Fig. 3. One of many shortest
maximum parsimony phylograms (simplified) for Boesenbergia
and other representatives of Zingiberaceae based on plastid trnK data analysis, rooted using
Siphonochilus J.M.Woods & Franks as suggested by Kress et al. (2002). The position of the
genus Boesenbergia, the
B. longiflora clade and the taxa sampled for this study are indicated.
•
indicate branches with MP Bootstrap values ≥50%.
53
Boesenbergia longiflora and related taxa
tree topologies were consistent between data partitions. In cases where tree topology
differed, branch support for the conflicting topology was reviewed. If branch support
was weak (<65% BS, <0.95 PP) the data for those taxa were analysed in combination.
As stated above, the ITS data for one taxon included two very different copies. In
another instance, there was strong positional conflict among the nuclear and plastid
phylogenies for one sample. Disparate copies were included in all individual data
set analyses, but only one copy was included in combined data analyses. The ITS
copy retained for combined analyses was selected to best match to the chloroplast
phylogeny.
Maximum parsimony and maximum likelihood (ML) analyses were conducted
in PAUP* (version 4.0b10, Swofford 2002). Heuristic search methods were conducted
in each case with 1000 random addition replicates for MP and 10 random addition
replicates for ML, each with tree bisection and regrafting (TBR) branch swapping.
Maximum Parsimony analyses were conducted under Fitch parsimony criteria (Fitch
1971). Maximum likelihood analyses were conducted using model parameters selected
by jModelTest (version 0.1.1 available at http://darwin.uvigo.es; Posada 2008, Guindon
& Gascuel 2003) under both the Akaike Information Criterion (AIC; Akaike 1974) and
the Bayesian Information Criterion (BIC; Schwarz 1978). All Maximum Likelihood
trees were saved. Both AIC criterion and BIC criterion models were analysed for all
data partitions, but only the trees from the AIC analyses will be discussed as both
methods produced nearly identical trees.
Branch support was estimated using parsimony bootstrap (BS) in PAUP* and
posterior probabilities (PP) in MrBayes v3.2.1 (Ronquist & Huelsenbeck 2003).
MrBayes analyses were conducted through the CIPRES portal (Cyberinfrastructure
for Phylogenetic Research; Miller et al. 2010) and used partitioned data whenever
appropriate (partitions = 5
ʹtrnK IGS + 3ʹtrnK IGS, matK, ITS) and were run in triplicate
to ensure convergence. Bootstrap values were based on 1,000 pseudoreplicates, each
with 100 random addition replicates, TBR branch swapping, saving a maximum of
10 trees per random addition replicate and hold=4 trees. The number of generations
necessary to estimate posterior probabilities varied depending upon the dataset and the
time required to reach stasis (average standard deviation of the splits frequency <0.01).
Results and discussion
A number of exploratory analyses were conducted to evaluate the consistency of the
phylogenetic hypotheses generated by the different data partitions. As stated above,
some of the data generated from the nuclear ribosomal partition were noisy, indicating
the presence of divergent ITS copies within a sample. Although it is not common,
multiple copies of ITS have been detected in a number of genera and species of
Zingiberaceae including Alpinia (Liu et al. 2009), and Cornukaempferia (L. Prince
pers. obs.), and multiple ITS copies per individual is prevalent in Curcuma
(Záveská
et al. 2012) and Kaempferia (L. Prince, pers. obs.). The presence of multiple, strikingly
different copies can be explained by a number of different evolutionary histories, a
54
Gard. Bull. Singapore 65(1) 2013
few of which are presented here. The ITS region is present in hundreds (or thousands)
of copies, often on multiple, different chromosomes. Processes affecting the utility
of ITS for phylogenetic reconstruction were reviewed by Álvarez & Wendel (2003)
including a discussion of the processes described below.
One explanation for the detection of multiple different copies in one organism
is that the multiple copies were present in the ancestor and those copies are being
maintained through time. Another explanation is recent genetic drift of some copies
due to relaxed evolutionary constraints. Given the importance of this region for the
functionality of the organism, it would not be likely unless there has been duplication
(perhaps due to polyploidisation). Chromosome counts have not been made for the
sample of
B. longiflora (Kress 03-7305, US) included here, so perhaps this is a plausible
explanation. Yet another possibility is recent gene flow between closely related taxa.
If hybridisation happened in the recent past, there is a possibility that the two different
copies have not yet had the time to undergo concerted evolution. If this process is
incomplete, each parental copy will be recovered and, in a phylogenetic analysis, the
different copies would each cluster with one of the (putative) parents. Alternatively,
chimeric ITS sequences might be detected.
No matter the source of multiple, different copies of ITS present within any given
organism, there is ample indication that most organisms undergo concerted evolution
of the ITS region, a process by which the different copies are homogenised across the
genome resulting in a single (or at least dominant) copy per organism. The length of
time for complete homogenisation of the ITS of any given organism is unknown and
likely highly variable. Although there has been a great deal of speculation regarding
concerted evolution and homogenisation in purported hybrid taxa of flowering plants,
our best understanding of the actual time required is gleaned from experiments with
artificial hybrids (e.g., Armeria: Fuertes Aguilar et al. 1999, Feliner et al. 2001;
Hieracium
: Mráz et al. 2011) and in recently derived polyploidy taxa (e.g., Spartina:
Ainouche et al. 2004; Helictotrichon: Winterfeld et al. 2009; Oryza: Ying et al. 2010).
Nuclear ribosomal ITS data
The aligned data matrix included 86 potentially parsimony-informative characters (43
for ingroup only). As indicated above, several samples required cloning to obtain clean
ITS sequences. Two other samples, indicated by an asterisk in Fig. 4, should be cloned
to clarify a few polymorphisms. Maximum parsimony analysis produced over 100,000
shortest trees and could not be run to completion. One of the 37 shortest ML trees is
shown in Fig. 4.
The ML phylogram presented in Fig. 4 identifies a moderately supported clade
that includes all samples of B. kerrii and B. collinsii plus one of the
B. longiflora
clones (88% BS; 1.00 PP). Both B. kerrii and B. collinsii are resolved as monophyletic
with moderate to strong support. The other ingroup clade is moderately to strongly
supported (75% BS; 0.98 PP), and includes all samples of B. kingii, B. hamiltonii,
B. maxwellii and two clones of
B. longiflora. Resolution within the clade is poor and
internal branches are generally poorly supported although most of the B. kingii samples
and clones cluster together with 70% BS and 1.00 PP.
55
Boesenbergia longiflora and related taxa