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