WMTTEN ET AL. Relationships of Zygopetalinae


sites, the percentage of sites that are variable, the
number of trees, number of steps, consistency index
(CI) excluding uninformative characters, and retention
index (RI) for each separate and combined analysis.
Alignment was unambiguous for matK and was not
problematic for ITS and trnL-F. The trnL-F alignment
contained indels up to 50 bases long, but these were
usually easily aligned and often were autoapomorphic.

matK-The matK matrix was the least variable of the
three and yielded the least resolution. The trimmed
amplified region ranges from 1314 to 1323 base pairs
(bp); the aligned length is 1341 bp and contains four
indels ranging in length from three to nine bp (not scored
as characters). Two indels are autapomorphic; one indel
of nine bp occurs in two of the three sampled species
of Huntleya, and another nine-base indel occurs in four
of the eight taxa of the Bollea/Pescatorea clade. The
matrix contains 154 potentially parsimony-informative
characters. Heuristic search (Fitch criterion) yielded
2115 trees (L=472, CI=0.53 excluding uninformative
characters here and below, RI=0.81. The few clades
with high bootstrap support (Fig. 1) are usually genera
or clades within genera.

trnL-F-Fifteen accessions from ten taxa repeatedly
yielded double bands or heterogeneous PCR products
that produced mixed sequences (suggestive of multiple
copies of this region) and were excluded from the
trnL-F analyses. The excluded taxa were: Ackermania
estradae, Chaubardiella pubescens, Chaubardiella
tigrina, Chondrorhyncha andreettae, Chondrorhyncha
aff. rosea, Chondroscaphe flaveola, Koellensteinia
boliviensis, Otostylis lepida, Neogardneria murrayana,
and Stenia bismarkii. Five additional samples
(Batemannia lepida, Galeottia burkei two accessions,
G. ciliata, and G. colombiana) yielded clean sequences,
with several deletions and many mutations not present
in congeners, resulting in the placement of these five
taxa on a relatively long branch (ca. 65 steps) relative to
their congeners in the shortest trees. This long branch
is suggestive of possible paralogy within the trnL-F
region; future studies will include cloning of trnL-
F PCR products to clarify problems due to multiple
copies.

The amplified trnL-F region ranges in length from
968 (Promenaea stapelioides) to 1154 (Warczewiczella
discolor) base pairs (bp). The aligned trnL-F matrix is
1358 bp long, and includes 829 bases of the trnL intron,
the 3' trnL exon (51 bp), 439 bp of the intergenic spacer,
and 25 bp of the 5' end of trnF. Two indels (26 bp of


intron and 9 bp of spacer) were judged unalignable
and were excluded from the analysis. Heuristic search
(Fitch criterion) yielded 9310 trees (L=443, CI=0.64,
RI=0.87. In the bootstrap consensus (Fig. 2), Warrea
warreana is weakly supported as sister to all other taxa,
and the Zygopetalum grade (as defined in the combined
analysis, Figs 4-5) + Cryptarrhena form an unresolved
polytomy basal to Dichaea + Huntleya clade. Within
the Huntleya grade, Huntleya and Chaubardia are
successively basal to a moderately supported (83% BS)
clade of all remaining Huntleya clade. Relationships
within this core Huntleya grade are poorly resolved and
many clades are weakly supported.

ITS rDNA-The aligned ITS rDNA matrix is 838 bp
in length: 110 bp of the 18S region, ITS 1 (235 bp),
the 5.8S gene (164 bp), ITS 2 (267 bp), and 62 bp of
the 26S region. The heuristic search (Fitch criterion)
yielded 857 equally parsimonious trees of 949 steps
(CI=0.54, RI=0.86).

The ITS bootstrap consensus (Fig. 3) is the most
highly resolved of the three data sets and is highly
congruent with the plastid data set. The Zygopetalum
grade (as defined in the combined analysis, Figs 4-5) +
Cryptarrhena again form a basal polytomy, but many
clades are highly supported: Koellensteinia + Otostylis
+ Paradisanthus; Zygopetalum + Neogardneria +
Pabstia; Galeottia + Batemannia; and Warrea +
Warreopsis. However, support for the Huntleya clade
+ Dichaea is weak (64% BS). Within the strongly
supported Huntleya clade (90% BS), Huntleya and
Chaubardia are strongly supported as basal to the
core Huntleya clade (Chondrorhyncha caquetae to
Cochleanthes flabelliformis). Many clades within
this core Huntleya clade are moderately to strongly
supported, including monophyletic Chondroscaphe
(100% BS) and Kefersteinia (99% BS). However,
many genera are not supported as monophyletic, e.g.,
Chondrorhyncha, Cochleanthes, Stenia, Bollea, and
Pescatorea. In the latter two genera, the lack of support
for monophyly is due to low sequence divergence.
In other genera (e.g., Chondrorhyncha), sequence
divergence is high and the species form several highly
divergent and well supported clades.

Combined analysis-omparison of bootstrap
consensus trees for analyses of both data sets revealed no
hard incongruence, i.e., clades that are highly supported
in one analysis that conflict with different and highly
supported clades in the others (\\ illiamnis ,l. 2001). We
therefore performed a combined analysis of both data


LANKESTERIANA5(2):