequences from outgroup monocots are included, alignment becomes a problem, and many base pairs must be excluded from the analysis due to uncertainty about homology. When these regions are excluded, valuable characters, which vary within the Arecaceae, are also excluded, and the result is too few variable characters to resolve relationships among the palms.

Loss of variable characters is critical because there are relatively few potentially parsimony-informative characters, and outgroup analysis was therefore rejected in some studies of Arecaceae (Asmussen, Baker, and Dransfield, 2000 ; Baker et al., 2000) . Based on a single distantly related monocot outgroup taxon (Dioscorea), Nypa is resolved as sister to the remaining members of the palm family in a cladistic analysis of plastid DNA restriction site data (Uhl et al., 1995 ). This result has been the argument for rooting topologies of phylogenetic analyses for these noncoding regions on Nypa (Asmussen, Baker, and Dransfield, 2000 ; Baker et al., 2000) .

One approach for estimating the root within palms is to sequence a length-conserved coding region, such as the rbcL gene, which can be easily aligned across all green plants (Källersjö et al., 1998 ). The rbcL sequences are then combined with rps16 intron and trnL-trnF sequences, and the nonalignable outgroup rps16 intron and trnL-trnF sequences (the full sequences) were treated as missing data. The expectation is that rbcL and alignable rps16 intron sequences will give information about the root and the combined rbcL, rps16 intron, and trnL-trnF sequences will provide resolution at internal nodes within the palm family, especially towards the root. The addition of sequences from yet another plastid DNA region to the current rps16 intron and trnL-trnF data may also add further resolution and improve support within the palm family. However, caution should be taken when using this approach because the relatively high percentage of missing data among outgroups could have a spurious effect on the topology of the ingroup and the effect of missing data should therefore be evaluated by running an analysis in which taxa with missing data are excluded.

The plastid genome is inherited as a unit and therefore not subject to recombination, which implies that rbcL sequences should be readily combined with rps16 intron and trnL-trnF sequences (Soltis and Soltis, 1998 ). Within the plastid genome the number of variable sites may differ among regions according to function, e.g., coding vs. noncoding DNA, and at a smaller scale according to position, e.g., codon positions of protein-coding genes and stem and loop positions of noncoding DNA (Soltis and Soltis, 1998 ). In general, noncoding DNA (introns and intergenic spacers) have been thought of as more variable than coding regions due to fewer constraints (Taberlet et al., 1991 ; Albert et al., 1994 ; Soltis and Soltis, 1998 ). However, recent studies in which the results of cladistic analyses of coding and noncoding regions are compared show that the variable sites in rbcL change faster than those of the noncoding trnL-trnF region but that trnL-trnF has more variable sites than rbcL (Chase et al., 2000 ; Richardson et al., 2000 ). This might be explained by noncoding regions being constrained by their own secondary structures. Most variation in noncoding regions, especially length variation, is confined to the loops, while variation in protein-coding genes is mostly confined to third codon positions (Albert et al., 1994 ; Chase et al., 1995b ; Chase and Albert, 1998 ). The few and well-marked length mutations in protein-coding genes make alignment straightforward (particularly for rbcL in which only a few length mutations at the 3' end of the exon have been detected), whereas there is no a priori hypothesis of homology for individual nucleotide positions associated with length-variable noncoding DNA, and homology assessment is therefore more complicated (Doyle and Davis, 1998 ). Phylogenetic analyses are dependent on data alignment, and it is therefore important that the evolutionary events, which cause length variation, are recognized and used during the alignment of length-variable sequences.

It is not only the variation measured as the number of variable sites and indels that is important when evaluating a DNA sequence as molecular marker for a group of plants. The usefulness of the variable characters is important too and that can be measured as tree length, retention index (amount of similarity in a data matrix that can be interpreted as synapomorphy on the cladogram), and number of supported clades, particularly those present in each of the trees resulting from individual data sets. The tree length can be important when evaluating how useful a DNA region is for phylogeny reconstruction, because faster evolving sites will result in more steps per variable character (more homoplasy) and therefore longer trees, and these extra steps may add more resolution to the trees (Hillis, 1998 ).

The purpose of this study was: (a) to investigate the position of the root in the palm family based on outgroup comparison, (b) to resolve further the relationships within the family, and (c) to compare the usefulness of coding and noncoding plastid DNA sequences by evaluating the number of parsimony-informative characters, tree lengths, retention indices, and number of supported clades.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 LITERATURE CITED
 
Sampling
Ninety-four palm species were included in the analyses, representing 92 of the 189 currently accepted genera (Dransfield and Uhl, 1998 ), 34 of the 38 subtribes, and all tribes in the classification of Uhl and Dransfield (1987 ; Appendix). The rbcL sequences from 24 monocot outgroups were downloaded from GenBank (http://www.ncbi.nlm.nih.gov) and aligned with the newly produced palm rbcL sequences, and the rps16 intron was sequenced from the 13 most closely related of these 24 outgroups (Appendix). Due to alignment problems the rps16 intron sequences were not included for the more divergent monocot outgroups. Sequences of the trnL-trnF region for monocot outgroups could not be aligned satisfactorily with the palm trnL-trnF sequences due to high levels of insertions/deletions. The rps16 intron sequences for the 11 most distantly related outgroup taxa and the trnL-trnF sequences for all outgroups were treated as missing data in the combined analysis.

DNA extraction, PCR, and nucleotide sequencing
Total genomic DNA was extracted from fresh or silica gel dried plant material using the DNeasyTM Plant Mini Kit (Qiagen, Crawley, West Sussex, UK) or the 2x CTAB method of Doyle and Doyle (1987) followed by purification on cesium chloride/ethidium bromide gradients (1.55 g/mL). All samples were vouchered with herbarium specimens (Appendix).

The rbcL sequences were amplified in two pieces from total genomic DNA using the primer rbcL-1F (5'-ATG TCA CCA CAA ACA GAA AC-3') with rbcL-724R (monocot specific, 5'-TCG CAT GTA CCY GCA GTT GC-3'), and rbcL-636F (5'-GCG TTG GAG AGA TCG TTT CT-3') with rbcL-reverse (5'-TCC TTT TAG TAA AAG ATT GGG CCG AG-3'). All primers are positioned within the rbcL gene except rbcL-reverse, which is in a downstream ribosomal control site (Olmstead et al., 1992 ; Fay, Swensen, and Chase, 1997 ; Fay et al., 1998 ). PCR (polymerase chain reaction) reactions (100 µL) were prepared on ice by combining 72.5 µL ddH2O, 10 µL 10x DNA polymerase buffer, 12 µL 25 µmol/L MgCl2, 2 µL 10 mmol/L each dNTP, 1 µL 0.4% BSA, 1 µL of each primer (100 ng/µL), 0.5 µL 5 u/µL taq DNA polymerase (Promega, Madison, Wisconsin, USA), and template DNA. The reactions were covered with oil and amplifications were conducted on a Perkin-Elmer Cetus DNA Thermal Cycler 480 (Foster City, California, USA) programmed as follows: one cycle at 94°C for 3 min, 30 cycles of 94°C for 1 min, 48°C for 1 min, and 72°C for 1 min and 30 sec, and a final cycle at 72°C for 5 min. The resulting PCR products were checked on a 1% agarose gel with ethidium bromide and purified using the QIAquickTM PCR Purification Kit (Qiagen) with 35% guanidinium chloride ((NH2)2C:NH.HCl). Amplification and sequencing primers for the rps16 intron and trnL-trnF regions were those of Taberlet et al. (1991) and Oxelman, Lidén, and Berglund (1997) and the protocols for amplification followed Asmussen, Baker, and Dransfield (2000) and Baker et al. (2000) .

Purified PCR products were sequenced using the ABI PRISMTM BigDyeTM Terminator Cycle Sequencing Ready Reaction Kit (Perkin-Elmer). The PCR amplification primers were used also as sequencing primers. Cycle-sequencing reactions (10 µL) were prepared on ice by combining 1 µL terminator mix, 3 µL 2.5x cycle-sequencing buffer (200 mmol/L trizma base, 5 mmol/L MgCl2, pH 9.0), 1 µL primer (5 ng/µL), and 5 µL cleaned PCR product. Cycle-sequencing was conducted on a Perkin-Elmer GeneAmp PCR System 9600 machine programmed as follows: 26 cycles of 96°C for 10 sec, 50°C for 5 sec, and 60°C for 4 min.

Cycle-sequencing products were cleaned by ethanol precipitation. One microliter 3 mol/L sodium acetate (NaOAc, pH 4.6), 25 µL 100% ethanol, and the 10 µL cycle-sequencing product were added to an eppendorf tube, vortexed, left on ice for 15 min, and centrifuged at 13 000 rpm for 25 min. The supernatant was discarded, 300 µL 70% ethanol added and the samples centrifuged for 15 min at 13 000 rpm. This step was repeated one more time, after which the samples were left in a drying oven (70°C) for 45 min. The cleaned cycle-sequencing products were analyzed on a PE Applied Biosystems 377 automated DNA sequencer (Perkin Elmer). Each base position in the forward and reverse sequences was checked and assembled using the program SequencherTM 3.0 (Gene Codes Corp, Ann Arbor, Michigan, USA).

Sequence alignment and indel coding
Initial automated alignments of consensus sequences were performed with the MegAlign program (Lasergene software package, DNASTAR Inc., Madison, Wisconsin, USA) and followed by refinement by hand (matrices are available from the first author on request). The alignment of rbcL sequences was straightforward because there was no length variation. For the length-variable rps16 intron and trnL-trnF sequences, indels were identified and coded as present or absent using the following rules: (1) direct repeats of neighboring sequences (one to many base pairs) were added to the matrix as a single event and if the repeat included base pair substitutions the bases of the repeats were excluded from the data matrix because homology of nucleotides could not be assessed; (2) overlapping repeats were coded separately if they could be ascribed to different events; (3) inversions were identified and added to the matrix as a single event and subsequently the bases of the inversion were excluded from the data matrix to avoid introducing multiple characters from what had been determined a single event; (4) long strings of As, Ts, or ATs of different lengths (microsatellites caused by slip-strand mispairing) were excluded because they have been shown to be variable within species (Kelchner and Wendel, 1996 ; Vendramin et al., 1996 ; Kelchner and Clark, 1997 ). The aligned rbcL, rps16 intron, and trnL-trnF sequence matrices (with indel characters added) were analyzed separately, and then the three combined matrices (including indel characters) were analyzed together.

Statistics
Because outgroups were not alignable for the trnL-trnF sequences all topologies resulting from analyses of separate data sets were rooted on the species from subfamily Calamoideae to compare CI (consistency index), RI (retention index), and number of supported clades among the three data sets. Subfamily Calamoideae was chosen as an outgroup from within the Arecaceae because it was positioned as sister to the rest of the family in the analysis of combined rbcL, rps16, and trnL-trnF sequences from palms and outgroups. Tree length, CI, RI, and clade support were calculated and compared for the four data sets (three separate and one combined). MacClade 4.04 (Maddison and Maddison, 1992 ) was used to calculate the distribution of variable sites, number of changes per variable site, and the A + T and C + G content. Transition–transversion ratios and consistency and retention indices for transitions and transversions, respectively, were calculated in PAUP* using a step matrix (Bayer et al., 1999 ). All the statistical comparisons were based on matrices of palm taxa only, i.e., outgroups were excluded, because there were no outgroup sequences for trnL-trnF and 11 of the rps16 intron outgroups.

Cladistic analyses
The separate and combined data sets were analyzed by Fitch parsimony (Fitch, 1971 ; unordered, equally weighted characters) using PAUP* version 4.0b2a (Swofford, 1998 ). Because of zero-length branches resulting from inadequate number of informative characters the separate analyses resulted in many trees, and heuristic searches could not be run to completion. Therefore the following search strategy was used. One thousand replicate searches were conducted using the tree-bisection-reconnection (TBR) branch-swapping algorithm with steepest descent and MULPARS in effect but holding only five trees per step to minimize time spent swapping on suboptimal trees. A round of TBR swapping was performed on the trees collected during the 1000 replicates, up to 5000 trees were collected, and these trees were swapped to completion. Support for clades was calculated by conducting 1000 replicates of the bootstrap with subtree pruning-regrafting (SPR) swapping. No more than 15 trees were saved in each replicate, and only groups that appeared in >50% of the trees were retained.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 LITERATURE CITED
 
Compositions of rbcL, rps16 intron, trnL intron, and trnL-trnF intergenic spacer matrices
The base composition of rbcL was a typical gene profile with almost equal ratios of the bases. The rps16 intron and the trnL and trnF sequences had more As and Ts relative to Cs and Gs, which is typical of noncoding DNA, and all regions had more transitions than transversions (Table 1). Consistency and retention indices were higher for transitions. Most steps on the rbcL tree originated from third codon positions, which also had the highest CI and RI. The rbcL sequences produced more steps per character (3.43 steps per character; max. = 18 steps per character) than the rps16 intron (1.71 steps per character; max. = 5 steps per character) and the trnL-trnF (1.77 steps per character; max. = 7 steps per character) regions (Table 2). Variable sites were more or less evenly distributed along the sequence for all three regions, and no obvious "hot spots" were present.


View this table:
[in this window]
[in a new window]
  		Table 1. Phylogenetic information of rbcL, rps16 intron, and separate trnL intron and trnL-trnF intergenic spacer sequences. The data are based on Fitch parsimony analyses with outgroups excluded. Only characters included in the analyses are included in calculations of transitions and transversions. All characters are included in the calculations of sequence composition (A + T, C + G, and A + T/C + G).

 

View this table:
[in this window]
[in a new window]
  		Table 2. Phylogenetic information of rbcL, rps16 intron, trnL-trnF, and combined sequences. The data are based on Fitch parsimony analyses with outgroups excluded

 
Analysis of rbcL
Only the coding region of rbcL was included in the alignment (1428 base pairs), and the first 30 base pairs were excluded from the analysis because most sequences lacked these positions (primer annealing region). The data matrix thus consisted of 1398 positions, of which 99 (7.08%) were autapomorphic and 92 (6.58%) were potentially parsimony-informative (Table 2). The more than 5000 equally parsimonious trees had 316 steps with a CI of 0.34 (excluding autapomorphies) and a RI of 0.69 (Table 2).

Analysis of rps16 intron
The length of the rps16 intron sequences in palms ranged from 710 (Kerriodoxa elegans) to 958 (Guihaia argyrata) base pairs. This is the entire intron except for the first 22 base pairs (because it is a primer annealing site). The data matrix of palms and 13 monocot outgroups consisted of 1556 positions, 785 of which were included in the analyses. The included characters consisted of 130 (16.56%) autapomorphic and 74 (9.43%) potentially parsimony-informative characters (Table 2). Sixty-five gaps varying from 1 to 143 base pairs in length were introduced in the palm part of the alignment, and indel coding resulted in an additional 11 characters. There were no indels shared among palms and outgroups and indels present in outgroups only were not coded because the homology assessments were complicated and the relationship of outgroups was not the focus of this study. The >5000 equally parsimonious trees had 145 steps with a CI of 0.72 (excluding autapomorphies) and a RI of 0.89 (Table 2).

Analysis of trnL-trnF
The length of the trnL-trnF sequences in palms ranged from 743 (Roystonea oleracea) to 919 base pairs (Iriartea deltoidea and Wettinia hirsuta). The data matrix consisted of 1337 positions, 1003 of which were included in the analysis. There were 110 (10.97%) autapomorphic and 63 (6.28%) potentially parsimony-informative characters (Table 2). Forty-seven gaps varying from 1 to 124 base pairs in length were introduced in the alignment, and indel coding resulted in an additional 11 characters. The 3' exon of trnL was 49 base pairs long in all species except Leopoldinia pulchra, which had a 15 base pair repeat, and there were no potentially parsimony-informative characters within the coding region. The >5000 equally parsimonious trees had 131 steps with a CI of 0.64 and a RI of 0.89 (Table 2).

Analyses of combined palm rbcL, rps16 intron, and trnL-trnF data sets
Fitch parsimony analysis of combined rbcL, rps16 intron, and trnL-trnF sequences also produced >5000 equally parsimonious trees, which had 636 steps, a CI of 0.47, and a RI of 0.76 (Table 2).

Comparison of the four analyses
The strict consensus trees from the three separate analyses all had partially resolved internal nodes (Fig. 1A–C). These relationships are more resolved in the strict consensus tree of combined data (Fig. 1D). Each separate data set yielded too few characters (92, 85, and 74 for rbcL, rps16 intron, and trnL-trnF, respectively) to provide high resolution and support, but the combination of data much improved the resolution and support for clades. The rps16 intron topology included about twice as many clades (29) with bootstrap support as the rbcL data set (15) and also more than the trnL-trnF data set (23). The combined tree performed best (41 supported clades), in particular the number of clades with high bootstrap support (>85%) was improved (16 clades vs. 5, 9, and 7; Table 2). The relative variation (number of potentially parsimony-informative characters per number of base pairs included in the analysis x 100%) were 6.6, 10.8, and 7.4% for rbcL, rps16 intron, and trnL-trnF, respectively, but the tree lengths (316 steps) resulting from analysis of rbcL data were much longer than tree lengths of rps16 intron and trnL-trnF trees (145 and 131, respectively). The high variation of individual characters of rbcL was also reflected in the CI and RI, which were lower for rbcL than for rps16 intron and trnL-trnF (Table 2). The length of the trees resulting from analysis of the combined data sets (636 steps) was longer than the sum of the tree lengths resulting from analyses of the three separate data sets (316 + 145 + 131 = 592 steps). The contribution from each data set to the difference of 44 steps (636 – 592 steps) was calculated by mapping each separate data set onto the combined tree and calculating the tree lengths. The difference between the trees resulting from the separate rbcL analysis and the combined data sets was 24 steps; and the differences between the combined data set and rps16 and trnL-trnF, respectively, were 11 and 9 steps.



View larger version (82K):
[in this window]
[in a new window]
  		Fig. 1. Strict consensus trees of 5000 equally parsimonious trees resulting from Fitch parsimony analyses of separate and combined rbcL, rps16 intron, and trnL-trnF data sets excluding outgroups. Subfamily Calamoideae is designated as outgroup, and bootstrap percentages are given above the branches. (A) rbcL sequences. (B) rps16 intron sequences. (C) trnL-trnF sequences. (D) Combined rbcL, rps16 intron, and trnL-trnF sequences

 
Analyses of combined palm and outgroup rbcL, rps16 intron, and trnL-trnF data sets
The combined matrix had 622 potentially parsimony-informative characters and consisted of rbcL sequences for palms and 24 outgroups, rps16 intron sequences for palms and 13 outgroups, and trnL-trnF sequences for palms only. The >5000 equally parsimonious trees collected in the Fitch analysis were 2325 steps long and had a CI of 0.41 (excluding autapomorphies) and a RI of 0.59. The tree was well resolved and had many well-supported clades (Fig. 2). The palm family was on a long branch relative to other major groups of monocots, but the internal and many terminal branches of the palm family were short relative to those of the outgroups (Fig. 3). The topologies resulting from combined analyses with and without outgroups were identical except for three differences: (1) the polytomy of subfamily Coryphoideae members was more resolved in the analysis including outgroups, in which the clade of Borasseae, Corypha, and Caryoteae was sister to the remaining palms except Calamoideae and Nypa (Figs. 1D and 2); (2) the clade of Nannorrhops, Kerriodoxa, and Chuniophoenix was more resolved in the analysis including outgroups; and (3) the clade of subtribe Thrinacinae pro parte sister to Sabal bermudana was more resolved in the analysis without outgroups (Figs. 1D and 2).



View larger version (46K):
[in this window]
[in a new window]
  		Fig. 2. Strict consensus tree of 5000 equally parsimonious trees resulting from Fitch parsimony analysis of combined palm and outgroup rbcL, rps16 intron, and trnL-trnF sequences. Bootstrap percentages for the clades are given above the branches, and nine clades are labeled for discussion in the text. The classification of Dransfield and Uhl (1998) is indicated to the right

 


View larger version (39K):
[in this window]
[in a new window]
  		Fig. 3. A tree with proportional branch lengths representing one of the 5000 equally parsimonious trees resulting from Fitch analysis of the combined matrix. The classification of Dransfield and Uhl (1998) is indicated

 
The monophyletic subfamily Calamoideae (100% bootstrap support; clade 1) was resolved as sister to the rest of the palms. Nypa fruticans from the monospecific subfamily Nypoideae (labeled 2) was sister to the remaining palms excluding Calamoideae (Fig. 2). However, there was <50% bootstrap support for the position of Calamoideae and Nypa, except in the combined analysis with outgroups excluded in which there was weak bootstrap support (71%) for the clade of all palms excluding Calamoideae and Nypoideae (Fig. 1D). The clade of Borasseae, Caryoteae, and Corypha (clade 3) was supported by 79% bootstrap and resolved as sister to the remaining palms except Calamoideae and Nypa. Tribe Borasseae was monophyletic (98% bootstrap support), and subtribe Coryphinae was nonmonophyletic because one species, Corypha, was sister to the Borasseae and the remaining members of Coryphinae (Nannorrhops, Kerriodoxa, and Chuniophoenix) constituted a well-supported (98% bootstrap) monophyletic group, the position of which was unresolved (clade 5; Fig. 2). The Caryoteae (Arecoideae) was well-supported (97% bootstrap) and monophyletic (Fig. 2). There was a polytomy consisting of four individually well-resolved clades (clade 5, 6, 7, and a monophyletic group consisting of clades 8 and 9) plus one species, Phoenix reclinata (labeled 4). Clades 3, 4, 5, 6, and 7 were species from the currently recognized subfamily Coryphoideae in addition to the Caryoteae (Arecoideae). Clade 6 was well-supported (89% bootstrap) and consisted of subtribe Thrinacinae pro parte and Sabal bermudana (Sabalinae). Clade 7 had 80% bootstrap support and included the remaining members of Thrinacinae (51% bootstrap support) and all Livistoninae (Fig. 2).

There was weak bootstrap support (62%) for the clade comprising subfamilies Ceroxyloideae, Phytelephantoideae, and Arecoideae except the Caryoteae (clades 8 and 9; Fig. 2). The Phytelephantoideae were monophyletic (74% bootstrap support) and constituted a clade (<50% bootstrap support) with the two monophyletic tribes, Cyclospatheae (100% bootstrap support) and Ceroxyleae (98% bootstrap support), of the Ceroxyloideae (clade 8). The remaining large clade (clade 9) was weakly supported (65% bootstrap) and consisted of all species from Arecoideae (except Caryoteae) in addition to tribe Hyophorbeae (Ceroxyloideae). Thirteen subtribes from Arecoideae not previously included in a molecular phylogeny of the palm family were all positioned within this clade (Wettiniinae, Leopoldiniinae, Malortieinae, Dypsidinae, Lemurophoenicinae, Roystoneinae, Cyrtostachydinae, Oncospermatinae, Sclerospermatinae, Masoalinae, Beccariophoenicinae, Elaeidinae, and Bactridinae). Tribe Iriarteeae was monophyletic (99% bootstrap support) and sister to a clade (<50% bootstrap support) of the remaining members of clade 9. Reinhardtia simplex (monogeneric Malortieinae) received <50% bootstrap support as sister to Podococcus barteri (monogeneric Podococceae). Roystonea regia (monogeneric Roystoneinae) was sister to the Hyophorbeae (Ceroxyloideae). Tribe Cocoeae was monophyletic but with <50% bootstrap support. Orania lauterbachiana (monogeneric Oraniinae) was sister (<50% bootstrap support) to a clade of tribe Geonomeae and all subtribes of tribe Areceae except Oraniinae, Malortieinae, Roystoneinae, and Sclerospermatinae. The Geonomeae were sister to subtribe Euterpeinae (Areceae), and Manicaria (monogeneric Manicariinae) and Leopoldinia (monogeneric Leopoldiniinae) constituted a monophyletic group within the core clade of tribe Areceae, but with <50% bootstrap support for either of the relationships. Subtribes Ptychospermatinae and Linospadicinae (Areceae) were each monophyletic (74 and 100% bootstrap support, respectively), and subtribes Iguanurinae, Archontophoenicinae, Arecinae, and Masoalinae were nonmonophyletic (Fig. 2).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
LITERATURE CITED
 
Usefulness of rbcL, rps16 intron, trnL intron, and trnL-trnF intergenic spacer
When the usefulness of the three DNA regions was evaluated, the length of the sequences was not taken into account because the rbcL exon is longer than individual rps16 intron and trnL-trnF sequences, but the alignments of the last two were longer than rbcL. Since the indels were coded and included in the analysis, variable characters did not originate only from base pair substitutions but also from indels, which are dependent on the alignment. The usefulness of each sequence was evaluated in terms of number of variable characters, tree length (homoplasy), RI, number of supported clades, number of supported clades also present in the combined analysis, and number of extra steps required to reconstruct the combined trees from the individual matrices. Analyses of the two noncoding regions, rps16 intron and trnL-trnF, resulted in phylogenies with higher RI and more clades with bootstrap support than rbcL, despite the fact that the noncoding regions produced somewhat fewer potentially parsimony-informative characters. In addition, the trees resulting from separate analyses of rps16 intron and trnL-trnF displayed more supported clades (23 and 19 clades, respectively) than the rbcL tree (14 clades). The CI for the large angiosperm phylogeny based on rbcL is highest for second codon positions (Chase and Albert, 1998 ; Chase and Cox, 1998 ), whereas the Arecaceae had the highest CI for third-codon positions, which is also what Bayer et al. (1999) found for Malvales. The palm rbcL sequences had most steps at third-codon positions and the highest RI for third-codon positions, which is in agreement with the rbcL angiosperm results (Chase and Albert, 1998 ).

The transition/transversion ratio for rbcL (1.15) was lower than that of the phylogeny of angiosperm families based on rbcL (1.72; Chase and Albert, 1998 ). The lower transition/transversion ratios for the two noncoding sequences trnL and trnF (1.07 and 1.28) are in agreement with most studies in which transition/transversion ratios for both coding and noncoding sequences are available (e.g., Cox et al., 1997 ; Pridgeon et al., 1997 ), whereas the transition/transversion ratio for rps16 intron was rather high (1.60). The CI and RI for transitions were generally higher than CI and RI for transversions, which is also what other studies on angiosperm families report (Bayer et al., 1999 ; Chase et al., 2000 ; Richardson et al., 2000 ).

Overall the rps16 intron was the most useful single sequence because the resulting trees had the highest RI and greatest number of supported clades. Not surprisingly, the rps16 intron tree was also the most coincident with the combined tree. The trnL-trnF sequences were almost as useful as the rps16 intron but had fewer potentially parsimony-informative characters and therefore resulted in a less resolved strict consensus tree. The rbcL sequences had more parsimony-informative characters, and each individual character evolved faster than those of the rps16 intron, but the RI and number of supported clades were considerably lower. However, it should be taken into consideration that the results of the rps16 intron analysis were partly related to the fact that a number of highly variable indel-rich regions were unalignable due to uncertainty about homology and therefore excluded from the analysis.

The position of the root in palms
The following discussion of systematics of the palm family is based on the strict consensus tree resulting from Fitch parsimony analysis of combined palm and outgroup rbcL, rps16 intron, and trnL-trnF data sets (Fig. 2). Only clades that did not appear in previous molecular phylogenetic studies of Arecaceae (Uhl et al., 1995 ; Asmussen, Baker, and Dransfield, 2000 ; Baker et al., 2000 ) are discussed in detail.

Monophyly of the Arecaceae has been supported in several of the large monocot and angiosperm phylogenies based on morphology and molecules (Chase et al., 1993, 1995a, b, 2000 ; Linder and Kellogg, 1995 ; Stevenson and Loconte, 1995 ). These studies include a few (5–8) often arbitrarily chosen species from the palm family and consequently do not sample enough clades within the Arecaceae to provide reasonable evidence for the early radiations of the palms. However, it is evident from these large angiosperm and monocot analyses that the Arecaceae are part of the commelinoid group and the closest relatives of the palm family are members of Poales (including Bromeliaceae; Chase et al., 2000) , Commelinales (including Haemodoraceae and Hanguanaceae), Zingiberales, and Dasypogonaceae (including Calectasiaceae; Chase et al., 1993, 1995b, 2000 ; Linder and Kellogg, 1995 ; Stevenson and Loconte, 1995 ; APG, 1998 ). The root of the palm family has never been properly explored in a cladistic context, either based on molecular or morphological, micromorphological, and phytochemical characters. Uhl et al. (1995) included one outgroup taxon, Dioscorea, in their cladistic analysis of 67 palm species and that resulted in Nypa being positioned as sister to the rest of the Arecaceae. The outgroup choice of a single rather distantly related species, Dioscorea, could have affected the ingroup topology because many evolutionary changes, which may be represented by groups between Dioscorea and Arecaceae, were not detected and spurious attractions between Dioscorea and various clades within the palms might have affected the results. Two recent papers on palm family phylogeny use the results of Uhl et al. (1995) and defined Nypa as sister to the rest of the palms because the sequences they used for cladistic analyses could not be aligned with outgroup taxa (Asmussen, Baker, and Dransfield, 2000 ; Baker et al., 2000) . These trees are important contributions to classification within Arecaceae but provide no hypotheses for the position of the root.

The hypothesis for the position of the root within the palms between Nypa and the rest of the Arecaceae is supported by the fact that Nypa possesses character states considered plesiomorphic within the palms and the presence of old fossil records of the species (Uhl et al., 1995 ; Uhl and Dransfield, 1999 ). Nypa, furthermore, shares character states with Calamoideae and Coryphoideae, two subfamilies that are also considered early radiations within the palms (Moore, 1973 ; Moore and Uhl, 1982 ; Dransfield and Uhl, 1990, 1998 ). However, most of the arguments for the position of Nypa as sister to the rest of the palms can equally be used in favor of this same position of the Calamoideae or the Coryphoideae.

In this study, the Calamoideae were resolved as the sister to the rest of the palms, but with <50% bootstrap support. The large number of outgroups representing most major clades of the monocots and special effort to align rps16 intron sequences for as many outgroups as possible were essential in reaching this result, which seems unaffected by the choice of outgroup taxa. In the study of Uhl et al. (1995) the Calamoideae are sister to all remaining palms except Nypa. Nypa was defined as sister to the remaining palms in the studies of Asmussen, Baker, and Dransfield (2000) and Baker et al. (2000) and hereby the node between Nypa and the Calamoideae becomes the root. So these studies do not contribute information about the mutual position of Nypa and the Calamoideae relative to the remaining members of the palm family. However, trnL-trnF sequences were missing for all outgroups in this study, and this might influence the topology of the family, in particular the clades that are positioned near the root. So, in future molecular studies of the palms, it is important to add sequences for which outgroup species can be aligned with ingroup taxa to confirm the position of subfamily Calamoideae as sister to the rest of the palms.

Relationships within palms
The circumscription of subfamily Coryphoideae has been little debated since Moore (1973) described the "coryphoid line," which he divided into three groups, coryphoid, phoenicoid, and borassoid palms (Dransfield, Ferguson, and Uhl, 1990 ). This classification was formalized by Dransfield and Uhl, and the three groups are now the tribes Corypheae, Phoeniceae, and Borasseae (Moore, 1973 ; Dransfield and Uhl, 1986, 1998 ; Uhl and Dransfield, 1987, 1999 ). However, analyses based on molecular data show that the Coryphoideae is nonmonophyletic, and Uhl and Dransfield (1999) suggested that some genera may represent relicts of early radiations of palms and be more closely related to other palms. In this study clades 3, 4, 5, 6, and 7 consisted of taxa from Coryphoideae, but there was <50% bootstrap support along the grade of these clades, and it only required an additional four steps to force the monophyly of the Coryphoideae. Clade 6 of Thrinacinae pro parte and Sabal (monogeneric Sabalinae; 89% bootstrap support) is present in all molecular phylogenies of Arecaceae (Uhl et al., 1995 ; Asmussen, Baker, and Dransfield, 2000 ; Baker et al., 2000 ). Clade 3 (79% bootstrap support) of Borasseae, Corypha, and Caryoteae (Arecoideae) is present in the combined analysis of rps16 intron and trnL-trnF data sets, but with the remaining Coryphinae members (clade 5) positioned as sister to the clade of Borasseae and Corypha (Asmussen, Baker, and Dransfield, 2000) . Clade 7 of Livistoninae and some Thrinacinae (80% bootstrap support) is present in the study of plastid DNA restriction site data and in the combined analysis of rps16 intron and trnL-trnF (Uhl et al. 1995 ; Asmussen, Baker, and Dransfield, 2000 ; Fig. 2).

Phoenix reclinata (monogeneric Phoeniceae) is included in subfamily Coryphoideae by Uhl and Dransfield (1987) , but its relationships within the Coryphoideae have always been problematic (Barrow, 1999a, b ). Most authors have placed Phoenix in subfamily Coryphoideae near Sabalinae and Thrinacinae of the Corypheae (Moore, 1973 ; Uhl and Dransfield, 1987 ; Dransfield and Uhl, 1998 ; Barrow, 1999a, b ), but no molecular phylogenetic analysis has resolved the relationships of tribe Phoeniceae (Uhl et al., 1995 ; Asmussen, Baker, and Dransfield, 2000 ; Baker et al., 2000 ). Phoenix was also unresolved at a near basal polytomy within the palms in the strict consensus trees of rps16, trnL-trnF, and combined data sets (Figs. 1B–D and 2). In the strict consensus tree of rbcL data Phoenix was resolved as sister to Sabal bermudana (Fig. 1A). Furthermore, Phoenix was positioned as sister to Ceroxyloideae, Arecoideae, and Phytelephantoideae in many of the individual trees resulting from analysis of the combined data set (Fig. 3). In summary, there are two hypotheses for the position of Phoenix, either as sister to the clade of Sabal and some Thrinacinae members or as sister to a major clade of Ceroxyloideae, Phytelephantoideae, and Areceae, but these hypotheses need to be evaluated with additional molecular and morphological data.

It has been clear since the study of Uhl et al. (1995) that subfamily Ceroxyloideae as circumscribed by Uhl and Dransfield (1987) is polyphyletic because the Hyophorbeae are nested within subfamily Arecoideae (Asmussen, Baker, and Dransfield, 2000 ; Baker et al., 2000) . However, in the current study the novel evidence of relationships between tribes Cyclospatheae and Ceroxyleae (both Ceroxyloideae) and subfamily Phytelephantoideae (Fig. 2, clade 8) was presented. There was <50% bootstrap support for this relationship, so a consensus on Cyclospatheae, Ceroxyloideae, and Phytelephantoideae will await more molecular data and further studies into morphological, anatomical, and biochemical characters. Martius (1824) put Pseudophoenix and the phytelephantoids together in his classification of palms. Barfod (1991) investigated the position of the Phytelephantoideae and described synapomorphies shared by Phytelephantoideae and Arecoideae, Ceroxyloideae, and Coryphoideae. Subfamily Phytelephantoideae is characterized by many morphological synapomorphies, but the position has been unresolved in previous molecular phylogenies, except for a combined analysis of morphological and plastid DNA restriction site data, in which the Phytelephantoideae are sister to tribe Ceroxyleae (Uhl et al., 1995 ). Moore (1973) included Phytelephantoideae in his "arecoid line" with Ceroxyloideae and Arecoideae. In fact Moore's "arecoid line" is equivalent to the combined clade of Ceroxyloideae, Phytelephantoideae, and Arecoideae in this study (Fig. 2, clades 8 and 9). Moore (1973) also placed Caryoteae (Arecoideae) outside the clade of Ceroxyloideae, Phytelephantoideae, and Arecoideae as the "caryotoid line," and this position outside Arecoideae is congruent with recent molecular results including this study (Moore, 1973 ; Uhl et al., 1995 ; Asmussen, Baker, and Dransfield, 2000 ; Baker et al., 2000 ). The molecular evidence indicates that the Caryoteae does not belong to subfamily Arecoideae, despite a wide range of shared morphological characters, and that the tribe is sister to the Borasseae (Coryphoideae), with which they bear little morphological resemblance.

Tribe Hyophorbeae (Ceroxyloideae) and subfamily Arecoideae (except Caryoteae) constituted clade 9 (65% bootstrap support). The Arecoideae are relatively unresolved and undersampled in previous molecular systematics studies, but here we sampled all tribes (6) and subtribes (22). The addition of another sequence matrix and more taxa from Arecoideae resulted in a more resolved tree, but few clades had high bootstrap support due to the low levels of divergence. The Hyophorbeae were monophyletic (80% bootstrap support), which is consistent with previous studies, but the resolution of Roystonea (not previously included) as sister to Hyophorbeae (<50% bootstrap support) was a novel finding. Subtribe Roystoneinae is monogeneric and although the relationships are not obvious it is considered well placed within Areceae (Uhl and Dransfield, 1987 ; Zona, 1996 ). The indicated sister relationship between Roystoneinae and Hyophorbeae deserves more attention in future studies. The clade of Geonomeae and all subtribes of Areceae except Sclerospermatinae, Roystoneinae, and Malortieinae is new, but received <50% bootstrap support. The species of Geonomeae are placed in their own tribe based on a number of morphological synapomorphies (Uhl and Dransfield, 1987 ; Asmussen, 1999a, b ) and the sister relationship (<50% bootstrap support) of the two South American taxa, Geonomeae and Euterpeinae (Areceae), warrants further investigation. Perhaps the Geonomeae are better placed as a subtribe within tribe Areceae, but before taxonomic changes are made a better supported estimate of relationships within the group of Geonomeae and core Areceae members is needed. It is also important to add more representatives for the large Iguanurinae (27 genera), Oncospermatinae, and Arecinae.

Conclusion
The two noncoding regions rps16 intron and trnL-trnF had somewhat fewer parsimony-informative sites than the rbcL gene, but the hypotheses of relationships based on rps16 intron and trnL-trnF sequences, respectively, received higher retention indices and had more clades with bootstrap support than the hypothesis of relationships based on rbcL sequences, in addition to requiring fewer extra steps to recover the combined trees from the separate data matrices.

The most important new systematic result of this study was the indication that subfamily Calamoideae is sister to the rest of the Arecaceae. This is the first time the position of the root in the palm family has been extensively evaluated with numerous outgroups representing most major monocot groups. In general, the trees based on combined rbcL, rps16 intron, and trnL-trnF data sets showed more resolution along the internal nodes within the palms than previous molecular phylogenies of Arecaceae. Another important result with systematic implications was the indication that subfamily Phytelephantoideae appears to be related to the two subfamily Ceroxyloideae tribes, Cyclospatheae and Ceroxyleae. Finally, in spite of collecting a large amount of data, a robust resolution of higher level relationships within the palms is not achieved. To reach a robust phylogeny, upon which we can base a new classification of the palm family, we need another two or three DNA regions that are alignable with outgroups and are at least as variable as the rps16 intron sequences. It is desirable that one of the additional DNA regions is from the nuclear genome, so the present hypothesis based on plastid DNA can be tested using an independent data set.


View this table:
[in this window]
[in a new window]
  		Appendix. Voucher and data base information for taxa used in this study. Herbarium acronyms are given in parentheses. Sequences downloaded from the EMBL/GenBank/DDBJ databases and used in the study are given with a literature reference. Whenever it was not possible to get identical outgroup species a closely related taxon was chosen, and the first species given is the rbcL sequence and the second species the rps16 intron sequence. There are no outgroup sequences for trnL-trnF and 11 rps16 intron taxa

 

View this table:
[in this window]
[in a new window]
  		Appendix. Continued

 

View this table:
[in this window]
[in a new window]
  		Appendix. Continued

 

   FOOTNOTES
 
1 The authors thank Dr. John Dransfield, Dr. Bill Baker, and Carl Lewis for helpful comments on the manuscript and Jim Clarkson, Jeff Joseph, and Martyn Powell for technical assistance. Dr. Sasha Barrow kindly provided trnL-trnF sequences from some coryphoid palms. This research was supported by a grant from The Carlsberg Foundation to Conny B. Asmussen. Back

2 Author for correspondence (e-mail: con{at}kvl.dk ). Back

3 Jodrell Laboratory, Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3DS, UK. Back


   LITERATURE CITED
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Albert V. A. A. Backlund K. Bremer M. W. Chase J. R. Manhart B. D. Mishler andK. C. Nixon 1994 Functional constraints and rbcL evidence for land plant phylogeny. Annals of the Missouri Botanical Garden 81: 534-567[CrossRef][ISI]

APG (Angiosperm Phylogeny Group). 1998 An ordinal classification for the families of flowering plants. Annals of the Missouri Botanical Garden 85: 531-553[CrossRef][ISI]

Asmussen C. B. 1999a Toward a chloroplast DNA phylogeny of the Geonomeae (Palmae). In A. Henderson and F. Borchsenius [eds.], Evolution, variation, and classification of palms. Memoirs of The New York Botanical Garden 83: 121-129

———. 1999b Relationships of tribe Geonomeae (Arecaceae) based on plastid rps16 DNA sequences. Acta Botanica Venezuelica 22: 65-76

———,W. J. Baker andJ. Dransfield 2000 Phylogeny of the palm family (Arecaceae) based on rps16 intron and trnL-trnF plastid DNA sequences. In K. L. Wilson and D. A. Morrison [eds.], Systematics and evolution of monocots, 525–537. CSIRO Publishing, Collingwood, Victoria, Australia

Baker W. J. C. B. Asmussen S. Barrow J. Dransfield andT. A. Hedderson 2000 A phylogenetic study of the palm family (Palmae) based on chloroplast DNA sequences from the trnL-trnF region. Plant Systematics and Evolution 219: 111-126[CrossRef][ISI]

———,T. A. Hedderson andJ. Dransfield 2000 Molecular phylogenetics of subfamily Calamoideae (Palmae) based on nrDNA ITS and cpDNA rps16 intron sequence data. Molecular Phylogenetics and Evolution 14: 195-217[CrossRef][ISI][Medline]

Barfod A. S. 1991 A monographic study of the subfamily Phytelephantoideae (Arecaceae). Opera Botanica 105: 1-73

Barrow S . 1999a A monograph of Phoenix L. (Palmae: Coryphoideae). Kew Bulletin 53: 513-575[CrossRef]

———. 1999b Systematic studies in Phoenix L. (Palmae: Coryphoideae). In A. Henderson and F. Borchsenius [eds.], Evolution, variation, and classification of palms. Memoirs of the New York Botanical Garden 83: 215-223

Bayer C. M. F. Fay A. Y. D. Bruijn V. Savolainen C. M. Morton K. Kubitzki W. S. Alverson andM. W. Chase 1999 Support for an expanded family concept of Malvaceae within a recircumscribed order Malvales: a combined analysis of plastid atpB and rbcL DNA sequences. Botanical Journal of the Linnean Society 129: 267-303[CrossRef]

Chase M. W. andV. A. Albert 1998 A perspective on the contribution of plastid rbcL DNA sequences to angiosperm phylogenies. In D. E. Soltis, P. S. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants II, 488–507. Kluwer Academic Publishers, Boston, Massachusetts, USA

———,A. Y. de Bruijn G. Reeves A. V. Cox P. J. Rudall M. A. T. Johnson andL. E. Eguiarte 2000 Phylogenetics of Asphodelaceae (Asparagales): an analysis of plastid rbcL and trnL-F DNA sequences. Annals of Botany 86: 935-956[Abstract/Free Full Text]

———, and A. V. Cox 1998 Gene sequences, collaboration and analysis of large data sets. Australian Systematic Botany 11: 215-229[CrossRef][ISI]

———,M. R. Duvall H. G. Hills J. G. Conran A. V. Cox L. E. Eguiarte J. Hartwell M. F. Fay L. R. Caddick K. M. Cameron andS. Hoot 1995a Molecular phylogenetics of Lilianae. In P. J. Rudall, P. J. Cribb, D. F. Cutler, and C. J. Humphries [eds.], Monocotyledons: systematics and evolution, 109–137. Royal Botanic Gardens, Kew, Richmond, Surrey, UK

———,D. E. Soltis P. S. Soltis P. J. Rudall M. F. Fay W. J. Hahn S. Sullivan J. Joseph M. Molvray P. J. Kores T. J. Givnish K. J. Sytsma andJ. C. Pires 2000 Higher-level systematics of the monocotyledons: an assessment of current knowledge and a new classification. In K. L. Wilson and D. A. Morrison [eds.], Systematics and evolution of monocots, 3–16. CSIRO Publishing, Collingwood, Victoria, Australia

———,D. W. Stevenson P. Wilkin andP. J. Rudall 1995b Monocot systematics: a combined analysis. In P. J. Rudall, P. J. Cribb, D. F. Cutler, and C. J. Humphries [eds.], Monocotyledons: systematics and evolution, 685–730. Royal Botanic Gardens, Kew, Richmond, Surrey, UK

———, et al. 1993 Phylogenetics of seed plants: an analysis of nucleotide sequences from the plastid gene rbcL.. Annals of the Missouri Botanical Garden 80: 528-580[CrossRef][ISI]

Cox A. V. A. M. Pridgeon V. A. Albert andM. 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. D. andJ. I. Davis 1998 Homology in molecular phylogenetics: a parsimony perspective. In D. E. Soltis, P. S. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants II, 101–131. Kluwer Academic Publishers, Boston, Massachusetts, USA

———, and J. L. Doyle 1987 A rapid DNA isolation procedure from small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11-15

Dransfield J. I. K. Ferguson andN. W. Uhl 1990 The coryphoid palms: patterns of variation and evolution. Annals of the Missouri Botanical Garden 77: 802-815[CrossRef][ISI]

———, and N. W. Uhl 1986 An outline of a classification of palms. Principes 30: 3-11

———, and ———. 1990 The coryphoid palms: patterns of variation and evolution. Annals of the Missouri Botanical Garden 77: 802-815

———, and ———. 1998 Palmae. In K. Kubitzki [ed.], The families and genera of vascular plants, vol. IV, Flowering plants: monocotyledons, 306–389. Springer-Verlag, Berlin, Germany

Duvall M. R. M. T. Clegg M. W. Chase W. D. Clark W. J. Kress H. G. Hills L. E. Eguiarte J. F. Smith B. S. Gaut E. A. Zimmer andG. H. Learn 1993a Phylogenetic hypotheses for the monocotyledons constructed from rbcL sequences data. Annals of the Missouri Botanical Garden 80: 607-619[CrossRef][ISI]

———,G. H. Learn L. E. Eguiarte andM. T. Clegg 1993b Phylogenetic analysis of rbcL sequences identifies Acorus calamus as the primal extant monocotyledon. Proceedings of the National Academy of Sciences, USA 90: 4641-4644[Abstract/Free Full Text]

Fay M. F. C. Bayer W. S. Alverson A. Y. de Bruijn andM. W. Chase 1998 Plastid rbcL sequence data indicate a close affinity between Diegodendron and Bixa.. Taxon 47: 43-50[CrossRef][ISI]

———,S. M. Swensen andM. W. Chase 1997 Taxonomic affinities of Medusagyne oppositifolia (Medusagynaceae). Kew Bulletin 52: 111-120[CrossRef]

Fitch W. M. 1971 Toward defining the course of evolution: minimal change for a specific tree topology. Systematic Zoology 20: 406-416[CrossRef][ISI]

Gaut B. S. B. R. Morton B. C. McCaig andM. T. Clegg 1996 Substitution rate comparisons between grasses and palms: synonymous rate differences at the nuclear gene Adh parallel rate differences at the plastid gene rbcL.. Proceedings of the National Academy of Sciences, USA 93: 10 274-10 279[Abstract/Free Full Text]

Hillis D. M. 1998 Taxonomic sampling, phylogenetic accuracy, and investigator bias. Systematic Biology 47: 3-8

Hiratsuka J. H. Shimada R. Whittier T. Ishibashi M. Sakamoto M. Mori C. Kondo Y. Honji C.-R. Sun B.-Y. Meng Y.-Q. Li A. Kanno Y. Nishizawa A. Hirai K. Shinozaki andM. Sugiura 1989 The complete sequence of the rice (Oryza sativa) chloroplast genome: intermolecular recombination between distinct tRNA genes accounts for a major plastid DNA inversion during the evolution of the cereals. Molecular and General Genetics 217: 185-194

Källersjö M. J. S. Farris M. W. Chase B. Bremer M. F. Fay C. J. Humphries G. Petersen O. Seberg andK. Bremer 1998 Simultaneous parsimony jackknife analysis of 2538 rbcL DNA sequences reveals support for major clades of green plants, land plants, seed plants and flowering plants. Plant Systematics and Evolution 213: 259-287[CrossRef][ISI]

Kelchner S. A. andL. G. Clark 1997 Molecular evolution and phylogenetic utility of the chloroplast rpl16 intron in Chusquea and the Bambusoideae (Poaceae). Molecular Phylogenetics and Evolution 8: 385-397[CrossRef][ISI][Medline]

———, and J. F. Wendel 1996 Hairpins create minute inversions in non-coding regions of chloroplast DNA. Current Genetics 30: 259-262[CrossRef][ISI][Medline]

Linder H. P. andE. A. Kellogg 1995 Phylogenetic patterns in the commelinid clade. In P. J. Rudall, P. J. Cribb, D. F. Cutler, and C. J. Humphries [eds.], Monocotyledons: systematics and evolution, 473–496. Royal Botanic Gardens, Kew, Richmond, Surrey, UK

Maddison W. P. andD. R. Maddison 1992 MacClade version 3.04. Sinauer, Sunderland, Massachusetts, USA

Martius C. F. P. von. 1824 Palmarum familia ejusque genera. Lindauer, Munich, Germany

Moore H. E. 1973 The major groups of palms and their distribution. Gentes Herbarum 11: 27-141

———, and N. W. Uhl 1982 Major trends of evolution in palms. Botanical Review 48: 1-69

Nishizawa Y. andA. Hirai 1987 Nucleotide sequence and expression of the gene for the large subunit of rice ribulose 1,5-biphosphate carboxylase. Japanese Journal of Genetics 62: 389-395[CrossRef][ISI]

Olmstead R. G. H. J. Michaels K. M. Scott andJ. D. Palmer 1992 Monophyly of the Asteridae and the identification of their major lineages inferred from DNA sequences of rbcL.. Annals of the Missouri Botanical Garden 79: 249-265

Oxelman B. M. Lidén andD. Berglund 1997 Chloroplast rps16 intron phylogeny of the tribe Sileneae (Caryophyllaceae). Plant Systematics and Evolution 206: 393-410[CrossRef][ISI]

Pridgeon A. M. R. M. Bateman A. V. Cox J. R. Hapeman andM. W. Chase 1997 Phylogenetics of subtribe Orchidinae (Orchidoideae, Orchidaceae) based on nuclear ITS sequences. 1. Intergeneric relationships and polyphyly of Orchis sensu lato. Lindleyana 12: 89-109

Richardson J. E. M. F. Fay Q. C. B. Cronk D. Bowman andM. W. Chase 2000 A molecular phylogenetic analysis of Rhamnaceae using rbcL and trnL-F plastid DNA sequences. American Journal of Botany 87: 1309-1324[Abstract/Free Full Text]

Smith J. F. W. J. Kress andE. A. Zimmer 1993 Phylogenetic analysis of the Zingiberales based on rbcL sequences. Annals of the Missouri Botanical Garden 80: 620-630[CrossRef][ISI]

Soltis D. E. andP. S. Soltis 1998 Choosing an approach and an appropriate gene for phylogenetic analysis. In D. E. Soltis, P. S. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants II, 1–42. Kluwer Academic Publishers, Boston, Massachusetts, USA

Stevenson D. W. andH. Loconte 1995 Cladistic analysis of monocot families. In P. J. Rudall, P. J. Cribb, D. F. Cutler, and C. J. Humphries [eds.], Monocotyledons: systematics and evolution, 543–578. Royal Botanic Gardens, Kew, Richmond, Surrey, UK

Swofford D. L. 1998 PAUP*: phylogenetic analysis using parsimony (*and other methods), version 4. Sinauer, Sunderland, Massachusetts, USA

Taberlet P. L. Gielly G. Pautou andJ. Bouvet 1991 Universal primers for amplification of three non-coding regions of chloroplast DNA. Plant Molecular Biology 17: 1105-1109[CrossRef][ISI][Medline]

Uhl N. W. andJ. Dransfield 1987 Genera Palmarum: a classification of palms based on the work of H. E. Moore, Jr. International Palm Society and L. H. Bailey Hortorium, Lawrence, Kansas, USA

———, and ———. 1999 Genera Palmarum after ten years. In A. Henderson and F. Borchsenius [eds.], Evolution, variation, and classification of palms. Memoirs of the New York Botanical Garden 83: 245-253

———, ———, J. I. Davis M. A. Luckow K. S. Hansen andJ. J. Doyle 1995 Phylogenetic relationships among palms: cladistic analyses