Evolutionary History of ‘‘Early-Diverging’’ Eukaryotes: The Excavate
Taxon Carpediemonas is a Close Relative of Giardia
Alastair G. B. Simpson,* Andrew J. Roger,* Jeffrey D. Silberman,† Detlef D. Leipe,‡
Virginia P. Edgcomb,§ Lars S. Jermiin,\ David J. Patterson,\ and Mitchell L. Sogin§
*Department of Biochemistry, Dalhousie University, Halifax, Canada; †UCLA Institute of Geophysics & Planetary Physics
and Department of Microbiology & Immunology; ‡National Center for Biotechnology Information, National Library of
Medicine, Bethesda; §Josephine Bay Paul Center for Comparative Molecular Biology and Evolution, Marine Biological
Laboratory, Woods Hole; and \School of Biological Sciences, University of Sydney
Diplomonads, such as Giardia, and their close relatives retortamonads have been proposed as early-branching
eukaryotes that diverged before the acquisition-retention of mitochondria, and they have become key organisms in
attempts to understand the evolution of eukaryotic cells. In this phylogenetic study we focus on a series of eukaryotes suggested to be relatives of diplomonads on morphological grounds, the ‘‘excavate taxa’’. Phylogenies of small
subunit ribosomal RNA (SSU rRNA) genes, a-tubulin, b-tubulin, and combined a- 1 b-tubulin all scatter the
various excavate taxa across the diversity of eukaryotes. But all phylogenies place the excavate taxon Carpediemonas as the closest relative of diplomonads (and, where data are available, retortamonads). This novel relationship
is recovered across phylogenetic methods and across various taxon-deletion experiments. Statistical support is strongest under maximum-likelihood (ML) (when among-site rate variation is modeled) and when the most divergent
diplomonad sequences are excluded, suggesting a true relationship rather than an artifact of long-branch attraction.
When all diplomonads are excluded, our ML SSU rRNA tree actually places retortamonads and Carpediemonas
away from the base of the eukaryotes. The branches separating excavate taxa are mostly not well supported (especially in analyses of SSU rRNA data). Statistical tests of the SSU rRNA data, including an ‘‘expected likelihood
weights’’ approach, do not reject trees where excavate taxa are constrained to be a clade (with or without parabasalids and Euglenozoa). Although diplomonads and retortamonads lack any mitochondria-like organelle, Carpediemonas contains double membrane-bounded structures physically resembling hydrogenosomes. The phylogenetic
position of Carpediemonas suggests that it will be valuable in interpreting the evolutionary significance of many
molecular and cellular peculiarities of diplomonads.
Introduction
From an evolutionary perspective diplomonads are
one of the most interesting and enigmatic groups of eukaryotes. Diplomonads (and their close relatives, retortamonads; Silberman et al. 2002) are small, flagellated
unicells that lack mitochondria and tend to have poorly
developed endomembrane systems (Brugerolle 1991).
Most are parasites, including the best known, the human
pathogen Giardia lamblia. Early phylogenies of small
subunit ribosomal RNA (SSU rRNA) genes and elongation factor proteins placed Giardia as one of the earliest diverging lineages in the eukaryotic evolutionary
tree (Sogin et al. 1989; Hashimoto et al. 1994; Hashimoto et al. 1995). These results (and later studies incorporating other genes and taxa) supported the idea that
diplomonads had diverged before the acquisition of mitochondria (Cavalier-Smith 1983). The possibility that
diplomonads are early-diverging, ‘‘primitive’’ eukary1 Research conducted at: The Josephine Bay Paul Center for Comparative Molecular Biology and Evolution, Marine Biological Laboratory, Woods Hole, and at the Department of Biochemistry, Dalhousie
University.
Abbreviations: BS, bootstrap support; K2P, Kimura two parameter
model; ML, maximum-likelihood; MLdist, maximum-likelihood distance; Pars, parsimony; SSU rRNA, small subunit ribosomal RNA;
TBR, tree bisection-reconnection.
Key words: Archezoa, diplomonads, jakobids, eukaryote evolution, mitochondria, protists.
Address for correspondence and reprints: Alastair G. B. Simpson,
Department of Biochemistry, Dalhousie University, Halifax, Nova
Scotia, Canada B3H 4H7. E-mail: [email protected].
Mol. Biol. Evol. 19(10):1782–1791. 2002
q 2002 by the Society for Molecular Biology and Evolution. ISSN: 0737-4038
1782
otes fuelled an explosion of interest in their molecular
and cellular biology, culminating in the Giardia genome
project (McArthur et al. 2000).
In recent years, controversy has surrounded the
evolutionary significance of diplomonads. Several genes
from Giardia (chaperonin 60, cpn60; pyridoxal-59-phosphate-dependent cysteine desulfurase, IscS; heat shock
protein 70, hsp70; triose phosphate isomerase, TPI; and
valyl-tRNA synthetase) and another diplomonad, Spironucleus (cpn60), are candidates for having mitochondrial origins, based on their phylogenetic affinity with
nuclear-encoded mitochondrial genes from other eukaryotes (Keeling and Doolittle 1997; Hashimoto et al.
1998; Roger et al. 1998; Horner and Embley 2001; Morrison et al. 2001; Tachezy, Sánchez, and Müller 2001).
At face value these data suggest that mitochondria were
present in ancestors of diplomonads. But the robustness
of these cases varies. Eukaryotic TPI and valyl-tRNA
synthetase lack a clear phylogenetic affinity with orthologs from a-proteobacteria, the closest relatives of
the mitochondrial symbiont. Perhaps these genes entered
early eukaryotic cells by independent lateral transfers
from other prokaryotes. Mitochondrial hsp70 from most
eukaryotes has a close phylogenetic affinity to hsp70
from a-proteobacteria, but the orthologous gene from
Giardia branches separately from these mitochondrial
forms in rigorous analyses and could represent a separate lateral transfer to diplomonads (Morrison et al.
2001). Only in the cases of cpn60 and probably IscS do
the genes from diplomonads form strong clades with
well-founded mitochondrial forms from other eukaryotes (Roger et al. 1998; Horner and Embley 2001; Tach-
Carpediemonas is Related to Diplomonads
ezy, Sánchez, and Müller 2001). Given the lack of cytological relics of mitochondria in diplomonads, alternative models have been proposed whereby even the
most promising putative mitochondrial genes from diplomonads were also acquired through other lateral gene
transfers or were transferred from a mitochondrial symbiont before its stable incorporation as an eukaryotic
organelle (Sogin 1997; Doolittle 1998; Chihade et al.
2000; Margulis, Dolan, and Guerrero 2000). These scenarios retain a ‘‘premitochondrial’’ status for diplomonads based on the assumption that they are basal eukaryotes. Further complicating the issue are indications that
the basal portions of many eukaryotic gene phylogenies
(including SSU rRNA and elongation factor proteins)
may be artificial collections of phylogenetically misleading sequences (Philippe and Adoutte 1998; Hirt et
al. 1999; Stiller and Hall 1999). This suggests that the
evidence for diplomonads being basal eukaryotes is unreliable but fails to demonstrate that diplomonads are
not early diverging.
A critical impediment to resolving the controversy
is our poor understanding of the immediate relationships
of diplomonads and retortamonads to other eukaryotes.
A close relationship with the amitochondriate, hydrogenosome-bearing parabasalids is most commonly entertained, based on some molecular phylogenies (Embley and Hirt 1998; Baldauf et al. 2000; Cavalier-Smith
2000). But cytoskeletal morphology studies suggest that
diplomonads and retortamonads might be closely related
to some or all of a different array of protists; ‘‘core’’
jakobids, Malawimonas, Heterolobosea, Trimastix and
Carpediemonas (O’Kelly 1993; O’Kelly, Farmer, and
Nerad 1999; Simpson and Patterson 1999; Simpson,
Bernard, and Patterson 2000; Simpson and Patterson
2001). Collectively these groups are termed the ‘‘excavate taxa’’, named for the shared possession of a distinctive form of feeding groove (Simpson and Patterson
1999). All excavate taxa other than diplomonads and
retortamonads have mitochondria, albeit sometimes with
unusually bacterial-like (primitive) mitochondrial genomes (Lang et al. 1997; Gray et al. 1998), or they
possess double membrane-bounded organelles that have
been proposed to be mitochondrial homologs (Brugerolle and Patterson 1997; O’Kelly, Farmer, and Nerad
1999; Simpson and Patterson 1999; Simpson, Bernard,
and Patterson 2000). Excavate taxa have not been systematically examined by molecular phylogenetic methods, with just three or four of the seven known groups
represented in any one analysis (Dacks et al. 2001;
Edgcomb et al. 2001; Archibald, O’Kelly, and Doolittle
2002; Cavalier-Smith 2002; Silberman et al. 2002).
The small excavate flagellate Carpediemonas lives
in oxygen-poor sediments and contains double membrane-bounded organelles of unknown function (Simpson and Patterson 1999; Bernard, Simpson, and Patterson 2000). Here we report the first gene sequences from
Carpediemonas: SSU rRNA, a-tubulin, and b-tubulin.
We also report SSU rRNA sequences for several core
jakobids, and for Malawimonas, the organism that, aside
from having true mitochondria, displays perhaps the
greatest morphological similarity to Carpediemonas
1783
(O’Kelly, Farmer, and Nerad 1999; Simpson and Patterson 1999). Our SSU rRNA alignment is the first molecular data set to include all seven known excavate taxa.
Our phylogenetic analyses suggest that Carpediemonas
is the sister group to the diplomonads 1 retortamonads
clade.
Materials and Methods
Gene Amplification and Sequencing
Carpediemonas membranifera and Jakoba incarcerata were grown as described previously (Simpson
and Patterson 1999; Bernard, Simpson, and Patterson
2000; Simpson and Patterson 2001), and genomic DNA
were extracted using a Puregene kit (Gentra Systems).
Reclinomonas americana (ATCC 50283) genomic DNA
was isolated from a cell pellet obtained from the American Type Culture Collection (Rockville, MD) by standard SDS lysis, proteinase K digestion, and organic extractions. DNA from Jakoba libera and Malawimonas
jakobiformis was provided kindly by Franz Lang (Université de Montréal). Sequences were amplified by PCR
(see below). Reclinomonas americana SSU rRNA genes
were cloned into M13 mp18 and mp19. All other amplification products reported were cloned in pGEMt or
pGEMt easy (Promega Biotech). Bidirectional sequencing was performed using Li-Cor 4000 or 4200 automated sequencers, confirming earlier manual sequencing
in the case of R. americana.
For C. membranifera, SSU rRNA genes were amplified as two overlapping fragments using the 59-end
primer ‘‘A’’ (Medlin et al. 1988) with the internal primer
CCGTCAAATYCTTTAAGTTTC, and the 39-end primer ‘‘B’’ (Medlin et al. 1988) with the internal primer
‘‘Dip-F’’ (Silberman et al. 2002). Annealing temperatures were 45–508C. Pools of 13 and 11 clones of the
respective fragments were sequenced, with heterogeneity detected at six sites. For J. incarcerata, J. libera, M.
jakobiformis, and R. americana, SSU rRNA genes were
amplified using primers A and B (annealing temperature
378C). For J. incarcerata, eight clones were pooled for
sequencing, with seven heterogeneous sites being found.
Heterogeneities were recorded as ambiguities. Ten
clones were pooled for sequencing for J. libera and M.
jakobiformis. Five and eight M13 mp18 and mp19 R.
americana clones, respectively, were pooled for sequencing. No heterogeneity was detected for the latter
three taxa.
Near-complete (;90%) a- and b-tubulin genes
from C. membranifera were amplified using universal
primers (Edgcomb et al. 2001) in the presence of acetamide (5% w/v) at annealing temperatures of 45–508C.
Two and six near-identical clones of a- and b-tubulin,
respectively, were examined; one of each was completely sequenced.
Phylogenetic Analysis—SSU rRNA
The new SSU rRNA gene sequences were aligned
by eye with sequences from a diversity of other eukaryotes, and two prokaryotic outgroups (45 taxa in total).
Within well-accepted groups, taxa that contribute short
1784
Simpson et al.
terminal branches were selected wherever possible.
Some groups, notably microsporidia, were omitted altogether. Initial phylogenetic analyses used 717 sites. A
second alignment of 1,018 sites excluded prokaryote,
diplomonad, and Entamoeba sequences. Alignments are
available at http://hades.biochem.dal.ca/Rogerlab. All
analyses were performed using PAUP* 4.0b4 (Swofford
2000), as described below.
For the maximum-likelihood (ML) analysis, the Tajima-Nei substitution model was selected, with a ‘‘fourcategory discrete G distribution plus invariable sites’’
allowance for among-site rate heterogeneity, after likelihood ratio tests in MODELTEST v3.0 (Posada and
Crandall 1998). The parameters were estimated in
PAUP* from a maximum-parsimony tree. Rate heterogeneity was estimated to be relatively strong (a 5
0.631; proportion of invariable sites 5 0.194). Treespace
was explored with 20 random taxon additions, with tree
bisection-reconnection (TBR) rearrangements. A bootstrap analysis with 108 replicates was performed (neighbor-joining starting trees, with TBR).
Three minimum-evolution distance analyses were
performed. First, ML estimates of pairwise distances
(MLdist) were calculated using the same model as for
the full ML analysis but with eight categories of variable
sites. Second, the LogDet method was used, after removal of constant sites. Third, the simple Kimura two
parameter (K2P) model was used. In each case treespace
was searched using 100 random taxon additions with
TBR, and a 1,000 replicate bootstrap analysis was performed (five random additions and TBR).
A parsimony (Pars) analysis with a 1,000 replicate
bootstrap was performed using the same search strategies as in the distance analyses.
This series of analyses was repeated with three restricted data sets. In the first, Giardia and Entamoeba
sequences were removed. In the second, all diplomonads
and Entamoeba were removed. In the third, prokaryotes,
diplomonads, and Entamoeba were all excluded (retaining 37 eukaryote sequences), but the 1,018 site alignment was examined. All parameter estimations and
searches were performed as mentioned above (ML bootstraps included 100–109 replicates).
One ‘‘long branch’’ that was excluded from consideration in the analyses described above was the oxymonad Pyrsonympha sp. This sequence branched
strongly with Trimastix in a previous study (Dacks et
al. 2001). To examine whether this Trimastix-oxymonad
relationship was robust to the addition of more excavate
taxa, we added the Pyrsonympha sequence to our 37
taxa, 1,018 sites data set. The inclusion of Pyrsonympha
mandated the exclusion of another eight sites. Using this
data set, we performed the same series of analyses as
described above, excepting ML.
We used the 37 taxa, 1,018 sites data set to assess
the evidence for polyphyly of the excavate taxa using
two methods for examining the plausibility of alternative tree topologies, Shimodaira-Hasegawa (SH) tests
(Shimodaira and Hasegawa 1999) and the ‘‘expected
likelihood weights’’ method of Strimmer and Rambaut
(Strimmer and Rambaut 2002). Preliminary to these
tests, using PAUP* 4.0b8, we performed a heuristic
search (one random addition sequence, with TBR) using
the same ML model as before, and saved the 5,000 best
trees found. The ML tree found was the same as in our
original analysis. We also searched for the ML trees under two topological constraints. In the first, the excavate
taxa were constrained to form a clade. In the second,
excavate taxa plus Euglenozoa and parabasalids were
constrained to form a clade (previous data suggests that
Euglenozoa and parabasalids are related to Heterolobosea and diplomonads, respectively (Baldauf et al. 2000)
and could fall ‘‘within’’ a clade of excavate taxa).
For the SH approach, the test implementation in
PAUP* 4.0b8 was used, using the ‘‘RELL’’ method with
1,000 bootstrap replicates. First, the two ML-constrained
trees were compared with the overall ML tree (i.e., three
trees in total). Then, these three ML trees were compared together with the next 100 best trees found in the
unconstrained search and then with the next 500 best
trees. For contrast, the three ML trees only were also
compared using one-tailed Kishino-Hasegawa tests.
There is concern about the utility of SH tests because they appear conservative (i.e., unlikely to reject)
relative to other tests and can give very different results
depending on which trees are considered (Andersson
and Roger 2002; Silberman et al. 2002). But the most
well-known alternative, parametric bootstrapping, may
be overly prone to reject in practice (Strimmer and Rambaut 2002) and is also very expensive computationally.
Another approach is the expected likelihood weights
method (Strimmer and Rambaut 2002). This test distributes a ‘‘likelihood weight’’ across a set of test trees,
averaged over a number of bootstrap replicates, considering the minimal collection of trees that contribute a
given fraction of the total likelihood weight as the confidence set. In principle, the test requires the consideration of all reasonable trees, a condition often intractable
in practice. For this study we approximated this condition, while keeping within sensible computational
bounds by using the best trees from the ML bootstrap
analysis to provide a broad sample from the landscape
of ‘‘reasonable trees’’. For the test we therefore included
the three ML trees (see above) together with the best
trees from 100 bootstrap replicates. We considered using
a greater number of MLdist bootstrap trees but found
that they conferred substantially lower likelihood on the
original data than did the bulk of the ML bootstrap trees,
and thus could not be well sampling the best trees under
the full ML criterion. For the test, 1,000 bootstrap replicates were generated using SEQBOOT (Felsenstein
2000), and the likelihoods for all 103 considered trees
were calculated for each replicate using PAUP*, using
the same model as described previously but with the
parameter values for each replicate optimized on a Jukes
and Cantor distance neighbor-joining tree. The PAUP
file was created using a Perl script, elw.pl. Expected
likelihood weights for each tree were then calculated
using calcwts.pl (elw.pl and calcwts.pl are available on
request from A.J.R.; [email protected]).
Carpediemonas is Related to Diplomonads
1785
FIG. 1.—Small subunit ribosomal RNA phylogenies of eukaryotes. New sequences are denoted by large type. Excavate taxa are shaded.
ML BS values are given on internal branches. Unmarked branches have ,50% BS. Estimated-rate scale bar applies for both phylogenies. A.
ML topology for the ‘‘full’’ 45 taxon, 717 site analysis (2ln L 5 9872.93). The apical region (5 eukaryotic crown) of the tree is denoted (M.
balamuthi falls in the basal portion with other methods). Bootstrap support values for the Carpediemonas 1 retortamonads 1 diplomonads
clade with all methods are displayed in the box. B. ML topology for the 717 site analysis with all diplomonads (and Entamoeba) excluded (2ln
L 5 8127.35). Bootstrap support values for the Carpediemonas 1 retortamonads clade with all methods are boxed.
Phylogenetic Analysis—Tubulins
The a- and b-tubulin genes from C. membranifera
were converted to inferred amino acids that were aligned
by eye with homologs from other eukaryotes. Taxa were
selected to sample diversity broadly, while excluding
highly divergent sequences. Notable omissions include
microsporidia and (other) long-branching fungi, leaving
chytrids to represent fungi sensu lato. All excavate taxa
except retortamonads and Trimastix were represented. A
combined a- 1 b-tubulin data set was similarly constructed from organisms where both genes are known.
To improve taxon representation, four ‘‘composite taxa’’
were also included: ‘‘parabasalid’’ (Monotrichomonas
sp. a-tubulin and Trichomonas vaginalis b-tubulin);
‘‘stramenopile’’ (a-tubulin from Pelvetia fastigiata, and
b- tubulin from Pythium ultimum); ‘‘Cercomonas’’
(Cercomonas sp. ATCC 50319 a-tubulin and Cercomonas sp. ATCC 50316 b-tubulin) and ‘‘Leishmania’’
(Leishmania donovani a-tubulin and L. mexicana btubulin).
All three data sets were analyzed by ML using
PROML 3.6a (Felsenstein 2000), with a model incorporating four site-rate categories. The rates approximated a G distribution estimated by TREE-PUZZLE 4.0.2
(Strimmer and von Haeseler 1996). The PAM substitution matrix was used (the only matrix available with
PROML 3.6a). Treespace was explored by 20 random
taxon additions with ‘‘global rearrangements’’. Bootstrap analyses with 250 replicates were also performed
(random taxon addition, with rearrangement).
For the distance analyses (MLdist), ML estimates
of pairwise distances were calculated by TREE-PUZZLE using the JTT substitution matrix, with eight categories of site-rates approximating a G distribution.
FITCH (Felsenstein 2000) was used to explore treespace
(Fitch-Margoliash criterion; 100 random additions, with
rearrangement). Bootstrap analyses (1,000 replicates)
were performed using PUZZLEBOOT 1.03 (http://
www.tree-puzzle.de/puzzleboot.sh) and FITCH (five additions, with rearrangement).
Parsimony analyses (with 1,000 replicate bootstraps) were completed in PAUP* using identical search
strategies to the SSU rRNA analyses.
All analyses of tubulin data were repeated with sequences from ‘‘hexamitid’’ diplomonads (Hexamita and
Spironucleus) excluded.
Results
Small Subunit Ribosomal RNA
As with previous studies, the trees from the 45 taxon SSU rRNA data set can be viewed as having two
regions, an ‘‘apical region’’ of mostly short branches
(widely referred to as the ‘‘eukaryotic crown’’) and a
‘‘base’’ of long branches (fig. 1A). The apical region
includes opisthokonts (represented by an animal, a fungus, and a choanoflagellate), Viridiplantae, alveolates,
stramenopiles, haptophytes, cryptomonads, the so-called
‘‘Cercozoa’’ (represented by a testate amoeba and Thaumatomonas sp.), rhodophytes, Acanthamoeba, Apusomonas, Trimastix and, in the ML analysis only, the pe-
1786
Simpson et al.
Table 1
Bootstrap Support for the Carpediemonas 1 Diplomonads
1 Retortamonads Clade in the SSU rRNA Analyses
ML MLdist LogDet K2P
All taxa . . . . . . . . . . . . . . . . . . . .
No Giardia, Entamoeba . . . . . . .
No diplomonads, Entamoeba. . .
No prokaryotes, diplomonads,
Entamoeba. . . . . . . . . . . . . . . .
Pars
92
99
98
58
74
78
71
81
75
48
79
86
52
54
74
98
99
99
100
90
lobiont Mastigamoeba balamuthi. The base includes
Mycetozoa (represented by myxogastrid and dictyostelid
slime molds), Entamoeba, parabasalids, heteroloboseids,
diplomonads, and retortamonads. Parabasalids are usually the most basal eukaryotic branch (51%–86% bootstrap support [BS]), although the ML tree itself does not
have this topology. Opisthokonts and Cercozoa always
form a weak clade (,55% BS). Diplomonads and retortamonads always form a strong clade (.90% BS).
Otherwise, relationships among the ‘‘major groups’’ are
poorly resolved.
Of the new sequences, those from Reclinomonas,
J. libera, J. incarcerata and Malawimonas all fall within
the apical region. Reclinomonas and J. libera form a
strongly supported clade with all methods (94%–100%
BS). Malawimonas forms a very weak clade with Trimastix with ML and Pars (,25% BS). Neither clade
clusters with the other or with J. incarcerata nor do they
show any particular affinity for any other group of
eukaryotes.
In contrast, Carpediemonas falls within the base,
grouping specifically with diplomonads and retortamonads with all methods. This clade is weakly supported in
Pars, K2P, and MLdist analyses (48%–58% BS), but
support is quite strong with LogDet (71%) and likelihood (92%; see table 1).
Two factors are focused upon as potential sources
of error in SSU rRNA analyses; base composition heterogeneity and highly divergent sequences. Both might
be relevant to the placement of Carpediemonas in our
analysis because sequences from Giardia are G1C rich,
and sequences from diplomonads all contribute very
long branches. We therefore performed two subsidiary
‘‘717 site’’ analyses. First, sequences from Giardia were
excluded (along with the G1C poor Entamoeba). Second, all other diplomonads were also excluded, leaving
the shorter branching retortamonads to represent the diplomonads 1 retortamonads clade. In both analyses the
optimal trees are very similar to the first analysis, but
the BS for Carpediemonas as sister to diplomonads-retortamonads increases with all methods, climbing to
over 70% in all cases except one Pars run (table 1).
Results are very similar when Entamoeba is retained
(data not shown). Notably, when all diplomonads are
excluded the ML tree loses the classical ‘‘crown plus
base’’ structure, with stramenopiles appearing in the
most basal position (fig. 1B). Most of the long branches
form a cluster, but Carpediemonas 1 retortamonads instead form a very weakly supported clade with the shortbranching excavate taxa Trimastix and Malawimonas.
In the final analyses prokaryotic outgroups were
excluded (in addition to diplomonads and Entamoeba),
permitting more sites to be included but making it impossible to root the tree (figure not shown). Resolution
of the relationships amongst most major groups remains
poor, except that the support for a clade of opisthokonts
and Cercozoa increases (55%–83% BS), and alveolates
are recovered as a strong clade with all methods (70%1
BS). Nonetheless, the Carpediemonas 1 retortamonads
grouping is always strongly supported (90%1 BS; table
1), although it still does not fall with other excavate
taxa. The BS for the J. libera 1 Reclinomonas clade is
100% with all methods, but J. incarcerata still branches
separately. Malawimonas and Trimastix form a very
weak clade with ML and LogDet only. When the oxymonad Pyrsonympha is included this sequence forms a
strong clade with Trimastix to the exclusion of other
excavate taxa and other eukaryotes (67% BS with Pars,
81%–95% BS with other methods). The branching order
and support values amongst eukaryotes are otherwise
very similar to those from other analyses.
Using the SH test, we further examined evidence
for the polyphyly of excavate taxa by comparing (1) the
overall ML tree (2ln L 11922.75), (2) the best tree in
which excavate taxa alone form a clade (2ln L
11948.03), and (3) the best tree in which excavate taxa,
parabasalids, and Euglenozoa collectively form a clade
(2ln L 11940.44). This test does not reject either alternative tree (2 and 3; P 5 0.165 and 0.266, respectively).
When more ‘‘good’’ trees are considered the margin
from rejection increases further (with 100 next best trees
added, P 5 0.291 and 0.451, respectively; with 500 next
best trees, P 5 0.394 and 0.578, respectively). It seems
likely that SH tests would fail to reject the alternative
trees irrespective of which other trees were included in
the test. More familiar (but technically invalid for this
question—Goldman, Anderson, and Rodrigo 2000) individual Kishino-Hasegawa tests also fail to reject the
alternative trees (P 5 0.155 and 0.221, respectively).
With the expected likelihood weights test, the overall ML tree receives the highest score of the 103 trees
included. Alternative tree 3 is included in the 46% confidence set, whereas alternative tree 2 is included in the
82% confidence set. As such, the test does not reject
either alternative tree, although given our crude sampling of ‘‘reasonable treespace’’ the result for tree 2 at
least should be considered tentative.
Tubulins
The basic topology of all the tubulin trees can be
viewed as comprising two regions, an ‘‘animals, chytrids
(fungi), diplomonads and parabasalids’’ cluster and a
‘‘plants plus other well-recognized groups of protists’’
cluster (figs. 2 and 3). Previously sequenced excavate
taxa aside from diplomonads (i.e., Heterolobosea, core
jakobids and Malawimonas) mostly fall as separate lineages within the ‘‘plant-protist’’ cluster. But the J. incarcerata a-tubulin sequence instead falls either as the
sister to parabasalids (ML) or to diplomonads 1 Carpediemonas (MLdist and Pars—see below), always with
Carpediemonas is Related to Diplomonads
1787
FIG. 2.—Tubulin phylogenies. Excavate taxa are shaded. ML BS values are shown. Unmarked branches have ,50% BS. Support values
for the Carpediemonas 1 diplomonads clade with all methods are displayed in the boxes. Estimated-rate scale bar applies for both phylogenies.
The phylogenies are unrooted. A. a-tubulin, ML topology (2ln L 5 5011.23). B. b-tubulin, ML topology (2ln L 5 4854.69).
,50% BS. In the b-tubulin and combined tubulin analyses Reclinomonas and J. libera form a fairly strong
clade (.90% BS with b-tubulin, 65%–89% with combined tubulin) but fall separately to J. incarcerata. Reclinomonas and J. libera branch separately in a-tubulin
analyses.
Carpediemonas is placed as the sister group to diplomonads with all data sets and methods examined.
This position is very strongly supported (.90% BS) in
all a-tubulin and combined analyses. Support is weaker
in the b-tubulin analyses (51%–71% BS; table 2). This
diplomonads 1 Carpediemonas clade clusters with the
parabasalids in the b-tubulin analyses, and, weakly, in
the ML analysis of the combined data set (table 2). Although not present in the best trees, this clustering also
attracts some BS in the MLdist and Pars analyses of the
combined data set (table 2).
The Carpediemonas branches are fairly long but
are markedly shorter than those of the hexamitid diplomonads Hexamita and Spironucleus. When the hexamitids are excluded, leaving the less divergent Giardia
sequences to represent diplomonads, all methods return
very similar best trees. In the a-tubulin and combined
analyses the bootstrap values for the diplomonads 1
Carpediemonas clade remain very high with all methods
(90%1). For b-tubulin, bootstrap values decline with
Pars but improve with MLdist and, particularly, with
likelihood (table 2). By contrast, BS for a sister group
relationship between parabasalids and the diplomonads
1 Carpediemonas clade declines with all methods or
remains negligible in the case of a-tubulin (table 2).
Discussion
The Positions of the Excavate Taxa
This study expands the sampling of SSU rRNA
genes from excavate taxa such that all recognized major
Table 2
Bootstrap Support for the Carpediemonas 1 Diplomonads
Clade (Diplo, Carpe) and the Parabasalids 1
Carpediemonas 1 Diplomonads Clade (Para, Diplo,
Carpe) in the Tubulin Analyses
(DIPLO, CARPE)
FIG. 3.—Combined a- 1 b-tubulin gene phylogeny. Excavate
taxa are shaded. Unrooted ML topology depicted (2ln L 5 9652.59).
ML BS values are shown. Unmarked branches have ,50% BS. Support values for the Carpediemonas 1 diplomonads clade with all methods are boxed.
a-Tubulin, all taxa . . . .
a-Tubulin, no hex . . . .
b-Tubulin, all taxa . . . .
b-Tubulin, no hex . . . .
Combined, all taxa. . . .
Combined, no hex . . . .
(PARA, DIPLO, CARPE)
ML MLdist Pars
ML MLdist
Pars
93
90
56
71
95
98
,5*
,5*
66
42
52
43
,5*
,5*
77
67
22*
18*
100
100
51
56
99
100
94
96
71
51
98
97
,5*
,5*
78
43
44*
34*
NOTE.—‘‘No hex’’ signifies data sets where hexamitid diplomonads are
excluded.
* Signifies the absence of the clade from the best trees for an analysis.
1788
Simpson et al.
groups are now represented. This is the first molecular
data set with such coverage. Our addition of sequences
from Carpediemonas gives good (but not comprehensive) coverage of excavate taxa for the a- and b-tubulin
data sets.
Broadly speaking, our results are congruent with
recent molecular phylogenetic studies. The close relationship between retortamonads and diplomonads and
between J. libera and R. americana in SSU rRNA gene
trees confirms recent examinations with more limited
taxon sampling (and unpublished sequences in the latter
case) (Cavalier-Smith 2000; Silberman et al. 2002). The
remoteness of Malawimonas and J. incarcerata from the
J. libera 1 Reclinomonas clade in both SSU rRNA and
tubulin trees agrees with recent tubulin phylogenies
(Edgcomb et al. 2001). The marked nonmonophyly of
the excavate taxa (forming at least four separated clades)
mirrors all previous molecular phylogenetic studies that
include several excavate taxa, with one recent, unconfirmed exception (Cavalier-Smith 2002).
The molecular evidence for polyphyly of the excavate taxa should be viewed cautiously. The branches
separating excavate taxa usually receive only weak BS
and are often not consistent across different data sets
and methods. For example, considering ML analyses,
the only branches separating excavate taxa that receive
greater than 50% BS are branches in tubulin trees linking Carpediemonas, diplomonads, and either Malawimonas or J. incarcerata (not both) to opisthokonts.
There is no measurable support for such relationships
with our SSU rRNA data (where opisthokonts instead
form a moderately well-supported clade with Cercozoa).
When used, statistical tests of alternative tree topologies
did not reject trees in which excavate taxa form a clade
(with or without Euglenozoa and parabasalids). It is possible that the apparent polyphyly of the excavate taxa is
due to analysis artifacts caused by the wide variation in
inferred evolutionary rate of the sequences examined. It
is also possible that the divergences between some excavate taxa were extremely ancient or rapid, beyond the
effective resolving power of our data sets. The disparity
between molecular trees indicating polyphyly and morphological arguments that excavate taxa are descended
from a common excavate ancestor (Simpson and Patterson 1999; Simpson, Bernard, and Patterson 2000) requires further exploration.
The Position of Carpediemonas
Our analyses strongly suggest that Carpediemonas
is the closest known relative of diplomonads and retortamonads. The relationship is recovered with a variety
of molecular markers, analysis methods, taxon sets, and
alignments. For all data sets studied, diplomonads, retortamonads, and Carpediemonas constitute ‘‘long
branches’’; however, several considerations suggest their
clustering is not a long-branch attraction artifact. First,
other similarly long branches fail to attract Carpediemonas to the exclusion of diplomonads and retortamonads (in some cases where diplomonads are excluded, the
retortamonads 1 Carpediemonas cluster actually falls
separately from other long branches—fig. 1B). Second,
in most analyses, support for the position of Carpediemonas is higher with methods expected to be more resilient to long-branch artifacts (e.g., ML with models of
among-site rate variation), compared with more susceptible methods (e.g., Pars, K2P distances). If anything,
the opposite result might be expected if long-branch attraction was primarily responsible for the position of
Carpediemonas. Third, removal of the most divergent
diplomonad sequences almost always increases support
for the position of Carpediemonas or maintains it at
very high levels. Again, it might be expected that support would decrease if the association were due to longbranch attraction.
Before this study, parabasalids were seen by many
as the best candidate for a sister group to diplomonads
and retortamonads. This proposal was based on several
protein phylogenies, none of which included any other
excavate taxa except Heterolobosea (Embley and Hirt
1998; Roger et al. 1998; Roger 1999; Baldauf et al.
2000). With the data sets examined here, parabasalids,
at closest, are the sister group to the diplomonads-retortamonads 1 Carpediemonas clade. Even this position
receives support only from the b-tubulin and combined
tubulin data sets. Furthermore, exclusion of the longbranch Spironucleus sequences from these data sets
markedly reduces the support for this position for parabasalids, hinting at an influence of long-branch attraction. It would not be surprising if evidence emerged
placing other excavate taxa inside the minimal clade
containing diplomonads and parabasalids.
The Mitochondrial Status of Diplomonads
Carpediemonas has double membrane-bounded organelles that structurally resemble hydrogenosomes-mitochondria (Simpson and Patterson 1999). We can envisage two simple scenarios for the symbiotic origin of
these organelles. First, they could be descended from
the same symbiosis as mitochondria. Second, they could
be degenerate descendants of a different prokaryotic
symbiont. At present we lack the genetic and biochemical data to directly distinguish between these two possibilities. But similar enigmatic organelles in other amitochondriate eukaryotes, such as the hydrogenosomes of
parabasalids, some ciliates and some chytrids, and the
mitosome of Entamoeba, have all turned out to be mitochondrial homologs when examined in detail (Roger
1999; Dyall and Johnson 2000; van der Giezen et al.
2002). It would seem most likely that this may also be
the case for Carpediemonas. As such, there is now a
candidate for a cytological mitochondrial relic in the diplomonads 1 retortamonads 1 Carpediemonas clade in
addition to the possible genetic relics (e.g., cpn60, IscS)
identified previously (Roger et al. 1998; Horner and Embley 2001; Tachezy, Sánchez, and Müller 2001). Provided there was a single origin for the mitochondrial
organelle in eukaryotes (Gray, Burger, and Lang 1999),
strong evidence that the Carpediemonas organelle is descended from the same symbiosis would make premitochondrial scenarios for diplomonads (including ‘‘tran-
Carpediemonas is Related to Diplomonads
1789
sient symbiont’’ models such as the one suggested by
Chihade et al. 2000) very difficult to sustain.
lutionary position and biology of Carpediemonas is well
warranted.
The Deep-Branching Status of Diplomonads
Supplementary Information
The molecular phylogenetic argument that diplomonads are ‘‘early-diverged’’ eukaryotes originated
with SSU rRNA trees in which Giardia was one of the
earliest emerging eukaryote lineages. It has been argued
that this position may be an artifact of long-branch attraction (Philippe and Adoutte 1998). But no previous
phylogenetic analysis of ‘‘all alignable sites’’ that we
know of has been able to ‘‘overcome’’ the inferred artifact and place diplomonads topologically remote from
prokaryote outgroups, even when BS for this basal position, and for the whole tree backbone, is low (as in
our analyses; fig. 1A). This has helped sustain the argument that the diplomonads really are very deepbranching eukaryotes.
Retortamonads and Carpediemonas have markedly
less divergent SSU rRNA genes than diplomonads, inviting their use as surrogates for diplomonads in phylogenetic analyses. When diplomonads are excluded, our
ML analysis places retortamonads 1 Carpediemonas
deeply embedded within the short-branching eukaryotes,
topologically remote from the prokaryotes, although the
tree backbone remains poorly supported (fig. 1B). The
consistently deep placement for diplomonads in SSU
rRNA trees thus does not seem to be a universal property of the entire diplomonads 1 retortamonads 1 Carpediemonas clade, illustrating the weakness of the ‘‘diplomonads early’’ case with this marker.
The sequences reported here have been deposited
with GenBank (accession nos. AY117416–AY117422).
A Window into the Evolution of Diplomonads
By virtue of its inferred phylogenetic position, Carpediemonas could be an important organism in understanding the evolution of the molecular and cellular attributes of diplomonads. The little that is known about
Carpediemonas hints at a much more ‘‘typical’’ eukaryote than are diplomonads or retortamonads. Carpediemonas is free-living, whereas most diplomonads and retortamonads are parasites or endocommensals, and it is
possible that the entire clade is ancestrally endobiotic
(Siddall, Hong, and Desser 1992). Carpediemonas has
a well-developed endomembrane system, including a
normal-looking Golgi apparatus (Simpson and Patterson
1999), a component that is cryptic in diplomonads and
retortamonads (Brugerolle 1991; Gillin, Reiner, and
McCaffery 1996). Most interestingly, although Carpediemonas lacks classical mitochondria it does possess
double membrane-bounded organelles. No similar organelle of any kind has been detected in diplomonads
or retortamonads. An intriguing possibility consistent
with our phylogenies is that Carpediemonas approaches
an intermediate stage in a process of mitochondrion loss
by the diplomonad lineage. The limited data available
make this suggestion speculative at present. But given
the potential for insight into the process by which mitochondria may have been discarded by the canonical
amitochondriate eukaryotes, greater attention to the evo-
Acknowledgments
Thanks to M. Holder for computational assistance
and to J. Andersson and J. Dacks for critical comments.
This work was supported by a NAI Astrobiology grant
NCC2-1054 and a NASA Exobiology Grant NG5-4895
to M.L.S., an Australia Research Council grant to D.J.P.,
and a NSERC (Canada) operating grant 227085-00 to
A.J.R. A.G.B.S. is the recipient of a Canadian Institutes
of Health Research postdoctoral fellowship. J.D.S. is
supported by the NASA Astrobiology Institute.
LITERATURE CITED
ANDERSSON, J. O., and A. J. ROGER. 2002. A cyanobacterial
gene in nonphotosynthetic protists—an early chloroplast acquisition in eukaryotes? Curr. Biol. 12:115–119.
ARCHIBALD, J. M., C. J. O’KELLY, and W. F. DOOLITTLE. 2002.
The chaperonin genes of jakobid and jakobid-like flagellates: implications for eukaryotic evolution. Mol. Biol.
Evol. 19:422–431.
BALDAUF, S. L., A. J. ROGER, I. WENK-SIEFERT, and W. F.
DOOLITTLE. 2000. A kingdom-level phylogeny of eukaryotes based on combined protein data. Science 290:972–977.
BERNARD, C., A. G. B. SIMPSON, and D. J. PATTERSON. 2000.
Some free-living flagellates (Protista) from anoxic habitats.
Ophelia 52:113–142.
BRUGEROLLE, G. 1991. Flagellar and cytoskeletal systems in
amitochondrial flagellates: Archamoeba, Metamonada and
Parabasala. Protoplasma 164:70–90.
BRUGEROLLE, G., and D. J. PATTERSON. 1997. Ultrastructure
of Trimastix convexa Hollande, an amitochondriate anaerobic flagellate with a previously undescribed organization.
Europ. J. Protistol. 33:121–130.
CAVALIER-SMITH, T. 1983. A 6-kingdom classification and a
unified phylogeny. Pp. 1027–1034 in W. SCHWEMMLER, and
H. E. A. SCHENK, eds. Endocytobiology. de Gruyter, Berlin.
———. 2000. Flagellate megaevolution: the basis for eukaryote diversification. Pp. 361–390 in B. S. C. LEADBEATER,
and J. C. GREEN, eds. The flagellates; unity, diversity and
evolution. Taylor and Francis, London.
———. 2002. The phagotrophic origin of eukaryotes and phylogenetic classification of Protozoa. Int. J. Syst. Evol. Microbiol. 52:297–354.
CHIHADE, J. W., J. R. BROWN, P. R. SCHIMMEL, and L. RIBAS
DE POUPLANA. 2000. Origin of mitochondria in relation to
evolutionary history of eukaryotic alanyl-tRNA synthetase.
Proc. Natl. Acad. Sci. USA 97:12153–12157.
DACKS, J. B., J. D. SILBERMAN, A. G. B. SIMPSON, S. MORUYA,
T. KUDO, M. OHKUMA, and R. REDFIELD. 2001. Oxymonads
are closely related to the excavate taxon Trimastix. Mol.
Biol. Evol. 18:1034–1044.
DOOLITTLE, W. F. 1998. You are what you eat: a genetic transfer ratchet could account for bacterial genes in eukaryotic
nuclear genomes. Trends Genet. 14:307–311.
DYALL, S. D., and P. J. JOHNSON. 2000. Origins of hydrogenosomes and mitochondria: evolution and organelle biogenesis. Curr. Opin. Microbiol. 3:404–411.
1790
Simpson et al.
EDGCOMB, V. P., A. J. ROGER, A. G. B. SIMPSON, D. KYSELA,
and M. L. SOGIN. 2001. Evolutionary relationships among
‘‘jakobid’’ flagellates as indicated by alpha- and beta-tubulin phylogenies. Mol. Biol. Evol. 18:514–522.
EMBLEY, T. M., and R. P. HIRT. 1998. Early branching eukaryotes? Curr. Opin. Genet. Dev. 8:624–629.
FELSENSTEIN, J. 2000. PHYLIP (phylogeny inference package).
Version 3.6a. Distributed by the author, University of Washington, Seattle.
GILLIN, F. D., D. S. REINER, and J. M. MCCAFFERY. 1996. Cell
biology of the primitive eukaryote Giardia lamblia. Ann.
Rev. Microbiol. 50:679–705.
GOLDMAN, N., J. P. ANDERSON, and A. G. RODRIGO. 2000.
Likelihood-based tests of topologies in phylogenetics. Syst.
Biol. 49:652–670.
GRAY, M. W., G. BURGER, B. F. LANG et al. (15 co-authors).
1999. Mitochondrial evolution. Science 283:1476–1481.
GRAY, M. W., B. F. LANG, R. CEDERGREN et al. 1998. Genome
structure and gene content in protist mitochondrial DNAs.
Nucleic Acids Res. 26:865–878.
HASHIMOTO, T., Y. NAKAMURA, T. KAMAISHI, F. NAKAMURA,
J. ADACHI, K. OKAMOTO, and M. HASEGAWA. 1995. Phylogenetic place of mitochondrion-lacking protozoan, Giardia lamblia, inferred from amino acid sequences of elongation factor 2. Mol. Biol. Evol. 12:782–793.
HASHIMOTO, T., Y. NAKAMURA, F. NAKAMURA, T. SHIRAKURA,
J. ADACHI, N. GOTO, K. OKAMOTO, and M. HASEGAWA.
1994. Protein phylogeny gives a robust estimation for early
divergences of eukaryotes: phylogenetic place of a mitochondria-lacking protozoan, Giardia lamblia. Mol. Biol.
Evol. 11:65–71.
HASHIMOTO, T., L. B. SANCHEZ, T. SHIRAKURA, M. MÜLLER,
and M. HASEGAWA. 1998. Secondary absence of mitochondria in Giardia lamblia and Trichomonas vaginalis revealed
by valyl-tRNA synthetase phylogeny. Proc. Natl. Acad. Sci.
USA 95:6860–6865.
HIRT, R. P., J. M. LOGSDON, B. HEALY, M. W. DOREY, W. F.
DOOLITTLE, and T. M. EMBLEY. 1999. Microsporidia are
related to fungi: evidence from the largest subunit of RNA
polymerase II and other proteins. Proc. Natl. Acad. Sci.
USA 96:580–585.
HORNER, D. S., and T. M. EMBLEY. 2001. Chaperonin 60 phylogeny provides further evidence for secondary loss of mitochondria among putative early-branching eukaryotes.
Mol. Biol. Evol. 18:1970–1975.
KEELING, P. J., and W. F. DOOLITTLE. 1997. Evidence that eukaryotic triosephosphate isomerase is of alpha-proteobacterial origin. Proc. Natl. Acad. Sci. USA 94:1270–1275.
LANG, B. F., G. BURGER, C. J. O’KELLY, R. CEDERGREN, G.
B. GOLDING, C. LEMIEUX, D. SANKOFF, M. TURMEL, and
M. W. GRAY. 1997. An ancestral mitochondrial DNA resembling a eubacterial genome in miniature. Nature 387:
493–497.
MARGULIS, L., M. F. DOLAN, and R. GUERRERO. 2000. The
chimeric eukaryote: origin of the nucleus from the karyomastigont in amitochondriate protists. Proc. Natl. Acad.
Sci. USA 97:6954–6959.
MCARTHUR, A., H. MORRISON, J. NIXON et al. (15 co-authors).
2000. The Giardia genome project database. FEMS Microbiol. Lett. 189:271–273.
MEDLIN, L., H. J. ELWOOD, S. STICKEL, and M. L. SOGIN.
1988. The characterization of enzymatically amplified eukaryotic 16S-like rRNA-coding regions. Gene 71:491–499.
MORRISON, H. G., A. J. ROGER, T. G. NYSTUL, F. D. GILLIN,
and M. L. SOGIN. 2001. Giardia lamblia expresses a protobacterial-like DnaK homolog. Mol. Biol. Evol. 18:530–
541.
O’KELLY, C. J. 1993. The jakobid flagellates: structural features of Jakoba, Reclinomonas and Histiona and implications for the early diversification of eukaryotes. J. Eukaryot.
Microbiol. 40:627–636.
O’KELLY, C. J., M. A. FARMER, and T. A. NERAD. 1999. Ultrastructure of Trimastix pyriformis (Klebs) Bernard et al.:
similarities of Trimastix species with retortamonad and jakobid flagellates. Protist 150:149–162.
PHILIPPE, H., and A. ADOUTTE. 1998. The molecular phylogeny
of Eukaryota: solid facts and uncertainties. Pp. 25–56 in G.
H. COOMBS, K. VICKERMAN, M. A. SLEIGH, and A. WARREN, eds. Evolutionary relationships among protozoa. Chapman and Hall, London.
POSADA, D., and K. A. CRANDALL. 1998. MODELTEST: testing the model of DNA substitution. Bioinformatics 14:817–
818.
ROGER, A. J. 1999. Reconstructing early events in eukaryotic
evolution. Am. Nat. 154:S146–S163.
ROGER, A. J., S. G. SVÄRD, J. TOVAR, C. G. CLARK, M. W.
SMITH, F. D. GILLIN, and M. L. SOGIN. 1998. A mitochondrial-like chaperonin 60 gene in Giardia lamblia: evidence
that diplomonads once harbored an endosymbiont related to
the progenitor of mitochondria. Proc. Natl. Acad. Sci. USA
95:229–234.
SHIMODAIRA, H., and M. HASEGAWA. 1999. Multiple comparisons of log-likelihoods with applications to phylogenetic
inference. Mol. Biol. Evol. 16:1114–1116.
SIDDALL, M. E., H. HONG, and S. S. DESSER. 1992. Phylogenetic analysis of the Diplomonadida (Wenyon, 1926) Brugerolle, 1975: evidence for heterochrony in protozoa and
against Giardia lamblia as a ‘‘missing link. ’’ J. Protozool.
39:361–367.
SILBERMAN, J. D., A. G. B. SIMPSON, J. KULDA, I. CEPICKA,
V. HAMPL, P. J. JOHNSON, and A. J. ROGER. 2002. Retortamonad flagellates are closely related to diplomonads- implications for the history of mitochondrial function in eukaryote evolution. Mol. Biol. Evol. 19:777–786.
SIMPSON, A. G. B., C. BERNARD, and D. J. PATTERSON. 2000.
The ultrastructure of Trimastix marina Kent, 1880 (Eukaryota), an excavate flagellate. Europ. J. Protistol. 36:229–
252.
SIMPSON, A. G. B., and D. J. PATTERSON. 1999. The ultrastructure of Carpediemonas membranifera (Eukaryota) with reference to the excavate hypothesis. Europ. J. Protistol. 35:
353–370.
———. 2001. On core jakobids and excavate taxa: the ultrastructure of Jakoba incarcerata. J. Eukaryot. Microbiol. 48:
480–492.
SOGIN, M. L. 1997. History assignment: when was the mitochondrion founded? Curr. Opin. Genet. Dev. 7:792–799.
SOGIN, M. L., J. H. GUNDERSON, H. J. ELWOOD, R. A. ALONSO,
and D. A. PEATTIE. 1989. Phylogenetic significance of the
kingdom concept: an unusual eukaryotic 16S-like ribosomal
RNA from Giardia lamblia. Science 243:75–77.
STILLER, J. W., and B. D. HALL. 1999. Long-branch attraction
and the rDNA model of early eukaryotic evolution. Mol.
Biol. Evol. 16:1270–1279.
STRIMMER, K., and A. RAMBAUT. 2002. Inferring confidence
sets of possibly misspecified gene trees. Proc. R. Soc. Lond.
B. 269:137–142.
STRIMMER, K., and A. VON HAESELER. 1996. Quartet puzzling:
a quartet maximum-likelihood method for reconstructing
tree topologies. Mol. Biol. Evol. 13:964–969.
SWOFFORD, D. L. 2000. PAUP*: phylogenetic analysis using
parsimony (*and other methods). Version 4. Sinauer Associates, Sunderland, Mass.
Carpediemonas is Related to Diplomonads
TACHEZY, J. L. B. SÁNCHEZ, and M. MÜLLER. 2001. Mitochondrial type iron-sulfur cluster assembly in the amitochondriate eukaryotes Trichomonas vaginalis and Giardia
intestinalis, as indicated by the phylogeny of IscS. Mol.
Biol. Evol. 18:1919–1928.
VAN DER GIEZEN, M., D. J. SLOTBOOM, D. S. HORNER, P. L.
DYAL, M. HARDING, G. P. XUE, T. M. EMBLEY, and E. R.
1791
S. KUNJI. 2002. Conserved properties of hydrogenosomal
and mitochondrial ADP/ATP carriers: a common origin for
both organelles. EMBO J. 21:572–579.
GEOFFREY MCFADDEN, reviewing editor
Accepted June 11, 2002