J Mol Evol (2005) 60:726–735
DOI: 10.1007/s00239-004-0164-y
Fungi Evolution Revisited: Application of the Penalized Likelihood Method to a
Bayesian Fungal Phylogeny Provides a New Perspective on Phylogenetic
Relationships and Divergence Dates of Ascomycota Groups
Ana Carolina B. Padovan, Gerdine F.O. Sanson, Adriana Brunstein, Marcelo R.S. Briones
Departamento de Microbiologia, Imunologia e Parasitologia, Universidade Federal de São Paulo, São Paulo, Brazil
Received: 26 May 2004 / Accepted: 15 December 2004 [Reviewing Editor: Dr. Nicolas Galtier]
Abstract. The depiction of evolutionary relationships within phylum Ascomycota is still controversial
because of unresolved branching orders in the radiation of major taxa. Here we generated a dataset of
166 small subunit (18S) rDNA sequences, representative of all groups of Fungi and used as input in a
Bayesian phylogenetic analysis. This phylogeny suggests that Discomycetes are a basal group of filamentous Ascomycetes and probably maintain
ancestor characters since their representatives are
intermingled among other filamentous fungi. Also,
we show that the evolutionary rate heterogeneity
within Ascomycota precludes the assumption of a
global molecular clock. Accordingly, we used the
penalized likelihood method, and for calibration we
included a 400 million-year-old Pyrenomycete fossil
considering two distinct scenarios found in the literature, one with an estimated date of 1576 Myr for the
plant–animal–fungus split and the other with an
estimated date of 965 Myr for the animal–fungus
split. Our data show that the current classification of
the fossil as a Pyrenomycete is not compatible with
the second scenario. Estimates under the first scenario are older than dates proposed in previous
studies based on small subunit rDNA sequences but
support estimates based on multiprotein analysis,
suggesting that the radiation of the major Ascomycota groups occurred into the Proterozoic era.
Correspondence to: Adriana Brunstein; email: adriana@ecb.epm.br
Key words: Ascomycota evolution — Bayesian
inference — Penalized likelihood — Divergence dateestimation — Fossil constraint
Introduction
Molecular sequences have been used to infer evolutionary relationships among the main groups of Ascomycota phyla, namely, filamentous Ascomycetes or
Euascomycetes, Saccharomycetales, and Archiascomycetes (Bruns et al. 1992; Berbee and Taylor 1993;
Gargas and Taylor 1995; Kurtzman 2000; Lumbsch
2000; Redecker et al. 2000; Berbee 2001). These studies
propose that morphological characters are often
insufficient to infer the evolutionary history of this
phylum because morphological similarity is in most
cases caused by convergence (Berbee and Taylor 1993).
Phylogenies based on molecular characters have been
useful in providing clues to the location of taxa with
uncertain taxonomic position (e.g., mitosporic fungi).
The most used marker is the nuclear small subunit
(SSU) rDNA (or 18S rDNA) (Bruns et al. 1992; Berbee
and Taylor 1993; Kurtzman 2000), although a few
studies have explored other genomic regions or protein
sequences (Liu et al. 1999; Heckman et al. 2001;
Hedges et al. 2004). The concerted evolution of SSU
rDNA copies within genomes precludes paralogy for
this gene and makes it a useful tool for phylogenetic
inference and subsequent molecular dating. These
points cannot be a priori guaranteed for other genes.
727
Several traditional taxa such as Pyrenomycetes,
Plectomycetes, Archiascomycetes, and Saccharomycetales have been confirmed by studies based on
molecular data (Berbee and Taylor 1992; Gargas and
Taylor 1995; Berbee et al. 2000; Lumbsch 2000;
Berbee 2001). In other cases, like Discomycetes and
Loculomycetes, it is not clear whether molecular
characters are insufficient or if these groups do not
represent a natural assemblage of organisms (Gargas
and Taylor 1995; Berbee 1996). In addition, there are
two proposed scenarios concerning the systematics of
Euascomycetes, one favoring the basal position of
Discomycetes (Gargas and Taylor 1995; Lutzoni
et al. 2001) and the other favoring the basal position
of Pyrenomycetes (Berbee and Taylor 2001). Consequently, the Euascomycetes radiation is depicted as a
polytomy due to the lack of phylogenetic stability
obtained in the previously employed methodologies.
The earliest fungal fossils are from the Ordovician
and correspond to probable ancestors of Zygomycota
or Chytridiomycota (Redecker et al. 2000). These
fossils have been useful to calibrate divergence date
estimates of Fungi based on phylogenetic trees (Berbee and Taylor 2001). Unfortunately, there are few
fossil representatives of derivate Ascomycota groups.
Also, some groups of Fungi cannot be distinguished
from other organisms in the fossil records (Alexopoulos et al. 1996; Redecker 2002). Evidence of Basidiomycota is assigned to a 290-Myr fossil with
clamp connections (Dennis 1970). The oldest fossil
that resembles an extant derivate Ascomycota group
is 400 Myr old and presents characteristics of extant
Pyrenomycetes (Taylor et al. 1999). Given this scarcity of fossil evidence, molecular evolution studies
have been useful to estimate divergence times between fungi (Berbee and Taylor 1993, 2001; Heckman
et al. 2001). Berbee and Taylor (1993) first estimated
the divergence date between Ascomycota and Basidiomycota to be 390 Myr and the divergence date
between Euascomycetes and true yeasts (Saccharomycetales order) 310 Myr by analysis of SSU rRNA
sequences. In a posterior refinement these authors
included more sequences and adjusted the fossil calibrations to obtain respective estimates of 500 and
370 Myr (Berbee and Taylor 2001). Using a different
approach, which averaged distances among protein
sequences, Heckman et al. (2001) estimated the same
divergence dates to be 1200 and 1000 Myr respectively, and Hedges et al. (2004), using protein sequence data and molecular clock methods, estimated
these same dates to be 968 and 982 Myr, respectively,
placing the main fungus radiation events into the
Proterozoic era.
Because of the disparity presented by these divergence dates and the lack of support within Ascomycota, we analyzed SSU rDNA previously published
sequences using Bayesian phylogeny reconstruction
methods. We estimated divergence dates among the
major phylogenetic groups of Fungi using the 400Myr-old Rhynie chert Ascomycota fossil as a first
calibration point and considered two different scenarios in order to define a second calibration point.
In the first scenario we used the plant–animal–fungus
divergence time (1576 Myr) estimated by Wang et al.
(1999). Although Wang et al. (1999) used a secondary
calibration point (Shaul and Graur 2002) and Hedges
et al. (2004) used an approach based on concatenated
datasets that passed the consistency test (Shaul and
Graur 2002), no significantly different dates for the
split were observed. In the second scenario we used
the animal–fungus divergence time (965 Myr) estimated by Doolittle et al. (1996), which was also used
for date estimates of Berbee and Taylor (2001).
There were included in the analysis 169 taxa, of
which 166 are representative of all groups of Fungi, 2
are representatives of animal and plants, and 1 is the
outgroup. Due to the large number of taxa a Bayesian approach was employed since it possesses
advantages over other methods in terms of ability of
using complex models of evolution, ease of interpretation of the results, and computational efficiency
(Huelsenbeck et al. 2002).
Nucleotide substitution rate constancy in fungal
SSU rDNA evolution was not supported by the data,
even in small groups such as Plectomycetes (Berbee
and Taylor 1993; Kasuga et al. 2002). Because of
this evolutionary rate heterogeneity we used the
penalized likelihood method (Sanderson 2002), a
semiparametric approach that combines likelihood
(parametric) and the nonparametric penalty function (Sanderson 1997), implemented in r8s (http://
ginger.ucdavis.edu/r8s). The method relaxes the
stringency of a clock assumption and permits the
addition of calibration points (fossil records and time
constraints), which allowed us to estimate divergence
dates of Fungal radiation events without the imposition of a rate constant molecular clock.
Materials and Methods
A criterion of fungal taxa selection based on Alexopoulos et al.
(1996) was chosen. Our goal was to get at least one representative
of each Ascomycota group and some representatives of Basidiomycota, Chytridiomycota, and Zygomicota.
We used 276 complete sequences of SSU rDNA downloaded
from GenBank (http://ncbi.nlm.nih.gov/Genbank) that were
aligned using Clustal X (Jeanmougin et al. 1998) and adjusted by
eye using Seaview (Galtier et al. 1996). In order to recover the plant–
animal–fungus split there included the sequences of Clathrina
cerebrum (U42452) and Sphagnum cuspidatum (X80213), which
represent, respectively, phylum Porifera and phylum Streptophyta.
A stramenopile (Developayella elegans; (U37107)) was used as
outgroup. This outgroup was chosen due to its unambiguous position in the ribosomal tree of life (Patterson 1989; Leipe et al. 1994).
Some sequences that presented high levels of gaps and/or ambiguities were excluded from the analysis. After this trial process we have
728
obtained an alignment of 169 taxa with 1474 bp (gaps excluded) and
the fungal sequences are described in Table 1 according to their
taxonomic classification and respective accession numbers.
An unconstrained consensus phylogeny was inferred with Mr.
Bayes. A general time-reversible model of DNA substitution
(Rodrı́guez et al. 1990) with a c distribution to account for rate
variation among sites (Yang 1994) was used as input. As we had no
knowledge of the priors, uninformative ones were associated with
branch lengths and substitution parameters of the rate matrix and
all trees were considered equally probable a priori (Huelsenbeck
and Ronquist 2001).
As testing the molecular clock hypothesis in a Bayesian
framework is not straightforward, subsets of the total alignment
containing only specific groups, e.g., Pyrenomycetes and Saccharomycetales, were considered and subjected to maximum likelihood
analysis using PAUP* 4.0b10 (Swofford 1998). Maximum likelihood phylogenies were then inferred with and without enforcing a
molecular clock, and a likelihood ratio test could be performed for
each subset (Felsenstein 1988). The significance level was assessed
by comparing d = 2D, where D is the difference between the likelihood scores (Ln likelihood) of the trees with a v2 distribution with
n ) 2 degrees of freedom, where n is the respective number of taxa
of each subset.
The penalized likelihood method (Sanderson 2002) as implemented in r8s (http://ginger.ucdavis.edu/r8s) was used to estimate
divergence dates of Fungi using the Bayesian phylogeny with all
169 taxa. Confidence intervals for dates were estimated reapplying
the same procedure for 100 bootstrapped matrices obtained by
resampling the data using Seqboot as implemented in PHYLIP 3.5c
(Felsenstein 1993). The Pyrenomycete fossil (400 Myr) described by
Taylor et al. (1999) and the animal–plant–fungus divergence time
estimate (1576 Myr) by Wang et al. (1999) were used as calibration
points in the first scenario as, respectively, minimal and maximal
constraints to corresponding nodes, in order to scale rates and
times to real units. In the second scenario, the Pyrenomycete fossil
and the animal–fungus divergence time estimate (965 Myr) by
Doolittle et al. (1996) were used in the same way as above as calibration points.
Results
A tree built by the neighbor-joining method (Saitou
and Nei 1987) with uncorrected distances was used to
start the Monte Carlo Markov chain in Mr. Bayes.
The chain length was 2 million and eight simultaneous chains were run and sampled every 100 generations. A consensus tree was built from the 10,000
trees corresponding to the last 1 million generations
for which the likelihood scores had converged to a
stable value. During the run the parameters of the
GTR model and the shape parameter of the gamma
distribution were estimated by Mr. Bayes with
respective 95% confidence intervals. Nucleotide frequencies were fA = 0.2166, fC = 0.2220, fG =
0.2901, and fT = 0.2713; the parameters of the rate
matrix were R(A)C) = 1.5826, R(A)G) = 3.3136,
R(A)T) = 1.6374, R(C)G) = 0.9357, R(C)T) =
4.4185, and R(G)T) = 1 and the shape parameter of
the gamma distribution was a = 0.3837.
The consensus phylogeny is shown in Fig. 1, where
groups are distinguished according to the classification scheme described elsewhere (Alexopoulos et al.
1996). The posterior probabilities (pp) of the main
Table 1. Sampled fungal taxon classification according to Alexopoulos et al. (1996) and respective accession numbers
Taxon
Ascomycota
Saccharomycetales
Arxula terrestris
Brettanomyces anomalus
Candida albicans
Candida anatomiae
Candida dubliniensis
Candida glabrata
Candida intermedia
Candida ishiwadae
Candida karawaiewii
Candida krusei
Candida lusitaniae
Candida maltosa
Candida parapsilosis
Candida populina
Candida rhagii
Candida sequanensis
Citeromyces matritensis
Debaromyces hansenii
Dekkera bruxellensis
Holleya sinecauda
Kluyveromyces delphensis
Kluyveromyces lactis
Kluyveromyces marxianus
Kluyveromyces polysporus
Kluyveromyces thermotolerans
Pichia anomala
Pichia capsulata
Pichia methanolica
Saccharomyces bayanus
Saccharomyces cerevisiae
Saccharomyces dairensis
Saccharomyces kluyveri
Saccharomyces rosinii
Saccharomyces sp.
Saccharomyces transvaalensis
Saccharomycodes ludwigii
Saccharomycopsis capsularis
Saccharomycopsis fibuligera
Williopsis saturnus
Zygosaccharomyces cidri
Zygosaccharomyces mrakii
Zygosaccharomyces rouxii
Archiascomycetes
Pneumocystis carinii
Protomyces lactucae–debilis
Protomyces macrosporus
Taphrina camea
Taphrina communis
Taphrina mirabilis
Taphrina nana
Taphrina populina
Taphrina pruni (1/2)
Taphrina robinsoniana
Taphrina virginica
Saitoella complicate
Schizosaccharomyces pombe
Discomycetes
Alectoria sarmentosa
Anamylopsora pulcherrima
Blumeria graminis sp. bromi (1/2)
Accession No.
AB000663
X83828
X53497
AB018159
X99399
X51831
X89518
AB018167
AB018168
M55528
M55526
D14593
AY055855
AB018171
AB018172
AB018173
AB018176
X62649
X58052
U53443
X83823
X51830
X89523
X69845
X89526
X58054
AB018178
AB018181
X97777
Z73326
X99527
Z75580
X99524
X99525
X99522
X69843
X69847
U10409
Y11318
X91085
X90757
X90758
L27658
D14164
D85143
AB000948
AB000949
AB000954
AB000955
D14165
AB000956/AB000957
AB000958
AB000960
D12530
X54866
AF140233
AF119501
AB033475/AB033476
(Continues)
729
Table 1. Continued.
Taxon
Blumeria graminis sp. hordei
Cladonia rangiferina
Cookeina tricholoma
Cyphelium inquinans
Cyttaria darwinii
Diploschistes rampoddensus
Diploschistes thunbergianus
Glaziella aurantiaca
Graphium rubrum
Graphis scripta
Graphium silanum
Helvetia lacunosa
Heterodea muelleri
Lasallia rossica
Lecanora disperse
Leifidium tenerum
Metus conglomeratus
Microsphaera friestii
Microstoma floccosum
Nanoscypha tetraspora
Neophyllis melacarpa
Orbilia fimicola
Otidea onotica
Peltula obscurans
Pertusaria saximontana
Peziza echinospora
Phillipsia domingensis
Pilophorus cereolus
Pithya cupressina
Placopsis gelida
Pseudopithyella minuscula
Psora decipiens
Sarcoscypha custriaca
Sclerotinia sclerotiorum
Spathularia flavida
Sphaerophorus globosus
Stereocaulon paschale
Terfezia terfezioides
Thelebolus stercoreus
Tuber gibbosum
Umula hiemalis
Verpa bohemica
Wynnella silvicola
Wynnea sp.
Loculomycetes
Aureobasidium pullulans
Botryosphaeria ribis
Capronia mansonii
Catapyrenium lachneum
Coccodinium bartschii
Dothidea insculpta
Herpotrichia diffusa
Jahnula siamensiae
Leptosphaeria doliolum
Leptosphaeria maculans
Mycosphaerella mycopappi
Myriangium duriaei
Phaeococcomyces exophialae
Pyrenophora trichostoma
Plectomycetes
Arachnomyces kanei
Arachnomyces minimus
Ascosphaera apis
Table 1. Continued.
Accession No.
AB033480
AF184753
AF006311
U86695
U53369
AF274111
AF274112
Z49753
AB007660
AF038878
AB007661
U42654
AF184754
AF088238
L37535
U70959
AF184755
AB033478
AF006313
AF006314
AF117981
AF006307
AF006308
AF282913
AF113720
AF006309
AF006315
AF184756
AF006316
AF119502
AF006317
AF184759
AF006318
X69850
230239
AF117983
AF140236
AF054900
U49936
U42663
Z49754
U42645
U42655
AF006319
M55639
U42477
X79318
AF412410
U77668
U42474
U42484
AF438180
U43447
U04233
U43449
AY016347
X80709
U43459
AF525308
AJ315167
M83264
(Continues)
Taxon
Aspergillus fumigatus
Aspergillus nomius
Blastomyces dermatitidis
Coccidioides immitis
Eremascus albus
Eurotium rubrum
Histoplasma capsulatum
Merimbla ingelheimensis
Penicillium chrysogenum
Pyrenomycetes
Amphisphaeria umbrina
Ascovaginospora stellipala
Ceratocystis fimbriata
Chaetomium elatum
Colletotrichum gloeosporiodes
Cornuvesica falcata
Cryphonectria parasitica
Halosarpheia spartinae
Hypocrea lutea
Kionochaeta spissa
Microascus cirrosus
Nais inornata
Neurospora crassa
Ophiostoma piliferum
Podospora anserina
Pseudallescheria boydii
Sordaria fimicola
Verticillium dahliae
Xylaria carpophila
Basidiomycota
Boletus satanas
Cantharellus tubaeformis
Cronartium ribicola
Cryptococcus neoformans
Cryptococcus podzolicus
Eocronartium muscicola
Exobasidium vaccinii
Lentinellus ursinus
Microbotryum violaceum
Tilletia caries
Tulostoma macrocephala
Ustilago hordei
Chytridiomycota
Allomyces macrogynus
Blastocladiella emersonii
Chytridium confervae
Spizellomyces acuminatus
Zygomycota
Mucor mucedo
Mucor racemosus
Accession No.
M60300
AB008404
AF320010
M55627
M83258
U00970
X58572
D14408
M55628
AF225207
U85087
U43777
M83257
M55640
AY271797
L42441
AF352076
D14407
AB003790
M89994
AF050482
X04971
AJ243295
X54864
U43914
X69851
U33637
Z49785
M94337
AF026636
M94338
X60183
AB032645
AY123323
AJ271380
U59081
U77062
U00972
AF026625
U00973
AMU23936
M54937
M59758
M59759
X89434
X54863
nodes of the tree, the location of the 400-Myr fossil,
and the plant–animal–fungus split are also depicted
in this figure.
The Bayesian phylogeny confirmed Plectomycetes,
Pyrenomycetes, Saccharomycetales, and Archiascomycetes as monophyletic groups as previously proposed (Berbee and Taylor 1992; Bruns et al. 1992).
Paraphyletic Loculomycetes and polyphyletic Discomycetes, as observed in our tree, are also supported
730
Fig. 1. SSU rDNA fungal phylogeny inferred by Bayesian analysis corresponding to the consensus of 10,000 trees. Values of
posterior probabilities are shown at nodes of interest. The main
groups of Fungi are depicted, namely, Chytridiomycota, Zygomycota, Basidiomycota, and Ascomycota, represented by Archiascomycetes, Saccharomycetales order (true yeasts), and the class
forms of Euascomycetes (filamentous Ascomycetes), which are
Discomycetes, Pyrenomycetes, Loculomycetes, and Plectomycetes.
In the current paper Discomycetes and Loculomycetes are subdivided. Polyphyletic Discomycetes are subdivided into Discomycetes
(1) (mostly operculate Discomycetes), Discomycetes (2) (inoperculate Discomycetes and Erysiphales order), and lichen-forming
Discomycetes. Paraphyletic Loculomycetes are subdivided into
Loculomycets (1) and Loculomycetes (2). Sequences from C.
cerebrum (Porifera) and S. cuspidatum (Streptophyta) were included to recover the 1576-Myr-old plant–animal–fungus split.
D. elegans (Stramenopile) is the outgroup. The 400-Myr fossil record that is used as a minimal age constraint to the date of Pyrenomycetes radiation is indicated by its own picture.
731
Fig. 2. Schematic phylogenies locating the estimated dates for the
main events of fungal radiation using the calibration date of 1576
Myr for the plant–animal–fungus split (a) and the calibration date
of 965 Myr for the animal–fungus split (b). Triangles extending
from numbered nodes indicate the radiation of extant phyla Chy-
tridiomycota, Zygomycota, and Basidiomycota and the main
groups of Ascomycota. Nodes are numbered from 1 to 24 and node
dates taken from Tables 2 and 3, respectively. A representation of
the Rhynie chert fossil is mapped to the corresponding 400-Myr
point.
by previous studies (Bruns et al. 1992; Gargas and
Taylor 1995; Berbee 1996).
There are three distinct groups of Discomycetes: Discomycetes (1), Discomycetes (2), and lichenforming Discomycetes. The first two groups are
monophyletic and the third is polyphyletic. The first
group is basal among Euascomycetes and contains
mostly Pezizales representatives, with the exception
of Orbilia fimicola, which belongs to Orbiliales
(Eriksson et al. 2004). The second one is a sister
group of Pyrenomycetes and most of its representatives are Erysiphales (Eriksson et al. 2004). All but
two (Leifidium tenerum and Peltula obscurans) selected lichen-forming Discomycetes form the third
732
Table 2. Divergence date estimates and 95% confidence intervals,
as calculated by the penalized likelihood method, of fungal clades
depicted in Fig. 2a
Table 3. Divergence date estimates and 95% confidence intervals,
as calculated by the penalized likelihood method, of fungal clades
depicted in Fig. 2b
Node
Estimated date (Myr)
95% CI
Node
Estimated date (Myr)
95% CI
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
803.24
376.48
533.44
816.03
653.55
677.61
922.02
871.44
928.24
1027.68
156.19
1241.99
753.40
887.66
847.59
883.19
899.23
930.37
971.86
1072.09
1147.78
1206.47
1286.61
1422.89
801.67–1078.88
312.49–517.73
473.85–715.03
1053.00–836.37
591.79–850.16
613.43–1087.60
922.62–1130.17
817.03–1092.15
770.45–1188.76
966.20–1166.62
110.98–204.24
1152.70–1406.97
733.68–1007.54
929.69–1136.51
853.13–1076.05
888.32–1105.01
900.14–1114.14
921.10–1141.47
955.22–1162.58
1051.23–1244.23
1107.74–1306.29
1164.78–1338.95
1215.94–1410.88
1269.69–1434.08
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
429.25
214.92
307.61
453.36
400.00
343.43
529.11
526.95
548.76
668.67
107.65
835.30
427.38
517.07
476.78
499.34
511.90
539.08
569.51
657.11
723.86
786.32
857.85
893.10
395.68–598.87
177.39–289.37
265.70–409.26
433.88–597.77
400.00–484.83
277.84–560.86
497.11–659.19
473.33–648.32
448.07–689.08
630.48–728.38
74.55–140.40
781.93–910.32
400.69–570.36
494.07–645.25
453.29–615.69
473.81–634.62
486.58–646.50
510.15–666.37
535.43–685.85
632.06–755.74
690.55–811.57
756.57–847.60
812.31–908.89
857.44–927.77
group, and most representatives are Lecanorales
(Eriksson et al. 2004).
Loculomycetes (1) are composed of Capronia
mansonii, Phaeococcomyces exophialae, and Catapyrenium lachneum and form a monophyletic clade with
L. tenerum (pp=1). This grouping of some Loculomycetes with Plectomycetes has already been observed (Berbee 1996).
Archiascomycetes appeared as a sister group of
other Ascomycota (pp=1), which is consistent with a
previous analysis (Alexopoulos et al. 1996), and their
monophyly is highly supported (pp=0.97).
The molecular clock hypothesis was rejected for
the trees corresponding to the considered subsets with
p values less than 0.05 (results not shown). This has
already been suggested by others (Berbee and Taylor
1993; Kasuga et al. 2002), and therefore we have
discarded the use of local molecular clocks methods
(Hasegawa et al. 1989; Cooper and Penny 1997;
Rambaut and Bromham 1998; Bromham and Hendy
2000; Yoder and Yang 2000). For large philogenies
there is a vast number of ways to assign different rates
to distinct subtrees, and we could therefore choose
groups to assign specific rates a priori. However, this
would certainly introduce bias in our analysis.
The penalized likelihood method was used to
estimate divergence dates. This method requires an
optimum smoothing value (k) that minimizes the
prediction error of the penalized likelihood function
and is given by the minimum score of a performed
cross-validation procedure (Sanderson 2002). For
our dataset, the minimum value of the cross-validation was attained for k = 12,589.58. In the first
scenario the node in which the Ascomycota fossil is
located (Fig. 1) was constrained to have a minimum
age of 400 Myr and the root of the tree (the point
corresponding to the split plant–animal–fungus in
Fig. 1) was constrained to be no older than 1576
Myr. In the second scenario the animal–fungus split
was constrained to be no older than 965 Myr. Fig. 2
shows the estimated dates of nodes for both scenarios in schematic phylogenies and the corresponding confidence intervals are shown in Tables 2
and 3.
Discussion
High support for the major groups of Ascomycota
and respective splits was obtained from a large taxa
sample and a great number of analyzed positions
through the incorporation of a full model of nucleotide substitution. We chose to exclude taxa instead
of excluding positions to maintain the gene integrity
and consider as much as possible the evolutionary
information of the molecule.
The Bayesian phylogeny supports the basal position
of the order Pezizales of Discomycetes (Discomycetes
[1] in Fig. 1) among Euascomycetes that has already
been observed (Gargas and Taylor 1995; Berbee 1996,
733
2001; Lumbsch 2000). We stress that all basal Discomycetes, excluding Orbilia fimicola, are operculate,
which could indicate that this is the ancestral form of
the ascus dehiscence. This grouping of O. fimicola
among Pezizales has already been observed, and consequently it has been purged from order Helotiales and
reclassified as Orbiliales (Eriksson et al. 2004). Other
representatives of Discomycetes that are interspersed
among Pyrenomycetes, Plectomycetes, and Loculomycetes are inoperculate or lichen-forming.
The Discomycetes radiation also suggests that they
could still maintain other characters that resemble
those of ancestral forms. As we see in Fig. 2, all lichen-forming representatives, including Discomycetes and Loculomycetes, descend from node 17, which
excludes the possibility that Pyrenomycetes, inoperculate Discomycetes, and Erysiphales descend from a
lichen-forming ancestor and contradicts the scenario
proposed by Lutzoni et al. (2001). In addition to this,
the estimated dates for node 17 and node 14, which
stand for the common ancestor of Pyrenomycetes and
Discomycetes (2), are extremely close to each other
(see Tables 2 and 3), which suggests that these extant
groups evolved in parallel.
Node 19 in Fig. 2 is the common ancestor of Euascomycetes, and although it is improbable that it
was lichen-forming, there remains the possibility that
it could make associations since the descendent
groups englobe mycorrhyzal fungi (Discomycetes [1]),
plant pathogens (Pyrenomycetes and Erysiphales),
human opportunistic pathogens (Plectomycetes and
Loculomycetes), and animal pathogens (Loculomycetes) (Alexopoulos et al. 1996; Berbee 2001).
Bayesian inference became a widely used method
of phylogenetic inference because it allows rapid
analysis of large datasets and incorporates full
models of sequence evolution. This method is
essential when considering taxa with ancient divergence dates, such as the Fungi, in which homoplasy
is very frequent. The Bayesian analysis also has the
advantage of not being restrictive to a unique best
tree. Consequently, even in the presence of alternative topologies with comparable likelihood scores,
support is assigned to specific clades by means of
posterior probabilities that represent the probability
that a clade is true given the data, the model, and
the priors (Larget and Simon 1999). Bayesian support values are an alternative to nonparametric
bootstrapping (Felsenstein 1985), although the
measures cannot be directly compared (Douady et
al. 2003) since they consider different data feature
(Alfaro et al. 2003). As pointed out by Wilcox et al.
(2002), Bayesian support values provide closer estimates of clades accuracy than those provided by
nonparametric bootstrap estimates. Besides, the last
procedure would require much more time and
computational resources.
Bayesian analysis provided strong support for the
main Euascomycetes radiation events. This could not
be accomplished in previous analyses of SSU rDNA
sequences that used other types of phylogenetic
inference methods (Gargas and Taylor 1995; Berbee
et al. 2000; Berbee and Taylor 2001). Even the
Bayesian analysis that placed Discomycetes as basal
among Euascomycetes (Lutzoni et al. 2001) did not
present enough support for subsequent splits among
others Euascomycetes groups.
Saccharomycetales form a monophyletic group
(pp=1), but families within this group do not seem to
do so. According to Kurtzman (2000) there are 11
families in Saccharomycetales. In our tree the majority
of yeasts belongs to Saccharomycetaceae and there are
few members of other families (Arxula terrestris and
Candida spp. [Candidacea], Saccharomycopsis capsularis and Saccharomycopsis fibuligera [Saccharomycopsidacea], Holleya sinecauda [Eremotheciaceae], and
Saccharomycodes ludwigii [Saccharomycodacea]). The
main splits in Saccharomycetaceae did present high
posterior probabilities (pp>0.89; not shown) even
with non-Saccharomycetaceae representatives interspersed among them. Consequently, the phylogeny
does not support the current family level taxonomic
scheme (Kurtzman 2000).
The impossibility of assuming a global molecular
clock in our data was fully justified. Therefore, the
penalized likelihood method developed by Sanderson
(2002) was more appropriate because it allows that
every lineage evolve at a different rate, applying a
penalty that prevents large variations among rates in
the phylogeny.
Fossil record data are informative of minimum
ages of nodes (Doyle and Donoghue 1993) and consequently constrain these nodes to a time interval,
providing boundaries for divergence dates (Sanderson 1997; Cutler 2000). Incorporation of these
uncertainties is a great advantage of constrained
optimization methods.
In Fig. 2a we sketch a topology to show our estimated dates that are described in Table 2. These
dates place most Ascomycota radiation events in the
Proterozoic era. Our estimated dates for splits among
major groups of Fungi could be directly compared
and corroborate estimates of Heckman et al. (2001)
and Hedges et al. (2004) that were based on multiprotein analysis. Particularly comparisons inside
Ascomycota are not meaningful since lack of Discomycetes and Loculomycetes in both analyses provided different topologies and relationships.
The schematic phylogeny shown in Fig. 2b locates
the estimated dates detailed in Table 3. Although in
this scenario the estimated dates of Ascomycota
radiation events are more recent than those estimated
in the first scenario, they do not accommodate the
dates of Berbee and Taylor (2001). Their estimates
734
were obtained under the assumption of a global
clocklike evolution, which leads to very conservative
dates that do not take into account the rate variation
among lineages observed in the SSU rDNA of Fungi.
Our analysis shows that even when the second scenario
is considered, the date for node 5, corresponding to the
fossil, is precisely 400 Myr. This opens the possibility
that the fossil classification as a Pyrenomycete is
incorrect, as previously suggested by Berbee and
Taylor (2001). Nevertheless, even with the 400-Myr
constraint, all major events of Ascomycota radiation
are systematically older than the dates estimated based
on maximum parsimony methodology used by Berbee
and Taylor (2001). Accordingly, if node 5 (in Fig. 2b)
was not constrained to a minimum of 400 Myr, more
recent date intervals would certainly have been obtained. Therefore, the acceptance of this second scenario imposes a reclassification of the fossil record.
The use of a derivate group fossil, instead of a basal
group, as a constraint allows more unbiased divergence dates estimates. This counterbalances the
uncertainty associated with the substitution rate variation considered by penalized likelihood. Our date
estimates, based on SSU rDNA, suggest that the origin
of the main groups of Fungi occurred between the
Middle and the Late Proterozoic, which is supported
by independent evidence based on multiprotein analysis (Heckman et al. 2001; Hedges et al. 2004).
Although phylogenetic relationships among Ascomycota representatives are extremely sensitive to
taxa sampling and analyzed positions, we could suggest an alternative scenario concerning ancestry, dating, and relationships among the main fungus groups.
Acknowledgments.
We thank Beatriz Schnabel for excellent
technical assistance and help with the artwork. A.C.B.P. received a
fellowship from FAPESP (Brazil). G.F.O.S. was supported by a
Howard Hughes Medical Institute Post-Doctoral Fellowship
(USA). A.B. received grants from FAPESP (Brazil) and M.R.S.B.
received grants from FAPESP and CNPq (Brazil) and the International Research Scholars Program of the Howard Hughes
Medical Institute (USA).
References
Alexopoulos CJ, Mims CW, Blackwell MM (1996) Introductory
mycology. John Wiley & Sons, New York
Alfaro ME, Zoller S, Lutzoni F (2003) Bayes or bootstrap? A
simulation study comparing the performance of Bayesian
Markov chain sampling and bootstrapping in assessing phylogenetic confidence. Mol Bio Evol 20:255–266
Berbee ML (1996) Loculoascomycete origins and evolution of filamentous ascomycetes morphology based on 18S rRNA gene
sequence data. Mol Biol Evol 13:462–470
Berbee ML (2001) The phylogeny of plant and animal pathogens in
the Ascomycota. Physiol Mol Plant Pathol 59:165–187
Berbee ML, Taylor JW (1992) Detecting morphological convergence in true fungi, using 18S rRNA gene sequence data. BioSystems 28:117–125
Berbee ML, Taylor JW (1993) Dating the evolutionary radiations
of true fungi. Can J Bot 71:1114–1127
Berbee ML, Taylor JW (2001) Fungal molecular evolution: gene
trees and geologic time. In: McLaughlin D, McLaughlin E,
Lemke P (eds) The Mycota. Springer-Verlag, Berlin, pp 229–245
Berbee ML, Carmean D, Winka K (2000) Ribosomal DNA and
resolution of branching order among the Ascomycota: How
many nucleotides are enough? Mol Phylogenet Evol 17:337–344
Bromham L, Hendy MD (2000) Can fast early rates reconcile
molecular dates with the Cambrian explosion? Proc Roy Soc
Lond Ser B Biol Sci 267:1041–1047
Bruns TD, Vilgalys R, Barns SM, Gonzales D, Hibbett DS, Lane
DJ, Simon L, Stickel S, Szaro TM, Weisburg WG, Sogin ML
(1992) Evolutionary relationships within the Fungi: Analyses of
nuclear small subunit rRNA sequences. Mol Phylogenet Evol
1:231–241
Cooper A, Penny D (1997) Mass survival of birds across the Cretaceous–Tertiary boundary: Molecular evidence. Science
275:1109–1113
Cutler DJ (2000) Estimating divergence times in the presence of an
overdispersed molecular clock. Mol Biol Evol 17:1647–1660
Dennis R (1970) A middle Pennsylvanian Basidiomycete mycelium
with clamp connections. Mycologia 62:578–584
Doolittle RF, Feng DF, tsang S, Cho G, Little E (1996) Determining divergence times of the major kingdoms of living
organisms with a protein clock. Science 271:470–477
Douady C, Delsuc F, Boucher Y, Ford Doolittle W, Douzery E
(2003) Comparison of Bayesian and maximum likelihood
bootstrap measures of phylogentic reliability. Mol Biol Evol
20:248–254
Doyle J, Donoghue M (1993) Phylogenies and angiosperm diversification. Paleobiology 19:141–167
Eriksson OE, Baral HO, Currah RS, Hansen K, Kurtzman CP,
Rambold G, Laessoe T (2004) Outline of Ascomycota—2004.
Myconet 10:1–89
Felsenstein J (1985) Confidence limits on phylogenies: An approach using the bootstrap. Evolution 39:783–791
Felsenstein J (1988) Phylogenies from molecular sequences: Inference and reliability. Annu Rev Genet 22:521–565
Felsenstein J (1993) PHYLIP: Phylogeny inference package. Seattle, WA
Galtier N, Guoy M, Gautier C (1996) SeaView and Phylo_win, two
graphic tools for sequence alignment and molecular phylogeny.
Comput Appl Biosci 12:543–548
Gargas A, Taylor JW (1995) Phylogeny of Discomycetes and early
radiations of the apothecial ascomycotina inferred from SSu
rDNA sequence data. Exp Mycol 19:7–15
Hasegawa M, Kishino H, Yano T (1989) Estimation of branching
dates among primates by molecular clocks of nuclear DNA
which slowed down in hominoidea. J Hum Evol 18:461–476
Heckman DS, Geiser DM, Eidell BR, Stauffer RL, Kardos NL,
Hedges SB (2001) Molecular evidence for the early colonization
of land by fungi and plants. Science 293:1129–1133
Hedges SB, Blair JE, Venturi ML, Shoe JL (2004) A molecular time
scale of eukaryote evolution and the rise of complex multicellular life. BMC Evol Biol 4:2
Hibbet DS, Grimaldi D, Donoghue MJ (1995) Cretaceous mushrooms in amber. Nature 377:487
Huelsenbeck J, Ronquist F (2001) MRBAYES: Bayesian inference
of phylogenetic trees. Bioinformatics 17:754–755
Huelsenbeck J, Larget B, Miller R, Ronquist F (2002) Potential
applications and pitfalls of Bayesian inference of phylogeny.
Syst Biol 51:673–688
735
Jeanmougin F, Thompson J, Gouy N, Higgins D, Gibson T (1998)
Multiple sequence alignment with Clustal X. Trends Biochem
Sci 23:403–405
Kasuga T, White TJ, Taylor JW (2002) Estimation of nucleotide
substitution rates in Eurotiomycete Fungi. Mol Biol Evol
19:2318–2324
Kurtzman CP (2000) Systematics and taxonomy of yeasts. In:
Ernst J, Schmidt A (eds) Dimorphism in human pathogenic and
apathogenic yeasts. Karger, Basel, pp 1–14
Larget B, Simon D (1999) Markov chain Monte Carlo algorithms
for the Bayesian analysis of phylogenetic trees. Mol Biol Evol
16:750–759
Leipe DD, Wainright PO, Gunderson JH, Porter D, Patterson DJ,
Valois F, Himmerich S, Sogin ML (1994) The stramenopiles from
a molecular perspective: 16S-like rRNA sequences from Labyrinthuloides minuta and Cafeteria roenbergensis. Phycologia
33:369–377
Liu YJ, Whelen S, Hall BD (1999) Phylogenetic relationships
among ascomycetes: Evidence from an RNA polymerase II
subunit. Mol Biol Evol 16:1799–1808
Lumbsch H (2000) Phylogeny of filamentous ascomycetes. Naturwissenschaften 87:335–342
Lutzoni F, Pagel M, Reeb V (2001) Major fungal lineages are derived from lichen symbiotic ancestors. Nature 411:937–940
Patterson DJ (1989) Stramenopiles: Chromophytes from a protistan perspective. In: Green JC, Leadbeater BSC, Diver WL (eds)
The chromophyte algae problems and perspectives. Clarendon
Press, Oxford, pp 357–379
Rambaut A, Bromham L (1998) Estimating divergence dates from
molecular sequences. Mol Biol Evol 15:442–448
Redecker D (2002) New views on fungal evolution based on DNA
markers and the fossil record. Res Microbiol 153:125–130
Redecker D, Kodner R, Graham LE (2000) Glomalean fungi from
the Ordovician. Science 289:1920–1921
Rodrı́guez F, Oliver JL, Marı́n A, Medina JR (1990) The general
stochastic model of nucleotide substitution. J Theor Biol
142:485–501
Saitou N, Nei M (1987) The neighbor-joining method: a new
method for reconstructing phylogenetic trees. Mol Biol Evol
4:406–425
Sanderson MJ (1997) A nonparametric approach to estimating
divergence times in the absence of rate constancy. Mol Biol
Evol 14:1218–1231
Sanderson MJ (2002) Estimating absolute rates of molecular evolution and divergence times: a penalized likelihood approach.
Mol Biol Evol 19:101–109
Shaul S, Graur (2002) Playing chicken (Gallus gallus): Methodological inconsistencies of molecular divergence date estimates
due to secondary calibration points. Gene 300:59–61
Swofford DL (1998) PAUP*. Phylogenetic analysis using parsimony (*and other methods). Sinauer Associates, Sunderland,
MA
Taylor T, Hass H, Kerp H (1999) The oldest fossil ascomycetes.
Nature 399:648
Wang DYC, Kumar S, Hedges SB (1999) Divergence time estimates for the early history of animal phyla and the origin of
plants, animals and fungi. Proc Roy Soc London Ser B Biol Sci
266:163–171
Wilcox TP, Zwickl DJ, Heath TA, Hillis DM (2002) Phylogenetic
relationships of the dwarf boas and a comparison of Bayesian
and bootstrap measures of phylogenetic support. Mol Phylogenet Evol 25:361–371
Yang Z (1994) Maximum likelihood phylogenetic estimation from
DNA sequences with variable rates over sites: approximate
methods. J Mol Evol 39:306–314
Yoder AD, Yang Z (2000) Estimation of primate speciation
dates using local molecular clocks. Mol Biol Evol 17:1081–
1090