Fig. 4. One of 37 best, Maximum Likelihood phylograms (-ln=2218.749) for
Boesenbergia
longiflora and closely related taxa based on analysis of nuclear
ribosomal ITS data for all
samples and clones. Support values (Bootstrap/Posterior Probability) for critical branches are
shown above branches. Bold typeface highlights strongly supported positions of the different
Boesenbergia longiflora copies generated by cloning. Asterisks indicate samples that should
also be cloned to clarify a few polymorphisms. Underlined samples are placed in conflicting
positions on the chloroplast phylogeny. Note: lower branches were pruned to fit page and are
not to scale.
56
Gard. Bull. Singapore 65(1) 2013
Fig. 5. Single best Maximum Likelihood phylogram (-ln=3953.067) for
B. longiflora and
closely related taxa based on analysis of chloroplast trnK intron data. Support values (Bootstrap/
Posterior Probability) for critical branches are shown above branches. Numbers in bold typeface
indicate strong support. Underlined samples are placed in conflicting positions on the nuclear
ribosomal phylogeny. Note: lower branches were pruned to fit page and are not to scale.
57
Boesenbergia longiflora and related taxa
The identification of at least two disparate ITS copies for
B. longiflora requires
further discussion. Only one sample of
B. longiflora was available for inclusion in
this study. This sample, M11P48
was from cultivated plants grown by the first author,
but originally collected in Burma (Kress 03-7305, US). Multiple clones representing
one
B. longiflora ITS type were distributed among members of the B. kingii clade
(ITS copy “A”) while the other
B. longiflora copy fell sister to the
B. kerrii clade
(ITS copy “B”) as is shown in Fig. 4. Although a number of tests have been proposed
to distinguish between the processes of hybridisation and incomplete lineage sorting
(Joly et al. 2009) all have limitations. We were restricted here by the single sample for
B. longiflora, but hopefully, more samples will be available for future analyses.
The majority of the polymorphic samples were identified as B. kingii based
on morphology. Six of the twelve samples required cloning to generate clean ITS
sequences. These samples fall into two categories. The first group included samples
with multiple ITS sequence types, but those sequences were all quite similar with only
a few different bases or slight length variations. Analyses of the ITS data partition
placed all of these sequences in the B. kingii sensu lato clade as described above
(B. hamiltonii + B. kingii + B. maxwellii
), and with moderate to high support (75%
BS, 0.98 PP) but with poor resolution within the clade (Fig. 4). Further, most of the
B. kingii
samples clustered with moderate support (70% BS, 1.00 PP). Among the
ingroup members, B. kingii is the most broadly distributed species and is the most
genetically diverse taxon.
Chloroplast trnK intron (including the protein coding
matK gene)
A total of 36 potentially parsimony-informative characters (15 for ingroup only) were
used to generate almost 100,000 maximum parsimony trees. However, maximum
likelihood analyses resulted in a single best tree (Fig. 5). The likelihood tree had a
topology identical to several of the MP trees. As with the results of the ITS analyses,
both MP and ML analyses identified two clades within the ingroup, one including
samples of B. collinsii + B. kerrii +
B. longiflora + B. maxwellii (65% BS; 0.85 PP)
and a clade comprising B. hamiltonii and B. kingii
(63% BS; 0.83 PP). Three taxa were
resolved and supported as monophyletic: B. collinsii
(65% BS, 0.94 PP) B. kerrii (63%
BS, 0.94 PP) and B. maxwellii
(99% BS, 1.00 PP). The relationship of B. maxwellii
to B. hamiltonii and B. kingii differed here relative to the ITS results, but only with
weak (65% BS; 0.85 PP) support. Sequences for two samples were not available for
this data set, B. hamiltonii M3212 and B. maxwellii MP1450. The two samples with
names underlined in Fig. 4, B. kingii M11P77 and M11P78, were identical and clearly
supported as members of the B. kingii group in the phylogeny produced by analyses of
the nuclear ITS dataset. Results of the chloroplast data analyses (Fig. 5) clearly (88%
BS, 1.00 PP) placed these two samples in a clade with B. hamiltonii.
Combined analyses
The combined data set required a number of samples to be excluded. Because the ITS
data required cloning for some samples, results of the ITS MP analyses were used to
select the clone with the shortest branch for each sample (as suggested by Beilstein et
58
Gard. Bull. Singapore 65(1) 2013
al. 2008) for subsequent use in the combined analyses. This was done only when all
clones for a particular sample either formed a monophyletic lineage or clustered within
a moderately or well-supported lineage (BS≥75%). In the instance of B. kingii samples
M11P77 and M11P78
, there was positional conflict between the nuclear and plastid
datasets. Those samples were excluded from the combined analysis. As mentioned,
there was only a single sample of
B. longiflora available for this study. The nuclear
data set included two distinctly different copies, A and B. Rather than totally exclude
this critical taxon, we selected the nuclear copy B that best matched the chloroplast
phylogeny, while recognising we do not know which better represents the evolutionary
history of the species.
Maximum parsimony analysis produced 468 shortest trees. Maximum likelihood
analyses produced three equally likely trees, one of which is shown in Fig. 6. Both
MP and ML analyses identified two clades within the in-group most similar to those
of the ITS data analyses, a B. kingii clade including samples of B. hamiltonii + B.
kingii + B. maxwellii and a
B. longiflora sensu lato clade including samples of B.
collinsii + B. kerrii +
B. longiflora. The B. kingii sensu lato clade received 62% BS and
0.78 PP. Unlike the ITS data analyses results, the combined data analyses produced
highly resolved topologies within this clade. Significantly, B. maxwellii is resolved as
monophyletic (83% BS; 1.00 PP). Boesenbergia hamiltonii and B. kingii samples are
grouped together with 51% BS and 0.95 PP. Boesenbergia hamiltonii was also resolved
as monophyletic, but with weak support (<50% BS, 0.86 PP). The B. longiflora sensu
lato clade was strongly supported (95% BS; 1.00 PP) with significant internal structure
and support. Boesenbergia collinsii
was supported as monophyletic (100% BS; 1.00
PP) as was B. kerrii
(93% BS, 1.00 PP). The single sample of B. longiflora was sister
to B. kerrii
(64% BS, 0.64 PP), but this placement should be viewed with caution.
Multiple accessions for each taxon (except
B. longiflora) were included to assess
genetic variation within and among the newly identified taxa. The amount of sequence
variation within each taxonomic clade was generally less than the amount of variation
between clades as is shown by the phylograms in Figs. 4
-6. Despite the higher level of
sequence diversity within B. kingii, it is supported as monophyletic. The results of both
maximum parsimony and maximum likelihood analyses of DNA data presented above
consistently identified five mostly strongly supported clades within the B. longiflora
complex: B. collinsii, B. hamiltonii, B. kerrii, B. kingii and B. maxwellii. The data also
support two sister relationships, one between B. collinsii and B. kerrii, and one between
B. hamiltonii and
B. kingii. The
position of
B. longiflora and
B. maxwellii are less clear.
The relationship of
B. longiflora to these taxa is uncertain due to polymorphism in the
nuclear (ITS) data set, but it is clearly closely related to them. Boesenbergia maxwellii
jumps between the two clades depending upon data set examined. Finally, additional
data are needed to test between various hypotheses (incomplete concerted evolution
due to genetic drift post polyploidisation, hybridisation, or incomplete lineage sorting)
to explain the high level of genetic variation detected in the nrITS of B. kingii. Overall,
although there are a number of questions remaining, the molecular data allowed for the
delineation of six taxa.
59
Boesenbergia longiflora and related taxa