1 Mycologia, 96(2), 2004, pp by The Mycological Society of America, Lawrence, KS Phylogenetic relationships of Gomphillaceae and Asterothyriaceae: evidence from a combined Bayesian analysis of nuclear and mitochondrial sequences Robert Lücking 1 Department of Botany, Field Museum of Natural History, 1400 S. Lake Shore Drive, Chicago, Illinois Bryan L. Stuart Department of Zoology, Field Museum of Natural History, 1400 S. Lake Shore Drive, Chicago, Illinois H. Thorsten Lumbsch Department of Botany, Field Museum of Natural History, 1400 S. Lake Shore Drive, Chicago, Illinois Abstract: The phylogeny and systematic position of Gomphillaceae was reconstructed using a combined Bayesian analysis of nuclear LSU rdna and mitochondrial SSU rdna sequences. Twenty-four partial sequences of 12 taxa (11 Gomphillaceae and one Asterothyriaceae) plus two new sequences of Stictis radiata (Ostropales outgroup) were generated and aligned with the corresponding sequences retrieved from GenBank, resulting in an alignment of 82 taxa that was analyzed using a Bayesian approach with Markov chain Monte Carlo (B/MCMC) methods. Our results confirm Gomphillaceae sensu Vezda and Poelt plus Asterothyriaceae to be a monophyletic group, with an unresolved relationship between the two families. Placement of Gomphillaceae and Asterothyriaceae within Ostropales sensu Kauff and Lutzoni, as sister of Thelotremataceae, also is strongly supported. Alternative hypotheses placing Gomphillaceae in Lecanorales (Cladoniaceae), Agyriales (Baeomycetaceae) or within bitunicate Ascomycota (Arthoniomycetes, Chaetothyriomycetes, Dothideomycetes) were rejected with our dataset. After recent synonymization of Dimerella with Coenogonium (Ostropales: Coenogoniaceae), we propose the new combination Coenogonium pineti (one of our Ostropales outgroup taxa in this analysis). Key words: foliicolous lichens, Lecanoromycetes, mitochondrial small-subunit rdna, nuclear large subunit rdna, systematics Accepted for publication August 26, Corresponding author. INTRODUCTION The ascomycete families Gomphillaceae and Asterothyriaceae form a medium-size group of mainly tropical, crustose microlichens (Lücking 1999, Vezda 1979, Vezda and Poelt 1987). Numerous species grow on leaves and form an important part of the tropical diversity in the phyllosphere (Lücking 2001). The Gomphillaceae morphologically are remarkable for the presence of peculiarly shaped conidiomata that are called hyphophores, while many Asterothyriaceae are characterized by a unique thallus cortex (Henssen and Lücking 2002, Vezda 1979). Despite their importance for tropical lichen diversity and their morphological peculiarities, the phylogeny and taxonomy of Gomphillaceae and Asterothyriaceae largely has been unsettled. The Gomphillaceae consist of almost 300 species currently classified into 14 genera (Lücking 1997, Vezda and Poelt 1987). While most species grow on living leaves, taxa occurring on bark of vascular plants, over bryophytes, on soil and rock surfaces and even lichenicolous species also are known in this group (Lücking 1997, Lücking and Kalb 2002, Lücking and Sérusiaux 1998, Vezda and Poelt 1987). Members of this family are characterized by apothecioid ascomata with hemiangiocarpous development, a hamathecium consisting of thin, strongly gelatinized and richly branched and anastomosing paraphyses, and nonamyloid asci corresponding to the annelasceous type (Lücking 1997, Vezda and Poelt 1987). Apothecial morphology is variable, ranging from sessile and biatorine (e.g., Gyalideopsis, Echinoplaca) to vertically elongate (Gomphillus) or immersed-erumpent and zeorine apothecia (Calenia, Gyalectidium) (FIG. 1). The conidiomata, the socalled hyphophores, are usually stipitate and produce conidia ( diahyphae ) at their tips, but many variations of this basic scheme occur and derived types even resemble disk-shape diaspores or campylidioid conidiomata (FIG. 2). Both apothecial morphology and hyphophore type are employed for the delimitation of genera in the family. Gomphillaceae originally was based on a single, monospecific taxon, Gomphillus calycioides, an enigmatic lichen usually growing over bryophytes and characterized by elongate apothecia with very long asci and filiform ascospores (Nylander 1860, Hafell- 283
2 284 MYCOLOGIA FIG. 1. Morphological variation of apothecia in selected genera of Asterothyriaceae (A) and Gomphillaceae (B H) A. Asterothyrium longisporum (immersed-erumpent with recurved lobules). B. Gomphillus ophiosporus (vertically elongate). C. Gyalideopsis vulgaris (sessile). D. Tricharia albostrigosa (sessile). E. Tricharia longispora (shortly stipitate). F. Echinoplaca atrofusca (adnate and spot-like). G. Calenia aspidota (immersed-erumpent). H. Aulaxina submuralis (carbonized).
3 LÜCKING ET AL: PHYLOGENY OF GOMPHILLACEAE 285 FIG. 2. Morphological variation of hyphophores in selected genera of Gomphillaceae. A. Calenia monospora (stipitate with apical bunch of diahyphae visible). B. Tricharia cuneata (spatulate). C. Calenia aspidota (setiform). D. Tricharia dilatata (hand-shaped). E. Echinoplaca gemmifera (resembling disk-shaped isidia). F. Gyalideopsis hyalina (resembling campylidia of Pilocarpaceae). G. Gyalectidium filicinum (squamiform wirh horns). H. Hippocrepidea nigra (lunular).
4 286 MYCOLOGIA TABLE I. Species and specimens of lichenized and nonlichenzied Ascomycota used in the current study. Taxa for which sequences have been newly obtained are in boldface. F denotes voucher specimens deposited at the Field Museum of Natural History Species Specimen GenBank acc. no. nulsu mtssu Absconditella sphagnorum AY AY Adelolecia pilati AY AY Agonimia tristicula AY AY Ainoa mooreana AY AY Arthonia dispersa AY AY Aspergillus flavus AF AFU29214 Aspergillus nidulans AF V00653 Asterothyrium longisporum Costa Rica (Cartago: Orosi), Lücking s.n. (F, AY AY sample No. 4) Aulaxina quadrangula Costa Rica (Puntarenas: Las Cruces, Lücking AY AY s.n. (F, sample No. 66) Baeomyces placophyllus AF AY Beauveria bassiana AF AB Berlesiella nigerrima AY AY Bryophagus gloeocapsa AF AY Cainia graminis AF AF Calenia monospora Costa Rica (Cartago: Orosi), Lücking s.n. (F, AY AY sample No. 1a) Calenia phyllogena Costa Rica (Heredia: La Selva), Lücking s.n. AY AY (F, sample No. 32) Calicium viride AF AY Caloplaca flavorubescens AY AY Capnodium citri AY AF Capronia mansonii AY AF Cephalotheca sulfurea AF AF Ceramothyrium carniolicum AY AF Cladonia rangiferina AY AY Coenogonium pineti AY AY Combea mollusca AY AY Dendrographa minor AY AY Diploschistes cinereocaesius AY AY Diploschistes muscorum AY AY Diploschistes rampoddensis AF AF Diploschistes thunbergianus AF AF Dothidea ribesia AY AY Echinoplaca diffluens Mexico (Veracruz: Lost Tuxtlas), Herrera et AY AY al s.n. (F, sample No. M12) Echinoplaca epiphylla Mexico (Veracruz: Lost Tuxtlas), Herrera et AY AY al s.n. (F, sample No. M13) Echinoplaca leucotrichoides Costa Rica (Heredia: La Selva), Lücking s.n. AY AY (F, sample No. 18) Echinoplaca lucernifera Costa Rica (Puntarenas: Monteverde), Lücking AY AY s.n. (F, sample No. 59a) Eurotium rubrum AY AF Glyphium elatum AF AF Gomphillus ophiosporus Costa Rica (Puntarenas: Las Alturas), Will- AY AY Wolf 10006a (F, sample No. 101) Gyalecta jenensis AF AF Gyalectidium imperfectum Costa Rica (Cartago: Orosi) Lücking s.n. (F, sample No. 2) AY AY341372
5 LÜCKING ET AL: PHYLOGENY OF GOMPHILLACEAE 287 TABLE I. Continued Species Specimen GenBank acc. no. nulsu mtssu Gyalideopsis sp. nov. Costa Rica (Puntarenas: Altamira), Nelsen AY AY a (F, sample No. 103) Lecania cyrtella AY AY Lecanora intumescens AY AY Lecidella meiococca AY AY Lepraria usnica AY AY Lobaria pulmonaria AF AF Myriangium duriaei AY AY Nephroma bellum AY AY Neurospora crassa M38154 Z34001 Ochrolechia balcanica AF AF Ochrolechia parella AF AF Ochrolechia tartarea AY AY Orceolina antarctica AF AY Orceolina kerguelensis AF AF Penicillium chrysogenum AF Z23072 Pertusaria albescens AF AF Pertusaria corallina AY AY Pertusaria scaberula AF AF Pertusaria subventosa AY AY Physcia aipolia AY AY Placopsis bicolor AY AY Placopsis gelida AY AY Pyrrhospora quernea AY AY Raciborskiomyces longisetosum AY AY Ramonia sp. AY AY Schismatomma pericleum AF AY Scoliciosporum umbrinum AY AY Speerschneidera euploca AY AY Steinera glaucella AY AY Stictis radiata AY AY Stictis radiata Costa Rica (Cartago: Irazú), Will-Wolf s.n. (F, AY AY sample No. 100) Stylodothis puccinioides AY AF Thelotrema lepadinum AY AY Thelotrema suecicum AY AY Trapelia coarctata AF AY Trapelia placodioides AF AF Trapeliopsis flexuosa AF AY Trapeliopsis granulosa AF AF Tricharia longispora Costa Rica (Heredia: La Selva), Lücking s.n. AY AY (F, sample No. 37) Xanthoria parietina AF AY Xylaria hypoxylon AF AF Xylographa vitiligo AY AY ner 1984, Vezda 1979, Vezda and Poelt 1987). Because its apothecia superficially resemble podetia, Gomphillus was placed close to genera such as Cladonia and Baeomyces (Räsänen 1943, Sato 1954), although Nylander (1860), Santesson (1952) and Jahns (1970) had observed that the apothecia are vertically elongate and what was erroneously interpreted as stipe represented a part of the hymenium. Based on similarities in hamathecium structure and ascus type, Vezda (1979) suggested that Gomphillus calycioides is related to four genera previously included in Asterothyriaceae (Santesson 1952), viz. Calenia, Gyalectidium, Echinoplaca and Tricharia. This view was supported by the discovery of hyphophores in a second species of the genus, Gomphillus americanus (Vezda and Poelt 1987). The genera pro-
6 288 MYCOLOGIA ducing hyphophores and having a hamathecium composed of anastomosing paraphyses subsequently were transferred from Asterothyriaceae to the resurrected Gomphillaceae, while taxa lacking hyphophores and having unbranched paraphyses were retained in Asterothyriaceae s.str. (Eriksson and Hawksworth 1987, Vezda and Poelt 1987). The systematic positions and the homogeneity of the two families have been questioned by various authors. Hafellner (1984, 1988) interpreted the asci of Gomphillus as being fissitunicate and separated the genus into an independent order Gomphillales. Based on the ascus type, he suggested a close relationship to bitunicate ascomycetes that currently are placed in Arthoniomycetes, Dothideomycetes and Chaetothyriomycetes (Eriksson 2001). As regards the Asterothyriaceae, some genera were transferred from that family to a separate family Solorinellaceae (Vezda and Poelt 1990), while Psorotheciopsis was included in Megalosporaceae (Vezda 1973) and Asterothyrium itself was suggested as belonging in Thelotremataceae (Aptroot in Aptroot et al 1994). Anatomical, ontogenetic and phenotype-based phylogenetic evidence, however, suggest that Gomphillaceae and Asterothyriaceae sensu Vezda and Poelt (1987) are monophyletic and best placed in Ostropales (Henssen and Lücking 2002, Lücking 1997, 1999). Recent molecular analyses indicate that the circumscription of this order needs clarification. In the most recent Outline of the Ascomycetes (Eriksson et al 2003), Gyalectales (including Gyalectaceae and Coenogoniaceae) and Ostropales (including Asterothyriaceae, Graphidaceae, Odontotremataceae, Phaneromycetaceae, Solorinellaceae, Stictidaceae and Thelotremataceae) are listed separately and Gomphillaceae are included among Ascomycota: Families of uncertain positions. However, Kauff and Lutzoni (2002) showed that Gyalectales are nested within and form part of Ostropales, which was confirmed in a subsequent analysis by Lumbsch et al (2004). To clarify the uncertain phylogenetic relationships of Gomphillaceae and to test the alternative relationships suggested by various authors, we gathered molecular data of representatives of this family and the Asterothyriaceae. For this purpose, we targeted the nuclear LSU (nulsu) and the mitochondrial SSU (mtssu) region of the ribosomal DNA because combined analyses of these two genes have been used successfully in previous approaches to the phylogeny of Lecanoromycetes (Lumbsch and Schmitt 2002, Lumbsch et al 2004). We chose a Bayesian approach that allows efficient analysis of datasets while employing complex nucleotide substitution models in a parametric statistical framework (Huelsenbeck et al 2001, Larget and Simon 1999). Bayesian phylogenetics also allows simultaneous estimation of uncertainty in the phylogenetic topography, as well as hypothesis testing of alternative topographies, because posterior probabilities of alternative trees can be calculated (Huelsenbeck et al 2000). Our Ostropales outgroup taxa includes the widespread lichen Dimerella pineti (Coenogoniaceae). Because the genus Dimerella recently has been synonymized with Coenogonium, which was confirmed by a molecular phylogenetic analysis (Kauff and Lutzoni 2002, Lücking and Kalb 2000), we propose the new combination Coenogonium pineti in this paper. MATERIALS AND METHODS Taxon sampling. Sequence data of the nulsu rdna and mtssu rdna were collected from a total of 82 euascomycetes. Twenty-four new sequences were obtained from 12 species, as listed in TABLE I. Taxa were sampled to ensure that representatives of the major clades within Lecanoromycetes and taxa of the major classes of euascomycetes were included in the study. Molecular methods. Small samples prepared from freshly collected and frozen herbarium specimens were deep frozen at 80 C for 30 min and ground with sterile plastic pestles. Total genomic DNA was extracted using PureGene Animal Tissue DNA Isolation Protocol (Gentra Systems Inc.). Nuclear LSU rrna was amplified by the polymerase chain reaction (PCR; 95 C 3 min, then 35 cycles of 95 C 45 s, 54 C 45 s, 72 C 1 min) using the forward primers LIC15R, LR0R, LIC24R (Miadlikowska et al 2002, Miadlikowska and Lutzoni 2000, Rehner and Samuels 1994) or ALR1 (Döring et al 2000), and the reverse primer LR3 (Vilgalys and Hester 1990). Mitochondrial SSU rrna was amplified by PCR (95 C 3 min, 55 C 1 min, 72 C 1 min, then 35 cycles of 94 C 1 min, 55 C 1 min, 72 C 1 min) using the primers mrssu1 (Zoller et al 1999) and MSU7 (Zhou and Stanosz 2001). Adding 5 L of purified, 10 mg/ml bovine serum albumin (BSA, New England BioLabs Inc.) to 25 L total PCR reactions greatly improved amplification success. PCR products were electrophoresed in a 1% low-melt agarose TALE gel stained with ethidium bromide and visualized under ultraviolet light. The bands containing DNA were excised and agarose was digested from bands using GELase (Epicentre Technologies). PCR products were sequenced with the amplifying primers in both directions by direct double-strand cycle sequencing using Big Dye version 1 chemistry (Perkin Elmer). Cycle sequencing products were precipitated with ethanol and 3 M sodium acetate and sequenced with a Prism 3100 Genetic Analyzer (ABI). Sequences were edited with Sequencher version 4.1 (Genecodes). About 300 bp of the 5 part of the nulsu could be generated for representatives of Gomphillaceae and Asterothyriaceae. This limitation probably was due to a splicosomal intron specific to these two families, starting at about position 350 of the 5 part of the nulsu, and this phenomenon is currently under further investigation by us.
7 LÜCKING ET AL: PHYLOGENY OF GOMPHILLACEAE 289 TABLE II. Probabilities of five phylogenetic null hypotheses being correct. Each test is based on a B/MCMC tree sample of 1000 trees. Probabilities significant at 0.1% are denoted *** Null hypothesis Gomphillaceae placed in Arthoniomycetes Gomphillaceae placed in Chaetothyriomycetes Gomphillaceae placed in Dothideomycetes Gomphillaceae placed in Agyriales Gomphillaceae placed in Lecanorales Probability 0.00*** 0.00*** 0.00*** 0.00*** 0.00*** Sequence alignments. The mtssu dataset contains sequence portions that are highly variable. Because standard multiple alignment programs, such as Clustal (Thompson et al 1994), become less reliable when sequences are highly divergent, we instead have used an alignment procedure employing a linear Hidden Markov Model (HMM) for the alignment, as implemented in the software SAM (Hughey and Krogh 1996; compbio/sam.html). Sequences of 82 species (TABLE II) were aligned separately for the two genes. Regions that could not be aligned with statistical confidence were excluded from the phylogenetic analysis. Phylogenetic analysis. The alignment was analyzed using the programs PAUP* 4.0b10 (Swofford 2003) and MrBayes 3.0 (Huelsenbeck and Ronquist 2001). The polarity of characters was assessed selecting four representatives of Dothideomycetes as outgroup because this group repeatedly was found as basal within inoperculate euascomycetes in recent phylogenetic estimates of euascomycetes (e.g., Liu et al 1999, Lumbsch et al 2002). The data were analyzed using a Bayesian approach (Huelsenbeck et al 2000, Larget and Simon 1999). Posterior probabilities were approximated by sampling trees using a Markov chain Monte Carlo (MCMC) method. The posterior probabilities of each branch were calculated by counting the frequency of trees that were visited during the course of MCMC analysis. The program MrBayes was used to sample trees. The analysis was performed assuming the general time-reversible model (Rodriguez et al 1990), including estimation of invariant sites and assuming a discrete gamma distribution with six rate categories (GTR I G) for the single-gene and the combined analyses. The nucleotide substitution model was selected with a likelihood ratio test (Huelsenbeck and Crandall 1997) with the program Modeltest (Posada and Crandall 1998). No molecular clock was assumed. Initial runs were conducted, starting with random, NJ or ME trees to check the number of simultaneous MCMC chains necessary to avoid being trapped on local optima. For this, the separate initial analyses were run with generations with an increasing number of chains (starting with four). When the separate analyses converged at a similar likelihood value, it was assumed that the number of chains was sufficient. This was the case with eight chains. To allow an additional range of security, we have chosen to run the analyses employing 12 simultaneous chains that started with a random tree. The analyses started with a random tree and was run with generations. Eleven of these chains were heated. During its search in the universe of trees, a cold chain might become stuck in isolated peaks. To circumvent this, heated chains that can jump to other areas in the universe of trees run simultaneously. These heated chains act as scouts to enable the cold chain to escape local optima. Every 100th tree sampled was saved into a file. We plotted the log-likelihood scores of sample points against generation time using Microsoft Excel and determined that stationarity was achieved when the log-likelihood values of the sample points reached a stable equilibrium value (Huelsenbeck and Ronquist 2001). The initial 2000 trees that showed a linear increase in likelihood values were discarded as burn-in before stationarity was reached. Using PAUP*, majority-rule consensus trees were calculated from trees sampled after reaching likelihood convergence to calculate the posterior probabilities of the tree nodes. Unlike nonparametric bootstrap values (Felsenstein 1985), these are estimated probabilities of the clades under the assumed model (Rannala and Yang 1996) and hence posterior probabilities equal to and above 95% are considered significant supports. Phylogenetic trees were drawn using TreeView (Page 1996). We used a Bayesian approach to examine the heterogeneity in phylogenetic signal between the two data partitions (Buckley et al 2002). For the two genes and the concatenated analyses, the set of topologies reaching 0.95 posterior probability was estimated. The combined analysis topology then was compared for conflict with the 0.95 posterior intervals of the single gene analyses. If no conflict was evident, it was assumed that the two datasets were congruent and could be combined. If conflict was evident, the two datasets were interpreted as incongruent and thus the concatenated analysis might be potentially misleading (Bull et al 1993). Five hypothesized phylogenetic relationships of Gomphillaceae expressed in recent publications were tested as null hypotheses using a MCMC tree sampling procedure as described above. For hypothesis testing, a run as described above was performed with the same settings as in the estimation of the phylogeny. One thousand trees at the equilibrium state per null hypothesis were used from this analysis. The probability of the null hypothesis being correct is calculated by counting the presence of this topology in the MCMC sample (Lewis 2001, Lumbsch et al 2004). The frequency of trees in the MCMC sample agreeing with the null hypothesis was calculated using the filter command in PAUP* with constraints used to describe the null hypothesis. The constraints were constructed so that only the single node of interest was resolved. To examine the possibility that the inferred phylogenetic relationships were due to long-branch attraction (Felsenstein 1978), we employed a 2 -test for deviant nucleotide composition using TREE-PUZZLE (Strimmer and von Haeseler 1996) and a relative-rate test using RRTREE (Robinson-Rechavi and Huchon 2000). RESULTS We generated a total of 12 new mitochondrial SSU rdna and 12 new nuclear LSU rdna sequences for
8 290 MYCOLOGIA this study (TABLE I). The sequences were aligned with 70 mtssu and 70 nulsu rdna sequences obtained from Genbank (TABLE I) to produce a matrix of 270 unambiguously aligned nucleotide position characters in the nu LSU and 766 in the mt SSU dataset. One hundred thirty-three characters in the nu LSU and 656 in the mt SSU dataset were variable. The Bayesian approach for testing datasets for incongruence indicated that the topology of the majorityrule consensus tree from the combined analysis lies within the 0.95 posterior intervals for the two separate datasets (data not shown). This is consistent with the hypothesis that the two partitions have evolved along the same underlying topology and hence a combined analysis was performed. The combined alignment is available in TreeBASE (http:// herbaria.harvard.edu/treebase/). The likelihood parameters in the sample of the combined analysis (values of the separate analyses not shown) had these average values ( one standard deviation): base frequenices (A) ( 0.008), (C) ( 0.005), (G) ( 0.006), (T) ( 0.008), rate matrix r(ac) ( 0.104), r(ag) ( 0.219), r(at) ( 0.201), r(cg) ( 0.103), r(ct) ( 0.389), r(gt) 1.0 ( 0.0), gamma shape parameter alpha ( 0.032), and the proportion of invariable site p(invar) ( 0.092). In the majority-rule consensus tree of sampled trees (FIG. 3), the currently accepted classes, such as Lecanoromycetes or Sordariomycetes (Eriksson et al 2003), are monophyletic with strong support (posterior probability [pp] 1.0 for all classes). Chaetothyriomycetes and Eurotiomycetes appear as a sister group of Lecanoromycetes, but this relationship lacks support. The Lecanoromycetes includes two major clades, one comprising Ostropales sensu lato and the other Lecanorales, Pertusariales and Agyriales. The latter group, however, again lacks support. Ostropales sensu lato is strongly supported (pp 1.0). Within this order, Stictidaceae (Stictis, Absconditella) appear basal, while monophyletic Thelotremataceae (Diploschistes, Thelotrema) are strongly supported (pp 1.0). Gyalectaceae (Bryophagus, Xerothrema, Gyalecta) appear paraphyletic when including Bryophagus, whereas Coenogoniaceae is represented by a single species only (Coenogonium pineti [Schrad. ex Ach.] Lücking & Lumbsch comb. nov.; Lecidea pineti Schrad. ex Ach., Lich. Univ.: ; Acharius, Syn. Lich.: ; Dimerella pineti [Schrad. ex Ach.] Vezda, Lich. Sel. Exs. (Pruhonice) 52, No ). Gomphillaceae plus Asterothyriaceae form a monophyletic lineage within Ostropales sensu lato (pp 1.0), and this lineage is sister of Thelotremataceae, supported by pp of The only representative of Asterothyriaceae, Asterothyrium longisporum, is nested within Gomphillaceae. The chiefly nonlichenized Stictidaceae show a sister-group relationship with other taxa of Ostropales sensu lato. To evaluate the potential presence of long-branch attraction, we performed a 2 -test and a relative-rate test. All sequences included in the study passed the 2 -test (P for Asterothyriaceae/Gomphillaceae, P for other euascomycetes), indicating that none of the sequences had a significantly deviating nucleotide composition. The results of the relative-rate tests showed that the Asterothyriaceae/Gomphillaceae and Thelotremataceae clades do not differ significantly in their substitution rate from other Lecanoromycetes. The results were not significant in all three cases examined (P for Asterothyriaceae/Gomphillaceae versus Thelotremataceae, P for Thelotremataceae versus other Lecanoromycetes excluding Asterothyriaceae/Gomphillaceae, P for Asterothyriaceae/Gomphillaceae versus other Lecanoromycetes excluding Thelotremataceae). DISCUSSION Our analysis confirms that Gomphillaceae and Asterothyriaceae (here represented by the single species Asterothyrium longisporum) are closely related and form a monophyletic lineage that is part of the Ostropalean clade in Lecanoromycetes. These results correspond well to previous studies on the anatomy and ontogeny of Gomphillaceae and Asterothyriaceae, including phenotype-based phylogenetic approaches (Aptroot and Lücking 2003, Dennetière and Péroni 1998, Henssen and Lücking 2002, Lücking 1997, 1999, Vezda 1979, Vezda and Poelt 1987). They also support the utility of phenotype-based analyses for hypothesis-building, even in lichen-forming fungi that are notorious for their variable and often homoplasious morphological characters. It also is clear from the analysis that Gomphillus is related closely to other members of Gomphillaceae sensu Vezda and Poelt (1987) and does not form an isolated member of this family, although its very elongate asci and ascospores are different from the clavate to ovoid asci and ellipsoid to cylindrical ascospores found in all other genera. We were unable to confirm a sister group relationship between the Asterothyriaceae and Gomphillaceae, as assumed in previous contributions (Henssen and Lücking 2002, Lücking 1997, 1999). However, because only one representative of the first family could be included in this analysis, this might be an artifact of insufficient taxon sampling. Generic delim-
9 LÜCKING ET AL: PHYLOGENY OF GOMPHILLACEAE 291 FIG. 3. Majority-rule consensus tree based on trees from a B/MCMC tree sampling procedure. Posterior probabilities equal or above 0.95 indicated at branches. Ordinal and/or class placement of taxa indicated at margin.
10 292 MYCOLOGIA itation within Gomphillaceae is also in flux (Lücking 1997) and is being studied by us using a larger set of mtssu and nulsu data. Taxa included here that are currently assigned to the genera Calenia and Echinoplaca accordingly do not form monophyletic groups in our analysis. However, our data suggest that taxa with sessile or adnate, biatorine apothecia (Gyalideopsis, Gomphillus, Tricharia, Echinoplaca) are derived from those with immersed-erumpent, zeorine or carbonized apothecia (Calenia, Gyalectidium, Aulaxina). This contradicts previous hypotheses about the evolution of the group (Lücking 1997) but is in accordance with our present results that Gomphillaceae plus Asterothyriaceae are sister of Thelotremataceae (here represented by Thelotrema and Diploschistes), which are characterized by immersed-erumpent, zeorine apothecia. Indeed, the sister-group relationship of Asterothyriaceae plus Gomphillaceae with Thelotremataceae is supported significantly. However, the branch leading to the Gomphillaceae is unusually long, suggesting that this relationship might be due to long-branch attraction (Felsenstein 1978). If two unrelated lineages have had an accelerated substitution rate compared to other included groups in a study, they will have accumulated characters that will distance them from other taxa in an analysis, resulting in a false clustering based on convergences (Swofford et al 1996). However, the results of the 2 -test and the relative-rate tests reject such an assumption. The Asterothyriaceae/Gomphillaceae clade and the Thelotremataceae do not differ significantly in their nucleotide composition and substitution rate from the other Lecanoromycetes. Our studies thus confirm placement of Gomphillaceae/Asterothyriaceae as a further clade within Ostropales sensu lato (Kauff and Lutzoni 2002), as previously suggested by Lücking (1997) and Henssen and Lücking (2002). This order originally was restricted to the chiefly nonlichenized Stictidaceae and allies, while lichenized Thelotremataceae and Graphidaceae were kept in a separate order Graphidales (Sherwood 1977). Recent molecular studies have not demonstrated only that Graphidales but also Gyalectales, with the two families Gyalectaceae and Coenogoniaceae, form part of Ostropales (Kalb et al pers comm 2003, Kauff and Lutzoni 2002, Lumbsch et al 2004, Winka et al 1998). Thus, Ostropales, in its present circumscription, consists of four lineages: (i) Stictidaceae and allies (Ostropales s.str.), (ii) Gyalectaceae/Coenogoniaceae (former Gyalectales), (iii) Thelotremataceae/Graphidaceae (former Graphidales), and (iv) Gomphillaceae/Asterothyriaceae (former Gomphillales). In all available analyses, Stictidaceae and allies, which include a few lichenized forms (Absconditella, Conotrema) but are otherwise nonlichenized, appear to be basal within the order and either monophyletic (Lumbsch et al 2004) or paraphyletic. Gyalectaceae/ Coenogoniaceae are related most closely to Thelotremataceae/Gomphillaceae and appear either paraphyletic (Lumbsch et al 2004) or monophyletic (Kauff and Lutzoni 2002), depending on whether Bryophagus is included here or in the Stictidaceae lineage. The Thelotremataceae/Graphidaceae clade always appears monophyletic in different studies (Kalb et al pers comm 2003, Kauff and Lutzoni 2002, Lumbsch et al 2004) and so does the previously unexplored Gomphillaceae/Asterothyriaceae clade in our study. Experience with Lecanoromycetes has shown that initially paraphyletic lineages eventually turn out to be monophyletic in more detailed studies with higher taxa and character resolution (Lumbsch et al 2004), and this cannot be excluded for Ostropales sensu lato, in which case the previously distinguished orders Gyalectales, Graphidales and Gomphillales could be reinstated or more appropriately be used at the subordinal level. This would correspond to the situation in Lecanorales sensu lato, where Peltigerineae and Teloschistineae currently are listed as suborders (Eriksson et al 2003) but could also be treated as orders parallel to Lecanorales sensu stricto. ACKNOWLEDGMENTS We wish to thank Jutta Buschbom (Chicago) for advice with sequencing of lichens. The sampling of specimens from which new sequences were generated was supported by the National Science Foundation (Grant DEB to Robert Lücking), the Mexican CONACYT (Grant V to María Herrera-Campos and Robert Lücking) and the Deutsche Forschungsgemeinschaft (grants LU 597/1-1-LU 597/ 4-1 to Robert Lücking). DNA extraction and sequencing at the Field Museum s Pritzker Laboratory for Molecular Systematics and Evolution were supported by start-up funds of the Field Museum to Robert Lücking. LITERATURE CITED Aptroot A, Kärnefeldt I, Tibell L Discussion 6: Caliciales, Graphidales, and Teloschistales. In: Hawksworth DL, ed. Ascomycete systematics. Problems and perspectives in the nineties (NATO ASI Series, Series A: Life Sciences, Vol. 29). Plenum Press, New York and London. p , Lücking R Phenotype-based phylogenetic analysis does not support generic separation of Gyalidea and Solorinella (Ostropales: Asterothyriaceae). Biblioth Lichenol 86: Buckley TR, Arensburger P, Simon C, Chambers GK