Fig. 1.
Gastrochilus longiflora (
B. longiflora). Watercolour on paper by Vishnupersaud (1829)
from the Wallich Collection, Kew. Reproduced with permission of The Board of Trustees of
the Royal Botanic Gardens, Kew.
49
Boesenbergia longiflora and related taxa
To determine which taxon a
B. aff. longiflora specimen belonged, the location, habitat,
plant height, flower colour and underground morphology were important. Here, the
“Additional specimens examined” under each described species are determined based
on a combination of these parameters. The name
B. longiflora (see “Taxonomy”
below) is lectotypified.
During 2008
-2012, c. 25 live rhizome divisions of
B. aff. longiflora were
obtained from a variety of sources: botanical gardens, research institutions and various
personal collections. Each accession was grown by the first author in Hawaii, USA
under controlled, shade-house conditions. Details on the morphology and phenology
were recorded over several growing seasons. DNA samples with a corresponding
herbarium specimen were taken from each accession.
In 2009
-
2012 field observations were made in Thailand with botanists from
BK, the Thailand Institute of Scientific and Technological Research and Kasesart
University. Similarly in India, observations were conducted with personnel from the
Botanical Survey India (Shillong), the Assam Forestry Department and the Nagaland
University. Digital images of all Boesenbergia
observed in the field and in cultivation
were recorded for later reference.
Early in the research, all Boesenbergia species known to have radical
inflorescences were evaluated for possible inclusion in the B. longiflora complex. Based
on molecular data (Mood et al., unpublished) B. basispicata K.Larsen ex Sirirugsa,
B. tenuispicata K.Larsen and B. trangensis K.Larsen belong to other Boesenbergia
clades. DNA samples were not available for B. angustifolia (Hallier f.) Schltr., B.
phyllostachya (Gagnep.) K.Larsen and
B. siphonantha (King ex Baker) M.Sabu,
Prasanthk. & Škorničk. Consequently, their phylogenetic position is undetermined at
this time.
The final taxonomy presented here represents an incorporation of data from the
phylogenetic analyses, field observations and detailed study of the cultivated plants.
The descriptions are ordered with the yellow-flowered species first, followed by the
white-flowered species. The distribution map (Fig. 20) was constructed by mapping
types and selected specimens of each species (these are marked by an asterisk following
the herbarium abbreviations in the list of additional specimens examined).
Molecular phylogeny. Phylogenetic analyses were conducted to complement and
assist in taxonomic decisions required to elucidate relationships among species of
the
B. longiflora clade as defined here. Taxon selection was based on an unpublished
broader analysis of plastid trnK data for Boesenbergia relative to other genera of
Zingiberaceae (authors in prep.). The bulk of the data for non-Boesenbergia samples
are from the earlier family-wide analyses of Kress et al. (2002) with a few additional
sequences from recently described genera such as from Kress et al. (2010) and Leong-
Škorničková et al. (2011). The data for the specimens cited in these references can
be found at http://www.ncbi.nlm.nih.gov/genbank/ (GenBank 2011).
A representative
shortest maximum parsimony (MP) phylogram is shown in Fig. 3. Only names of
relevant Boesenbergia samples are indicated on the phylogram.
50
Gard. Bull. Singapore 65(1) 2013
Fig. 2. Hand-coloured lithograph published in Curtis’s Botanical Magazine 69, t. 4010, 1843
as Gastrochilus
longiflora. Reproduced with permission of The Board of Trustees of the Royal
Botanic Gardens, Kew. (Identified by Baker (1890) as G. jenkinsii. Described here as B.
hamiltonii.)
51
Boesenbergia longiflora and related taxa
DNA was extracted from silica-dried leaf material or
leaf tissue from herbarium
specimens using standard methods as described in Kress et al. (2002). Thirty-seven
samples of Boesenbergia were used in the DNA sequencing analyses. Two samples
of B. pulcherrima (Wall.) Kuntze, the generic type, were designated as outgroup taxa.
Two other samples, B. plicata var. lurida (Ridl.) Holttum and B. plicata var. plicata
(Ridl.) Holttum were also included. The remaining samples were all members of the
B. longiflora clade
. All samples were designated by a collection or
accession number
and vouchered (Appendix B).
Both nuclear and plastid data were collected. The nuclear ribosomal ITS (nrITS)
region was amplified using the 18S-F and 26S-R primers (Prince 2010) and Phusion
high fidelity polymerase (New England BioLabs, Ipswich, Massachusetts, USA)
with 5X GC Buffer per the manufacturer’s instructions at an annealing temperature
of 62ºC. The same primers were used for subsequent sequencing. The plastid trnK
region was amplified and sequenced using conditions described in Kress et al. (2002)
or Prince & Kress (2006). Two newly designed primers were synthesized to assist with
sequencing of some difficult samples (5Fb: CTCTATGGATTTTCAAGGAT and 5Rb:
AGACCAAAATGAAAATAATA). All samples were direct sequenced on a 3130xl
Genetic Analyser at Rancho Santa Ana Botanic Garden (Claremont, California, USA).
Noisy sequences (electrophoregrams showing more than a few polymorphic
sites) in the nrITS data were cloned using TOPO TA cloning kits (Invitrogen, now Life
Technologies, Grand Island, New York, USA) with four to eight clones sequenced per
sample. A small number of the trnK sequences were also noisy (due to polymerase
“stutter” caused by a mononucleotide T repeat around 450bp). Those samples were
re-amplified using Phusion polymerase, which has been shown to outperform other
polymerases for this particular problem (Fazekas et al. 2010) and re-sequenced.
Individual sequences of each specimen were edited and a consensus sequence
was generated in Sequencher v4.9 (Gene Codes Corporation, Ann Arbor, Michigan,
USA). The consensus sequence was trimmed (18S and 26S data pruned from the
ITS1+5.8S+ITS2 region, removal of amplification primer sequences from the trnK
region) exported and aligned manually in Se-Al v2.0a11 (1996; available from A.
Rambaut, Oxford, England, UK at http://tree.bio.ed.ac.uk/software/seal). Alignment
was relatively straightforward due to the high degree of sequence similarity and length
for the taxa involved. Ambiguous regions were generally restricted to the monomeric
“T” repeat in the 5
ʹ trnK–matK intergenic spacer (IGS). All newly generated data have
been deposited in GenBank and are available under accession numbers JX992748–
JX992840 (Appendix B).
Data were analyzed under parsimony and likelihood criteria by genomic data
partition, first independently, and later in combination. The decision to combine data
partitions is often based upon the results of an incongruence-length difference, or ILD
test (Farris et al. 1995). We did not perform this test as it has been suggested to be
too conservative (Cunningham 1997, Struck 2007). We expected the test to fail due
to the presence of two very different ITS copies detected for at least one sample in
the ITS data partition. The decision to combine data partitions was instead based on a
combination of tree topology congruence and branch support values. In most instances
52
Gard. Bull. Singapore 65(1) 2013