Conservation patterns in angiosperm rDNA ITS2 sequences
Conservation patterns in angiosperm rDNA ITS2 sequences
Mark A.
Hershkovitz*
and
Elizabeth A.
Zimmer
Laboratory of Molecular Systematics, MRC534, Smithsonian Institution,
Washington
, DC 20560,
USA
Received December 18, 1995;
Revised and Accepted June 18, 1996
ABSTRACT
The two internal transcribed spacers (ITS1 and ITS2) of nuclear ribosomal DNA
have become commonly exploited sources of informative variation for
interspecific-/intergeneric-level phylogenetic analyses among angiosperms and other eukaryotes.
We present an alignment in which one-third to one-half of the ITS2 sequence is alignable above the family level in
angiosperms and a phenetic analysis showing that ITS2 contains information
sufficient to diagnose lineages at several hierarchical levels. Base
compositional analysis shows that angiosperm ITS2 is inherently GC-rich, and that the proportion of T is much more variable than that for
other bases. We propose a general model of angiosperm ITS2 secondary structure
that shows common pairing relationships for most of the conserved sequence
tracts. Variations in our secondary structure predictions for sequences from
different taxa indicate that compensatory mutation is not limited to paired
positions.
INTRODUCTION
The two internal transcribed spacers (ITS1 and ITS2) of nuclear ribosomal DNA
(rDNA) have emerged from the status of `alternative' phylogenetic marker (
1
) to become one of the more widely applied DNA sequences in angiosperm molecular
systematics (reviewed in
2
). At this writing, GenBank contains ITS sequences from ~800 angiosperm species, a figure likely to increase dramatically over the
next few years. Likewise, ITS is increasingly used in lower-level phylogenetics in other organisms, e.g. fungi (
3
), algae (
4
,
5
) and insects (
6
-
9
).
These spacer sequences have been reputed to contain inadequate signal for
phylogenetic analyses at the interfamilial and deeper levels (
1
). Notwithstanding, sequence motifs conserved throughout angiosperms have been
recognized (
10
), and Hershkovitz and Lewis (
11
) have demonstrated that ITS2 has retained sufficient phylogenetic signal to
discriminate among sequences from green algae, fungi and seed plants (conifers
and angiosperms). The purpose of the present paper is to document ITS2 sequence
conservation among angiosperms, and to consider the implications for
phylogenetic inference, secondary structure determination, functional
constraints and molecular evolution. ITS1 will be considered in a separate
paper (Hershkovitz and Zimmer, in preparation) because it is structurally and
functionally unrelated to ITS2 and exhibits a more complex pattern of sequence
conservation.
MATERIALS AND METHODS
Taxon sampling
In order to evaluate ITS2 sequence/structural conservation and the degree of
phylogenetic signal at familial levels and deeper, we compared sequences from
75 species representing available phylogenetic diversity among angiosperms and
including multiple members of several well-represented orders and families (Table
1
;
12
-
17
). Sequences reported for
Mimulus
(Scrophulariaceae;
18
) were examined but determined to be green algal in origin (
11
). Sequences from
Arceuthobium
(Viscaceae) were not included because, like other DNA sequences from parasitic
plants, they were exceptionally divergent from those of angiosperms in general
(
19
). The
Arceuthobium
sequences do not appear to be fungal or algal contaminants, however; when
included in parsimony and distance analyses of eukaryotic 5.8S sequences (data
not shown), they emerge as a highly divergent branch among the seed plants.
DNA extraction, amplification and sequencing
Sequences for members of the angiosperm order Caryophyllales were determined as
part of an ongoing analysis of relationships within the order. The ITS regions
were amplified from CTAB-extracted (reviewed in
20
) genomic DNA using the primer pairs: (i) ITS4 (
21
) and ITS5 modified for plants (GGAAGGAGAAGTCGTAACAAGG); or (ii) 26A (
22
) and Nnc18S10 (bases 4-21 of the modified ITS5). Cloned or direct amplification products were
sequenced in both directions using either manual or automated protocols. For
the manual protocol, single-stranded DNA (both coding and non-coding) was derived via asymmetric amplification of 2 [mu]l crude double-stranded amplification product using the ITS4 or N18L18
(AAGTCGTAACAAGGTTTC) primer. 7-deaza-dGTP was substituted for dGTP in order to relax secondary structure
in the single-stranded sequencing template. Asymmetric amplification products were
purified according to the Gene Clean protocol (Bio 101, Inc.) and sequenced
with the primers ITS3 (
21
), C58S (
22
), ITS4 and N18L18 following the Sequenase version 2.0 protocol (US
Biochemicals) for the 7-deaza-dGTP reagent kit and
35
S-dATP labeling. Sequencing reactions were electrophoresed and the gels
exposed to radiographic film according to standard protocols. Automated
sequencing used the same primer set and followed the dye-terminator cycle-sequencing protocol for the ABI model 373A sequencer (Applied
Biosystems, Inc.). Chromatograms were analyzed using Sequencher (Gene Codes,
Inc.). For apparently polymorphic taxa, double-stranded ITS amplification product was ligated and cloned following the
T/A cloning protocol (Invitrogen, Inc.) and double-stranded products amplified directly from recombinant colonies were sequenced as described above.
The ITS region of CTAB-extracted
Ravenala
madagascariensis
genomic DNA was amplified with restriction-site-tagged primers CLOA (
Kpn
I-26A) and CLOE (
Xba
I-N18L18) and the purified product ligated into doubly-digested pBluescript II KS
+
phagemid vector (Stratagene, Inc.), transformed into
Escherichia
coli
XL1B cells (Stratagene, Inc.) rendered heat-shock competent (
23
). Transformations were cultured on LB/X-gal/IPTG/ampicillin agar plates and white colonies cultured in
LB/ampicillin broth (
23
). The cloned vector was purified following the RPM Rapid Pure miniprep protocol
(Bio 101, Inc.), and the insert amplified and sequenced following the automated
sequencing protocol described above.
Sequence alignment
A taxonomically representative subset of the ITS2 sequences was aligned manually
with the aid of the GDE version 2.2 (
24
) multiple sequence editor. Regions of conserved versus variable sequence were
approximated by eye. Similarities among all angiosperms or recognized major
groups therein were considered as evidence for conservation.
Analysis
The objective of this analysis was to evaluate, on the basis of alignability,
the degree to which assorted angiosperm ITS2 sequences clustered according to
independently derived phylogenetic groupings. Alignability was estimated using the `guide-tree' feature in CLUSTAL W (
25
), which yields a neighbor-joining tree based on similarity scores for pairs of sequences aligned
optimally for given gap-opening and gap-extension penalties. The tree is not a phylogram, but a phenogram.
Nominally, the guide tree topology functions as a `phylogenetic template' for
successive sequence alignment, but it also provides a means for comparing,
without imposition of a fixed multiple alignment, all base positions in a set
of highly divergent and length-variable (i.e. poorly alignable) sequences. The most similar sequences
form terminal bifurcations in the tree, and the internal branches cluster
mutually more similar sequences. The branch lengths, however, are distance-distorted relative to evolutionary divergence. For example, an inferred
optimal multiple alignment of the sequence tracts AGAGAA, AGGAA and AAGAA might
not maintain the optimal pairwise alignments. Each shorter sequence differs
from the longer by a single insertion/deletion (indel), but not at the same
position. This implies that the shorter sequences differ from each other by two
indels rather than a single substitution. The guide tree branch lengths,
however, reflect the optimal pairwise alignments, even if these are
evolutionarily illogical. Nonetheless, the procedure will cluster sequences
sharing alignable motifs relative to those that do not. A guide tree generated
using randomly-generated sequences of comparable, and comparably variable, length would
be starlike, lacking significant internal branching. We generated trees using
the `slow, accurate' algorithm, and high (15-50) gap-opening penalties in order to detect conserved motifs combined with low (0-1) gap-extension penalties in order to extend unconserved sequence
as necessary to accommodate conserved motif alignment.
CLUSTAL W does not output the pairwise percent similarity scores to a data file,
but these can be recovered (rounded to two decimal places) by capturing the
screen output as text, which can then be transferred to a word processing or
spreadsheet program.
Base composition of aligned conserved versus variable regions was calculated
using PAUP* (
26
).
Secondary structure prediction
Secondary structure was explored using the minimum free-energy (MFE) program MFOLD (
27
) in GCG version 8.1 (
28
). Unlike its predecessor (FOLD, see
27
), MFOLD provides all suboptimal foldings up to a user-specified free-energy limit. Sequences including up to 50 bases of each of the
flanking coding regions were folded at 25, 37 and 42oC. Structures within 5.7 kcal/mol of the optimal structure and differing
from each other by at least three window units were recovered using the CONNECT
option in PLOTFOLD (
28
), visualized using the Olsen format in LoopDLoop version 1.2a64 (
29
), and redrawn to emphasize similarities using CANVAS (Deneba Software, Inc.).
From among the multiple foldings generated for each sequence, a set of shared
structural features was inferred (see Results), and the sequences were
resubmitted to MFOLD constraining each as necessary to include the complete
set. All sequences were constrained, e.g. to force specified pairing relations
between the 5'-end of the 26S and 3'-end of the 5.8S regions. Only simple, canonical base
pairings, including G[middot]U, were considered.
RESULTS
Angiosperm ITS2 sequence alignment
Figure
1
identifies six regions of ITS2 (c1-c6) that are conserved in all sampled angiosperms or at least in major
groups therein. The six alternating variable regions (v1-v6) are, except among obviously similar sequences, arbitrarily aligned to
the longest sequence. Although c1 is variable across angiosperms, a 5' RYYR motif, especially ATCG, is common among higher dicots or `eudicots'
(
30
,
31
).
Canella
and monocots differ from eudicots at the 5'-end, notably in sharing a Y at the first position, but all
angiosperms share a C-rich 3'-end. The G (or A) at position 7 of some taxa, including
Caryophyllales, might align better with the G at position 10 of other
angiosperms. This would maintain Y-richness in the 3'-end of region c1. Region v1 becomes C- to G-rich moving 5' to 3', but is highly length variable (13-41 bases).
ITS2 downstream of c4 is highly variable in length (67-111 bases) and sequence, but two common elements occur. Our recognition
of c5 is based in part on its pairing with c4 in our secondary structure models
(see below). Alternative alignments of a given sequence in c5 might seem
equally likely, but most sequences contain an RYRYYRYRY motif in this region.
Region c6 is only tentatively designated because of its small size, the
occurrence in several sequences of an additional ACCC, RCCC or AYYY upstream or
downstream of the aligned motif, and its variable pairing relations in our
secondary structure models (see below). Establishing that the indicated motif
is conserved will require additional sampling and evaluation at lower
phylogenetic levels.
Despite poor alignability over all angiosperms, regions v1-v6 show conservation above the traditional familial level. For example,
v4 aligned in the sequences representing Portulacaceae (CIST_TWEE), Cactaceae
(MAIH_POEP), Didiereaceae (ALLU_PROC) and Basellaceae (ANRE_CORD; cf.
1
), and the 3'-end of v6 is more similar among Caryophyllales (Table
1
) than among angiosperms as a whole.
Phenetic analysis
Figure
2
illustrates a guide tree generated with gap-opening/-extension penalties of 17.5/0.2. Adjusting the parameters away from
these values yielded more starlike trees.
Base compositional patterns
High GC content among angiosperm ITS sequences has been noted previously (
2
). Figure
3
(data in Appendix 1) shows that the conserved regions are inherently GC-rich. GC content in the variable regions appears correlated with that in
the conserved regions but is more extreme and more erratic (Fig.
3
). In 24/30 samples, GC content is lower in the variable regions than in the
conserved.
Secondary structure
Preliminary inputs into MFOLD yielded up to 16 secondary structures for each of
the nine analyzed taxa. These varied markedly in the disposition of the coding
and conserved ITS2 regions. In order to infer a `consensus' structure, the
substructural disposition of the conserved regions were scored in all
structures generated for all taxa. The criteria for deriving a set of mutually
compatible consensus substructural features were, in order of preference: (i)
common presence among all structures generated for all taxa; (ii) presence in
at least one of the structures generated for each taxon; and (iii) presence in
at least one of the structures for a majority of taxa. The consensus features
are as follows: (i) pairing of the 3'-end of the 5.8S sequence with the 5'-end of the 26S sequence, as inferred experimentally in
yeast (
37
) and believed common to eukaryotes (
11
); (ii) pairing of the 3'-end of c1 with the 5'-end of c2; (iii) pairing of the 3'-end of c2 with the 5'-end of c3; (iv) a long stem
structure with c4 subterminal along the 5' flank; and (v) the 5'-end of c4 pairing with c5. All of these features co-occurred in only one MFOLD-generated structure, that for tomato
(LYCO_ESCU). At least one of the MFOLD-generated structures for each taxon included three or four of the five
substructural features, and we found that manual adjustments easily yielded the
remainder. Thus, sequences were resubmitted to MFOLD using constraints for
plausible base-pairing relations that would generate all five consensus substructural
features.
Figure
4
illustrates nine putative angiosperm ITS2 secondary structures. The pairing
relations of c1-5' and c6 vary. The former includes pairings with the ITS2-3' end (v6) in Figure
4
C, G and I. Figure
4
B and E includes c2-c6 pairing, but in Figure
4
E, this c2 tract also shows pairing potential with the unpaired region of c3.
Figure
4
E is the only structure lacking a stem in v6.
DISCUSSION
The present results reveal that ITS2 sequence is more conserved among
angiosperms than previously appreciated. Regions conserved at least among
eudicots averaged 47% of the ITS2 length (conserved/total bases; Appendix 1),
whereas 40% (c2-c4/total bases) is alignable across angiosperms. The previously-detected conserved motifs (
10
) included only ~10% of ITS2 length. Some variable-sequence regions exhibit structural conservation, e.g. the C- to G-rich transition in v1. At the same time, the present
results do not challenge the notion that ITS spacers are hypervariable relative
to other molecules. For example, the 5.8S sequence is essentially completely
alignable and typically only 10% divergent across angiosperms (Hershkovitz and
Lewis, submitted).
While we do not suggest that ITS is optimal or even suitable for angiosperm-wide phylogenetics, the present results do indicate that it contains
sequence tracts diagnostic at more, and more inclusive, phylogenetic levels
than previously demonstrated (cf.
2
). The clustering of higher monocots, eudicots and Caryophyllales in Figure
2
suggests not only that conserved, diagnostic motifs exist, but also that ITS2
is not mutationally saturated at such divergence levels. Thus, ITS sequences
might provide phylogenetic evidence auxiliary to that from other genes, and at
a relatively small cost considering its short length.
The principal shortcoming of ITS for deeper-level investigation is the lack of a rigorous, adequately quantifiable
analytical method applicable to highly length-variable sequence regions. The best developed methods (parsimony, minimum
evolution, maximum likelihood;
38
) all rely on
a priori
assumed homology of aligned sequence positions. The ITS2 data are not amenable
to such methods partly because of the paucity of phylogenetically informative
sites in the most confidently alignable portion of the conserved regions. The
guide tree method circumvents the requirement for
a priori
alignment, but it is a clustering rather than phylogenetic method. Moreover, it
yields no overall statistic for comparison to alternative trees, the data from
different genes cannot be directly combined in a single analysis, and the
robustness of individual groupings cannot be evaluated using such techniques as
bootstrap or decay analyses (cf.
38
). The value of the guide tree approach, however, is in its unique ability to
exploit alignability information in a rapid sequence similarity comparison.
This value might be optimized by the incorporation of corrections for
empirically-estimated substitution and base compositional biases and dynamic gap-opening and extension weights based on sequence divergence.
We also deferred conventional analyses of the ITS2 data because of the sparse
and sporadic sampling, which in turn affected confidence in the alignment. In
fact, inadequate sampling can obfuscate any phylogenetic method using any data
source (
39
). For example, we noted that the guide tree did not correctly cluster subf.
Asteroideae (the two Heliantheae plus
Senecio
) among the 10 Asteraceae sequences; indeed, neither did parsimony analysis of
seven Asteraceae
rbc
L sequences (
40
). The
rbc
L taxon sampling was smaller, but the sequence is ca. seven times longer than
ITS2 and perfectly alignable. Therefore, the guide tree method might have
performed better with broader sampling from among the ~20 000 Asteraceae species (
12
). Thus, while phylogenetically unsatisfactory guide tree results for members of
singly-represented groups, e.g.
Cucurbita
and
Betula
, might result from phylogenetic signal loss via mutational saturation or
convergence, they also simply may represent inadequate sampling.
For analyses at interspecific/intergeneric levels for which ITS sequences are
typically exploited, the elucidation of conservation patterns will facilitate
weighting and optimization of character-state reconstruction. Here, DNA sequence data provide a relatively large
number of characters and discrete character states, but they have not afforded
much `circumspection' for phylogenetic analysis, i.e., interpretability in view
of ample data from the broader group of interest and in the context of obvious
and potentially correlated developmental, physiological and/or environmental
variables. Such potential for circumspection is especially limiting for ITS
data given their high divergence, sporadic sampling, and the lack of even such
obvious variability correlates as codon position. While these problems can be
mitigated using techniques for estimating appropriate stochastic models of
sequence evolution (e.g. Kimura-2-parameter; summarized in
38
), empirically-observed patterns such as those elucidated here can be used to examine
site-specific constraints on sequence evolution, e.g. sites where C-T transitions are less likely to occur than at others.
The utility of the generalized model of angiosperm ITS2 secondary structure is 3-fold: (i) it provides a means for evaluating evidence for correlates of
RNA sequence evolution, e.g. compensatory mutation; (ii) it provides an
empirically testable model for functional analysis of the angiosperm ITS2; and
(iii) it underscores the limitations of predicting RNA secondary structure
using computer programs and single sequences. We will consider these points in
turn.
Compensatory mutations, or secondary mutations that maintain RNA base-pairing relations consequential to an initial mutation, are regarded as
correlated characters that should be downweighted in phylogenetic analyses (
41
). Differences among our (admittedly unproved) ITS2 models evidence structural
evolution not involving paired-position compensation, e.g. in the c4-c5 stem, which consistently recurred among MFOLD structures of
different taxa, but which varied in specific stem/bulge pattern and relative
disposition of the conserved motifs. In interspecific-level comparisons, Baldwin
et al
. (
2
,
42
) reported that mutations in
Calycadenia
(Asteraceae) also showed no evidence of paired-position compensation in the putative ITS2 secondary structure. These
results suggested that either compensatory mutation was not a significant
evolutionary force, or that compensation had involved mutations at non-paired positions (`cryptic [mutational] non-independence'), or that the secondary structural models were
incorrect (
2
). Olsthoorn
et al
. (
43
) provided evidence for compensatory mutation at non-paired positions (i.e. cryptic non-independence) during
in vivo
viral RNA evolution.
There are no clearly conserved primary ITS2 sequence motifs shared between
angiosperms, green algae and yeast (
11
). Likewise, our predicted angiosperm ITS2 secondary structures differ from
those experimentally deduced in yeasts (
37
,
44
) and proposed for green algae (
4
), as well as mosquitoes (
7
,
45
). In particular, the ITS2-3' sequence, highly variable in angiosperms, is conserved in yeasts,
and it pairs with a 5' conserved sequence structurally analogous to the angiosperm c2 region.
The yeast model, however, does include a long stem with a distal conserved
pairing region, analogous to the stem formed between c3 and c6 in the
angiosperm model and the c4-c5 pairing. Functional analyses of ITS2 in yeasts indicate that
evolutionarily conserved secondary structural motifs are critical for rDNA
processing (
44
). Thus, the evidence for secondary structural evolution suggests that the rDNA
processing mechanism has also been phylogenetically labile.
Our attempt to reconcile secondary structure with angiosperm-wide ITS2 sequence conservation illustrates the problem inherent in
inferring secondary structure from a single MFE folding of a single sequence.
For example, our models differ markedly from Yeh and Lee's (
37
) computer-generated MFE folding of mung bean (
Vigna radiata
) ITS2, which they characterized as `surprisingly similar' to their
biochemically-deduced yeast model. The putative similarity (and others claimed elsewhere
among ITS2 secondary structures) is an illusory effect of comparing `Squiggle'
diagrams (
28
), which force foldings to conform to limited geometric shapes and condense
interior bulges of any length to fit within the fixed length of a single paired-base space. Yeh and Lee's (
37
) mung bean model pairs c4 and c5, but it also pairs the conserved ITS2 5'-end (including c1-c2) with the hypervariable 3'-end. We submitted to MFOLD the
Vigna radiata
sequence, but did not recover a structure similar to Yeh and Lee's model.
Baldwin
et al
.'s computer-generated MFE folding of
Calycadenia
(Asteraceae;
2
,
42
) is similar to our model for rice (Fig.
4
G) in pairing c1 with c2, c2 with c3, c4 with c5, and c6 with c2. The pairing
indicated in
Calycadenia
between the c1-5' and v6-3', however, is not reflected in any of our models. We
found that the
Calycadenia
sequence can be folded readily to include all of the consensus substructural
elements, and the resulting structure (not shown) closely resembles our tomato
model (Fig.
4
A). While we found the MFE criterion to be useful for developing our model, in
only one case did we adopt a computer-generated MFE structure, and we found that multiple and often radically
different structures have similar minimum free-energy (cf.
2
). The limitation of MFE as a sole criterion in secondary structure prediction (
46
,
47
) is also underscored by experimental evidence for rDNA 5.8S structures having
subminimal free-energy in
Chlamydomonas
and yeast (
48
). Thus, constraints imposed by as yet unknown intermolecular interactions and
the cellular environment likely have a significant influence on secondary
structure.
The prospects for advancing ITS secondary structural resolution and
incorporation of this information into phylogenetic analysis will depend upon
progress using biochemical, simulation (
46
) and phylogenetic/statistical approaches (
47
,
49
). The last offer the advantage of detecting character correlations independent
of an
a priori
presumed secondary structural model. The limitation, however, is the
requirement for an adequate number of phylogenetically independent correlated
substitutions for a given pair of bases, i.e., from rare evolutionary events,
one cannot distinguish between inevitable, necessary compensation and
phylogenetic coincidence. We emphasize that the secondary structures presented
in Figure
4
are for heuristic purposes, but, pending application of more advanced methods
of secondary structural analyses, this simple correlation of structural with
sequence conservation provides a preliminary basis for broader consideration of
secondary structure in angiosperm ITS2 evolutionary analyses.
Besides phylogenetic and functional implications, the present results also
enhance the value of ITS as a paradigm for DNA sequence evolution. The
previously cited advantages of ITS included its high information content at
lower phylogenetic levels and ease of amplification in diverse eukaryotes (
2
). The present analysis demonstrates that ITS2 also exhibits conserved sequence
patterns diagnostic at many hierarchical levels and substantial alignability
across angiosperms. The combination of angiosperm-wide sequence conservation with species-level sequence variability renders ITS a unique window for examining
the behavior of a rapidly-evolving, homologous, non-coding DNA sequence through divergence times spanning relatively
ancient (90-130 million years;
50
) to the most contemporary.
ACKNOWLEDGEMENTS
We are indebted to Dave Swofford for access to and permission to publish results
using prerelease versions of PAUP* 4.0, and to NIH-National Cancer Research Institute, (Frederick, MD) for GCG access and
computer support. Caleb Gordon (Department of Ecology and Evolutionary Biology,
University of Arizona, Tucson, AZ) generated the
Ravenala
ITS2 sequence as part of an ongoing phylogenetic study of Zingiberales with EAZ
and John Kress (Smithsonian Institution). William J. Hahn (Columbia University,
NY) generously provided the
Isomeris
ITS2 sequence. We thank Bruce Baldwin and Louise Lewis for critical comments.
MAH was supported by a Smithsonian Postdoctoral Fellowship in Molecular
Evolution.
REFERENCES
1 Soltis,P.S. and Soltis,D.E. (1995) Ann. Missouri Bot. Gard., 82, 147.
2 Baldwin,B.G., Sanderson,M.J., Porter,J.M., Wojciechowski,M.F., Campbell,C.S. and Donoghue,M.J. (1995) Ann. Missouri Bot. Gard.,82, 247-277.
3 Vilgalys,R. and Sun,B.L. (1994) Proc. Natl. Acad. Sci. USA,91, 4599-4603.MEDLINE Abstract
4 Bakker,F.T., Olsen,J.L. and Stam,W.T. (1995) J. Mol. Evol.,40, 640-651.MEDLINE Abstract
5 Coleman,A.W., Suarez,A. and Goff,L.J. (1994) J. Phycol.,30, 80-90.
6 Campbell,B.C., Steffen-Campbell,J.D. and Werren,J.H. (1993) Insect Mol. Biol.,2, 225-237.
12 Bremer,K. (1994) Asteraceae: Cladistics and Classification. Timber Press, OR.
13 Campbell,C.S., Donoghue,M.J., Baldwin,B.G. and Wojciechowski,M.F. (1995) Am. J. Bot.,82, 903-918.
14 Chase,M.W., Duvall,M.R., Hills,H.G., Conran,J.G., Cox,A.V., Eguiarte,L.E., Hartwell,J., Fay,M.F., Caddick,L.R., Cameron,K.G. and Hoot,S.B. (1995) In Rudall,P.J., Cribb,P.J., Cutler,D.F. and Humphries,C.J. (eds), Monocotyledons: Systematics and Evolution. Royal Botanic Gardens, Kew, London, pp. 109-137.
15 Dahlgren,R.M.T., Clifford,H.T. and Yeo,P.F. (1985) The Families of the Monocotyledons. Springer-Verlag, Berlin.
16 Kellogg,E.A. and Linder,H.P. (1995) Phylogeny of the Poales. In Rudall,P.J., Cribb,P.J., Cutler,D.F. and Humphries,C.J. (eds), Monocotyledons: Systematics and Evolution. Royal Botanic Gardens, Kew, London, pp. 511-542.
17 Ko,S.C., O'Kane,S.L.,Jr and Schaal,B.A. (1993) Rhodora,95, 254-277.
28 Genetics Computer Group (1994) Program Manual for the Wisconsin Package, version 8. Genetics Computer Group, Madison, WI.
29 Gilbert,D.G. (1992) LoopDloop. Available via anonymous ftp to ftp.bio.indiana.edu
30 Chase,M.W., Soltis,D.E., Olmstead,R.G., Morgan,D., Les,D.H., Mishler,B.D., Duvall,M.R., Price,R.A., Hills,H.G., Qiu,Y.-L. et al. (1993) Ann. Missouri Bot. Gard., 80, 528-580.
31 Doyle,J.A., Donoghue,M.J. and Zimmer,E.A. (1994) Ann. Missouri Bot. Garden, 81, 419-450.
32 Hershkovitz,M.A. (1993) Ann. Missouri Bot. Gard., 80, 333-365.
33 Chase,M.W., Stevenson,D.W., Wilkin,P. and Rudall,P.J. (1995) In Rudall,P.J., Cribb,P.J., Cutler,D.F. and Humphries,C.J. (eds), Monocotyledons: Systematics and Evolution. Royal Botanic Gardens, Kew, London, pp. 685-730.
34 Manhart,J.R. and Rettig,J.H. (1994) In Behnke,H.-D. and Mabry,T.J. (eds), Caryophyllales, Evolution and Systematics. Springer-Verlag, Berlin, pp. 235-246.
35 Lidén,M. (1993) In Kubitzki,K. Rohwer,J.G. and Bittrich,V. (eds), The Families and Genera of Flowering Plants. Springer-Verlag, Berlin., Vol. 2, pp. 556-557.
36 Nickrent,D.L. and Soltis,D.E. (1995) Ann. Missouri. Bot. Gard., 82, 208-234.
37 Yeh,L.-C.C. and Lee,J.C. (1990) J. Mol. Biol., 211, 699-712.MEDLINE Abstract
38 Swofford,D.L., Olsen,G.J., Waddell,P.J. and Hillis,D.M. (1996) In Hillis,D.M. and Moritz,C. (eds), Molecular Systematics. Sinauer Assoc., Sunderland, Massachusetts, 2nd Ed., pp. 407-514.
39 Penny,D., Lockhart,P.S., Steel,M.A. and Hendy,M.D. (1994) In Scotland,R.W., Siebert,D.J. and Williams,D.M. (eds), Models in Phylogeny Reconstruction, Systematics Association Special, Vol. 52. Clarendon Press, Oxford, pp. 211-230.
40 Michaels,H.J., Scott,K.M., Olmstead,R.G., Szaro,T., Jansen,R.K. and Palmer,J.D. (1993) Ann. Missouri Bot. Gard., 80, 742-751.
G. W. Grimm and T. Denk ITS Evolution in Platanus (Platanaceae): Homoeologues, Pseudogenes and Ancient Hybridization
Ann. Bot.,
February 1, 2008;
101(3):
403 - 419.
[Abstract][Full Text][PDF]
H. Kathriarachchi, R. Samuel, P. Hoffmann, J. Mlinarec, K. J. Wurdack, H. Ralimanana, T. F. Stuessy, and M. W. Chase Phylogenetics of tribe Phyllantheae (Phyllanthaceae; Euphorbiaceae sensu lato) based on nrITS and plastid matK DNA sequence data
Am. J. Botany,
April 1, 2006;
93(4):
637 - 655.
[Abstract][Full Text][PDF]
R. A. Levin, N. R. Myers, and L. Bohs Phylogenetic relationships among the "spiny solanums" (Solanum subgenus Leptostemonum, Solanaceae)
Am. J. Botany,
January 1, 2006;
93(1):
157 - 169.
[Abstract][Full Text][PDF]
J. A. Tate, J. Fuertes Aguilar, S. J. Wagstaff, J. C. La Duke, T. A. Bodo Slotta, and B. B. Simpson Phylogenetic relationships within the tribe Malveae (Malvaceae, subfamily Malvoideae) as inferred from ITS sequence data
Am. J. Botany,
April 1, 2005;
92(4):
584 - 602.
[Abstract][Full Text][PDF]
R. A. Levin, K. Watson, and L. Bohs A four-gene study of evolutionary relationships in Solanum section Acanthophora
Am. J. Botany,
April 1, 2005;
92(4):
603 - 612.
[Abstract][Full Text][PDF]
A. N. Muellner, R. Samuel, M. W. Chase, C. M. Pannell, and H. Greger Aglaia (Meliaceae): an evaluation of taxonomic concepts based on DNA data and secondary metabolites
Am. J. Botany,
March 1, 2005;
92(3):
534 - 543.
[Abstract][Full Text][PDF]
P. E. Berry, W. J. Hahn, K. J. Sytsma, J. C. Hall, and A. Mast Phylogenetic relationships and biogeography of Fuchsia (Onagraceae) based on noncoding nuclear and chloroplast DNA data
Am. J. Botany,
April 1, 2004;
91(4):
601 - 614.
[Abstract][Full Text][PDF]
S. S. NEVES and M. F. WATSON Phylogenetic Relationships in Bupleurum (Apiaceae) Based on Nuclear Ribosomal DNA ITS Sequence Data
Ann. Bot.,
April 1, 2004;
93(4):
379 - 398.
[Abstract][Full Text][PDF]
K. Andreasen and B. G. Baldwin Reexamination of relationships, habital evolution, and phylogeography of checker mallows (Sidalcea; Malvaceae) based on molecular phylogenetic data
Am. J. Botany,
March 1, 2003;
90(3):
436 - 444.
[Abstract][Full Text][PDF]
C. A. Cote and B. A. Peculis Role of the ITS2-proximal stem and evidence for indirect recognition of processing sites in pre-rRNA processing in yeast
Nucleic Acids Res.,
May 15, 2001;
29(10):
2106 - 2116.
[Abstract][Full Text][PDF]
A. E. Schwarzbach and R. E. Ricklefs Systematic affinities of Rhizophoraceae and Anisophylleaceae, and intergeneric relationships within Rhizophoraceae, based on chloroplast DNA, nuclear ribosomal DNA, and morphology
Am. J. Botany,
April 1, 2000;
87(4):
547 - 564.
[Abstract][Full Text]
S. R. Downie, D. S. Katz-Downie, and K. Spalik A phylogeny of Apiaceae tribe Scandiceae: evidence from nuclear ribosomal DNA internal transcribed spacer sequences
Am. J. Botany,
January 1, 2000;
87(1):
76 - 95.
[Abstract][Full Text]
R. D. Noyes and L. H. Rieseberg ITS sequence data support a single origin for North American Astereae (Asteraceae) and reflect deep geographic divisions in Aster s.l.
Am. J. Botany,
March 1, 1999;
86(3):
398 - 412.
[Abstract][Full Text][PDF]
CiteULike 
Connotea 
Del.icio.us What's this?
