ANNUAL

REVIEWS Further
Quick links to online content
Annu. Rev. Eeol. Syst. 1991. 22:525--64
Copyright © 1991 by Annual Reviews Inc. All rights reserved

FUNGAL MOLECULAR
SYSTEMATICS
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

Thomas D. Bruns
by University of Winnipeg on 09/11/14. For personal use only.

Department of Plant Pathology, University of California, Berkeley, California 94720

Thomas J. White

Hoffmann-La Roche, 1145 Atlantic Ave., Alameda, California 94501

John W. Taylor

Department of Plant Biology, University of California, Berkeley, California 94720

KEY WORDS; evolution, phylogenetic analysis, taxonomy, sequence, fingerprint

MOLECULAR EVOLUTION OF FUNGI

The fungi comprise both members of the kingdom Fungi as we now recognize
it (Ascomycota, Basidiomycota, Zygomycota, and Chytridiomycota) and
fungal-like protists such as the Oomycota and the cellular and acellular slime
molds (Myxomycota and Acrasiomycota) . Treating this admittedly
polyphyletic assemblage as a group is useful because these organisms often
fill rather similar roles within ecosystems, and they have traditionally been
studied almost exclusively by mycologists and plant pathologists . Throughout
this review, Fungi will refer to the Kingdom, fungi to the organisms studied
by mycologists . The fungi , as thus defined, are of great importance for the
following reasons: (a) They are the primary decomposers in all terrestrial
ecosystems; (b) they are important symbiotic associates of vascular plants
both in mutualistic and parasitic relationships; (c) they constitute the over­
whelming majority of plant pathogens and as such have a tremendous eco-

525
0066-4 1 62/9 1 / 1 1 20-0525$02.00
 526 BRUNS, WHITE & TAYLOR

nomic impact (Several significant human pathogens are also fungi); (d) they
offer several well-developed genetic systems for molecular biologists (Sac­
charomyces cerevisiae, Neurospora crassa, Aspergillus nidulans), and (e)
they are crucial to the fermentation and biotechnology industries .
In spite of their importance, very little is known about evolutionary rela­
tionships within the fungi . Their simple and frequently convergent morpholo­
gy, their lack of a useful fossil record, and their diversity have been major
impediments to progress in this field. With the development of molecular
technique s , many new avenues are now available that allow us to circumvent
these obstacles .
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

Many techniques have been used to study evolution within the fungi . In this
review we only consider those that compare nucleic acids, and we have made
no attempt to cite all papers on this topic . Instead we have chosen those that
by University of Winnipeg on 09/11/14. For personal use only.

enable us to illustrate specific points about different methods , types of

analysis, modes of molecular evolution , or evolutionary relationships. We
attempt to answer the questions of "what is the appropriate technique for a
given level of divergence?" and "what are valid ways to analyze various types
of molecular data?" We summarize what is known of evolution within and
among the major divisions of fungi. Because only a small number of fungi
have been sampled and the preliminary data are often fragmentary or equivo­
cal , however, we have not proposed or endorsed any global classification
scheme for the fungi .

METHODS USED I N FUNGAL MOLECULAR

SYSTEMATICS

DNA-DNA Hybridization
The small size of fungal genomes and the fact that the proportion of repetitive
sequences is much smaller than in plants or animals would seem to make them
ideal for hybridization studies, but in fact the technique has very restricted
usefulness because of the manner in which fungal genomes evolve . Virtually
all studies in fungi have focused on the percentage of cross-hybridization
between total DNA extracts rather than the thermal stability of hybrids .
Several variations of both isotopic and spectrophotometric assays have been
used and are reported to yield reasonably similar results if the percentage of
cross-hybridization is greater than 90% . At lower percentages the results of
different methods vary dramaticall y (78 , 90). Perhaps the most interesting
result to emerge from such studies is the observation that the percentage of
DNA that cross-hybridizes between even closely related species is ex­
ceedingly low , typically less than 20%, while the percentage cross­
hybridization between individuals within a biological species is typically
greater than 90%. Species pairs exhibiting intermediate values are rarely
 FUNGAL MOLECULAR SYSTEMATICS 527

reported . Vilgalys & Johnson ( 1 57) point out that the low level of cross­
hybridization betwccn what appear to be closely relatcd species stands in stark
contrast to the gradual reduction in hybridization between distantly related
animals . The mechanisms that underlie this apparently rapid genomic turn­
over in fungi remain unknown and are an area ripe for further research.
Whatever the cause, the phenomenon effectively limits the method to ques­
tions of very close relationships (78, 89) . Some have used DNA hybridization
for questions beyond the species level by focusing solely on ribosomal RNA
genes . These highly conserved genes provide sufficient cross-hybridization
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

for comparisons of greater phylogenetic distances , but resolution from such

studies remains low (9 1 ) .
Even i n cases where the resolution i s sufficient, the options for analyzing
by University of Winnipeg on 09/11/14. For personal use only.

the hybridization distance values produced are more limited than for character
data, and statistical testing of the resulting trees is difficult (38) . Another
major drawback is the need for pairwise comparisons. For N taxa the number
of experiments equals N(N-I)I2. Thus, for a study with 25 taxa, 300 ex­
periments are necessary. This number assumes no replications or even recip­
rocal hybridization experiments , both of which are typically needed to verify
and improve the estimated hybridization values. As a result, large quantities
of DNA are necessary for hybridization studies; this requirement eliminates
the possibility of examining fungi that do not grow well in culture or produce
sufficient quantities of easily collected biomass in nature . A further drawback
is that the results of two different studies of related organisms are not readily
compared without additional experiments.
The sharp drop off in cross-hybridization has commonly been used to help
define species. This practice has proven to be of great utility in yeast
systematics where correlation with mating studies has generally supported the
concept that biological species can be recognized by levels of cross­
hybridizations greater than 80% (89, 125) . In filamentous fungi, defining
species with levels of cross-hybridization has been used less often, and at least
a few exceptions to strict correlation with ability to mate are known (33 , 157).
Even with the exceptions, however, cross-hybridization measurements
represent the only simple one-number comparison that seems to correlate
fairly well with ability to mate; for this reason alone the technique will
probably continue to be used.

Restriction Enzyme Analysis

RFLPS VERSUS MAPPING Restriction patterns are generated by cleavage of

DNA with restriction enzymes , followed by size separation of the resulting
fragments via gel electrophoresis. The fragment patterns may be compared in
two ways, either: (a) an RFLP (restriction fragment length polymorphism)
approach is used in which the pattern of fragments or the individual fragments
 528 B RUNS, WHITE & TAYLOR

themselves are analyzed , or (b) a mapping approach is employed by using the

pattern of fragments to deduce a map of the enzyme sites , which then become
the units of analysis along with mapped length mutations. Mapping is the only
way to determine the physical relationship of restriction fragments to each
other, and as a result it has been and will continue to be the tool of choice for
investiga1:ing structural variation in mitochondrial DNA (mtDNA). Mapping ,
however, is a time consuming and relatively error-prone process. In theory,
maps can be compared directly , but in practice , detailed alignment of maps
from different studies is often not possible without at least some dire<;t
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

fragment comparisons. The resulting resolution of phylogenetic relation­

ships is often marginal because unlike plants and animals , fungi lack a
large contiguous genomic region that is relatively free of length differences.
by University of Winnipeg on 09/11/14. For personal use only.

Length mutations (i.e. insertions and deletions) complicate mapping , and

restriction sites are prone to convergent loss and saturate quickly . For
these reasons mapping to make phylogenetic comparisons among fungi is
of l imited value.
RFLP analyses can be very fast, and although a physical comparison is
required, this can be done among 10 to 40 isolates simultaneously rather than
by pair-wise experiments . The limits depend only on the number of lanes in a
single gel . Problems exist with RFLPs , however, if one tries to use the
number of fragment differences to estimate the degree of difference or
nucleotide divergence or if one treats the individual fragments as independent
characters . The greatest problem exists if length mutations are present and if
multiple enzymes are surveyed (Figure 1). In this case a single mutation will
be count,ed as at least 2N differences where N equals the number of enzymes
surveyed.. For obvious reasons, length mutations also invalidate estimations
of nucleotide divergence. Even if no length mutations exist, the relationship
between the number of fragment differences and the number of mutations is
not very precise (Figure 2). Furthermore, even without length differences ,
fragments are clearly not independent characters because single site or length
changes always result in multiple fragment changes . This lack of in­
dependence should concern anyone employing a parsimony analysis , and it
could lead to erroneously high levels of confidence from bootstrapping. In the
absence of length mutations , nucleotide divergence can be calculated by the
method of Nei & Li ( 1 1 8). However, their formula for fragments employs no
correction for multiple hits and so only makes reasonable estimates if the level
of divergence is low (118). The only tests conducted with this formula were
on circular DNA molecules (both simulated and real) where unobserved
flanking regions do not exist; it is unclear how the estimates derived from
random probe analysis would compare .
How serious are these problems? Unfortunately, length mutations occur at
a high frequency in mtDNA and probably in nuclear DNA of fungi . Many
 FUNGAL MOLECULAR SYSTEMATICS 529

Figure 1

Taxon 1 Taxon 2
Fragment differences relative
----
to region without insert

insertion

A 2

B 3

c 3
�
N 2
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org
by University of Winnipeg on 09/11/14. For personal use only.

Figure 2
Site and fragment differences
probe
taxa
A B C D E

A 1 2 3 4

B 0 1 2 3

c 2 2 1 2

o 3 3 3 1

E 3 3 3 2

Figure 1 Problems with counting fragment difef rences

tion enzyme maps for a hypothetical region of taxon 1 are shown with sites for enzymes A to N
given on separate lines. A probe based on a cloned fragment of enzyme A i s used to visualize a
portion of the area (shaded). The same region from taxon 2 contains a single insert. This single
length difference would create RFLP differences in any enzyme surveyed. In the simplest case
(enzymes A & N), this would result in the loss of one fragment from the taxon I pattern and the
gain of a longer fragment in the taxon 2 pattern. An additional fragment difference appears if the
insert·contains sites for the given enzyme (B & C).

Figure 2 Problems with counting fragment difef rences

contributing problem. Maps are shown for the same enzyme in five different hypotheti cal taxa
(A-E); each differs from the proceeding by a single site change. The area visualized by a probe is
indicated. The matrix shows the number of site differences separating each taxon (top) and the
number of observable fragment differences (bottom). Note that single site changes always result
in 0, 2, or 3 changes in the number of fragments seen but additional site changes add either 0 or 1
additional fragments.

researchers have recognized the problem and tried to circumvent it by: (a)
making course-maps to check for length mutations ( 1 1 7) , (b) scanning probe
regions with only a single enzyme to eliminate the multiple counting problem
(47, 144), or (c) not analyzing the fragments as individual characters but
instead cataloging the fragment patterns from a single probe region as differ-
 530 BRUNS, WHITE & TAYLOR

ent allelic forms of a given genetic locus (74, 1 08, Il l ). This latter approach
is ideally suited to the investigation of population structure. The use of RFLPs
as fingerprints for strain identification is another major strength (see below).
For these purposes the use of RFLP analyses is likely to flourish.

TARGET REGIONS FOR RESTRICTION ENZYME ANALYSIS Mitochondrial

DNA has been widely used for evolutionary studies in fungi. Several features
contributl� to its popularity: (a) It has a useful size ( 176- 1 7 Kb), small enough
so that it can be mapped or compared by RFLPs in its entirety, but large
enough to supply many characters. (b) No evidence exists for methylation of
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

bases-thus a potentially confounding factor of nuclear DNA is avoided. (c)

The strongly A + T biased composition makes isolation of purified mtDNA
relatively easy (73) . (d) The high copy number of the mitochondrial genome
by University of Winnipeg on 09/11/14. For personal use only.

makes it possible to visualize restriction fragments easily by hybridization of

total DNA to mtDNA probes. Mitochondrial restriction fragments larger than
2Kb can often be visualized directly from stained gels of total DNA digested
with restriction enzymes with G-C four-base recognition sites (e.g. Haem,
efoI, HpaII). These approaches are ideal for large comparative studies where
crude DNA extracts from many isolates are the rule. (e) MtDNA is rich in
RFLPs at the intraspecific level ( 1 7, 42-44, 47, 144, 152) , and length
mutations are the major cause of the variation in all cases where these
polymorphisms have been mapped ( 17, 1 37, 1 5 2) . Length mutations cause
some analytical problems, but one major advantage is that, unlike site
changes which are unique to a specific enzyme, length difference can be
dctected by virtually any restriction enzyme (Figure 1 ) . (j) MtDNA is the best
studied genomic element in fungi. Five ascomycetous mitochondrial genomes
have been extensively sequenced: Aspergillus nidulans, Neurospora crassa,
Podospora anserina, Saccharomyces cerevisae, and Schizosaccharomyces
pombe (13, 27, 1 65 ) . Maps of many other genomes are published, and most
are summarized along with basic information about the structure and function
of these genomes (72, 1 65 ) .
Plasmids are extremely common in fungi ( 1 36) and potentially offer an­
other multicopy target region for mapping or RFLP studies, but the dis­
advantages of using plasmids generally will outweigh any advantages. They
are neither universally present nor conserved in sequence, and they may be
horizontally transmitted (26, 1 07).
The nuclear ribosomal DNA repeat unit (rDNA) has also been used ex.­
tensively for restriction enzyme studies in fungi, and many of these maps have
been recently summarized by Garber et al (45). In most eukaryotes, including
all true fungi, rDNA exists as a tandomly repeated array of the three largest
rRNA genes separated by transcribed and nontranscribed spacers. In Physa r­
um and Dictyosteiium, however, these genes are arrayed in an ex­
trachromosomal palindromic repeat (20, 25). Gene order appears to be un-
 FUNGAL MOLECULAR SYSTEMATICS 531

iversally conserved with the exception of the 5 S gene, which may or may not
be within the repeat and is reported to exist in the opposite orientation in at
least some Basidiomycetes (23) . The multiple copies of the repeat unit appear
to homogenize quickly via concerted evolution (4a), and thus they generally
behave like a single copy gene. A minor exception to this exists in some
plants, animals, and Oomycota where different copies within the tandem
repeat may vary in size owing to different copy numbers of small subrepeats
within the intergenic spacer (IGS) (82 , 1 3 1 , 106).
The gene-spacer-gene arrangement of the multicopy array is part of the
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

underlying reason that rDNA has been so popular. Portions of genic regions
are so highly conserved that all heterologous probes hybridize strongly to
them, yet other genic regions vary considerably at even moderate taxonomic
by University of Winnipeg on 09/11/14. For personal use only.

levels. The spacers , particularly the intergenic spacer (lGS), often vary
significantly even at the intraspecific level. Mapping studies have shown that
both site and length differences occur in the repeat and that the level of RFLP
variation often is similar to or slightly less than mtDNA (Table 1 ) .
Universally conserved sequences within the ribosomal genes also make
ideal priming targets for enzymatic amplification (see below); Vilgalys &
Hester ( 1 56) have recently taken advantage of this fact to speed up the
mapping of portions of the largc subunit rRNA. Their method makes mapping
only slightly more work than RFLP analysis, and for the additional effort one
should gain considerable resolution and flexibility in analytical methods.
Single-copy and multi-copy anonymous clones have been used with in­
creasing frequency in RFLP and mapping studies of fungi to examine dis­
persed genetic regions. Clones containing highly repeated DNA sequences
can be quickly identified by colony hybridization using total genomic DNA as
a probe (62, 1 1 1 , 1 39) . Single-copy regions are confirmed by probing south­
ern blots of restricted DNA with individual clones ( 1 1 1 ) .

Table 1 Relative levels of observed intraspecific RFLP variation 1

Random
Fungus single-copy clones MtDNA rDNA References

Agaricus bitorquis nt + + 68
Armillaria spp . nt +++ + 3 , 144
Coprinus cinereus + ++ + + 166, 34
Fusarium oxysporum nt +++ + 81
Laccaria spp. nt +++ + 46, 47
Neurospora spp . ++ + + 132,117, lSI, 152
Phytophthora spp. +++ + nt 44
Septoria tritici +++ + + 1 09, Bruce McDonald
personal communication

-+-similar levels; + +, slightly greater; + + +, much greater; nt, not tested

'comparisons are only made within species, not between
 532 BRUNS , WHITE & TAYLOR

Studies employing random clones have revealed an amazing amount of the

RFLP polymorphism both within and between species (Table I); virtually
every clone in some studies has been found to be polymorphic. (108). Many
of the polymorphisms , particularly between species, involve apparent loss of
the fragment or at least loss of ability to detect it via hybridization (44, 1 09 ,
R . Vilgalys, personal communication). Mapping o f the sites within s i x ran­
dom clones in Coprinus cinereus, by Wu et al (166), demonstrated that many
of the polymorphisms were caused by length mutations . They estimated that
about 20% of random clones of 17 kb would contain unique inserts. These
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

results are consistent with a model of frequent length mutation in the nuclear
genome and may help explain the rapid drop off seen in DNA-DNA
hybridization experiments discussed above.
by University of Winnipeg on 09/11/14. For personal use only.

Sequence Analysis
DIRECT SEQUENCING The use of DNA sequences for evolutionary studil�s
can overcome many of the problems associated with restriction enzyme
analysis . The large number of characters compared can substantially increase
the resolving power. One can also observe the mode of sequence variation,
i.e. whether a change is a transversion or transition, silent or selected, and can
measure the degree of nucleotide bias; these observations may be incorporated
into the phylogenetic analysis in various ways (see below) . Results from
different laboratories can be directly compared, and the publication of se­
quences and their deposition in electronic databases (GENBANK, EMBL)
facilitate: the confirmation of results and their application to other taxa without
the need to obtain strains or clones or to repeat experiments .
Until recently, however, recombinant DNA techniques used to obtain
sequence information were sufficiently difficult and laborious that the study
of large numbers of species or individuals required exceptional efforts . Two
methods--direct sequencing of ribosomal RNA and direct sequencing of
amplified DNA fragments-have circumvented the need for cloning and thus
dramatically reduced the time and effort required for comparative sequencing
studies .
The first rRNA to be sequenced extensively was the 5S rRNA ( 10 , 7 1 ,
158), but this gene is too small and evolves too rapidly to be suitable for the
study of most fungal relationships (60, 148) . Additionally, compensatory
substitutions that maintain RNA secondary structure and the existence of
multiple , independently evolving copics of the gcne in filamentous Ascomy­
cota, including Neurospora and Aspergillus ( 5 , 140), present further prob­
lems.
Direct sequencing of the two largest rRNA by primer extension has broader
applicability to systematic studies (61, 93) . The rRNA template can be easily
isolated in large quantities. This procedure has been used extensively for
 FUNGAL MOLECULAR SYSTEMATICS 533

determining the 1 6S rRNA sequences of prokaryotes ( 1 64) , and Lynn &

Sogin ( 1 OOa) have extended its use to the systematics of protists . A deficiency
of the method is that the sequence of only one strand of the nucleic acid is
obtained . This results in a relatively high frequency of errors (1-5%) because
ambiguities cannot be resolved by comparison with the opposite strand.
Another potential problem is that RNA sequences of some organisms have
extensive posttranscriptional sequence modifications or "editing" and do not
reflect the actual DNA sequence of the gene ( 1 49) .
In contrast to rRNA sequencing , the polymerase chain reaction (peR)
developed by Mullis 0 1 6 , 133) allows biologists to sequence DNA from
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

many species or individuals in a few days. Several methods are available for
direct sequencing of peR products. These include methods that generate and
sequence single strands (59, 67 , 1 1 4) or that directly sequence double strands
by University of Winnipeg on 09/11/14. For personal use only.

(22). Any of these approaches when working properly is ideally suited for
large comparative studies. The accuracy of direct sequencing of peR products
is improved over rRNA sequencing because both strands can be determined.
A second advantage is that regions other than nuclear rRNA genes are
accessible for sequence analysis . "Universal" primer sequences for both
nuclear and mitochondrial rRNA genes as well as the internal transcribed
spacer have been described for fungi ( 1 56, 1 60), and in animal systems
primers for cytochome b have also been developed (85). An underlying
assumption of scquence analysis is that the phylogeny of the region is a good
indicator of the phylogeny of the organisms. A good test of this assumption is
to compare the results from regions that are physically and functionally
unlinked; peR makes such a test feasible with little additional effort. Another
important advantage of peR is that only minute amounts of DNA are needed.
As a result, rare or obligately parasitic fungi are now accessible to molecular
systematists .
One concern regarding direct sequencing of amplified DNA involves errors
introduced during synthesis by the DNA polymerase. The initial observed
error frequency for the Taq DNA polymerase was as high as one substitution
per four hundred base pairs after 30 cycles of amplification ( 1 38). However,
conditions can be optimized to reduce this error frequency to less than one
substitution in 1 5 ,000 bp (49). Fidelity is usually of no. concern in direct
sequencing of amplified DNA (in contrast to sequencing single clones from
amplified DNA) because the sequence obtained is a consensus of all mole­
cules present in the reaction (59) . Thus, each individual misincorporation will
be represented only very infrequently in the population of DNA molecules to
be sequenced. Even if the error is introduced in the first cycle, using a single
molecule of target only 25% of the relevant band density on the sequencing
gel will reflect the erroneous base. Although cloning is slower and potentially
more error prone, individual clones may be worth sequencing when allelic
 534 BRUNS, WHITE & TAYLOR

variation is high, but several should be determined to confirm that the

differences found are not artifactual.

SELECTION O F APPROPRIATE REGIONS T O SEQUENCE There are several

factors to consider in selecting a region for sequence analysis (76) . Thes.e
include: (a) The region should be evolving at an appropriate rate for the
comparison of interest. Ideally this means that the region supplies enough
consistent differences to separate the taxa into statistically supported
monophyletic groups. Regions that are too conserved will provide too few
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

changes. Regions that are too variable will contain too many inconsistent
characters due to multiple substitutions at single positions, and alignment may
be an additional problem. (b) The region should be present ideally as a single
by University of Winnipeg on 09/11/14. For personal use only.

copy or should at least evolve like a single copy region (e. g . rDNA, mtDNA) .
The danger of multicopy regions is that different copies might be compared in
different species (i . e . paralogous comparisons) . (e) The region should have
the same function in all taxa. Evolution of a new function changes selective
pressures and therefore the rate of sequence change. (d) The effect of base
composition and codon bias should be examined. Both factors can distort
estimates of divergence (36, 142).
Virtually all sequence studies in fungi have focused on the rRNA genes.
Their popularity is caused primarily by the presence of universally conserved
regions that serve as ideal primer sites . Different regions of the mitochondrial
and nuclear rRNA genes diverge at different rates; thercfore if one wants to
get the maximum amount of information from a minimum amount of sequenc­
ing the question of which regions are most appropriate for a specific level of
comparison is important. To this end a preliminary comparison among six
different rRNA regions is relevant (Figure 3). The situation with rRNA genes
is complicated because both base substitutions and length mutations make
alignment of some regions ambiguous. Thus, as more distant comparisons are
made the number of alignable nucleotides decreases. This problem is most
acute in the mitochondrial genes (Figure 3); they are thus not particularly
useful for distant phylogenetic comparisons. At intermediate taxonomic
levels , however, the mitochondrial genes may have a higher proportion of
variable sites. This point can be seen by comparing the MS 1 I2 region of the
mitochondrial small subunit rRNA (SrRNA) gene and the NS3/4 region of the
nuclear SrRNA gene (Figure 3). This comparison is reasonably sound because
both regions overlap equivalent structural domains. Similar comparisons of
LrRNA genes would be useful particularly for the regions of the nuclear gene
that have been used for phylogenetic studies in fungi (56-58, 88, 1 67 , 1 68).
Protein coding genes have some advantages over rRNA genes and spacers
in that alignment of the sequence is less problematic. Protein sequences also
lend themselves to differential weighting of bases by codon position, and third
position sites can provide a relatively good estimate of the neutral
 FUNGAL MOLECULAR SYSTEMATICS 535

600
R B

500

.!J 400

.�
g= 300
.!/
�=
..
:; 200
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

100
by University of Winnipeg on 09/11/14. For personal use only.

ML7/8 ML5/6 MSl/2 NS3/4 NS5/6 NS7/B

mitochondrial Nuclear
Primer defined regions

Figure 3 Size and variability 0/ aliRnable sequences/rom six selected rRNA regions. Pairwise
comparisons of sequences of Suillus sinuspaulianus versus Rhizopogon subcaerulescens (R),
Boletus satanas (B), and Saccharomyces cerevisae (S) are shown. The regions are defined by
primer pairs described by White et al (160) from the mitochondrial LrRNA (ML), mitochondrial
SrRNA (MS), and the nuclear SrRNA (NS) genes. All regions were completely alignable
between Suillus and Rhizopogon except within MS112 in which 59 bp of the Suitlus gene could
not be unambiguously aligned with Rhizopogon. Basidiomycete sequences are from Bruns ct al
1 989 , 1990; T. D. Bruns, T. M . Szaro, unpublished; and the Saccharomyces sequences were
from GENBANK.

substitution rate . Unfortunately, no "universal" primers have yet been de­

veloped for fungal protein genes. Part of the problem is that many or most of
the well-characterized nuclear protein genes may be multicopy; for the rea­
sons mentioned above this makes them unattractive. Mitochondrial protein
genes have been favored in animal studies, but in fungi virtually all of the
large mitochondrial proteins, except for cytochrome oxidase subunit III ( 1 2 1 ),
are frequently interrupted by numerous large introns.

Electrophoretic Karyotyping

Recent advances in gel electrophoresis technology have enabled megabase­

sized DNA molecules to be efficiently separated. These techniques have
already been applied extensively to the separation of fungal chromosomes.
Mills & McCluskey's ( 1 1 2) recent review lists 25 species of fungi whose
electrophoretic karyotypes have been examined. One of the most surprising
results of these studies is the high frequency of chromosomal length poly­
morphisms observed within species; this adds a new dimension to the emerg­
ing picture of fungal genome fluidity but essentially eliminates the method for
phylogenetic studies above the population level.
 536 BRUNS , WHITE & TAYLOR

ANALYTICAL OPTIONS FOR MOLECULAR DATA

To facilitate our later discussion of phylogenetic studies we briefly outline

some of the methods currently used for analysis and recommend several
reviews for a comprehensive discussion of the subject (38, 1 43a, 1 50) . There
are two main categories of methods for inferring phylogenetic trees: those
based on distance, and those based on characters . Distance methods create
trees by reducing all information on genetic characters to pairwise estimates
of similarity or dissimilarity. Variations such as the UPGMA procedure
assume or at least work best with equal rates of change on all lineages,
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

while others such as the neighbor-joining method and maximum likelihood

are less constrained (38, 1 35). Character-based methods such as parsimony
search for the tree that minimizes the number of character state changes.
by University of Winnipeg on 09/11/14. For personal use only.

Methods of both types such as maximum likelihood, evolutionary parsimony,

and compositional statistics ( 1 42) are designed specifically for analysis
of sequence data and are based on specific assumptions about the manm:r
in which sequences evolve. Other more general methods can be adapted
to sequence analysis by using a molecular evolution model to calculate
the distances or by weighting certain types of characters or character state
changes. An important point to keep in mind is that trees do not "reconstruct
a phylogeny"-they are an inference of it. To paraphrase Felsenstcin (38):
The real question is not "what is the best tree?" but rather, "how mm:h
confidence can we place in a given phylogenetic estimate?" The latter
question is a statistical one, and several methods can be used to address
it. Bootstrapping (37) can be used to resample the data and assess the strength
of intemal branches. It has been used primarily with parsimony, but
version 3 .4 of PHYLIP allows it to be easily adapted to any method of anal­
ysis (39). A major advantage of bootstrapping is that it can be applied easily
to moderately complex trees, although with large numbers of taxa the com­
putational time becomes a major constraint. Several other methods-­
maximum likelihood, evolutionary parsimony, and the winning-sites method
(38, 92 , 1 24)--can be used to compare trees and ask if one is significantly
better than another. With these methods a small number of taxa are used
(typically four) in ordcr to reduce the total numbcr of trees comparcd to a
manageable number.
When the number of variable characters is sufficiently large and when most
characters have changed only once, all methods yield similar if not identical
results (32, 80, 1 34). Problems arise as characters become saturated with
mutations , and under these conditions methods often disagree and stati stical
evidence; conflicts with different methods (54). Similar problems occur when
multiple branches originate nearly simultaneously , such as during an adaptive
radiation. The best documented case involves the phylogeriy of the hominoid
 FUNGAL MOLECULAR SYSTEMATICS 537

primates, which cannot be resolved with an acceptable level of statistical

confidence despite large quantities of sequence data (69) .

ANALYSIS AT THE POPULATION LEVEL

The population level is of great interest to mycologists and especially plant

pathologists because it is a window into the process of speciation. The genetic
structure of pathogens may indicate their potential for development of new
races and fungicide-resistant strains . Sexual recombination, however, serves
to shuffle genetic markers , and allelic differences often are not fixed. There­
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

fore a dicotomously branching phylogenetic representation of relationships is

often inappropriate. If the species are asexual , however, phylogenetic
methods of analysis may be relevant.
by University of Winnipeg on 09/11/14. For personal use only.

RFLPs As Alleles
Thc use of RFLPs for genetic markers has now become a standard genetic
tool , and their use in fungi was recently reviewed by Michelmore & Hulbert
( 1 1 1 ) . Two later papers by these authors and their colleagues on Bremia
lactucae, an obligately parasitic Oomycete, provide a model for the use of
RFLPs as genetic markers in diploid fungal systems (74, 75) . For haploid
systems the recent work on Septaria tritici by McDonald & Martinez ( 1 08)
provides a useful example. Their work with RFLPs is unique in that it focuses
on the population structure within a single wheat field, and their results reveal
a high level of genetic variation, including differences between different
lesions on a single leaf. Their findings certainly demonstrate that for single­
copy random clones, much of the variation may be within populations. As
they point out this result raises questions about conclusions of other studies
that employ relatively small intraspecific samples or assume that worldwide
collections provide a good representation of population-level variation , but
this concern is probably most relevant to the particular technique used: RFLPs
of random clones.
A recently developed PeR-based method, termed RAPD (random ampli­
fied polymorphic DNA), avoids the need for restriction enzymes, blotting,
probing, or cloning and produces fragment differences similar to RLFP
analysis ( 1 62). Typically several fragments of varying intensity are produced
by amplification with a single short primer of arbitrary sequence . Standardiza­
tion of amplification conditions is very important because the fragment
pattern observed is highly sensitive to concentrations of Mg+2 , DNA
polymerase , primers and template DNAs, and cycling temperatures. Mapping
of individual fragments has demonstrated that they behave as simple Mendel­
ian loci ( 1 62). Several potential problems exist in the use of RAPDs . First,
alternative alleles are usually seen as the absence of the fragment; this means
 538 BRUNS, WHITE & TAYLOR

heterozygotes and homozygotes are often not distinguishable, Second, the

multiple fragments of a single RAPD sometimes map to the same or nearly the
same locus and so may not be independent loci ( 1 62) . Third, different band
intensities may make scoring patterns ambiguous . When used as markers for
constructing genetic maps these problems are all relatively minor because
segregation analysis can be used to clear up most of the ambiguities. If
RAPDs are used to analyze population genetics, however, these problems will
need to be addressed; the speed and convenience of the method virtually
ensures that they will.
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

Fingerprinting
For distinguishing strain or species differences a unique RFLP pattern can be
by University of Winnipeg on 09/11/14. For personal use only.

used as a fingerprint. For this purpose, clones containing repeated elements

have proven extremely useful. In Candida albicans and Magnaporthe grisea,
closely related species or different pathogen-defined portions of the same
species can be distinguished by their lack of dispersed copies of the repeat
element (62 , 1 39) . In both cases, intraspecies RFLP variation was very high.
In Magnaporthe grisea every rice pathogen had a distinct pattern; this finding
is particularly impressive given that isozymes found little or no variation in
these strains (62). In the case of Candida albicans, RFLP patterns actually
changed at measurable rates from serial culturing on rich media and apparent­
ly also changed in isolates taken over the course of infection in a single
individual (139). In Colletotrichum lindemuthianum a cloned repetitive ele­
ment revealed a conserved RFLP pattern within the species (R. J. Rodriguez ,
personal communication) , but a different pattern between species. Subsequent
sequence: analysis demonstrated that tandemly repeated copies of a degenerate
trinucleotide , CAN, were responsible for the multilocus hybridization pattern
CR. J. Rodriguez, unpublished) .
Multilocus fingerprints can also be produced by hybridization of simpk­
sequence synthetic oligonucleotide probes to minisatellite DNAs . This
approach avoids the necessity of cloning, and is beginning to be used widely
in animal systems (2, 1 55 ) . In fungal systems the approach has recently betm
used with Penicillium, Aspergillus, and Trichoderma ( l 1 Oa) .
While multilocus probes are very powerful for fingerprinting , numerous
problems exist if one tries to use similarity among fingerprints to discern
genctic relationships (99, ' 100) . Some of these problems would be less
significant in haploid asexual fungi because recombination and the problem of
ascertaining whether an individual is homo- or heterozygous would not be a
factor. Nevertheless, one can be misled by nonhomologous fragments with
similar electrophoretic migration and missing or unresolved alleles . All of the
difficulties with respect to analyzing fragment differences as characters also
apply to fingerprints . Furthermore, a recent study by McDonald & Martinez
( 1 09) has shown that little correlation exists between genetic distances ca1cu-
 FUNGAL MOLECULAR SYSTEMATICS 539

lated from allele frequencies of single copy RFLPs and from fragment differ­
ences visualized from multilocus probes. Even distances calculated from two
different multilocus probes did not correlate well ( 109) .
Strain and species differences frequently have been correlated with rDNA
RFLPs. For example, seven species of Sclerotina, six species of Canidia, and
1 5 different intraspecific groups of Rhizoctonia solan; can all be separated by
rDNA RFLPS (86, 1 0 1 , 1 57a) . In other cases such as Cenococcum geophilum
the amount of intraspecific variation appears to be considerably higher. This
may be an artifact caused by a broadly defined morphological species that
includes several distinct biological species (97), but the amount of intraspecif­
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

ic RFLP variation may also be highly variable among unrelated taxa. Differ­
ences in size and structure of the lOS spacer, the level of methylation, and the
amount of sexual recombination clearly could affect the level of intraspecific
by University of Winnipeg on 09/11/14. For personal use only.

rRNA variability.
Vilgalys & Hester ( 1 56) also used a portion of the rDNA repeat to distin­
guish the human pathogen Cryptococcus neoformans from three other spe­
cies, but their RFLPs were derived from a peR amplified portion of the
repeat. This approach is not only faster, it also avoids the potential complica­
tion of RFLPs caused by methylation differences. In addition, it allows one to
limit the survey to a portion of the rDNA that is evolving at the appropriate
rate for the comparison of interest.
MtDNA RFLPs have served as fingerprints , but their use is often most
straightforward at the strain rather than the species level because of the high
RFLP variability of the genome. Different levels of variation are often
observed among equivalent taxonomic levels of closely related fungi (44, 47,
1 5 1 ). The reasons for these differences in the level of intraspecific mtDNA
variation are unclear and certainly are complicated by human agricultural
practices, sample size differences, and the basis of the taxonomic categories.
If one is planning to use mtDNA RFLPs for fingerprinting , the take home
message is that the level of variation one might see below the species level
cannot be anticipated from previous studies but must be determined empir­
ically.

PHYLOGENETIC STUDIES

Owing to their economic importance, their laboratory tractability, or the

hopeless condition of their classification, a few fungi have been subjected to
intense molecular phylogenetic scrutiny by a variety of methods .

Yeasts
Clark-Walker and colleagues (24), have mapped mtDNA, determined its gene
order, and sequenced cytochrome oxidase subunit II (cox/l) genes in Dekkera
species and species of the anamorphic genera Brettanomyces and Eeniella.
 540 BRUNS , WHITE & TAYLOR

With these data, they have asked an important question about mtDNA evolu­
tion: Do phylogenetic relationships inferred from gene order or length muta­
tions agn:e with those based on nucleotide substitution?" They conclude that,
"neither length nor gene rearrangements are useful characteristics for es­
tablishing relationships between these molecules or the yeasts harboring
them." It is worth noting, however, that both mtDNA size and gene order in
these yeasts are at least consistent with their sequence data. Species with the
same gene order also have similar sized mtDNAs, and branch together on the
unrooted cox II tree. Two large rearrangements hold the large genome species
together; B . custersii retains this order, while additional unique rearrange­
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

ments occur in the two other lineages. What the sequence data have provided
is much greater resolution and a likely root for the tree. Although mtDNA
sizes make sense when reviewed on the coxIl tree , size differences are great
by University of Winnipeg on 09/11/14. For personal use only.

enough that length mutation mapping would be useful only between the two
Dekkera species. Their similar mtDNAs are like those of outbreeding Neuros­
pora species, among which mapped length mutations have been phylogeneti­
cally useful (151).
Clark-Walker et al (24) also recognize the problem of using phylogeny to
define taxa. They ask, "Can species and genera be defined by DNA similarity
coefficients?" and recommend lumping the two Dekkera species and putting
Eeniella back in Brettanomyces. Lumping sexually reproducing species with
very similar mtDNAs such as Dekkera may be premature because members of
different Neurospora biological species do have identical mtDNAs ( 1 5 1 ) .
Howcver, putting Eeniella in Brettannomyces seems logical because :it
branches within Brettanomyces and no mating-based species concept is possi­
ble with these asexually reproducing fungi .
The use of nuclear small subunit rRNA sequence to detennine fungal
phylogeny has been swiftly applied to yeast classification. Barns and col­
leagues (4b) used the gene to examine relationships among 1 0 of the most
commonly pathogenic species of Candida. Their results confirmed the hetero­
geneous nature of the taxon and produced the first estimate of phylogenetic
relationships within it. Their analysis included a filamentous ascomycet�:.
Aspergillus, and three additional ascomyceteous yeasts. Addition of Basi­
diomycota to the analysis should prove interesting, because it has long been
thought that Candida may include members of both divisions .
To obtain complete rRNA sequence for the many fungi required for sys­
tematics is both time consuming and expensive. Kurtzman & colleagues (57,
58, 88) have addressed this problem and argue that partial sequences of small
and large subunit nuclear rRNAs can provide enough characters to classify
ascomycete and basidiomycete yeasts . Determining the statistical support for
the resulting phylogenies would seem essential to this approach. Gueho et al
(57) considered the question of significance in an intuitive way when they
 FUNGAL MOLECULAR SYSTEMATICS 541

used 350 bp of large subunit and 338 bp of small subunit sequence to compare
basidiomycete yeasts, including animal dermatophytes in the genus Tricho­
sporon. They found that maximum likelihood analyses of each subunit alone
produced different trees, and the deep branches in the phylogeny based on the
combined sequence were marked to reflect this uncertainty. Alternative trees
can be compared using a likelihood ratio test (38); if this test had been used,
conclusions about the polyphyly of Trichosporon species might be stronger.
Instead, branch rearrangements in the uncertain region can produce a
monophyletic Trichosporon clade.
In a study of Sterigmatomyces sensu lato , Gueho et al (58) showed that the
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

four segregate taxa fell into two groups. Yamada and colleagues ( 1 67) took
the same approach and problem further and claimed that one of the segregate
genera, Tsuchiyaea. was distinct from another, Fellomyces, based on distance
by University of Winnipeg on 09/11/14. For personal use only.

analysis of partial small subunit rRNA sequence They made this claim
although their analysis of large subunit sequence contradicted it. In this case,
statistical tests are important. Of the 1 60 bp of small subunit rRNA sequence
given, only one site could be used to compare the taxa in question with the
outgroup, Sterigmatosporidium polymorphym--clearly not enough to es­
tablish statistical support. The creation of new taxa is a matter of opinion, but
if this approach is used to justify the new taxa ( 1 68) , statistical testing cannot
be ignored . Demonstrating statistical significance does not necessarily require
long sequences. Mitchell & colleagues ( 1 1 5) used parsimony analysis and
bootstrapping with the short 5 . 8S rRNA sequences of yeast-like basidiomy­
cetes to support morphologically based classification of Filobasidiella and
Filobasidium species.
Filamentous Ascomycetes

NEUROSPORA Neurospora has been the subject of a surprising number of

evolutionary studies, considering its reputation as a laboratory denizen . When
length mutations were used to compare four biological Neurospora species,
most of the variability, and hence the oldest divergences, were found in N.
crassa and N. intermedia, neither of which were monophyletic; N. tetrasper­
ma and N. sitophila formed monophyletic groups with less variability ( 1 5 1 ) .
This result is consistent with at least three different but not mutually exclusive
explanations: (i) the ancestral taxon is N. crassa / N. intermedia, and N.
tetrasperma and N. sitophila have diverged recently (ii) N. crassa and N.
intermedia became sympatric before reproductive isolation was complete and
have exchanged mtDNAs; (iii) the conflict is illusory because too few muta­
tions are seen to establish statistical significance . In support of the first
explanation, Taylor & Natvig ( 1 5 1 ) argue that among closely related species,
intraspecific diversity of clonally propagated molecules [i.e. mtDNA of N.
crassa (103) and N. tetrasperma (95)] should be greatest in the ancestral
 542 BRUNS, WHITE & TAYLOR

species . This same argument has been made for mammals ( 1 63). By this
argument, interpretation of mtDNA variation in closely related species must
account for variation present in the ancestral population before species di ­
verged. Only when extinctions have reduced intraspecific variation relative to
interspecific variation will phylogenetic trees be reliably supported. Nuclear
genomes in sexually reproducing fungi will also contain alleles present before
divergen(;e, but the frequencies of alleles should differ between populations .
The second explanation , hybridization following allopatric speciation, i s
supported by the interfertility observed between N. crassa and N. intermedia
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

( 1 20). The third explanation, that too few data exist, is true and would be best
addressed by sequencing studies , but to date no region with cnough variation
has been found. Anonymous clone nuclear DNA studies have not contradicted
by University of Winnipeg on 09/11/14. For personal use only.

the mtDNA tree, but with the exception of N. tetrasperma, neither have they
confirmed it because too few common isolates have been examined ( 1 1 7 ) .
Low resolution mapping indicated that length mutations were rare i n the
nuclear fragments , but sites outside the region of hybridization made in­
terpretation somewhat more difficult.
Other genera of the Sordariaceae have been compared to Neurospora, first
by DNA-DNA hybridization (33) which showed that they are unusually
closely related ascomycete genera (87) . Recently DNA sequence comparison
of nuclear ITS and mitochondrial SrDNAs ( 1 53) showed that the
sordariaceous genera are no more divergent in these DNA regions than are
species of Laccaria (48) or Suillus (6). This disparity of generic and specific
variation in the two groups cannot be ascribed solely to taxonomic opinion, as
it holds for biologically determined species of Neurospora and Laccaria .
Either ITS and mitochondrial SrDNA evolve more slowly in the Sordariaceae
than the Agaricales, or speciation occurs more rapidly in the Sordariaceae .
Those that favor a molecular clock with a constant rate , at least in higher
fungi , might look to differences in life histories of saprobic ascomycetes and
mycorrhizal basidiomycetes , or differences between monokaryon or dikaryon
dominated life cycles, for an explanation of different rates of speciation.

FUSARIUM The appeal of this fungus for evolutionary studies is obvious: It

is economically important, well suited to the laboratory, and no stranger to
taxonomic controversy. Jacobson & Gordon (77) surveyed mtDNA variability
within 78 strains of Fusarium oxysporumf. sp . melonis . This sample included
all known vegetative compatibility groups (VeGs) and host-defined races
within the taxon. Initial comparison by RFLPs revealed seven forms of the
mtDNAs . Each of these forms was subsequently mapped, and the restriction
site and kngth differences were then analyzed by both distance and parsimony
methods. The results of the analysis showed that yeGs were strongly corre­
lated with mtDNA type: six of the veGs exhibited unique mtDNA patterns ,
 FUNGAL MOLECULAR SYSTEMATICS 543

while two shared a single mtDNA pattern. This strong correlation stems in
part from the strictly asexual life cycle of this fungus , but it also suggests that
either the rate of evolution of new YCGs, which is presumably based on
nuclear genes , is similar to the rate of RFLP evolution in the mtDNA or that
mtDNA recombination may serve to homogenize variants within portions of
the population that retain the ability to fuse. The pattern of host-defined races,
however, was not strongly correlated with the inferred phylogenetic rela­
tionships , suggesting that they may have evolved both more rapidly and
convergently (77).
Guadet & colleagues used direct sequencing of two portions of the nuclear
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

LrRNA gene to examine phylogenetic relationships within 52 strains of eight

species of the genus Fusarium (56) . The results of their phylogenetic analysis
correlate well with traditional classification, based on both anamorphs (asex­
by University of Winnipeg on 09/11/14. For personal use only.

ual stages) and teleomorphs (sexual). One species, F. nivale, was unrelated,
but even this result is supported by prior taxonomic opinions. Their analysis ,
however, does not permit evaluation of the statistical strength of the con­
clusions. To examine this question we reanalyzed their data by a bootstrap
parsimony method, which shows that most of their conclusions are strongly
supported (Figure 4) . The removal of F. nivale from the genus Fusarium and
the monophylesis of two of the three teleomorphic genera and two of the five
anamorphic sections are all supported above the 95% level. Section Liseola is
the only supraspecific taxon not supported above the 80% level . The overall
strength of these branches combined with the correlation with prior taxonomic
opinions demonstrates that their analysis represents a very reasonable initial
estimate of evolutionary relationships within Fusarium . The usefulness of
partial sequences from nuclear LrRNA for phylogenetic analysis at in­
termediate to distant levels is also evident, and the database that they have
started will make it easy to add additional taxa such as the related anamorphic
genera Cylindrocladium, Cylindrocarpon, and Tubercularia .
Although the broad framework of relationships within Fusarium was clear­
ly an appropriate taxonomic level for the approach of Guadet and colleagues,
the minor sequence divergences revealed within and between closely related
species were generally insufficient to resolve these lower-level relationships.
Their analysis of multiple varieties and isolates, however, clearly shows that
although variation exists at virtually all levels , intraspecific variation does not
have a strong effect on tree topology at higher levels. This is an important
assumption of molecular systematic analysis that has not often been tested
thoroughly.
Questions about higher level ascomycete systematics have also been ad­
dressed, including the placement of enigmatic organisms such as Pneumocys­
tis (35) , as well as problems of fungi possessing what have been considered
convergent, reduced morphologies (e. g . Plectomycetes with their simple asci)
 544 lBRUNS, WHITE & TAYLOR

Anamorphs I TeieomoIphs

l
Fusarium oxysporum

F. oxysporum

k
83
var. bullJigenum
F. oxysporum
var. redolens
37
F. moni/iforme
F. moniliforme

97
55 G. fujikuroi

F. moniliforme
var. subglutinans
]1 ]..,
i3
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

F. graminearum

J1
98 G. zeae
70
F. culmorum
by University of Winnipeg on 09/11/14. For personal use only.

F. decemcellulare

85 F. decemcellulare

F. decemcellulare
96
F. solani var. coeruleum
F. sol ani f. sp. pisi

F. solani vaT. martii

vaT. cucurbitae

100
99 F. solan; vaT. minus

F. solani var. javanicum

N. haematococca
79

F. solan; var. radicicola

L-_______ Neurospora crassa

F. nivale

L-____. ______ Saccharomyces cerevisiae

Figure 4 Bootstrap analysis of phylogenetic relationships within Fusarium (Based on the data
of Guadet et ai-56). Relationships depicted are based on Parsimony analysis. Horizontal
distances c orrespond to the minimum number of inferred mutations; vertical distances are
arbitrary. Numbers on subterminal branches correspond to the number of trees from a bootstrap
sample size of 1 00 in which the clade existed as a monophyletic group.

or contradictory morphologies (e . g . Ophiostoma with its pyrenomycete peri­

thecium and plectomycete asci). Berbee & Taylor (9) analyzed I8S rDNA
sequences using parsimony analysis with bootstrapping to show that Ple<:­
tomycetes are a monophyletic group compared to Pyrenomycetes and yeasts,
and that, for Ophiostoma the ascocarp is a better indicator of phylogenetic
 FUNGAL MOLECULAR SYSTEMATICS 545

relationships than the ascus. Bowman et al . ( 1 2) examined Coccidioides

immitis, an anamorphic human pathogen claimed for the Ascomycetes and
Zygomycetes ( 1 29), and showed that its 1 8 SrRNA sequence groups it with
other ascomycete pathogens in the Plectomycetes.

Filamentous Basidiomycetes

Within the filamentous basidiomycetes , most phylogenetic studies have thus

far been limited to mushroom-forming taxa: agarics, boletes, and their gaster­
oid relatives . Most of the published studies of agarics have focused on
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

biological species complexes (i .e. mating-defined species that are difficult to

separate morphologically) or closely related morphological species. At both
levels lack of resolution has been a common problem with the techniques
by University of Winnipeg on 09/11/14. For personal use only.

used.
In the Armillaria mellea complex , both mtDNA RFLPs and rDNA maps
have been used to study relationships among the biological species of Armil­
laria, and both resulted in relatively limited resolution. For rDNA, ten
variable sites and two length differences were identified; based on these
differences the 1 3 North American and European taxa were placed into six
groups, and relationships among these groups were analyzed by compatibility
and parsimony analysis . The resulting unrooted network shows that differ­
ences between the six groups are limited to one or two site or length
differences, and four of the nine informative changes occur convergently (3) .
In the case of mtDNA the patterns between species are so dissimilar that
RFLP comparison is not possible (79 , 1 44) .
Relationships among Laccaria laccata, L . bicolor, and L . amethystina
were also examined with both rDNA and mtDNA RFLPs , with results very
similar to those of the Armillaria studies. Little variation was found in rDNA
RFLPS , while mtDNA RFLPS were so variable that most intraspecies com­
parisons were out of range (46, 47) . In both Armillaria and Laccaria,
intraspecific mtDNA variation was so common that it proved useful for strain
fingerprinting (47a, 1 45) .
In the Collybia dryophila complex, DNA-DNA hybridization was used to
study relationships ( 1 57). A virtual continuum of intermediate hybridization
percentages enabled estimation of phylogenetic relationships . The tree pro­
duced correlated well with mating-compatible groups, and divergence within
these potential breeding units provided evidence for allopatric speciation.
Intermediate hybridization percentages have also been reported among Eu­
ropean species of Armillaria, but these were not used for phylogenetic
inference (79) . In both the Collybia and the Armillaria studies , estimation
error associated with interspecific comparisons appears to be fairly high;
differences of greater than 10% existed , dependent on which isolates were
chosen for intraspecific comparison. This variation is typical of DNA hybridi-
 546 BRUNS, WHITE & TAYLOR

zation; whether it is biological or methodological in origin , it would certainly

have an effect on the confidence of inferred interspecific relationships.
Morphological species of Suillus were examined by mtDNA mapping (16,
17). The extreme size differences among mtDNAs of Suillus species made
detailed alignments of mtDNAs impossible. To overcome this problem fine­
structure maps of the mitochondrial rRNA genes were constructed and
aligned. This time-consuming approach yielded only 47 shared differences
among the 19 taxa investigated. Parsimony analysis of these data clearly
separated Suillus from the related genera and thereby resolved a long-standing
controversy concerning the position of Paragyrodon . Reanalysis of these data
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

by bootstrapping shows that this result is supported above the 99% level, and
the result has since been confirmed within a subsample of the taxa by
by University of Winnipeg on 09/11/14. For personal use only.

sequence analysis with the same level of confidence (15). Relationships

within Suillus, however were only defined by one to three site changes , and a
,

reanalysis of these data with bootstrapping shows that none of the branches
are supported at even the 95 % confidence level. Failure to resolve the
relationships within Suillus is caused primarily by the low sequence di­
vergence (0.3-2. 9%) found within the genus in the regions mapped. D irect
sequencing of peR products has now confirmed these low estimates in several
portions of the mitochondrial rRNA genes (14, 15) but has failed to produce a
robust estimate of evolutionary relationships within the genus (T . Bruns,
unpublished results) . Although the tree produced from the restriction site
mapping study was not statistically significant, its correlation with mtDNA
size and basidiocarp morphology suggests that it is at least a reasonable
estimate . The close placement of S. spraguei (pictus) and S. luteus was the
only really glaring conflict with morphology (14). In lieu of stronger molecu­
lar evidence the meaning of the apparent confliCt remains uncertain .
The question o f what technique is appropriate for analysis o f closely related
species of basidiomycetes remains largely unanswered. Direct sequencing of
the nuclear ITS spacers, however, is one possible approach. The region
differs by approximately 1-3% within a morphological species of Suillus and
by about 10% or more between species examined, yet most of the region is
alignable among what we now know to be the closest relatives of Suillus:
Rhizopogon, Gomphidius, Chroogomphus (6, 15, 1 8 ) . ITS variation among
Laccaria species also ranges from about 1-3% (48) Within Armillaria,
however, ITS sequence variation may be insufficient to resolve relationships
within alii species, but a portion of the lOS appears to be more useful (J .
Anderson , personal communication) .
Several ongoing studies based on rDNA mapping and on sequencing of
nuclear or mitochondrial rRNA genes are focused on intra- and inter-family
level relationships within the agarics and boletes (18, 126 , 130). The results
from these studies should soon provide a general framework of relationships,
 FUNGAL MOLECULAR SYSTEMATICS 547

but many family-level relationships are likely to remain unresolved. Two

published studies may portend the morphological surprises likely to arise from
ongoing works. The first utilized partial sequence data from mitochondrial
LrRNA gene to show that the false-truffle Rhizopogon is more closely related
to Suillus than the latter is to three other members of the Boletaceae ( 1 5 ) . This
result was supported at the 99% confidence level through bootstrap analysis
and was also corroborated by a shared mitochondrial gene order found in
Suillus and Rhizopogon. This close relationship between Suillus and Rhizopo­
gon can now also be supported with high confidence levels by data from two
other portions of the mitochondrial rRNA genes as well from the nuclear
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

SrRNA gene ( 14, 1 8 ; Figure 3). In this case the explanation for the dramatic
morphological shift had been proposed 60 years earlier ( 1 02) , but the close­
ness of the relationship between mushrooms and false-truffles revealed by the
by University of Winnipeg on 09/11/14. For personal use only.

molecular data was still startling. Work in the genus Coprinus has now shown
similar relationships between secotioid genera Montagnea and Podaxis and
mushrooms in the genus Coprinus (70) . Another morphological surprise
appeared in a recent study by Hibbett & Vilgalys (66). In this case the results
suggest that the genus Lentinus is polyphyletic and the closest relatives of L .
tigrinus are polypores. N o statistics were used t o evaluate the tree, and their
analysis was flawed by the use of fragments as characters and the likely
presence of length mutations in the region analyzed, but their result is also
reported to be strongly supported by sequence data (D. S. Hibbett, personal
communication) . Thc morphological implications of their conclusions are
certainly significant: the switch between gilled and tubular hymenophores had
to occur at least twice during the evolution of Lentinus. Similar results have
also been reported for the Boletales ( 1 6) . Taken together these results suggest
that the underlying developmental difference between gills and tubes may not
be very great, and evolutionary switching between these two types of hyme­
nophores may be relatively easy, at least within lineages that have the
developmental potential to make tubes.

Oomycota
Oomycota, particularly Phytophthora and Pythium, have been the focus of
mtDNA and rDNA studies aimed at the species level and below. The most
popular approach has involved mtDNA . In Oomycota, the molecule is circu­
lar (84) and evolves through a combination of nucleotide substitutions , length
mutations (2 1 , 42, 43 , 1 06) , and rearrangements ( 14 1 ) . Most taxa have a long
inverted repeat containing the rRNA genes (72 , 1 1 0) . Judging from the
systematic distribution and sizes of the repeat, and the presence of very short
inverted repeats in taxa that lack it (e . g . Phytophthora, Leptomitus, Apodach­
lya; l I D , 1 4 1 ) , the repeat is an ancestral feature of Oomycota, has been lost
independently at least twice, and has changed size significantly in several
 548 E RUNS , WHITE & TAYLOR

lineages. Independent support for the convergent loss of the inverted repeat in
Apodachlya and Phytophthora is provided by a mapping study of nuclear
rDNA (83) .
Inheritance of mtDNA has been claimed to be uniparental in Pythium
( 1 05a) and Phytophthora (40) . In Pythium, however, no confirmation that
mating actually occurred was offered , and in Phytophthora, recombination of
mtDNAs could not be ruled out as many of the smaIIer mtDNA fragments
were obsc:ured by nuclear DNA fragments .
Frequent occurrence of both length mutations and rearrangements, coupled
with uncertainty about the mode of inheritance , makes it difficult to expoit
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

Oomycete mtDNA variation for phylogenetic purposes. One straightforward

use of mtDNA RFLPs is for fingerprinting . An example is the correlation of
asexual "hyphal sweIIing" Pythium isolates and their closest sexual relatives
by University of Winnipeg on 09/11/14. For personal use only.

( l 05b) . For phylogenetic studies, Forster and colleagues (42, 43) examined
isolates of Phytophthura mega�perma, P. parasitica, and P. infestans by
comparison of restriction fragments . Because they ignored length mutation ,
their estimates of nucleotide substitution among isolates became meaningless ..
With mapped mutations , however, they show that P. megasperma j. sp .
glycinea and P. m f. sp . medicaginis are as distant from each other as P .
infestans i s from P . parasitica; from this comparison they conclude that P .
megasperma comprises more than one species. This conclusion assumes that
interspecific variation should bc constant throughout a genus. Yet when:
mtDNA variation has been studied within and among biological species, it
was anything but constant (47 , 1 5 1) .
Forster and colleagues (44) have extended their studies of restriction frag·­
ment patterns to show that the amount of mtDNA diversity is variable within
six Phytophthora species and to compare intraspecific phylogenies inferred
from mtDNAs and nuclear DNAs. Although length mutations were not
accounted for in using unmapped fragments to estimate diversity, and alterna..
tive branching patterns were not evaluated statistically, there was good agree ­
ment between phenograms based on mitochondrial and nuclear DNA for P .
capsici, although less agreement for P . parasitica .
An examination of morphological, physiological , isozyme, and mtDNA
variation in many isolates of Phytophthora cryptogea and P. drechsleri
provided convincing evidence of morphological and physiological mis··
identification of these fungi ( l 1 3) and again makes it clear that some species
harbor more variation than others . As a result, Mills and coIIeagues proposed
that groups of isolates with similar amounts of variability be called "molecula:r
species . " Here too, their phenograms are based on fragment comparison and
do not account for length mutations; and trees with alternative branching
patterns a:re not ruled out. In the absence of a clear understanding of the mode
of mitochondrial inheritance, with variation exaggerated due to length muta-
 FUNGAL MOLECULAR SYSTEMATICS 549

tions, with no statistical testing of alternative trees, and without a thorough

comparison of known biological species, it seems premature to define clades
as "molecular species."
Sequencing of PCR amplified elements of the rDNA repeat has been
applied to the former "morphological forms" of Phytuphthora palmivora (94).
Use of neighbor-joining ( 1 35) and parsimony with bootstrapped samples (37)
showed that P. palmivora and P. megakarya are significantly more closely
related to each other than either is to P. capsici or P. citrophthora when P.
cinnamomi is the outgroup. In line with the findings of Foster and colleagues,
variation within and among species is itself variable; ITS regions that easily
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

separated P. palmivora and P. megakarya were too similar to be used

confidently to separate , P. capsici and P. citrophthora . Again, should we
expect variation to be equal in all species?
by University of Winnipeg on 09/11/14. For personal use only.

Neglected Groups
Chytridiomycota and Zygomycota have received only a little attention. Based
on 18S rDNA sequence , the rumen anaerobe Neocallimastix branches within
the chytrids and is closest to Spizellomyces ( 1 1). Sequence analysis of l8S
rDNA were also used to show that Glomus. Gigaspora. and Endogone are
true fungi but not ascomycetes. Unfortunately. the lack of any zygomycete
sequences for comparison prevented testing of their putative rclationship (L.
Simon, pers. comm. ) . In the Zygomycota, rDNA RFLPs from the animal
dermatophyte Basidiobolus have been analyzed (119) . Here, the human
isolates show little variation compared to those isolated from other sources ,
but parsimony analysis of RFLPs used to understand Basidiobolus phylogeny
suffered from the lack of an outgroup and the presence of length mutations. In
Myxomycota and Acrasiomycota, the field is completely open.

Relationships Among the Major Groups

The big picture of the evolution of fungi and their place in biota has been a
preoccupation of mycologists since the classics (19). Sogin and colleagues
have obtained nuclear small subunit rRNA sequences from many organisms,
including fungi (4 1 , 147). They have shown with these sequences that the
organisms traditionally claimed by mycologists fail to form one monophyletic
branch on the eukaryotic evolutionary tree (Figure 5). The cellular and
plasmodial slime molds diverge early , in the region populated by protists, and
the Oomycota share a branch with chrysophyte algae. Farther up the tree,
three familiar groups diverge, the land plants , animals, and Fungi. Included in
the monophyletic kingdom Fungi are representatives of the divisions, Chytri­
diomycota, Ascomycota and Basidiomycota, and we anticipate that the
Zygomycota can be included when a sequence is published.
The distance method used by Sogin and colleagues to infer phylogenetic
 550 B RUNS, WHITE & TAYLOR

NUCLEAR SMALL-SUBUNIT RI BOSOMAL RNA

Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org
by University of Winnipeg on 09/11/14. For personal use only.

Bacteria and Archaea

Is It In the Kingdom Fungi? Answer Significant Branch

Physarum, Plasmodial Slime Mold No 8-E
Dictyostelium, Cellular Slime Mold No 8-0
Achlya, Oomycota No A-C
Graci/aria, Red Alga No A-C
Chytridium, Chytridiomycota Yes A-B
Zygomycota no data

Figure 5 An evolutionary tree based on parsimony analysis of 185 rDNA sequence. Th(:
statistical support for internal branches is discussed in the section on "Relationships Among the
Major Groups" and is summarized as a series of questions about membership in the kingdom
Fungi. Branch lengths are arbitrary. GenBank 1 8 S rDNA sequences from the following taxa wen:
used in the analysis. Fungi: Saccharomyces cerevisiae, Neurospora crassa, Spongipellis
( =Tyromyc€s) unicolor ( I I ) , Chytridium confervae ( I I et al?). Plants: Oryza sativa. Animals:
Xenopus laevis. Achlya bisexualis, Graci/aria lemaneiformis, Dictyostelium discoideum, and
Physarum polycephalum.

relationships from small subunit rRNA does not lend itself to explicit statisti..
cal tests of the robustness of internal branches and branching order. Evidence:
that more than one evolutionary history can be obtained for the same organ·­
isms using the same DNA sequence is found by comparing the nuclear
SrRNA tn�es of F6rster et al (41) and Hendriks et al (64, 65) , which conflict in
their placement of ciliates, sporozoa, plants , animals , and fungi . The sensitiv··
ity of these trees to sequence alignment is made clear by Hendriks and
colleagues (64), whose alignment of all eukaryote sequences fails to place
Fungi on a single clade, while alignment of only the higher eukaryotes does .
Understanding the branching order of the three groups of higher eukaryotes
 FUNGAL MOLECULAR SYSTEMATICS 551

is a puzzle of great general interest. Hasegawa et al (63) applied a maximum

likelihood method to sequences from both the large and small rRNA subunits
to show that plants diverged prior to Fungi and animals , but they noted that
this result is not significant at the 95% confidence limit. Gouy & Li (55) used
transformed distance and parsimony methods with an estimation of the stan­
dard error of branch lengths to compare rRNA sequence, tRNA sequence, and
several protein sequences in hopes of solving this puzzle of higher eukaryote
radiation. Their tRNA tests were not significant at the 95% level and favored
either an early fungal divergence or an early plant divergence, depending on
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

which prokaryote served as the outgroup. The amino aeid sequences favored
an early fungal divergence with 95% confidence but included GA3PDH,
which may not be an appropriate protein to answer such ancient questions (see
by University of Winnipeg on 09/11/14. For personal use only.

below) . The combined small and large subunit rRNA sequences also favored
an early fungal divergence with 95% confidence, but alternative trees in
which animals or plants diverge first required only 7 or 9 more character state
changes out of ca. 1 550 total changes . Perhaps the clearest and most interest­
ing fact about the radiation of higher eukaryotes is not the exact order but that
the radiation appears to have occurred very quickly compared to subsequent
evolution, so that each kingdom has essentially the same relationship to the
others .
Statistical testing of branching order and internal branches has become a
necessary feature of phylogenetic studies. As an illustration, we used three
methods amenable to statistical evaluation-winning sites ( 1 24) , evolutionary
parsimony (92), and parsimony analysis with bootstrapped samples (37)-to
examine the internal branches and branching order of small subunit rRNA
trees composed of organisms studied by mycologists (4 1 , 65) . We have
framed these tests as questions about the inclusion of various organisms in a
monophyletic fungal kingdom.
Do the plasmodial slime molds belong in the Fungi? Hasegawa et al (63)
used large and small subunit rRNA sequences with maximum likelihood to
show that Physarum polycephalum was outside the Fungi at well above the
95% confidence level. To include Physarum in the fungi would mean includ­
ing many protists , the groups of eukaryotic algae, the animals and the plants.
Clearly, plasmodial slime molds should not be included among true fungi .
Do the cellular slime molds belong in the Fungi? Again, Hasegawa et al
(63) used large and small subunit rRNA sequences and maximum likelihood
to show that Dictyostelium discoideum was not a member of the Fungi . Using
small subunit rRNA data, we made a winning sites test of Dictyostelium
against Neurospora and rice, with Physarum as the outgroup. There is
significant support for the branch uniting Neurospora and rice to the exclusion
of Physarum and Dictyostelium . To include Dictyostelium among the Fungi
would mean including the plants. Parsimony analysis of 1 00 bootstrapped
 552 BRUNS, WHITE & TAYLOR

samples of the same data supported the same branch at the 1 00% level. The
cellular slime molds are outside the Fungi .
Do the Oomycota belong in the Fungi? A winning site test of small subunit
rRNA sequence from Achlya bisexualis against Neurospora and rice, with
Dictyostelium as the outgroup did not support the branch uniting Achlya and
Dictyostelium at the 95% level. Parsimony analysis with 100 bootstrapped
samples of the same molecule for the same taxa plus Chytridium confervae,
yeast, and the basidiomycete Spongipellis (;:Tyromyces) unicolor, supported
the branch uniting Achlya, rice and Dictyostelium to the exclusion of others at
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

the 100% level . However, the branch uniting Achlya and Dictyostelium to the
exclusion of rice and the Fungi was not supported at the 95% confidenc��
level . The: possibility that Achlya could form a monophyletic branch with the
by University of Winnipeg on 09/11/14. For personal use only.

Fungi cannot be ruled out, but if the Oomycota were included in a fungal
branch they would diverge earlier than the Chytridiomycota. This broadened
fungal kingdom would necessarily include the Chrysophyte algae and di··
atoms.
Do Chytridiomycota belong in the Fungi? Although mycologists have long
claimed these flagellated organisms, some have allied them with the ciliates
( 104). FC'rster & colleagues (4 1 ) used distance methods and I 8S rRNA
sequence to show that a monophyletic fungal branch could include chytrids .
Bowman & colleagues ( 1) used neighbor joining and winning sites tests on
1 8S rDNA sequence to show that the branch uniting chytrids with the rest of
the Fungi, to the exclusion of a ciliate , was supported at the 95% level.
Chytrids are Fungi .
Do the red algae belong in the Fungi? This question has intrigued several
generations of mycologists with adherents on both sides of a debate over
phenotypic attributes (28). Proponents see the red algae as ancestors to the
Ascomycota. The relationship between the red algae and the Fungi based on
small subunit rRNA is similar to that of the Oomycota. A winning sites test of
sequence from Gracilaria lemaneiformis against Neurospora and rice with
D ictyostelium as the outgroup supported the branch uniting Dictyostelium and
Gracilaria, but not at the 95% level. If Chytridium was substituted for rice in
this test, the branch uniting Dictyostelium and Gracilaria was supported at the
99% level. Parsimony analysis with 100 bootstrapped samples gave the sam;:
result. The branch uniting the Fungi was supported 1 00%, but the branch
uniting plants with the Fungi to the exclusion of Gracilaria and Dictyosteliurn
was not �;upported at the 95% level . The possibility of red algae forming a
monophyletic group with the Fungi cannot be ruled out, but if they do they
must branch earlier than the Chytridiomycota and cannot be direct ancestors
to the Ascomycota. This conclusion supports that of a prior DNA hybridiza­
tion study of red algal affinities using cloned basidiomycete rDNA, which
showed that red algae were no closer to Ascomycota than were green algae
(9 1 ) .
 FUNGAL MOLECULAR SYSTEMATICS 553

Phylogenetic comparisons of nucleotide sequences of 5S rRNA (7 1 ) or

GA3PDH ( 146) have produced trees that would exclude from the Fungi the
entire Ascomycota or just the yeasts , or require broadening the Fungi to
include plants and animals. Winning sites tests of small subunit rRNA
sequences of yeast against Neurospora and either Xenopus or rice, with
Achlya (Oomycota) as the outgroup, support the branch uniting yeast and
Neurospora at the 99% level. Bowman & colleagues ( 1 1 ) used the winning
sites test and neighbor joining to show that Neurospora and the basidiomycete
Spongipellis (= Tyromyces) were closest relatives compared to a plant with a
ciliate as an outgroup. In addition , parsimony analysis of 1 00 bootstrapped
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

samples of Chytridiomycota, Ascomycota, and Basidiomycota with rice and

Xenopus supports the notion that the branch unites the Fungi against rice and
by University of Winnipeg on 09/11/14. For personal use only.

Xenopus at 99% . Therefore, with small subunit rRNA sequence, neither the
Ascomycota nor the Fungi are polyphyletic, compared to animals or plants.
This result is unsettling because the trees based on GA3PDH and 5S rRNA
contradict those based on morphological and physiological phenotype as well
as nuclear small subunit rRNA sequence. The difficulty of using 5S rRNA to
analyze older evolutionary divergences has already been discussed. The
apparent conflict between the I SS rDNA and GA3PDH sequence data is
worth examining further because, at our present level of sophistication,
analysis of protein coding genes is more complex than that of rDNA genes .
With GA3PDH, Smith ( 1 46) used evolutionary parsimony to analyze
nucleotide sequences from E. coli, yeasts, filamentous Ascomycota and
Basidiomycota, plants, and animals . In the tree resulting from this analysis,
the branch grouping E. coli and yeasts with plants against filamentous fungi
and animals was supported at the 99% level .
E. coli has been shown to have two genes for GA3PDH ( 1 ) , and it is
postulated that one of them, the one used by Smith, has come to E. coli by
horizontal transfer from a eukaryote (3 1 ) . If this hypothesis is true, it could
explain why yeast appeared to branch close to E. coli. It does not explain
Smith's demonstration that a branch uniting yeasts and plants to the exclusion
of filamentous ascomycetes and animals is supported by evolutionary parsi­
mony analysis.
The evolutionary rates of molecules must match the antiquity of the di­
vergences in question. This may not be the case for GA3PDH and higher
eukaryotes. Pairwise comparison of amounts of nucleotide substitution in
GA3PDH between Saccharomyces cerevisiae and Cochliobolus hetero­
strophus shows percentages of nucleotide substitution for the first, second,
and third codon positions of this protein coding gene to be 32%, 20% , and
55% respectively. Given that total substitution in GA3PDH between two
isolates of S. cerevisiae can be as high as 1 1 . 3 % , nucleotide replacement may
be near saturation and is well above the point where multiple mutations at
single sites should confound analysis and produce an anomalous result.
 554 B RUNS , WHITE & TAYLOR

In comparing animals, plants, yeasts , and filamentous ascomycetes, one

would expect that the same result would be obtained regardless of which
representative species were used. This is not the case with GA3PDH; 59% of
the possible four-taxa evolutionary parsimony tests were not significant.
Omitting the extremely variable third colon position resulted in 55% in­
significant results . An assumption of evolutionary parsimony is that nucleo­
tide substitution is unbiased, but GA3PDH in yeast is known to have extreme
codon bias (8) , a feature it shares with E . coli. Highly expressed genes in
fermentative organisms may be under selective pressures unknown to their
homologues in organisms capable of growing into new food sources. Nucle-·
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

otide composition is also biased, e.g. the percentage of cytosine is 22% in

Saccharomyces, 31 % in filamentous ascomycetes, and 24% in E. coli. Como.
by University of Winnipeg on 09/11/14. For personal use only.

positional statistics ( 1 42) can be applied to evolutionary parsimony when

nucleotidt: substitution bias is known to exist. In this case, it did not help;
analysis of all possible four-taxa groups with compositional statistics returnecl
69% insignificant results with all three codon positions, and 57% with the
third position omitted .

The high levels of nucleotide substitution, the codon and nucleotide sub­
stitution bias , and the dependence of significant evolutionary parsimony
results on specific groups of four taxa-all indicate that GA3PDH is e vo l v i n g
too rapidly to be useful in determining the divergence of plants, animals, and
fungi . Loomis & Smith (98) used a similar approach with eight different
protein coding genes to study the phylogeny of Dictyostelium. In a compari­
son of these genes in Dictyostelium and S. cerevisiae, all except the actin gene
had at least 40% of their amino acids substituted . The unusual relationships
proposed for Dictyostelium by these authors were based on parsimony analy­
sis and di stance analyses uncorrected for mUltiple hits . It is very likely that
these genes are not appropriate for studies of divergences as old as that of
Dictyostelium and Fungi or mamtnals.

SUMMARY

Figure 6 provides a visual summary of the way we perceive the match

between methods available and taxonomic levels for which they are appropri­
ate. The necessary caveat is that equivalent taxonomic categories of different
lineages often vary in terms of the level of molecular divergence; therefore ,
choosing an approach for a given study will almost always require an empiri­
cal survey. Throughout this review we have followed the lead of Felsenstein
(38) in �.tressing the importance of the statistical testing of phylogenetic
hypotheses, but as indicated in Figure 6 we believe that the usefulness of a
method is not always limited to levels that yield a statistically significant
result. Correlation with prior phylogenetic hypotheses based on morphology,
 FUNGAL MOLECULAR SYSTEMATICS 555

F I g u re 6 P o p u l a t ion
genetics p hy l oge n e t i c r e l a t i o n s h i p s
------ --------��-
Fam I l I e s Cl asses &
PopulatIOns Spe c i e s Genera & Orders DiVis ions Kingdoms

Total DNA hyb r i d i za t i on �

RFLPs

anonomous c l on e �

RAPDs ;>
mtDNA
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

rONA

Mapping

anonamous clones
by University of Winnipeg on 09/11/14. For personal use only.

mtDNA (whole m o l e c u l e )

r O N A spacers

rONA genes m t or nuc

Sequenc1ng

spacers ( I T S )

Mitochondrial rONA genes

Nuclear rONA genes

Pro t e i n genes

Moderately conServed c: :>

I I
Conserved c::

I I I

Figure 6 Qualitative summary ofmolecular methods and their useful ranges in fungal systemat­
ics. Dark stippling i ndicates the useful ranges of currently tested methods, no stippling indicates
the projected range of RAPDs and direct protein sequencing if the appropriate genes can be
identified. Black indicates ranges in which sufficient numbers of consistent characters might be
expected to yield statistically significant results.

ecology, physiology, or other molecular data may provide additional support

for molecular phylogenies . Indeed, it is in cases where molecular phylogenies
conflict with other available evidence that it is most important to address the
question of statistical significancc .
Figure 6 also illustrates two important gaps in our current methodology:
relationships among closely related species and among Divisions and King­
doms. At the former level many methods are available, but to date none have
yielded statistically significant results. Sequence analysis of IGS or perhaps
third position sites of a variable protein may be useful at this level , but
currently no studies are available in fungi to evaluate these options . At the
levelof Kingdoms and Divisions most attention has been focused on nuclear
ribosomal RNA genes, but analysis of conserved proteins is starting to be
used more frequently (5 1 ) .
 556 BRUNS , WHITE & TAYLOR

FUTURE PROSPECTS

Beyond establishing the evolutionary relationships among fungi, perhaps the

most important unanswered question involves establishing a timeframe for it.
The fungal fossil record is scanty , but minimum divergence dates for at least a
few modem groups can be obtained (30 , 1 22 , 1 54) . Indirect dating based on
specific associations could be used to obtain maximum dates of divergence for
other groups. For example, diversification within genera of mmen fungi is
unlikely to have preceded the divergence of the mminants . Similarly , taxa of
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

msts or other obligate parasites that are restricted to a given plant taxon might
be assumed to be no older than their hosts. Thus, rough divergence dates
could be coupled with sequence divergence observed in extant lineages to gt::t
by University of Winnipeg on 09/11/14. For personal use only.

initial calibration points for fungal molecular clocks . A second and more
direct method of clock calibration may also be possible. This would involve
obtaining both sequence data and isotopic age estimates from specimens
preserved in amber, or semi-fossilized mud (52, 123). If such specimens can
be unambiguously assigned to an extant group, the amount of sequence
divergenl;e per unit time can be calculated. A similar approach could be used
to examine divergence between fungal strains recovered from buried perigla­
cial soil of known dates and their closest relatives from extant arctic soils (7) ,
or perhaps from frozen Antarctic soils of greater antiquity .
Another area for future research involves estimates of fungal biodiversity
for those species that cannot be cultured in the laboratory . For example, the
diversity of soil fungi is clearly underestimated by current soil dilution plate
methods. The molecular approaches used for estimating diversity for
thermophilic bacteria and plankton should also prove useful for fungi (50,
1 59) .
The expanding capability to type strains and species of fungi molecularly
will have a profound eff�ct on the analysis of natural populations. Inheritance
of nuclear and mitochondrial genes will be easier to study in the field. The
work of Smith et al ( 1 45) provides a good example. The persistence, succe:s­
sion, and dispersal of individual strains, whether symbionts or pathogens, will
be accessible to study using hyphae or spores instead of cultured isolates . This
approach should be particularly useful for epidemiological investigations of
the spread of plant and human pathogens, e.g. , in determining whether an
infection is the result of expansion of a vimlent clone or importation of a m:w
strain . The application to fungi of PeR-mediated identification systems de­
veloped for medical purposes ( 1 6 1 ) has begun with a study of Phytophthora,
Laccaria, and Cryptococcus species (48 , 96, 1 56). Realizing the full potential
of this method will require its testing on many isolates.
At the level of experimental genetic s , it should be possible to estimate
recombination frequencies of linked genes and constmct genetic maps via
 FUNGAL MOLECULAR SYSTEMATICS 557

analysis of individual meiotic spores. By analogy to studies on individual

sperm from mammals, this may be the only way to obtain genetic maps for
some species where genetic crosses are impractical. This may also improve
our ideas on the relationship of physical to genetic distance in the fungal
genome. In the area of developmental biology, understanding the evolution­
ary relatedness of other fungal species to those that have been extensively
analyzed , i.e. Saccharomyces and Neurospora, may make it easier to identify
a gene' s homolog in a novel organism.
A final area involves nomenclature and classification . If all fungi can be
compared through their nucleic acids and placed on a single phylogenetic tree
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

(56) , do we need to maintain the Deuteromycota ( 1 27 , 1 28)?

ACKNOWLEDGMENTS
by University of Winnipeg on 09/11/14. For personal use only.

We thank all the people who sent us prepublication data and manuscripts, also
Barbara Bowman for editorial help, Tim S zaro for assistance with graphics
and the data analysis , and Eric Swan for helpful editorial comments . This
review was prepared with support from NSF (BSR-870039 1 , BSR-89 1 8454)
and NIH (RO I AI28545) .

Literature Cited

1 . Alefounder, P. R . , Perham, R. N. 1 989. in Aspergillus nidulans SS rRNA genes.

Identification, molecular cloning and se­ Curro Genet. 1 1 :571-73
quence analysis of a gene cluster encod­ 6. Baura, G. , Szaro, T. M . , Bruns, T. D.
ing the class II fructose 1 ,6-bisphosphate 1 99 1 . Gastroboletus laricinus is a recent
aldolase, 3-phosphoglycerate kinase and derivative of Suillus grevillei: molecular
a putativc second glyceraldchyde 3- evidence. Mycol. Soc. Am . Newsl.
phosphate dehydrogenase of Escherichia (Abstr.) 42:5
coli. Mol. Microbial. 3:723-32 7 . Beiswenger, 1 . M . , Christensen, M .
2. Ali, S . , Miiller, C. R . , Epplen J. T. 1 989. Fungi as indicators o f past en­
1 986 . DNA finger pri n tin g by oligonu­ v iron ments . Curro Res. Pleistocene 6:
cleotide probes specific for simple re­ 54-56
peats. Hum. Genet. 74:239-43 8 . Bennetzen, J. L . , Hall, B . D. 1 982.
3. Anderson, J. B . , Bailey, S. S . , Pukkila, Codon selection in yeast. J. B ioi. Chem.
P. J. 1n9. Variation in ribosomal DNA 257:3026-3 1
among biological species of Armillaria, 9. Berbee, M. L . , Taylor, J. W. 1 99 1 .
a genus of root infecting fungi. Evolu­ Ribosomal DNA evidence for two
tion 43: 1 652-62 classes of Ascomycetes based on fruiting
4a. Arnheim, N . , Krystal, M . , Schmickel, body type . Mycol. Soc. Am. Newsl.
R . , Wilson, G . , Ryder, 0 . , Zimmer, E. (Abstr.) 42:6
1 980. Molecular evidence for genetic 10. Blanz, P. A . , Gottschalk, M. 1986.
exchange among ribosomal genes on Systematic position of Septobasidium.
nonhomologous chromosomes in man Graphinla and other Basidiomycetes as
and apes. Proc. Natl. Acad. Sci. USA deduced on the basis of their 5S ribo­
77:7323-27 somal RNA nucleotide sequences. Sys­
4b. Barnes, S. M . , Lane, D. J . , Sogin, M . tem. Appl. Microbial. 8 : 1 2 1-27
L . , B ibeau, C . , Weisburg, W . G . 1 990. 1 1 . Bowman, B . , Taylor, J . W . , Brownlee,
Evolutionary relationships among A. G . , Lee, J . , Lu, S . -D. , White, T. J .
pathogenetic Candida species and rela­ 1 99 1 . Molecular evolution o f the fungi:
tives. J. Bacterial. 173:2250--5 5 relationship of the Basidiomycetes,
5. B artoszewski, S . , Borsuk, P . , Kern, I . , Ascomycetes and Chytridiomycetes.
B artnick, E. 1 987. Microheterogeneity Mol. Bioi. Evol. In press
 558 BRUNS, WHITE & TAYLOR

1 2 . Bowman, B . , Taylor, J. W. , White, T. R . A . 1 97 8 . Dictyostelium rDNA con­

J. 1 99 1 . Molecular evolution and detec­ sists of non-chromosomal palindromic
tion of human pathogenic fungi . Mycol. dimers containing 5s and 36s coding re­
Soc. Am. Newsl. (Abstr.) 42:7 gions. Chromosoma 70: 1 9-29
1 3 . B rown, T. A. , Waring, R. B., Scazzoc­ 26. Collins, R. A . , Saville, B . J . , 1 990.
chio, c . , Davies, R. W. 1 985 . The Independent transfer of mitochondrial
Aspergillus nidulans mitochondrial chromosomes and plasmids during un­
genome. Curro Genet. 9: 1 1 3- 1 7 stable vegetative fusion in Neurospora .
1 4 . Bruns, T . D . , Fogel, R . , Taylor, J . W . Nature 345 : 1 77-79
1 990. Amplification and sequencing of 27. Cummings, D. J . , McNally, K. L. ,
DNA from fungal herbarium specimens. D omenico , J. M . , Matsuura, E. T.
Mycologia 82: 1 75-84 1 990. The complete DNA sequence of
1 5 . Bruns, T. D. , Fogel, R. , White, T. J. , the mitochondrial genome of Podospora
Palmer, J. D. 1 989. Accelerated evolu­ anserina. Curro Genet. 1 7 :375-402
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

tion of a false-truffle from a mushroom 28. Demoulin, V. 1 98 5 . The red algal­

ancestor. Nature 339: 140-42 higher fungi phylogenetic link: the last
1 6 . Bruns, T. D. , Palmer, J. D. 1 989. ten years. BioSystems 1 8 :347-56
Evolution of mushroom mitochondrial 29. De leted in proof
by University of Winnipeg on 09/11/14. For personal use only.

DNA: Sui/Ius and related genera. J. 30. Dilcher, D. L. 1 965 . Epiphyllous fungi
Mol. Evol. 28: 349-62 from Eocene deposits in western Ten­
1 7 . Bruns, T. D. , Palmer, J. D . , Shumard, nessee, USA. Palaeontographica 1 1 6:
D. S. , Grossman, L 1 . , Hudspeth, M. I-54
E. S. 1 988. Mitochondrial DNAs of 3 1 . Doolittle , R . F . , Feng, D. F . , An der son ,
Sui/Ius: three fold size change in mole­ K. L. , A lberro , M . R. 1 990. A naturally
cules that share a common gene order. occurring horizontal gene transfer from a
Curro Genet. 13 :49-56 eukaryote to a prokaryote . J. Mol. Evo!.
1 8 . Bruns, T. D . , Szaro, T. M. 1990. 3 1 : 383-88
Phylogenetic relationships within the 32. Duncan, T , Phillips, B . , W agner ,
K.
Boletales: molecular perspectives. In W. H. Jf. 1 980. A comparison of
Fourth International Mycological Con­ branching diagrams derived by various
gress Abstracts, ed. A. Reisinger, A . phenetic and cladistic methods. SysL
B resinsky p. 1 2. (Abstr. ) Regensburg
, Bot. 5:264-93
Germ,my: Bot. Inst. Univ. Regensburg 33. Dutta, S. K. 1 976. DNA homologies
1 9 . Buller, A. H. R. 1 9 1 4 . The fungus lore among heterothallic speci es of Neuros­
of the Greeks and Romans. Trans. Br. pora. Mycologia 68:388-401
Mycol. Soc. 5 : 2 1-66 34. Economou, A . , Castleton, L. A. , 1 989.
20. Cambell, G. R. , Littau, V. c . , Melera , Polymorphism of the mitochondrial L­
P. W . , Allfrey, V. G. , Johnson. E. M . RNA gene of the basidiomycete Copri­
1 979. Unique sequence arrangement of nus cinereus. Curro Genet. 1 6:4 1-46
ribosomal ge ne s i n the pal i ndromic 3 5 . Edman , J. C . , Kovacs, J. A . , Masur,
rDNA. molecule of Physarum po/yceph­ H . , Santi, D. V . , Elwood, H . J . , Sogin,
alum . Nucl. Acids Res. 6: 1 433-47 M. L . 1 988. Ribo somal RNA sequence
2 1 . Carter, D. A . , Archer, S. A . , Buck, K . shows Pneumocystis carinii to be a
W. , S haw. D . S . , Shattock , R . C . 1 990. member of the fungi. Nature 334:5 1 9-
Restriction fragment length polymor­ 22
phisms of mitochondrial DNA of Phy­ 36. Edward s , S. V _ , Arctander, P . , Wilson,
tophthora infestans. Mycol. Res. A. C. 199 1 . Mitochondrial resolution of
94: 1 1 23-28 a deep branch in the genealogical tree for
22. Casanova, J . -L . , Pannetier, C . , Jaulin, perching birds. Proc. R . Soc. Lond.
c . , K ourilsky, P . 1 990. Optimal condi­ 243:99- 1 07
tion for directly sequencin g double­ 37. Felsenstein, J. 1 985. Confidence in­
stranded PCR products with sequenase. tervals on phylogenies: an approach us­
Nucl. Acids Res. 1 8:4028 ing the bootstrap. Evolution 39:783-9 1
23 . Cassi dy 1. R . , Pukkila, P. J . , 1 987.
. 3 8 . Felsenstein , J. 1 98 8 . Phylogenies from
Inversion of 5S ribosomal RNA genes molecular sequences: i n ference and
within the genus Coprinus. Curro Genet. reliability. Annu. Rev. Genet. 22: 5 2 1 -
1 2 :33-36 65
24. Clark-Walker, G. D. , Hoeben, P. , Pla­ 39. Felsenstein, J. 1 99 1 . PHYLIP 3 . 4 . Dt!­
zinska, A . , Smith, D. K . , Wimmer, E . partment of Genetics SK-50, Univ.
H. 1 987. Application of mitochondrial Wash . , Seattle, Wash. 98 1 95
DNA analysis to yeast systematics. 40. Forster, H. , Coffey, M. D. 1 990. Mat­
Stud. Mycol . 30:259-66 ing behavior of Phytophthora parasiti­
25 . Cockburn, A . F. , Tayler, W. c . , Firtel , ca: evidence for sexual recombination in
 FUNGAL MOLECULAR SYSTEMATICS 559

oospores using DNA restriction frag­ implications for the origin of eukaryotes.
ment length polymorp h isms as gen etic Proc . Natl. Acad. Sci. USA 86:666 1 -65
markers . Exp . Mycol. 1 4:35 1 -59 52. Golenberg, M. E . , Giannasi, D. E . ,
41. Forster, H . , Coffey, M. D . , Elwood, Clegg, M . T . , Smiley , C. J . , Durbin,
H . , Sogin, M . 1. 1 990. Sequence analy ­ M . , et al. 1 990. Chloropla�t DNA se­
sis of the small subunit ribosomal RNAs quences from a Miocene Magnolia spec­
of three zoosporic fungi and implications ie s. Nature 334:656-58
for fungal evolution. Mycologia 82:306- 53. Deleted in proof
12 54. Gouy, M . , Li, W . -H . 1 989. Phylogenet ­
42. Forster, H . , Kinscherf, T. G . , Leong, S. ic analysis based on rRNA sequences
A . , Maxwe l l , D . P. 1988. Estimation of supports the archaebacterial rather than
relatedness between Phytophthora spe ­ the eocyte tree . Nature 339 : 1 45-47
cies by analysis of mitochondrial DNA. 55. Gouy, M . , Li, W-H. 1 989. Molecular
phylogeny of the kingdoms Animalia,
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

Mycologia 80: 466-78

43 . Forster, H . , Kinscherf, T. G . , Leong, S. Plantae, and Fungi. Mol. Bioi. Evol.
A., Maxwell, D . P. 1989. Restriction 6 : 1 09-22
fragment length polymorphisms of the 56. Guadet, J . , Julien J . , Lafay, J. F . , B ry­
mitochondrial DNA of Phytophthora goo, Y. 1 989. Phylogeny of some
by University of Winnipeg on 09/11/14. For personal use only.

megasperma isolated from soybean, Fusarium species, as determined by

alfalfa and fruit trees . Can. J. Bot. large-subunit rRNA sequence compari­
67:529-37 son. Mol. Bioi. Evol. 6:227-42
44. Forster, H . , Oudemans, P . , Coffey , M . 57. Gueho, E . , Kurtzman, C. P . , Peterson,
D . 1 990. Mitochondrial and nuclear S. W . , 1989. Evolutionary affinities of
DNA diversity within six species of Phy­ heterobasidiomycetous yeast estimated
tophthora . Exp . Mycol. 14: 1 8-3 1 from 1 8S and 25S ri bosomal RNA se­
45 . Garber, R. C . , Turgeon , B. G . , Selker, quence divergenc e . System. Appl. Mic­
E. U . , Yoder, O. C. 1 988. Organi zation robial. 1 2:230-36
of ribosomal RNA genes in the fungus 58. Gueho, E . , Kurtzman, C. P . , Peterson,
Cochliobolus heterostrophus. Curro S. W. 1990. Phylogenetic relationships
Genet. 1 4 :573-82 among species of Sterigmatomyces and
46. Gardes , M . , Fortin, J. A . , Mueller, G . Fellomyces as determined from partial
M . , Kropp , B . R . 1 990. Restriction rRNA sequences. Int. J. Syst. Bact.
fragment length polymorphisms in the 40:60-65
nuclear ribosomal DNA of four Laccaria 59. Gyll en s ten, U. B . , Erlich, H. A. 1 98 8 .
spp: L . hicolor, L . laccata , L . proxima , Generation of single-stranded DNA by
and L. amethystina. Phytopathology the pol ymerase chain reaction and its
80: 1 3 1 2- 1 7 appl ication to direct sequencing of the
47 . Gardes, M . , Mueller, G . M . , Fortin, J . HLA-DQa locus. Proc. Nat!. Acad. Sci.
A . , Kropp, B . R . 1 99 1 . Mitochondrial USA 85:7652-56
DNA polymorphisms in Laccaria hico ­ 60. Hal an ych K. M. 1 99 1 . 5S ribosomal
,

lor, L. laccata, L. proxima, and L . RNA sequences inappropriate for

amethystina . Mycol. Res. 95:206-16 phylogenetic rewnstruction. Mol. Bioi.
48. Gardes , M . , White, T. J . , Fortin, J. A . , Evol. 8:249-53
Bruns, T . D . , Taylor, J. W . 1 99 1 . 61. Hamby, R. K . , Sims, 1. E . , IsseI, 1.
Identification of indigenous and in­ E . , Zi mmer, E. A. 1 98 8 . Direct ribo­
troduced symbi otic fungi in ectomycor­ somal RNA sequencing : optimization of
rhizae by amp lificatio n of nuclear and extraction and sequencing methods for
mitochondrial ribosomal DNA. Can. J. work with higher plants. Plant Bioi.
Bot. 69: 1 80-90 Rep. 6 : 1 75-92
49. Gelfand, D. H . , White, T. J. 1 990. 62. Hamer, J. E . , Farral l, 1 . , Orbach , M .
Thermostahle DNA polymerases . In J . , Valent, B . , Chumley, F. G . 1 989.
PCR Protocols: A Guide To Methods Host spec ie s- specifi c conservation of a
and Applications, ed. M. A. Innis, D . family of repeated DNA sequences in
H. Gelfand, J. S . Sninsky , T . J . White . the genome of a fungal plant p atho gen .

16: 1 29-1 4 1 . Ne w York: Academic . 482 Proc . Natl. Acad. Sci. USA 86:9981-85
pp. 63 . Hasegawa, M . , Iida, Y., Yano, T-A. ,
50. Giovannoni , S. J . , Britschgi, T. B . , Takaiwa, F. , Iwabuchi , M. 1 985.
Mo yer , C . L . , Field, K . G . 1 990. Ge­ Phylogenetic relationships among eu ­
netic diversity in Sargasso sca bacterio­ karyotic kingdoms inferred from ribo­
plankton. Nature 345:60-63 somal RNA sequences. J. Mol. Eval.
51. Gogarten, J. P . , Kibak, H . , Dittrich, P . , 22:32-38
Taiz, 1 . , Bowman, E . J . , e t al. 1 989. 64. Hendriks , 1 . , De B aere, R . , Van de
Evolution of the vaculolar H+ -ATPase: Peer, Y . , Neefs , J . , Goris, A . , D e
 560 B RUNS, WHITE & TAYLOR

Wachter, R. 1 99 1 . The evolutionary po­ the lettuce downy mildew fungus , Bre·
sition of the rhodophyte Porphyra umbi­ mia lactucae. Mol. PI. Microbe In··
licalis and the basidiomycete Leucospor­ teraet. I: 1 7-24
idium scottii among other eukaryotes as 76. Irwin, D. M . , Wilson, A. C. 1 99 1
deduced from complete sequences of Limitations o f molecular methods for e s..
small ribosomal subunit RNA. 1. Mol. tablishing the phylogeny of mammals,
Evol. 32: 1 67-77 with special reference to the position of
65. Hendriks, L . , Goris, A . , Neefs , I.-M., elephants. In American Museum of Nat ­
Van ele Peer, Y., Hennebert, G., De ural History Symposium on Mammalian
Wachter, R. 1 989. The nucleotide se­ Phylogeny ed. F. S. Szalay, M. 1 .
,

quence of the small ribosomal subunit Novacek, M . C . McKenna. New York:

RNA of the yeast Candida albicans and Springer Verlag. In press
the evolutionary position of the fungi 77. Jacobson, D. J . , Gordon, T. R. 1 990.
among the Eukaryotes. Syst. Appl. Mi­
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

The variability of mitochondrial DNA as

crobial. 1 2 :223-29 an indicator of the relationships between
66. Hibbett, D. S . , Vilgalys, R. 1 99 1 . Evo­ popUlations of Fusarium oxysporum
lutionary relationships of Lentinus to the J.sp. melonis. Mycol. Res. 94:734-44
Polyporaceae: Evidence from restriction 78. Jahnke, K . -D . , Bahnweg, G . 1986 .
by University of Winnipeg on 09/11/14. For personal use only.

analy�;is of enzymatically amplified Assessing natural relationships in basid­

Ribosomal DNA. Mycologia 83 : ln press iomycetes by DNA analysis. Trans. Br.
67. Higuchi, G . R . , Ochman, H . 1989. Pro­ Mycol. Soc. 87 : 1 75-9 1
duction of single-stranded DNA tem­ 79. Jahnke, K . -D . , B ahnweg, G . , Worrall ,
plates by exonuclease digestion follow­ J. J. 1 987 . Species delimitation in the
ing the polymerase chain reaction. Nucl. Armillaria mellea complex by analysis
Acids Res. 1 7:5�65 of nuclear and mitochondrial DNAs.
68. Hintz, W. E. A . , Anderson , J. B . , Trans. Br. Mycol. Soc. 88:572-75
Horgen, P . A . 1988. Relatedness of the 80. Kim. J., Burgman, M. A. 1 988.
three species of Agaricus inferred from Accuracy of phylogenetic-estimation
restriction fragment length polymor­ methods under unequal evolutionary
phism analysis of the ribosomal DNA rates. Evolution 42:596-602
repeat and mitochondrial DNA . Genome 81. Kistler, H. c . , Bosland, P. W . , Benny,
32: 1 73-78 U. , Leong, S. , Williams, P. H . 1 987 .
69. Holmquist, R . , Miyamoto, M. M . , Relatedness of strains of Fusarium oxy­
Goodman , M . 1 98 8 . Higher-primate sporum from Crucifers measured by ex­
phylogeny-why can't we decide? Mol. amination of mitochondrial and ribo­
Bio i . Evol. 5:20 1 - 1 6 somal D NA . Phytopathology 77: 1 289-
70. Hopple, J . S . J r . 1 990. Phylogenetic re­ 93
lationships within the genus Coprinus 82. Klassen, G. R . , Buchko, J . 1 990. Sub­
based on molecular and morphological repeat structure of the intergenic region
evidence. See ref. 1 8 , p 25 in the ribosomal DNA of the
7 1 . Hori, H . , Osawa, S. 1 987. Origin and oomycetous fungus Pythium ullimum.
evolution of organisms as deduced from Curr o Genet. 1 7 : 1 25-27
5S ribosomal RNA sequences. Mol. 83. Klassen, G. R . , McNabb, S. A . , Dick,
Bioi. Evol. 4:445-72 M . N. 1 987. Comparison of physical
72. Hudspeth, M. E. S. 1 99 1 . The fungal maps of ribosomal DNA repeating un its
mitochondrial genome-a broader per­ in Pythium, Phytophthora and Apodaeh ­
spective. In Handbook of Applied lya . 1. Gen. Microbiol. 1 3 3 :2953-59
Mycology, Vol. IV , Biotechnology. pp 84. Klimczak, L. J . , Prell, H. H. 1 984.
2 1 3-4 1 . New York: Marcel Dekker Isolation and characterization of
7 3 . Hud5,peth, M. E. S . , Shumard, D. S . , mitochondrial DNA of the oomycetous
Tatti , K . M . , Grossman, L . I . 1980. fungus P hytophthora illfestans. Curro
Rapid purification of yeast mitochondri­ Genet. 8 :323-26
al DNA in high yield. Biochim . Biophys . 85 . Kocher, T. D . , Thomas, W. K . , Meyer,
Acta 6 1 0:22 1-28 A" Edwards, S. V . , PHabo, S . , et al.
74 . Hulbert, S. H . , Hott, T . W . , Legg, E. 1 989. Dynamics of mitochondrial DNA
1 . , Lincoln, S . E . , Lander, E. S . , evolution in animals: amplification and
Michelmore, R . W . 1 988. Genetic anal­ sequencing with conserved primers.
ysis of the fungus, Bremia lactucae, us­ Proc, Natl, Acad. Sci, USA 86;6196-
ing restriction fragment length poly­ 6200
morphisms. Genetics, 1 20:947-58 86. Kohn , L. M . , Petsche, D. M . , Bailey,
7 5 . Hulbert, S. H . , Michelmore, R. W. S. R . , Novak, L. A . , Anderson, J . B.
1 988. D NA restriction fragment length 1 988. Restriction fragment length poly­
polymorphism and somatic variation in morphisms in nuclear and mitochondrial
 FUNGAL MOLECULAR SYSTEMATICS 561

DNA o f Sclerotinia species. Phytopath­ 98. Loomis, W. F., Smith, D . W . 1 990.

ology 7 8 : 1 047-5 1 Molecular phylogeny of Dictyostelium
87. Kurtzman, C. P. 1985. Molecular taxon­ disco ideum by protein sequence com­
amy of the fungi. In Gene Manipula­ parison . Proc. Natl. Acad. Sci. USA .
tions in Fungi, ed. J. W. Bennett, L. L. 87:9093-97
Lasure, pp. 35-63. Orlando, Fla: Aca­ 99. Lynch, M. 1 988. Estimation of related­
demic ness by DNA fingerprinting . Mol. Bioi.
88. Kurtzman, C. P. 1989. Estimation of Evol. 5 :584-99
phylogenetic distances among ascomyce­ 1 00. Lynch, M. 1990. The similarity index
tous yeasts from partial sequencing of and DNA fingerprinting Mol. Bioi.
ribosomal RNA. Yeast 5:S35 1-S354 Evol. 7(5):478-84
89. Kurtzman, C. P . , Phaff, H. J . , Meyer, 1 00a. Lynn, D. H . , Sogin, M. L. 1 988.
S. A. 1983. Nucleic acid relatedness Assessment of phylogenetic rela­
among yeasts. In Yeast Genetics Fun­ tionships among ciliated protist s using
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

damental and Applied Aspects. ed. J. F. parial ribosomal RNA sequences derived
T. Spencer, D . M. Spencer, A . R. W . from reverse transcriptase. Biosystems
Smith, pp 139-66. N e w York: Springer­ 2 1 :249--54
Verlag 1 0 1 . Magee, B. B . , D' Souza, T. M . , Magee,
by University of Winnipeg on 09/11/14. For personal use only.

90. Kurtzman, C. P . , Smiley, M . J . , John ­ P . T. 1987. Strain and species identifica­

son, C. J . , Wickerham, L. J . , Fuscon, tion by restriction fragment length poly­
G. B. 1 980. Two new and closely rc­ morphisms in the ribosomal DNA repeat
lated heterothallic species, Pichia amy­ of Candida species. 1. Bact. 1 69: 1 639-
lophi/a and Pichia mississippiensis; 43
Characterization by hybridization and 102. Malen<;on, G. 1 93 1 . La serie des Aster­
deoxyribonucleic acid reassociation. Int. ospores. Trav . crypto. ded a L . Mangin.
J. Syst . Bacteriol . 30:208-1 6 1 :337-96
91. Kwok, S . , White, T . J . , Taylor, J . T. 1 03. Mannella, C. A Pittenger, T. H . , Lam­
. •

1 986. Evolutionary relationships be­ bowitz, A. M. 1979. Transmission of

tween fungi , red algae, and other simple mitochondrial deoxyribonucleic acid in
eucaryotes inferred from total DNA hy­ Neurospora crassa sexual crosses. 1.
bridizations to a cloned Basidiomycete Bacteriol. 1 37: 1 449--5 1
ribosomal DNA. Exp . Mycol. 10: 1 96- 1 04 . MarguliS, L. , Schwartz, K. V. 1 988.
204 Five Kingdoms; An Illustrated Guide To
92. Lake, J. A. 1987. A rate-independent the Phyla of Life on Earth . New York:
technique for analysis of nucleic acid W. H. Freeman 2nd ed.
sequences: evolutionary parsimony. 105a. Martin , F . N . 1 989. Maternal in­
Mol. Bioi. Evol. 4 : 1 67-91 heritance of mitochondrial DNA in sex­
93. Lane , D. J . , Pace, B . , Olsen, G. J . , ual cros ses of Pythium sylvaticum . Curro
Stahl, D . A . , Sogin, M . L . , Pace, N . R . Genet. 1 6 :373-74
1 985. Rapid determination o f 1 6S ribo­ 1 05b. Martin, F. N. 1 989. Taxonomic classifi­
somal RNA sequences for phylogenetic cation of asexual isolates of Pythium 111-
analyses. Proc. Natl. Acad. Sci. USA timum based on cultural characteristics
82:6955-59 and mitochondrial DNA restriction pat­
94. Lee, S. B . , Taylor, J. W. 199 1 . Phy­ terns. Exp. Mycol. 14:47-56
logeny of five Phytophthora species as 1 06. Martin, F. N. 1 99 1 . Variation in the
determined by ribosomal DNA internal ribosomal DNA repeat unit within single
transcribed spacer sequence comparison oospore isolates of the genus Pythium .
Mol. Bioi. Evol. In press Genome 33:585-91
95. Lee, S. B . , Taylor, J. W. 1 99 1 . Unipa­ 1 07 . May, G . , Taylor, J. W . 1 989. In­
rental inheritance and replacement of dependent transfer of mitochondrial
mitochondrial DNA in Neurospora tet­ plasmids in Neurospora crassa . Nature
rasperma . Genetics . Submitted 339:320--2 2
96 . Lee, S. B . , White, T. J . , Taylor, J. W . 108. McDonald, B . A., Martinez, J. P. 1 990.
1 99 1 . Detection o f species-specific DNA restriction fragment length poly­
DNA variability in Phytophthora by oli­ morphisms among Mycosphaerella gra­
gonucleotide hybridization to amplified minicola (Anamorph Septoria tririci)
ribosomal DNA. Phytopathology . Sub­ isolates collected from a single wheat
mitted field. Phytopathology 80: 1 368-73
97. LoBuglio, K. F . , Rogers, S. 0 . , Wang, 109. McDonald, B. A . , Martinez, J. P. 1 99 1 .
C. J. K. 1 990. Variation in ribosomal DNA fingerprinting of the plant
DNA among isolates of the mycorrhizal pathogenetic fungus Mycosphaerella
fungus Cenoccum geophilum Fr. See graminicola (Anamorph Septora triti­
Ref. 18, p. 330 cal. Exp. Mycol. 1 5 : 1 46-58
 562 BRUNS, WHITE & TAYLOR

1 1 0. McNabb, S. A. , Klassen, G. R. 1 988. Geological h istory of the Glomaceat:

UnifOlwity of mitochondrial DNA com­ with particular reference to mycorrhizal
plexity in Oomycetes and the evolution symbiosis . Symbiosis 7:1-36
of the inverted repeat. Exp . Mycol. 1 23 . Poinar, G. 0 . , Singer, R . 1990. Upper
1 2:233-42 Eocene gilled mushroom from the Do­
1 10a. Meyer, W . , Kock, A . , Beyermann, B . , minican Republic. Science 248: 1 099-­
Epplen, J. T . , Bomer, ? 1 99 1 . Differ­ nOI
entiation of species and strains among 1 24. Prager, E. M . , Wilson, A . C . 1 98 8 .
filamentous fungi by DNA printing. Ancient origin of lactalbumin from
Curro Genet. 1 9:239-42 lysozyme: analysis of DNA and amino
1 1 1 . Miche lmore, R. W . , Hulbert, S. H . acid sequences. 1. Mol. Evol. 27:326--3 5
1 98 7 . Molecular markers for genetic 1 25 . Price, C . W . , Fuson, G. B., Phaff, H. J .
analY5.is of phytopathological fungi. 1 97 8 . Genome comparison in yeast sys­
Annu . Rev. Phytopathol. 25:383-404 tematics: delimitation of species within
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

1 1 2 . Mills, D . , McCluskey, K. 1 990. the genera Schwanniomyces, Sacchar�

Electrophoretic k aryotypes of fungi : the omyces, D eba ryomyces , and Pichia.
new cytology. Mol. Pl. Microb. lnteract Microbiol. Rev. 42: 1 6 1 -9 3 .
3 : 3 5 1 ·-57 126. Rehner, S. A . , Bruns, T. D . , Vilga1Y5.,
by University of Winnipeg on 09/11/14. For personal use only.

1 1 3 . Mills, S. D . , Forster, H . , Coffey, M. D. R. 1 99Q. Mole<.:ular systematic in­

1 99 1 . Taxonomic structure of Phytoph­ vestigations in the Agaricales. See Ref.
thora cryptogea and P . drechsleri based 1 6 , p. 339 (Abstr) .
on isozyme and mitochondrial DNA 1 27 . Reynold�, D. R . , Taylor, J. W. 1 9 9 1 .
analysis. Mycol. Res. 95:3 1 -48 DNA specimens and the international
1 1 4 . Mitch ell, L. G . , Merril, C. R . , 1 989. code of botanical nomen c lature . Taxon
Affinity generation of single-stranded 40:3 1 1- 1 5
DNA for dideoxy sequencing following 1 2 8 . Reynolds, D . R . , Taylor, J . W . 1 992.
the polymerase chain reaction. Anal. Article 59: Reinterpretation or revision?
Biochem. 1 78:239-42 Taxon 42: In press
l I S . Mitchell, T. G . , White, T. 1 . , Tayl or 1 .
, 1 29 . Rippon , 1. W. 1 98 8 . Medical Mycology .
W . 1 99 1 . Comparison o f ribosomal Philadelphia: W. B . Saunders . 3rd. ed.
DNA sequences among the basidio­ 1 30. Rogers, S . 0. , Ammirati. J . F. , LoBu g ­
mycetous yeast genera, Cystofilobasi­ lio, K . , Rehner, S . A . , Bledsoe, C .
dium, Filobasidium and Filobasidiella. 1 990. Molecular taxonomy and phy­
1. Bacterial. Submitted logenetics of Basidiomycetes. (Abstr.)
1 1 6. Mullis, K. B . , Faloona, F. A. 1 987. See Ref. 1 8 , p. 339
Specific synthesis of DNA in vitro via a 1 3 1 . Rogers, S. 0 . , Honda, S . , Bendich, A .
polymerase-catalysed chain reaction. J . 1 986. Variation in the ribosomal R N A
Methods Enzymol. 1 55 : 335-50 genes among individuals of Vicia faba.
1 17 . Natvig, D . O . , Jackson , D. A . , Taylor, Plant Mol. Bioi. 6: 33 9-45
J. W. 1987. Random-fragment hybrid­ 1 32 . Russell, P. 1 . , Wagner, S . , Rodland, K .
ization analysis of evolution in the genus D . , Feinbaum, R . L . , Russel, J . P . , et
Neurospora: the status of four-spored al. 1 984. Organization of the ribosomal
strains. Evolution 4 1 : 1 003-2 1 ribonucleic acid genes in various wild­
1 1 8 . Nei, M . , Li, W-H. 1 979 . Mathematical type strains and wild-collected strains of
modd for studying genetic variation in Neurospora . Mol. Gen. Genet. 1 96:
terms of restriction endonuc\eases. 275-82
Proc. Natl. Acad. Sci. USA 76:5269-73 1 33 . Saiki, R. K. , Gelfand, D. H . , Stoffd,
1 1 9 . Nelsnn, R. T . , Yangcn, B . G . , Te S . , Scharf, S. J . , H i guc hi, R. G . , Horn,
Strake , D . , Cochrane B . J . 1 990. Genet­ G. T., Mullis, K . B . , Erlich, H. A.
ic studies in the genus Basidiobolus. II. 1 98 8 . Primer-directed enzymatic ampli­
Phylogenetic relationships inferred from fication of DNA with a thermostable
ribosomal DNA analysis. Exp. Mycol. DNA polymerase. Science 239:487-9 1
1 4: 1 97-206 1 34. Saitou , N . , Imanishi, T. 1 989. Relative
1 20 . Perkins, D. D . , Turner, B . C. 1988. efficiencies of the Fitch-Margoliash.
Neurospora from natural popUlations: maximum- parsimony, maximum-likeli­
toward the population biology of a hap� hood, minimum-evolution, and
loid eukaryote. Exp . Mycol. 1 2 : 9 1 - \ 3 1 neighbor-joining methods of phyloge­
1 2 1 . Phelps, L . G . , Burke, J . M . , Ullrich, R. netic tree construction in obtaining the
c., Novotny, C. P. 1 9 8 8 . Nucleotide correct tree. Mol. B ioi. Evol. 6:5 1 4-25
base sequence of the mitochondrial 1 3 5 . Saitou, N . , Nei, M . 1 987. The
com gene of Schizophyllum commune. neighbor-jnining methnd: a new method
Curr . Genet. 1 4 :401-403 for reconstructing phylogenetic trees.
1 22 . Pirozynski, K. A . , Dalpe, Y. 1 989 . Mol. BioI. Evol. 4:406-- 2 5
 FUNGAL MOLECULAR SYSTEMATICS 563

1 36. Samac, D. A . , Leong, S. A. 1 989. subunit ribosomal RNAs. Am. Zool.

Mitochondrial plasmids of filamentous 29:487-99
fungi: characteristics and use in 1 48 . Steele, 1<. P. , Holsinger, K. E . , Jansen,
transformation vectors . Mol. Plant­ R. K. , Taylor, D . W . 1 99 1 . Assess ing
Microbe Interact. 2 : 1 55-59 the reliability of 5S rRNA sequence data
1 37 . Sanders, J. P. M . , Heyting, c . , Verbett, for phylogenetic analysis in green
M. P . , Meijlink, F. C. P. W . , Borst, P. plants. Mol. Bioi. Evol. 8:240--48
1 977. The organization of genes in yeast 149. Stuart, K. , Feagin, 1. E. , Abraham, J .
mitochondrial DNA. III. Comparison of M . 1 989. RNA editing: the creation of
the physical maps of the mitochondrial nucleotide sequences in messenger
DNAs from three wild-type Sacchar­ RNA-a minireview. Gene 82( 1 ) : 1 55-
omyces strains. Mol. Gen . Genet. 1 57: 60
239-61 150. Swofford, D . L . , Olsen, G. J. 1 990.
1 38 . Scharf, S. J . , Horn, G. T . , Erlich, H . Phylogeny reconstruction. In Molecular
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

A . 1 986. Direct cloning and sequence Systematics, ed. D. M. Hillis, C .

analysis of enzymatically amplified Moritz, p p . 4 1 1-50 1 . Sunderland,
genomic sequences. Science 233: 1076- Mass: Sinauer
78 1 51. Taylor, 1. W . , Natvig, D. O. 1 989.
by University of Winnipeg on 09/11/14. For personal use only.

1 39. Scherer, S . , Stevens, D . A . 1 9!!!!. A Mitochondrial DNA and evolution of

Candida albicans dispersed, repeated heterothallic and pseudohomothallic
gene family and its epidemiologic ap­ Neurospora species. Mycal. Res. 93:
plications. Proc. Nat! . Acad. Sci. USA 257-72
8 5 : 1 425-56 152. Taylor, J . W . , Smolich, B . D . , May , G .
1 40. Selker, E. U . , Stevens, J. N . , Metzen­ 1 986. Evolution and mitochondrial
berg, R. L. 1 98 5. Heterogeneity of 5S DNA in Neurospora crassa . Evolution
RNA in fungal ribosomes. Science 40:7 1 6-39
227: 1 340--4 3 1 5 3 . Taylor, J. W . , White , T. J. 1 99 1 .
141. Shumard-Hudspeth, D . S . , Hudspeth, Molecular evolution of nuclear and
M. E. S. 1 990 . Genic rearrangements in mitochondrial ribosomal DNA region in
Phytopluhora mitochondrial DNA. Neurospora and othcr Sordariaccae and
Curro Genet. 1 7 :4 1 3- 1 5 comparison with the mushroom, Laccar·
1 42. Sidow, A . , Wilson, A . C. 1 990. Com­ ia FUllg. Genet. Newsl. 38:26
positional statistics: An improvement of 154. Taylor, T. N. 1 990. Fungal associations
evolutionary parsimony and its applica­ in the terrestrial paleoecosystem. TREE
tion to deep branches in the tree of life. 5:2 1-25
J. Mol. Evol. 3 1 :5 1-6!! 1 5 5 . Turner, B. J . , Elders, J. F . , Laughlin,
1 43a. Simon, C. 1 99 1 . Molecular systemat­ T. F. , Davis, W. P. 1 990. Genetic varia­
ics at the species boundary: exploiting tion in clonal vertebrates detected by
conserved and variable regions of the simple-sequence DNA fingerprinting.
mitochondrial genome of animals via di­ Proc. Natl. Acad. Sci. USA 87:5653-
rect sequencing from amplified DNA. In 57
Molecular Taxonomv, ' ed. G. M. 1 56. Vilgalys, R . , Hester, M. 1 990. Rapid
Hewitt, A. W . B . Johnston. Berlin: genetic identification and mapping of
Springer Verlag. In press enzymatically amplified ribosomal DNA
1 44. Smith, M. L. , Anderson, J. B. 1 989. from several Cryptococcus species. J.
Restriction fragment length polymor­ Bacteriol. 1 72(8):4238-46
phisms in mitochondrial DNAs of Armil­ 1 5 7 . Vilgalys, R. J . , Johnson, J . L. 1 987.
laria: identification of North American Extensive genetic divergence associated
biological species. Mycol. Res. 93:247- with speciation in filamentous fungi.
56 Proc. Natl. Acad. Sci. USA 84:2355-58
1 45 . Smith, M . L., Duchesne, L. c., Bruhn, 1 57a. Vilgalys, R . , Gonzalez , D . 1 990.
J. N . , Anderson, J. B . 1 990. Mito­ Ribosomal DNA restriction fragment
chondrial genetics in a natural popula­ length polymorphisms in Rhizoctonia
tion of the plant pathogen Armillaria . solalli. Phytopathology 80(2): 1 5 1-62
Genetics 1 26:575-82 1 5 8 . Walker, W. F . , Doolittle, W. F. 1 982.
1 46. Smith, T. L . 1 989. Disparate evolution Redividing the basidiomycetes on the
of yeasts and filamentous fungi indicated basis of 5S rRNA sequences. Nature
by phylogenetic analysis of glyceralde­ 299:723-24
hyde-3-phosphate dehydrogenase genes. 1 59. Ward, D . M . , Weller, R . , Bateson , M .
Proc. Natl. Acad. Sci. USA 86: M . 1 990. 1 6S rRNA sequences reveal
7063-66 numerous uncultured microorganisms in
147. Sogin, M. L. 1 989. Evolution of eu­ a natural community. Nature 345 :63-65
karyotic microorganisms and their small 1 60 . White, T. J . , Bruns, T. D . , Lee, S . ,
 564 BRUNS, WHITE & TAYLOR

Tay lor , J . W. 1990. Amplification and 1 65 . Wolf, K . , Del Giudice L. 1 98 8 . The

direct sequencing of fungal rihosomal variable mitochondrial genome of
RNA genes for phylogenetics. See Ref. Ascomycetes: organization, mutational
49, 38:3 1 5-22 alterations , and expression. Adv. Genet.
161. White, T. J . , Madej , R . , Persing , D. H. 25 : 1 85-308
1 99 1 . The polymerase chain reaction: 1 66 . Wu, M. M. J . , Cassidy, J. R . , Pukkila,
clinical applications. In Advances in P. J. 1 98 3 . Polymorphisms i n DNA of
Clinical Chemistry 29, ed. H. Spiegel. Coprinus cinereus. Curro Genet. 7 : 3 85-
New York: Academic Press. In press 92
1 62. Williams, J . G. K . , Kubelik, A. R . , 1 67 . Yamada, Y. , Kawasaki, H. , N akas e,
Livak, K . J . , Rafalski, J . A . , Tingey, S . T., Banno, I . 1 989. The phylogenetic
V . 1990. DNA polymorphisms ampli­ relationship of the conidium-fonning
fied by arbitrary primers are useful as anamorphic yeast genera Sterigmatomy­
genetic markers . Nucl. Acids Res. 1 8 : ces, Kurtzmanomyces, Tsuchiyaea and
Annu. Rev. Ecol. Syst. 1991.22:525-564. Downloaded from www.annualreviews.org

653 1--35 Fellomyces, and the teleomorphic yeast

1 63 . W i l s on , A . c., Cann, R. L . , Carr, S . genus Sterigmatosporidium on the basis
M . , George, M . . Gyllensten, U . B . , I-Iel of the partial sequences of 1 8S and 26S
m-Bychowski, K. M. Higuchi, R. G . , ribosomal ribonucleic acids. Agric. BioI.
by University of Winnipeg on 09/11/14. For personal use only.

Palurnbi , S . R . , Prager, E . M . , Sage. R . Chem. 53:2993-300 1

D . , Stoneking, M . 1 98 5 . M itochondrial 1 6 8 . Yamada, Y . , Nakagawa, Y . , Banno, I .
DNA and two perspectives on evolution­ 1 989. The phylogenetic relationship of
ary genetics . BioI. 1. Linnean Soc. Q9-equipped species of the hetero­
26:375-400 basidiomycetous yeast genera Rhodos­
164, Woese, C. R . , Kandler, 0 . , Wheelis, poridium and Leucosporidium based on
M. L. 1 990. Towards a natural system the partial sequences of 1 8S and 26S
of organisms: Propos al for the domains ribosomal ribonucleic acids: the pro­
Archaea, Bacteria, and Eucarya. Proc. posal of a new genus Kondoa . J. Gen.
Natl. Acad. Sci, USA 87:4576-79 Appl. Microbiol. 35:377-85