Effects of Sequence Alignment and Structural Domains of Ribosomal DNA
on Phylogeny Reconstruction for the Protozoan Family Sarcocystidae
N. B. Mugridge,* D. A. Morrison,* T. Jäkel,† A. R. Heckeroth,‡ A. M. Tenter,‡ and
A. M. Johnson*
*Molecular Parasitology Unit, University of Technology, Sydney, New South Wales, Australia; †Division of Parasitology,
University of Hohenheim, Stuttgart, Germany; and ‡Institut für Parasitologie, Tierärztliche Hochschule Hannover, Hannover,
Germany
Finding correct species relationships using phylogeny reconstruction based on molecular data is dependent on several
empirical and technical factors. These include the choice of DNA sequence from which phylogeny is to be inferred,
the establishment of character homology within a sequence alignment, and the phylogeny algorithm used. Nevertheless, sequencing and phylogeny tools provide a way of testing certain hypotheses regarding the relationship
among the organisms for which phenotypic characters demonstrate conflicting evolutionary information. The protozoan family Sarcocystidae is one such group for which molecular data have been applied phylogenetically to
resolve questionable relationships. However, analyses carried out to date, particularly based on small-subunit ribosomal DNA, have not resolved all of the relationships within this family. Analysis of more than one gene is
necessary in order to obtain a robust species signal, and some DNA sequences may not be appropriate in terms of
their phylogenetic information content. With this in mind, we tested the informativeness of our chosen molecule,
the large-subunit ribosomal DNA (lsu rDNA), by using subdivisions of the sequence in phylogenetic analysis
through PAUP, fastDNAml, and neighbor joining. The segments of sequence applied correspond to areas of higher
nucleotide variation in a secondary-structure alignment involving 21 taxa. We found that subdivision of the entire
lsu rDNA is inappropriate for phylogenetic analysis of the Sarcocystidae. There are limited informative nucleotide
sites in the lsu rDNA for certain clades, such as the one encompassing the subfamily Toxoplasmatinae. Consequently,
the removal of any segment of the alignment compromises the final tree topology. We also tested the effect of using
two different alignment procedures (CLUSTAL W and the structure alignment using DCSE) and three different
tree-building methods on the final tree topology. This work shows that congruence between different methods in
the formation of clades may be a feature of robust topology; however, a sequence alignment based on primary
structure may not be comparing homologous nucleotides even though the expected topology is obtained. Our results
support previous findings showing the paraphyly of the current genera Sarcocystis and Hammondia and again bring
to question the relationships of Sarcocystis muris, Isospora felis, and Neospora caninum. In addition, results based
on phylogenetic analysis of the structure alignment suggest that Sarcocystis zamani and Sarcocystis singaporensis,
which have reptilian definitive hosts, are monophyletic with Sarcocystis species using mammalian definitive hosts
if the genus Frenkelia is synonymized with Sarcocystis.
Introduction
For taxa that cannot be unequivocally classified
based on phenotypic markers, DNA sequencing provides another marker for the investigation of phylogenetic relationships (Olsen and Woese 1993). The relative
ease with which nucleotide sequences can be obtained
today allows one to compare a large number of taxa
based on nucleotide variations; hence, the phylogenetic
informativeness of DNA now has the potential to outweigh that of the morphology or biology of an organism.
However, the informativeness of the gene being used
requires consideration if one is to make assumptions regarding species relationships (Olsen and Woese 1993).
One way of ensuring a robust phylogeny in which the
species signal, rather than the gene signal, emerges is to
analyze numerous genes. Consequently, the DNA sequence chosen for the construction of a phylogeny is a
very important factor, due to the varying phylogenetic
Key words: Hammondia, large-subunit ribosomal DNA, Neospora, nucleotide sequence alignment, phylogeny, Sarcocystidae, Sarcocystis, Toxoplasma.
Address for correspondence and reprints: A. M. Johnson, Molecular Parasitology Unit, University of Technology, Sydney, Westbourne
Street, Gore Hill, NSW 2065, Australia. E-mail:
[email protected].
Mol. Biol. Evol. 17(12):1842–1853. 2000
q 2000 by the Society for Molecular Biology and Evolution. ISSN: 0737-4038
1842
content among genes. The phylogenetic informativeness
of a gene can be a factor of the function of its product
or a result of its length. In some cases, such as that of
the stems and loops of the ribosomal RNA, different
parts of the gene can have varying information contents
due to evolutionary selection pressures on secondary
structure and function (Sogin and Gunderson 1986; Olsen 1987, 1988; Olsen and Woese 1993).
The different methods available for building a phylogenetic tree can also affect the outcome of the analysis. Tree-building algorithms have different aims and
assumptions. Therefore, congruence between tree topologies resulting from the use of varying algorithms can
also be seen as a sign of a robust topology. The process
of sequence alignment is also crucial, as it has been
shown that nonhomologous alignment is the major cause
of erroneous phylogenies (Morrison and Ellis 1997).
Previous studies have shown that using a primary-structure alignment instead of a secondary-structure alignment can influence how well cladistic relationships
within the Sarcocystidae are resolved (Ellis and Morrison 1995).
Ribosomal RNA is a useful molecule for phylogenetic analysis because it holds certain advantages for
such studies. The conserved core regions that are interspersed among variable stretches of sequence assist in
Sequence Alignment and Structure Domains
the design of primers that can then be used to amplify
even the more divergent areas. This mosaic structure
also allows flexibility in experimental design of phylogenetic investigations so that closely related and less
closely related organisms can be studied (Sogin and
Gunderson 1986; Olsen 1987; Olsen and Woese 1993).
Study of the phylogenetic informativeness of the
large-subunit ribosomal DNA (lsu rDNA) helped us to
resolve some unanswered questions regarding the Sarcocystidae. This family includes heteroxenous (two-host
life cycle) coccidian parasites that have the ability to
form tissue cysts in intermediate hosts. The family is
usually divided into a range of different genera, such as
Sarcocystis, Frenkelia, Toxoplasma, Besnoitia, Hammondia, Neospora, and Cystoisospora (Frenkel, Mehlhorn, and Heydorn 1987; Levine 1988; Current, Upton,
and Long 1990; Tenter and Johnson 1997). Thus far,
taxa within this family have been classified based on
phenotypic characters, but recent results based on molecular data call for the revision of some of these genera.
Phenotypically, the Sarcocystidae are distinguished from
other coccidia by their heteroxenous life cycle, their
ability to form tissue cysts in the intermediate host, and
the morphology of their oocyst (disporous, tetrazoic).
Division into two or three subfamilies is also based on
phenotypic characters. The genera Sarcocystis and Frenkelia, which make up the subfamily Sarcocystinae, are
obligatorily heteroxenous and develop only inside their
hosts. The subfamily Toxoplasmatinae includes the genera Toxoplasma, Besnoitia, Hammondia, and Neospora,
which collectively consist of far fewer species than belong to the genus Sarcocystis alone. These species differ
from the Sarcocystinae in that they have a different type
of asexual development in the intermediate host, they
have an additional asexual phase of proliferation preceding sexual development in the definitive host, and
they sustain sporogony in the environment (Frenkel
1977; Levine 1988; Current, Upton, and Long 1990). A
third, monogeneric, subfamily, the Cystoisosporinae, has
been described based on the formation of monozoic
cysts in optional intermediate hosts (Frenkel et al. 1979).
However, the genus Cystoisospora has not been widely
accepted and is most frequently synonymized with the
genus Isospora in the family Eimeriidae.
Although the familial grouping of the Sarcocystidae has been widely accepted according to phenotypic
characters, the biological and morphological data regarding current generic groups are sparse and conflicting
(Tenter and Johnson 1997). Thus, in step with molecular
biology advances, this has resulted in the use of other,
more informative, markers, namely DNA sequences, for
classification or reconstruction of phylogeny within the
Sarcocystidae.
Phylogenetic research on the Sarcocystidae and
other members of the phylum Apicomplexa has largely
been based on the small-subunit ribosomal RNA (ssu
rDNA) gene sequence (Tenter and Johnson 1997; Carreno et al. 1998; Votýpka et al. 1998; Carreno and Barta
1999; Dole El et al. 1999; Holmdahl et al. 1999; Jenkins
et al. 1999). It is clear that the smaller size of the ssu
rDNA makes it more popular when longer sequences
1843
are technically harder to obtain, and the helical regions
of the ssu rDNA are phylogenetically more informative
than the nonhelical regions (Ellis and Morrison 1995).
Ssu rDNA failed to be sufficiently informative in some
cases, as evidenced by conflicting phylogenies obtained
for the Sarcocystidae (Jeffries et al. 1997; Tenter and
Johnson 1997; Votýpka et al. 1998). To resolve some
relationships and confirm others postulated from the ssu
rDNA analyses, we recently published work on the entire lsu rDNA (Mugridge et al. 1999a, 1999b), based on
the fact that this molecule is more informative than the
ssu rDNA (Olsen 1987; Cavalier-Smith 1989). This
work confirmed the paraphyly of the genus Sarcocystis,
unless Frenkelia was included, and the paraphyly of the
genus Hammondia, unless it was combined with Toxoplasma gondii and Neospora caninum. The results supported the hypothesis that the genus Frenkelia should be
incorporated into the genus Sarcocystis (Odening 1998)
and validated the biological evidence that showed N.
caninum to be closer to Hammondia heydorni, and T.
gondii to be closer to Hammondia hammondi (Mugridge
et al. 1999a).
Here, we investigate the information content of the
lsu rDNA gene through the subdivision of the nucleotide
sequence and the effect of this on three tree-building
methods, using members of the protozoan family Sarcocystidae as a model. We also include members of another family of coccidia, the Eimeriidae, which contains
homoxenous (one-host life cycle) parasites, such as the
genera Eimeria and Isospora. In addition, we address the
importance of the alignment of sequences by observing
the effect of using two methods, one of which we believe is superior in achieving homologous nucleotide
alignment.
We present a detailed study on the informativeness
of the lsu rDNA based on the analysis of particular domains and segments and their contribution in terms of
phylogenetic information for the family Sarcocystidae.
Using a larger data set of 21 taxa, we investigated the
informativeness of lsu rDNA segments corresponding to
secondary structures, which appeared to be hot spots for
nucleotide variation. The informativeness of random
segments of the molecule was also studied. Our data set
served as a model for the analysis of congruence of
results among different alignment strategies, alignment
parameters, and tree-building methods and the sensitivity of these methods to the subdivision of data.
In addition, our data set included two newly added
species of Sarcocystis, Sarcocystis zamani and Sarcocystis singaporensis, which use reptiles as definitive
hosts (Beaver and Maleckar 1981). This paper reports
the relationships of these species to Sarcocystis species
which use mammals as definitive hosts and also shows
the branching of the Toxoplasmatine species Besnoitia
besnoiti and of Isospora felis (syn. Cystoisospora felis),
which is often classified into the family Eimeriidae.
Materials and Methods
Genomic DNA
Genomic DNA of Sarcocystis arieticanis, Sarcocystis capracanis, Sarcocystis cruzi, Sarcocystis gigan-
1844
Mugridge et al.
tea, Sarcocystis miescheriana, Sarcocystis moulei, Sarcocystis muris, Sarcocystis neurona (SN-2 isolate), Sarcocystis tenella, Frenkelia glareoli, Frenkelia microti,
H. hammondi, H. heydorni, T. gondii (ME49 strain), and
I. felis was obtained as described previously (Granstrom
et al. 1994; Tenter, Luton, and Johnson 1994; Biñas and
Johnson 1998; Carreno et al. 1998; Votýpka et al. 1998;
Mugridge et al. 1999a). The isolates of S. zamani and
S. singaporensis have been described elsewhere (Jäkel
et al. 1999), and genomic DNA was extracted from purified cystozoites of these species as described for the
other Sarcocystis species (Tenter, Luton, and Johnson
1994). Besnoitia besnoiti genomic DNA was a generous
gift from Dr. T. de Waal, Onderstepoort Veterinary Institute, South Africa.
Polymerase Chain Reaction Amplification, Cloning,
and Sequencing
The lsu rDNA encompassing nucleotides 24–42 to
nucleotides 4337–4354 (murine numbering system) was
amplified from the 18 taxa listed above using five pairs
of primers. Primer sequences, PCR protocols, and cloning and sequencing methods, as well as the specificity
of the primers for members of the Sarcocystidae, were
as previously reported (Mugridge et al. 1999b).
Phylogenetic Analysis of the Family Sarcocystidae
Based on the Full lsu rDNA Sequence
The full-length lsu rDNA sequence was obtained
for the 18 taxa listed above. The shortest sequence was
3,212 bp (in T. gondii ME49 and H. hammondi), and
the longest was 3,500 bp (in S. arieticanis). Lsu rDNA
sequences of N. caninum and Eimeria tenella were obtained from GenBank (accession numbers AF001946
and AF026388). The T. gondii RH strain sequence was
a consensus of a GenBank submission (X75429) and the
sequence published by Ellis et al. (1998). We aligned
sequences according to their primary structure similarity
using CLUSTAL W, version 1.5, with default gap values
(Thompson, Higgins, and Gibson 1994) and according
to their secondary structure using the DCSE program by
De Rijk and De Wachter (1993), by way of observing
the influence of two alignment strategies on the resulting
trees. Both alignments were subjected to three different
tree-building methods: maximum parsimony (Farris
1970), neighbor joining (a distance method) (Saitou and
Nei 1987), and maximum likelihood (Fukami and Tateno 1989). These methods were implemented in the
computer programs PAUP, version 3.1.1 (Swofford
1990); neighbor in the PHYLIP, version 3.5c, package
(Felsenstein 1995); and fastDNAml, version 1.0.6 (Olsen et al. 1994), respectively. Maximum parsimony was
carried out using the stepwise-addition option and a heuristic search method with 10 random starts. The Kimura
two-parameter distance model was used for maximumlikelihood and neighbor-joining analyses. Maximum
likelihood used global rearrangements and a transition/
transversion ratio of 2.0, and each analysis was repeated
three times with the jumble option. Spatial variation in
evolutionary rates along the sequence was also investi-
gated for maximum likelihood using a heuristic procedure. The DNArates, version 1.0.3, program was used
to infer the evolutionary rate at each sequence position
for the full structure alignment, including the possibility
of invariant sites. These rates were then used as nine
possible rate categories using the weights and categories
options in fastDNAml. This analysis changed the branch
lengths of the resulting trees but not the branching order.
Therefore, only the analyses using the equal-rates option
are shown here.
A member of the genus Eimeria, E. tenella, which
is closely related to the Sarcocystidae, was chosen as
the outgroup for all of the analyses.
Informative Nucleotide Sites of the lsu rDNA
Molecule
To find the domains or stems of the lsu rDNA molecule that were contributing most to informativeness, we
subdivided the original full-length secondary-structure
alignment into several shorter alignments, each including the individual sequence segments to be investigated
through tree-building. First, we divided the structure
alignment into two separate alignments, one including
double-stranded, helical regions and the other including
single-stranded, nonhelical regions. Other potentially informative domains or segments were found using the
program MacClade, version 3.07. Trees from the secondary-structure alignment were examined using the
MacClade, version 3.07, program under the branch reconstruction and character changes options. The first option reconstructs branch changes in a parsimonious manner and lists the sites that support each branch (Maddison and Maddison 1992). The latter option charts the
minimum and maximum numbers of changes that could
have occurred at each nucleotide site for a particular
tree.
In this manner, segments seeming to contain a large
number of sites involved in change and branch formation were taken as probable informative segments for
phylogeny reconstruction within the Sarcocystidae. The
structure alignment was then subdivided to produce
alignments each containing one, or a combination, of
the chosen segments. Each of these alignments was independently subjected to the three tree-building
methods.
To test the informativeness of random segments of
the lsu rDNA, the entire sequence was divided in half
without any consideration of secondary structure. The
two resulting alignments thus represented the 59 end and
the 39 end of the lsu rDNA. Once again, these two regions were subjected to the three tree-building
alignments.
Results
Phylogenetic Analysis of Full-Length lsu rDNA
Sequences were aligned by two alignment strategies in order to observe the effect on tree building. An
example tree with bootstraps and branch lengths based
on full-length lsu rDNA and the maximum-likelihood
tree-building method is shown in figure 1.
Sequence Alignment and Structure Domains
1845
FIG. 1.—Phylogenetic relationships among the Sarcocystidae as inferred from the structure alignment of the full-length lsu rDNA and the
maximum-likelihood tree-building method for 21 taxa. Neospora caninum and the Toxoplasma gondii strains split the genus Hammondia;
Isospora felis is a sister taxon to the subfamily Sarcocystinae; Sarcocystis muris is a sister taxon to Sarcocystis neurona, Frenkelia glareoli,
and Frenkelia microti; Sarcocystis zamani and Sarcocystis singaporensis are placed within the genus Sarcocystis. The branch lengths are
proportional to the amount of inferred evolutionary change, and the numbers on the branches are the numbers of times that the branch was
supported in 100 bootstrap replicates.
The importance of the alignment step in phylogeny
is well established (Morrison and Ellis 1997). Any errors
in alignment will lead to potential errors in the trees. To
corroborate the importance of the resultant alignment,
we varied the gap opening and gap extension values in
the CLUSTAL W alignment (Wheeler 1995; Morrison
and Ellis 1997). This was done following the method of
Wheeler (1995), whereby the values increase logarithmically (gap extension range of 0.031–4, gap opening
range of 0.5–64), and using the experimental plan of
Morrison and Ellis (1997), which tests all orthogonal
combinations of values (72 in total). Results (not shown)
for this experiment clearly demonstrated that alignment
parameters could be influential. Eight different topolo-
gies were obtained from the 72 alignments, consisting
of varying gap opening and extension penalty values,
using PAUP as the tree-building method, and seven different topologies were obtained using fastDNAml.
The trees obtained from the full-length CLUSTAL
W sequence alignment using maximum parsimony,
maximum likelihood, and neighbor joining mostly supported the same phylogenetic groupings. The tree given
by maximum likelihood is shown in figure 2. The group
including the Toxoplasmatinae confirms our previous
findings (Mugridge et al. 1999a). That is, the genus
Hammondia is split by the T. gondii strains where H.
hammondi is located on a monophyletic clade with the
T. gondii RH and ME49 strains, while H. heydorni and
1846
Mugridge et al.
FIG. 2.—Phylogenetic relationships among the Sarcocystidae as
inferred from the CLUSTAL W alignment of the full-length lsu rDNA.
Hammondia heydorni and Neospora caninum are on sister branches;
Isospora felis is a sister taxon to the subfamily Toxoplasmatinae; Sarcocystis muris is part of a monophyletic group including Sarcocystis
neurona, Frenkelia glareoli, and Frenkelia microti; the Sarcocystis
species with reptilian definitive hosts (Sarcocystis zamani and Sarcocystis singaporensis) form a sister clade to the Sarcocystis species with
mammalian definitive hosts and the Frenkelia species.
N. caninum appear on sister branches. Besnoitia besnoiti
is the immediate relative to this group, followed by I.
felis. The genus Sarcocystis, too, demonstrates paraphyly as previously shown (Mugridge et al. 1999b). Frenkelia microti and F. glareoli demonstrate a close relationship with S. neurona and S. muris by forming the
monophyletic group {S. muris [S. neurona (F. glareoli,
F. microti)]}, which is a sister clade to the other Sarcocystis species with mammalian definitive hosts. Other
branchings within the genus Sarcocystis include the
monophyletic clades {S. cruzi [S. arieticanis (S. tenella,
S. capracanis)]} and [S. miescheriana (S. moulei, S. gigantea)]. The Sarcocystis species with reptilian definitive hosts (S. zamani and S. singaporensis) form a
monophyletic clade which is a sister group to the rest
of the genus Sarcocystis and the Frenkelia species.
The tree topology resulting from neighbor joining
is like that in figure 2 except that N. caninum and H.
heydorni form a monophyletic clade rather than being
sister branches; this results in the genus Hammondia
being split by N. caninum in addition to the two T. gondii strains. Maximum parsimony gave four equally parsimonious trees where the only varying branches were
those in the Toxoplasmatinae. This method found conflicting information within the clade [H. hammondi (T.
gondii RH, T. gondii ME49)] and conflicting information regarding whether the nearest sister taxon to this
clade is N. caninum or H. heydorni; however, the paraphyly of the genus Hammondia is supported.
Tree-building analysis on the secondary-structure
alignment of the full lsu rDNA sequences was more sensitive to the tree-building method used than was the
CLUSTAL W alignment. Figure 1 shows the tree topology given by maximum likelihood using the secondary-structure alignment. It differs from the maximum
likelihood tree based on the CLUSTAL W alignment in
the placement of I. felis, H. heydorni and N. caninum,
S. muris, and S. zamani and S. singaporensis. In figure
1, I. felis splits the subfamilies Sarcocystinae and Toxoplasmatinae. Neospora caninum splits the genus Hammondia, as in the tree from the CLUSTAL W alignment
using neighbor joining. Sarcocystis muris does not form
a monophyletic clade with the Frenkelia species and S.
neurona, but splits the latter from the rest of the Sarcocystis species. Sarcocystis zamani and S. singaporensis prove to be part of the genus Sarcocystis by branching after the Frenkelia species/S. neurona clade and S.
muris. Neighbor joining, when applied to the secondarystructure alignment, gave the same tree given by this
method using the CLUSTAL W alignment, which is also
essentially the same as that given by the maximum-likelihood method and the CLUSTAL W alignment (fig. 2).
Maximum parsimony, when applied to the secondarystructure alignment, resulted in eight equally parsimonious trees with a consensus showing polytomies within
the Toxoplasmatinae, and in the placement of I. felis.
Informative Sites of the lsu rDNA
Phylogenetic Analysis of Helical and Nonhelical
Regions
The results obtained by analysis of helical and nonhelical regions differed from each other in the inferred
branching of S. muris, I. felis, and the Toxoplasmatine
taxa. Variations in topology were found among treebuilding algorithms and between the two alignments.
Figure 3 summarizes the differences in topology between trees obtained from helical and nonhelical alignments using the three tree-building methods.
Informative Domains and Segments
The trees obtained using maximum parsimony,
neighbor joining, and maximum likelihood on the structure alignment were analyzed under the branch reconstruction and character change options in MacClade,
version 3.07. From this analysis, we observed the sites
which were involved in the formation of each branch on
a tree. All trees showed the same hot spots for nucleotide change under the character change options. Figure
4 shows part of a chart giving the amount of character
Sequence Alignment and Structure Domains
1847
FIG. 3.—Summary of results obtained using (A) the helical regions and (B) the nonhelical regions of the structure alignment in conjunction
with maximum parsimony, maximum likelihood, and neighbor joining. The only varying placements involve Sarcocystis muris, Isospora felis,
and the Toxoplasmatinae. The tree diagrams are of sections of full trees featuring groupings of interest.
change for the maximum-likelihood tree given by the
structure alignment. This section of the chart presents
the relatively larger amount of variation found in the D2
domain of the lsu rDNA which encompasses nucleotides
425–750 in the T. gondii RH strain. Table 1 lists the
segments of more change by their nucleotide positions
in the T. gondii RH strain lsu rDNA and describes the
secondary structures each encompasses. Small areas
within segments encompassing nucleotides 1426–1904
and 2047–2720 containing less change were included to
prevent the use of shorter stretches of sequence in subsequent tree analyses. The D3 segment was also added
to the D2 region for analysis because of its phylogenetically unrealistic length of 60 bases. However, the D2
segment was analyzed independently as well.
In total, 12 phylogenetic analyses were performed
(three cladistic methods for each of four sequence subsets). The D2 domain gave trees supporting the major
groupings found for the full-length analyses using maximum parsimony and neighbor joining, with the exception that S. cruzi was placed as a sister taxon to S. tenella instead of S. capracanis (fig. 5). The result from
maximum likelihood based on the D2 region shows this
and also changes the placement of S. zamani and S.
singaporensis to form a clade with S. miescheriana, S.
moulei, and S. gigantea (fig. 6). With respect to the Toxoplasmatinae, the D2 segment does not deviate from results given by the full-length analyses. The three trees
from the D2 region all place I. felis as a sister taxon to
the subfamily Sarcocystinae.
1848
Mugridge et al.
FIG. 4.—Part of a chart obtained with the program MacClade demonstrating the section of the lsu rDNA alignment which includes domain
D2, known to be high in nucleotide variation. Sections of the alignment containing a larger amount of nucleotide variation as seen here were
chosen for individual tree-building analysis to test the phylogenetic informativeness for the Sarcocystidae.
Adding D3 to the D2 domain did not change the
above results for any of the tree-building methods, except that neighbor joining placed S. cruzi and S. capracanis as monophyletic sisters, and maximum likelihood
did not place S. zamani and S. singaporensis in a clade
with S. moulei, S. miescheriana, and S. gigantea.
The other two segments from bases 1426–1904 and
2047–2720 gave trees that should not be counted as phylogenetically informative for the Sarcocystidae because
of groupings that intersperse the genus Sarcocystis with
members of the Toxoplasmatinae.
Results from the truncation of the full-length lsu
rDNA alignment into two alignments containing either
the 59 or the 39 end gave results more in accordance
with those from the full-length alignment. The maximum-likelihood trees from the 59 and 39 ends, respectively, are shown in figures 7 and 8. Maximum parsiTable 1
Nucleotide Regions of the lsu rDNA Demonstrating
Relatively More Extensive Change Using the Program
MacClade in Trees Derived from the DCSE Alignment
Area of High Change
(Toxoplasma gondii
RH
Nucleotide Numbering)
420–725 . . . . . . . . . . .
806–860 . . . . . . . . . . .
1426–1904 . . . . . . . . . .
2047–2720 . . . . . . . . . .
Secondary-Structure Features
Included in the Segment
Domain D2
Domain D3
Regions 31–41, domains D6, D7a, b
Region 49–63, domains D8, D9, D10
mony and neighbor joining also gave results in congruence with the full-length results, except for the neighborjoining tree of the 39 end, which split the T. gondii
strains with H. hammondi.
Discussion
Informative Regions of the lsu rDNA
Secondary-Structure Regions
Our investigation of the use of partial lsu rDNA
sequences to study the phylogeny of the Sarcocystidae
revealed that the entire length of the gene is needed to
obtain full phylogenetic informativeness. In order to analyze the influence of different structural segments of
the alignment on the final phylogeny, all trees were
compared based on tree topology only. Previous work
on the phylogeny of the Sarcocystidae using ssu rDNA
proposed that helical regions, which correspond to double helices, are more informative than the nonhelical
areas (Morrison and Ellis 1997). Nonhelical areas,
which are single-stranded regions forming globular domains, have a higher rate of change than helical areas
(Olsen and Woese 1993). Therefore, changes in the more
conserved helical regions that make up the backbone of
the molecule could be construed to be more significant
in the evolution of these taxa than changes in the fasterevolving nonhelical areas. However, these conserved areas, although useful in the alignment of sequences, may
sometimes contain little phylogenetic information for
closely related taxa. Our investigations into the amount
Sequence Alignment and Structure Domains
FIG. 5.—Phylogenetic relationships among the Sarcocystidae as
inferred from maximum-parsimony and neighbor-joining analyses of
the structure alignment of the D2 domain. Deviation from full-length
corresponding analyses is included.
of nucleotide substitutions throughout the lsu rDNA
molecule for the Sarcocystidae support the idea that
globular domains can have more changes, as evidenced
by the fact that the areas of highest variation observed
by us were present in the globular domains. As seen in
figure 4, the D2 domain is the area with the most
change. The D2 region of the lsu rDNA is only 305 bp
long in the T. gondii RH strain but has nevertheless been
used to infer phylogenies on a small number of taxa
belonging to the Sarcocystidae (Ellis et al. 1999). One
of the aims of this work was to determine whether this
site alone could be informative enough for the Sarcocystidae. It was observed that many of the changes involved in the formation of tree branches are indeed part
of the D2 domain but that informative sites involved in
resolving branches for very similar species or strains,
such as H. hammondi and the T. gondii RH and ME49
strains, also come from other areas of the molecule, including the 39 end. For example, there are only three
nucleotide variations supporting the branch grouping of
H. hammondi, T. gondii RH, and T. gondii ME49, and
only one of these is from the D2 domain. Therefore, it
is evident that in such cases the entire lsu rDNA sequence is more informative. Furthermore, from figures
1849
FIG. 6.—Phylogenetic relationships among the Sarcocystidae as
inferred from maximum-likelihood analysis of the structure alignment
of the D2 domain.
5 and 6, it can be seen that the D2 domain gave trees
which departed from two major groupings given by the
full-length analyses by placing S. cruzi as a sister taxon
to S. tenella instead of S. capracanis (fig. 5) and placing
S. zamani and S. singaporensis with the S. miescheriana,
S. moulei, and S. gigantea clade (fig. 6). The considerable number of site changes not within the D2 domain
supporting the various groupings within the genus Sarcocystis suggests that the entire length of the gene, as
opposed to the D2 domain alone, gives a more robust
result. The fact that the addition of the D3 domain to
the D2 domain did not add to the informativeness, and
that the other truncated regions of change gave questionable results, also supports this view.
Helical and Nonhelical Regions
Our results from the use of helical and nonhelical
parts of the secondary alignment for tree building also
demonstrated that subdivisions of the full-length gene
could have an effect on tree topology. In figure 3, it can
be seen that the placement of S. muris remained constant
for the two alignment subdivisions for the maximumparsimony and neighbor-joining tree-building methods
but that maximum likelihood had a different placement
1850
Mugridge et al.
FIG. 7.—Phylogenetic relationships among the Sarcocystidae as
inferred from maximum-likelihood analysis of the 59 end of the lsu
rDNA structure alignment.
FIG. 8.—Phylogenetic relationships among the Sarcocystidae as
inferred from maximum-likelihood analysis of the 39 end of the lsu
rDNA structure alignment.
Random Choice of Nucleotide Sites
for this species depending on the alignment. The branching of I. felis did not stay constant for the different treebuilding methods for each alignment, and furthermore,
subdividing the full sequence into helical and nonhelical
sections significantly reduced the resolution of the Toxoplasmatinae (fig. 3).
Previously, we reported that the nonhelical regions
of the lsu rDNA are as informative as the helical areas
(Mugridge et al. 1999b). This remains the case in this
study, as neither the helical nor the nonhelical alignments appear more informative than the other in conjunction with any of the tree-building algorithms. However, we note, not unexpectedly, that adding to the number of species (only nine were used in the previous
study) decreases the congruence of topologies obtained
from the different tree-building methods. For this work,
we added more Sarcocystis species and all current genera of the Toxoplasmatinae; however, the incongruities
remain with the placement of S. muris, as well as I. felis.
The paraphyly of T. gondii shown in the nonhelical
alignment is a clear result of the removal of informative
sites for the Toxoplasmatinae, which comprise only 4
nt.
Dividing the full-length secondary alignment in
half to produce 59 and 39 alignments also proved to be
unsatisfactory, as the 39 alignment split the T. gondii
strains using the maximum-likelihood and neighborjoining trees. Minor branch swappings diverging from
the full-length alignment results were also present for
these two alignments within the genus Sarcocystis. This
result also demonstrates that neither the 59 nor the 39
end of the gene is more informative than the other in
the final analysis.
Alignments
Recent work on the effect of sequence alignment
on the phylogeny of the genus Sarcocystis showed that
the resulting trees can be influenced by the choice of a
primary- or secondary-structure alignment (Ellis and
Morrison 1995). This work addressed the issue by investigating the phylogeny of the family Sarcocystidae
using two alignment strategies and three tree-building
methods in order to observe the effect this had on the
various groupings in the family. The DCSE alignment
of De Rijk and De Wachter (1993) was chosen for its
Sequence Alignment and Structure Domains
consideration of the secondary structure of the ribosomal RNA molecule. Since homology between nucleotides
of the aligned sequences is imperative for a true phylogenetic reconstruction, we believe that adhering to the
biological constraints of the molecule during the alignment process is more likely to achieve this aim. The
CLUSTAL W alignment algorithm (Thompson, Higgins, and Gibson 1994) aims at achieving the highest
nucleotide identity among sequences; hence, in some
parts of the alignment, nucleotides from different sequences may be positionally aligned because of their
similarity rather than their true homology. Any nonhomologous alignment will lead to an erroneous
phylogeny.
From the results obtained using the two full-length
alignments with the three tree-building programs, we
found that all of the trees derived from the CLUSTAL
W alignment had identical topologies except for variations in the placement of N. caninum as a sister taxon
to the genus Hammondia or as a sister taxon to H. heydorni, thereby splitting the genus Hammondia. The trees
resulting from the application of the structure alignment
gave inconsistencies among tree-building methods.
Again, the three differing groups involved S. muris, I.
felis, and the Toxoplasmatinae. In trees derived by the
maximum-parsimony and maximum-likelihood methods, which are both character-based methods, S. muris
splits a clade consisting of S. neurona and the Frenkelia
species from the rest of the Sarcocystis species. In contrast, neighbor joining, which deals with distance data,
placed S. muris in a clade with S. neurona and the two
Frenkelia species.
The branching of I. felis differed for the three treebuilding methods, with maximum parsimony and neighbor joining placing it as the sister taxon to all species
after E. tenella, while maximum likelihood placed it as
the sister taxon to the subfamily Sarcocystinae. With
regard to this last group, the three methods gave the
same topology, except that in the case of maximum parsimony, the clade involving the T. gondii ME49 and RH
strains and H. hammondi could not be resolved.
The question that arises is whether the result obtained from the CLUSTAL W alignment is the true evolutionary topology of the group under analysis. The congruence in results within the CLUSTAL W analyses
could be linked to the use of the lsu rDNA molecule.
This molecule is about double the length of the ssu
rDNA, and this increase in the available number of nucleotides for analysis may render any differences between alignment strategies in this case irrelevant. The
structure alignment contained many more gaps and areas
in which the alignment was not intuitively obvious, but
in keeping to secondary structure, this should in theory
be a more informative alignment, and the few differences found among the tree-building methods could be
the effect of these methods having different goals in
their tree-building processes. Areas in the alignment that
contain many gaps could be removed from the alignment with the aim of reducing inconsistencies among
trees. With respect to the lsu rDNA structure alignment,
most of the areas that have a larger amount of gaps
1851
correspond to the nonhelical regions of the molecule.
The results here pertaining to the helical segments of
the alignment (fig. 3) demonstrate that removing the regions with more gaps does not aid the congruency of
results among different tree-building methods. Furthermore, excluding such regions in this case means the removal of areas that contain informative nucleotide sites,
and this is particularly important for the genera Hammondia, Neospora, and Toxoplasma, which have a very
small number of informative sites in the lsu rDNA.
Using the three tree-building methods, however,
did show that the majority of the groups within the inferred phylogeny are very robust. For example, the major groupings within the subfamily Sarcocystinae remained the same in all of the full-length trees. The
placement of S. muris was dependent on the tree-building method as to whether it branched in a clade with S.
neurona and the two Frenkelia species or split the latter
from the rest of the Sarcocystis species. The other variation is due to the different alignments and concerns S.
singaporensis and S. zamani, which branch before the
other Sarcocystis species in trees from the CLUSTAL
W alignment but branch after the Frenkelia species/S.
neurona clade and S. muris in trees derived using the
structure alignment.
Congruence among trees with respect to the Toxoplasmatinae is also high, with the only exception being
that maximum parsimony provides eight equally parsimonious trees because it cannot resolve which are closest among T. gondii RH, T. gondii ME49, and H. hammondi. The other methods agree that the T. gondii
strains are closest, and in the case of parsimony, it is
probably just a case of the lsu rDNA not having sufficient characters to resolve this particular group using
this method. All of the full-length results, however, support our previous findings that N. caninum and H. heydorni are closely related, thus splitting the genus Hammondia. It can be seen in figure 1 that the local rearrangements in this group are due to the short branch
lengths between the genera Hammondia, Neospora, and
Toxoplasma, with even a branch length of 0 between H.
hammondi and the T. gondii strains.
Applying different alignment and tree-building
methods provides an opportunity to observe how the
data behave under different assumptions and aims. Literature on phylogeny algorithms remains focused on the
advantages and pitfalls of various methods, despite the
fact that as a taxonomic and evolutionary tool, phylogeny is becoming increasingly popular. Different methods
may or may not extract varying information from a
gene, and reducing the number of data (the number of
taxa or the length of a sequence) may add to the variation, as evidenced by our results using partial sequences. We have shown that using the entire lsu rDNA gives
a more informative topology of the Sarcocystidae, and
the use of different tree-building methods and even two
alignment algorithms has revealed that most of the
groupings within this family are robust.
Relationships Within the Sarcocystidae
The findings from this work support our previous
results regarding the genus Frenkelia and the subfamily
1852
Mugridge et al.
Toxoplasmatinae (Mugridge et al. 1999a, 1999b). All
the trees in this study using lsu rDNA sequences confirm
that the current genus Sarcocystis is split by the genus
Frenkelia. This has also been found in a study using ssu
rDNA sequences (Votýpka et al. 1998), and it has already been suggested that Frenkelia should be synonymized with Sarcocystis based on phenotypic characters
(Tadros and Laarman 1976; Odening 1998; Mugridge et
al. 1999b), which would render the current subfamily
Sarcocystinae monogeneric. New findings obtained here
show that the clades [S. miescheriana (S. moulei, S. gigantea)], {S. cruzi [S. arieticanis (S. tenella, S. capracanis)]}, [S. neurona (F. glareoli, F. microti)], and (S.
zamani, S. singaporensis), within the subfamily Sarcocystinae, appear to be very robust when exposed to
varying alignment methods and parameters and different
tree-building methods. The Sarcocystis species with reptilian definitive hosts (S. zamani and S. singaporensis)
are shown to form a lineage that is separated from the
other Sarcocystis species, which use mammals as definitive hosts, and from the Frenkelia species, which use
birds as definitive hosts. In contrast, as found in analyses
using ssu rDNA sequences (Jeffries et al. 1997; Votýpka
et al. 1998), the placement of S. muris also remained
the main variant within the Sarcocystinae based on the
lsu rDNA data.
The current study also confirmed previous analyses
based on lsu rDNA and ITS1 sequences (Ellis et al.
1999; Mugridge et al. 1999a), which showed the genus
Hammondia to be paraphyletic and showed species
placed in the genera Toxoplasma, Hammondia, and
Neospora to be very closely related. As described previously, this finding, together with the broad range of
phenotypic characters that are shared by these taxa, calls
for a revision of these taxa within the subfamily Toxoplasmatinae (Mugridge et al. 1999a). In contrast, the
branching of B. besnoiti as the nearest sister to these
taxa is a robust result in this study.
Isospora felis (syn. C. felis), which has the ability
to produce monozoic tissue cysts in optional intermediate hosts, is sometimes placed in the genus Cystoisospora, which is then placed in the monogeneric subfamily Cystoisosporinae of the Sarcocystidae (Frenkel
et al. 1979; Frenkel, Mehlhorn, and Heydorn 1987).
However, it is most frequently placed in the genus Isospora, which is a member of the family Eimeriidae and
contains several coccidia with differences in life cycle
and morphology (Levine 1988; Current, Upton, and
Long, 1990; Tenter and Johnson 1997). In our present
work, all of the full-length sequence results from the
CLUSTAL W alignment, and from the neighbor-joining
and maximum-parsimony methods using the structure
alignment, place I. felis as the sister taxon to a monophyletic group comprising the subfamilies Toxoplasmatinae and Sarcocystinae (fig. 2), while maximum likelihood in conjunction with the structure alignment suggested that I. felis splits the Sarcocystinae from the Toxoplasmatinae (fig. 1). A recent phylogenetic analysis
using ssu rDNA sequences of several Isospora species
showed the genus Isospora to be paraphyletic and identified different lineages within this genus which are in
accordance with morphologic similarities among their
life cycle stages (Carreno and Barta 1999). That study
suggests that a monophyletic group of I. felis and homoxenous Isospora species of mammals is a sister taxon
to the Toxoplasmatinae, whereas Isospora species of
birds are monophyletic with species of Eimeria. Further
studies involving lsu rDNA sequences of more species
of Isospora will be needed to validate the different lineages within the family Sarcocystidae and its subdivision into subfamilies and genera.
Supplementary Material
Nucleotide sequence data reported in this paper are
available in the GenBank, EMBL and DDJB databases
under accession numbers AF044250 (S. arieticanis),
AF076899 (S. tenella), AF076902 (S. miescheriana),
AF076903 (S. cruzi), AF092927 (S. neurona),
AF012883 (S. muris), AF012884 (S. moulei), AF012885
(S. capracanis), U85706 (S. gigantea), AF076900 (B.
besnoiti), AF044251 (F. glareoli), AF044252 (F. microti), AF076901 (T. gondii ME49), AF101077 (H. hammondi), AF159240 (H. heydorni), U85705 (I. felis),
AF237616 (S. zamani), and AF237617 (S.
singaporensis).
LITERATURE CITED
BEAVER, P. C., and J. R. MALECKAR. 1981. Sarcocystis singaporensis Zaman and Colley, (1975) 1976, Sarcocystis villivilliso sp. n., and Sarcocystis zamani sp. n.: development,
morphology, and persistence in the laboratory rat, Rattus n.
J. Parasitol. 67:241–256.
BIÑAS, M., and A. M. JOHNSON. 1998. A polymorphism in a
DNA polymerase alpha gene intron differentiates between
murine virulent and avirulent strains of Toxoplasma gondii.
Int. J. Parasitol. 28:1033–1040.
CARRENO, R. A., and J. R. BARTA. 1999. An eimeriid origin
of isosporoid coccidia with Stieda bodies as shown by phylogenetic analysis of small subunit ribosomal RNA gene
sequences. J. Parasitol. 85:77–83.
CARRENO, R. A., B. E. SCHNITZLER, A. C. JEFFRIES, A. M.
TENTER, A. M. JOHNSON, and J. R. BARTA. 1998. Phylogenetic analysis of coccidia based on 18S rDNA sequence
comparison indicates that Isospora is most closely related
to Toxoplasma and Neospora. J. Eukaryot. Microbiol. 45:
184–188.
CAVALIER-SMITH, T. 1989. Archaebacteria and Archezoa. Nature 339:100–101.
CURRENT, W. L., S. J. UPTON, and P. L. LONG. 1990. Taxonomy
and life cycles. Pp. 1–16 in P. L. LONG, ed. Coccidiosis of
man and domestic animals. CRC Press, Boca Raton, Fla.
DE RIJK, P., and R. DE WACHTER. 1993. DCSE, an interactive
tool for sequence alignment and secondary structure research. CABIOS 9:735–740.
DOLE EL, D., B. KOUDELA, M. JIRKŮ, V. HYPŠA, M. OBORNÍK,
J. VOTÝPKA, D. MODRÝ, J. R. ŠLAPETA, and J. LUKEŠ. 1999.
Phylogenetic analysis of Sarcocystis spp. of mammals and
reptiles supports the coevolution of Sarcocystis spp. with
their final hosts. Int. J. Parasitol. 29:795–798.
ELLIS, J. T., G. AMOYAL, C. RYCE, P. A. HARPER, K. A.
CLOUGH, W. L. HOMAN, and P. J. BRINDLEY. 1998. Comparison of the large subunit ribosomal DNA of Neospora
and Toxoplasma and development of a new genetic marker
Sequence Alignment and Structure Domains
for their differentiation based on the D2 domain. Mol. Cell.
Probes 12:1–13.
ELLIS, J. T., and D. A. MORRISON. 1995. Effects of sequence
alignment on the phylogeny of Sarcocystis deduced from
18S rDNA sequences. Parasitol. Res. 81:696–699.
ELLIS, J. T., D. A. MORRISON, S. LIDDELL, M. C. JENKINS, O.
B. MOHAMMED, C. RYCE, and J. P. DUBEY. 1999. The genus
Hammondia is paraphyletic. Parasitology 118:357–362.
FARRIS, J. S. 1970. Methods for computing Wagner trees. Syst.
Zool. 19:83–92.
FELSENSTEIN, J. 1995. PHYLIP (phylogenetic inference package). Distributed by the author, Department of Genetics,
University of Washington, Seattle.
FRENKEL, J. K. 1977. Besnoitia wallacei of cats and rodents:
with a reclassification of other cyst-forming isosporoid coccidia. J. Parasitol. 63:611–628.
FRENKEL, J. K., A. O. HEYDORN, H. MEHLHORN, and M. ROMMEL. 1979. Sarcocystinae: nomina dubia and available
names. Z. Parasitenkd. 58:115–139.
FRENKEL, J. K., H. MEHLHORN, and A. O. HEYDORN. 1987.
Beyond the oocyst: over the molehills and mountains of
coccidialand. Parasitol. Today 3:250–252.
FUKAMI, K., and Y. TATENO. 1989. On the maximum likelihood method for estimating molecular trees: uniqueness of
the likelihood point. J. Mol. Evol. 28:460–464.
GRANSTROM, D. E., J. M. MACPHERSON, A. A. GAJADHAR, J.
P. DUBEY, R. TRAMONTIN, and S. STAMPER. 1994. Differentiation of Sarcocystis neurona from eight related coccidia
by random amplified polymorphic DNA assay. Mol. Cell.
Probes 8:353–356.
HOLMDAHL, O. J., D. A. MORRISON, J. T. ELLIS, and L. T.
HUONG. 1999. Evolution of ruminant Sarcocystis (Sporozoa) parasites based on small ribosomal rDNA sequences.
Mol. Phylogenet. Evol. 11:27–37.
JÄKEL, T., Y. KHOPRASERT, S. ENDEPOLS, C. ARCHER-BAUMANN, K. SUASA-ARD, P. PROMKERD, D. KLIEMT, P. BOONSONG, and S. HONGNARK. 1999. Biological control of rodents using Sarcocystis singaporensis. Int. J. Parasitol. 29:
1321–1330.
JEFFRIES, A. C., B. SCHNITZLER, A. O. HEYDORN, A. M. JOHNSON, and A. M. TENTER. 1997. Identification of synapomorphic characters in the genus Sarcocystis based on 18S
rDNA sequence comparison. J. Eukaryot. Microbiol. 44:
388–392.
JENKINS, M. C., J. T. ELLIS, S. LIDDELL, C. RYCE, B. L. MUNDAY, D. A. MORRISON, and J. P. DUBEY. 1999. The relationship of Hammondia hammondi and Sarcocystis mucosa
to other heteroxenous cyst-forming coccidia as inferred by
phylogenetic analysis of the 18S SSU ribosomal DNA sequence. Parasitology 119:135–142.
LEVINE, N. D. 1988. The protozoan phylum Apicomplexa.
Vols. 1 and 2. CRC Press, Boca Raton, Fla.
MADDISON, W. P., and D. R. MADDISON. 1992. MacClade, Version 3.0. Analysis of phylogeny and character evolution.
Sinauer, Sunderland, Mass.
MORRISON, D. A., and J. T. ELLIS. 1997. Effects of nucleotide
sequence alignment on phylogeny estimation: a case study
of 18S rDNAs of Apicomplexa. Mol. Biol. Evol. 14:428–
441.
MUGRIDGE, N. B., D. A. MORRISON, A. R. HECKEROTH, A. M.
JOHNSON, and A. M. TENTER. 1999a. Phylogenetic analysis
1853
based on full-length large subunit ribosomal RNA gene sequence comparison reveals that Neospora caninum is more
closely related to Hammondia heydorni than to Toxoplasma
gondii. Int. J. Parasitol. 29:1545–1556.
MUGRIDGE, N. B., D. A. MORRISON, A. M. JOHNSON, K. LUTON, J. P. DUBEY, J. VOTÝPKA, and A. M. TENTER. 1999b.
Phylogenetic relationships of the genus Frenkelia: a review
of its history and new knowledge gained from comparison
of large subunit ribosomal ribonucleic acid gene sequences.
Int. J. Parasitol. 29:957–972.
ODENING, K. 1998. The present state of species-systematics in
Sarcocystis Lankester 1882 (Protista, Sporozoa, Coccidia).
Syst. Parasitol. 41:209–233.
OLSEN, G. J. 1987. earliest phylogenetic branchings: comparing rRNA-based evolutionary trees inferred with various
techniques. Cold Spring Harb. Symp. Quant. Biol. 52:825–
837.
———. 1988. Phylogenetic analysis using ribosomal RNA.
Methods Enzymol. 164:793–812.
OLSEN, G. J., R. MATSUDA, R. HAGSTROM, and R. OVERBEEK.
1994. fastDNAml: a tool for construction of phylogenetic
trees of DNA sequences using maximum likelihood. Comput. Appl. Biosci. 10:41–48.
OLSEN, G. J., and C. R. WOESE. 1993. Ribosomal RNA: a key
to phylogeny. FASEB J. 7:113–123.
SAITOU, N., and M. NEI. 1987. The neighbor-joining method:
a new method for reconstructing phylogenetic trees. Mol.
Biol. Evol. 4:406–425.
SOGIN, M. L., and J. H. GUNDERSON. 1986. Structural diversity
of eukaryotic small subunit ribosomal RNAs. Ann. N.Y.
Acad. Sci. 503:125–139.
SWOFFORD, D. L. 1990. PAUP: phylogenetic analysis using
parsimony. Version 3.1.1. Illinois Natural History Survey,
Champaign.
TADROS, W., and J. J. LAARMAN. 1976. Sarcocystis and related
coccidian parasites: a brief general review, together with a
discussion on some biological aspects of their life cycles
and a new proposal for their classification. Acta Leidensia
44:1–137.
TENTER, A. M., and A. M. JOHNSON. 1997. Phylogeny of the
tissue cyst-forming coccidia. Adv. Parasitol. 39:69–139.
TENTER, A. M., K. LUTON, and A. M. JOHNSON. 1994. Speciesspecific identification of Sarcocystis and Toxoplasma by
PCR amplification of small subunit ribosomal RNA gene
fragments. Appl. Parasitol. 35:173–188.
THOMPSON, J. D., D. G. HIGGINS, and T. J. GIBSON. 1994.
CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, positions-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673–4680.
VOTÝPKA, J., V. HYPŠA, M. JIRKŮ, J. FLEGR, J. VÁVRA, and J.
LUKEŠ. 1998. Molecular phylogenetic relatedness of Frenkelia spp. (Protozoa, Apicomplexa) to Sarcocystis falcatula
Stiles 1893: is the genus Sarcocystis paraphyletic? J. Eukaryot. Microbiol. 45:137–141.
WHEELER, W. C. 1995. Sequence alignment, parameter sensitivity, and the phylogenetic analysis of molecular data. Syst.
Biol. 44:321–331.
WILLIAM TAYLOR, reviewing editor
Accepted August 9, 2000