|Home | About | Journals | Submit | Contact Us | Français|
We present a comprehensive phylogeny derived from 5 genes, nucSSU, nucLSU rDNA, TEF1, RPB1 and RPB2, for 356 isolates and 41 families (six newly described in this volume) in Dothideomycetes. All currently accepted orders in the class are represented for the first time in addition to numerous previously unplaced lineages. Subclass Pleosporomycetidae is expanded to include the aquatic order Jahnulales. An ancestral reconstruction of basic nutritional modes supports numerous transitions from saprobic life histories to plant associated and lichenised modes and a transition from terrestrial to aquatic habitats are confirmed. Finally, a genomic comparison of 6 dothideomycete genomes with other fungi finds a high level of unique protein associated with the class, supporting its delineation as a separate taxon.
Multi laboratory collaborative research in various biological disciplines is providing a high level of interaction amongst researchers with diverse interests and backgrounds. For the mycological community, the “Assembling the Fungal Tree of Life” project (AFTOL) provided the first DNA-based comprehensive multigene phylogenetic view of the fungal Kingdom (Lutzoni et al. 2004, James et al. 2006). This has also made it possible to revise the classification of the fungi above the ordinal level (Hibbett et al. 2007). Subsequent work is focused on elucidating poorly resolved nodes that were highlighted in the initial DNA-based phylogeny (McLaughlin et al. 2009).
At the other end of the scale from the tree of life projects, taxon sampling with relatively small numbers of sequence characters are also progressing in various barcoding projects (Seifert et al. 2007, Chase et al. 2009, Seifert 2009). It remains important to link these two ends of the spectrum by also sampling intensively at foci of interest between barcoding and the tree of life. With this in mind it is the aim of this paper and subsequent ones in this volume to provide a broadly sampled phylogeny at class level and below for Dothideomycetes. This result is combined efforts and data from a diverse group of researchers to focus on systematic sampling, therefore developing a more robust fungal class wide phylogeny of Dothideomycetes. This is especially important as a framework for comprehending how fungi have evolved as they shift ecological habitats and adapt to new environments and nutritional modes.
It is apparent that the assemblage of fungi, now defined as Dothideomycetes, exemplifies a dynamic evolutionary history. This is by far the largest and arguably most phylogenetically diverse class within the largest fungal phylum, Ascomycota (Kirk et al. 2008). It contains a heterogeneous group of fungi that subsist in the majority of the niches where fungi can be found. The best-known members of the group are plant pathogens that cause serious crop losses. Species in the genera Cochliobolus, Didymella, Phaeosphaeria, Pyrenophora, Venturia, Mycosphaerella and Leptosphaeria, or their anamorphs, are major pathogens of corn, melons, wheat, barley, apples, bananas and brassicas respectively, in most areas of the world where they are cultivated. Other species are important pathogens in forestry e.g. species in the genera Botryosphaeria and Mycosphaerella and their anamorphs that attack economically important tree species.
Despite a large body of work containing taxonomic, phytopathological, genetic and genomic research, the majority of fungi hypothesised to be members of Dothideomycetes remain under-sampled within a systematic framework. Several studies performed during the course of the last four years have advanced our understanding of these fungi, but phylogenetic relationships of the saprobes, aquatic, asexual and lichenised species remain particularly poorly studied. Indeed, their conspicuous absence in phylogenetic analyses frustrates a broader understanding of dothideomycete evolution.
Dothideomycetes share a number of morphological characters with other fungal classes. It was recently formally described (Eriksson & Winka 1997) replacing in part the long-recognised loculoascomycetes (Luttrell 1955). This redefinition of the loculoascomycetes was mainly prompted by DNA sequencing comparisons of ribosomal RNA genes (Berbee & Taylor 1992, Spatafora et al. 1995) that was subsequently expanded and confirmed (Berbee 1996, Silva-Hanlin & Hanlin 1999, Lindemuth et al. 2001, Lumbsch & Lindemuth 2001). These early phylogenetic studies demonstrated that loculoascomycetes, as it was defined, is not monophyletic, although contrary views exist (Liu & Hall 2004). Nevertheless the majority of analyses have shown that some loculoascomycete taxa, such as the “black yeasts” in Chaetothyriales as well as the lichenised Verrucariales, reside within Eurotiomycetes as subclass Chaetothyriomycetidae (Spatafora et al. 1995, Winka et al. 1998, Geiser et al. 2006, Gueidan et al. 2008). The majority of the remaining loculoascomycete species are now placed in Dothideomycetes. Although finer morphological distinctions between the distantly related members of loculoascomycetes can be made, their synapomorphies remain elusive (Lumbsch & Huhndorf 2007). These findings all point to the fact that a number of loculoascomycete morphological characters are either retained ancestral traits or that they exhibit convergence due to similar selection pressures.
Traditionally the most important morphological characters used to define major groups in Ascomycota were the type of ascus, septation of ascospores, the morphology and development of the ascoma, as well as the structure and organisation of the centrum. Dothideomycetes (and previously, loculoascomycetes) have fissitunicate (or functionally bitunicate) asci, that emerge from ascolocular development in preformed locules within vegetative tissue, that represents the ascoma. The reproductive structures in ascolocular development are derived from cells before fusion of opposing mating types occurs and can contain one or several locules. This form of ascolocular development is in contrast to the ascohymenial development found in most other fungal classes. During ascohymenial development asci are generated in a hymenium and the reproductive structure is derived from cells after fusion of opposing mating types. The fissitunicate ascus has been described for more than a century, but the importance of ascolocular development was first emphasised in 1932 (Nannfeldt 1932). Importantly Nannfeldt's concepts were also the basis for the Santesson's integration of lichens into the fungal classification (Santesson 1952). In fissitunicate asci, generally, the ascospores are dispersed by the rupture of the thick outer layers (ectotunica) at its apex, allowing the thinner inner layer (endotunica) to elongate similar to a “jack in a box”. The elongated endotunica ruptures apically and releases the ascospores forcefully through the ascoma opening. The spores are then released in the air, or in aquatic species, under water. Building on this work and that of others (Miller 1949), Luttrell proposed Loculoascomycetes, synonymous to Nannfeldt's “Ascoloculares” (Luttrell 1955). Importantly, he proposed a correlation between fissitunicate asci and ascolocular development, also emphasising the importance of ascus morphology and dehiscence as well as the development of surrounding elements within the ascoma.
Although the concept of a group of fungi (including the Dothideomycetes) with fissitunicate asci and ascolocular development has been accepted by several authors, much less agreement could be found on ordinal definitions in the era before molecular characters. This ranged from proposing a single order (von Arx & Müller 1975) to three (Müller & von Arx 1962), five (Luttrell 1951, 1955) six (Barr 1979), or seven (Barr 1987). Luttrell initially described a number of important development types centered on descriptions of all tissues inside the ascoma (the centrum concept) and combined this with ascoma structure to define his five orders (Luttrell 1951, 1955). Of Luttrell's initial centrum concepts three are applicable to the Dothideomycetes as they are presently defined. Thus, the Pleospora type, the Dothidea type and the Elsinoë type centra correspond to the dothideomycete orders Pleosporales, Dothideales and Myriangiales, respectively. An important refinement to Luttrell's ideas was introduced with the concept of the hamathecium by Eriksson (Eriksson 1981). This is defined as a neutral term for sterile hyphae or other tissues between the asci in the ascoma (Kirk et al. 2008). For example, hamathecial types can include the presence or absence of pseudoparaphyses, which are sterile cells that extend down from the upper portion of the ascomatal cavity. They become attached at both ends, although the upper part may become free at maturity. Other important concepts introduced by Müller and von Arx (Müller & von Arx 1962) focused on the morphology of the ascoma opening and ascus shape. The Dothidea type centrum in the type species of Dothidea, D. sambuci illustrates several typical dothideomycete morphologies (Fig. 1). These include the thick-walled fissitunicate asci produced within a multilocular stroma.
The most recent dothideomycete class-wide morphological assessments were carried out by Barr (Barr 1979, 1987). Her subclasses were determined based on characters in the centrum, including the absence, presence and types of hamathecial tissues. Consistent with several earlier authors, Barr's ordinal classifications were based on ascomatal shape (perithecioid or apothecioid) and manner in which nutrients are obtained by the fungus (Barr 1987). In addition to these characters she emphasised the importance of finer distinctions in the hamathecium such as the shape and structure of the pseudoparaphyses (Barr 1979, 1987).
The introduction of molecular phylogenies for Dothideomycetes (Berbee 1996) provided an opportunity to verify the significance of various morphological characters used in the aforementioned classifications. The clearest correlation with a DNA sequence-based phylogeny was for the presence or absence of pseudoparaphyses, largely agreeing with the first orders proposed by Luttrell (Liew et al. 2000, Lumbsch & Lindemuth 2001). Barr's concept of applying the shape of the pseudoparaphyses to define orders was rejected by molecular phylogenies (Liew et al. 2000). This set the stage for more comprehensive analyses incorporating protein data, and resulted in the definition of two subclasses, Pleosporomycetidae (pseudoparaphyses present) and the Dothideomycetidae (pseudoparaphyses absent; Schoch et al. 2006). Numerous orders and other taxa remained unresolved outside of these two subclasses.
The most recent class level phylogenetic analyses combining sequences from protein coding genes with ribosomal RNA sequences fortified the view that Dothideomycetes is a monophyletic group (Schoch et al. 2009a, b). Furthermore, strong support was found for a sister relationship between Dothideomycetes and the lichenised class Arthoniomycetes (Lumbsch et al. 2005, Spatafora et al. 2006, Schoch et al. 2009a). This clade was recently defined as a rankless taxon “Dothideomyceta” (Schoch et al. 2009a, b). The Arthoniomycetes consists of a single order (Arthoniales) of lichens and lichenicolous fungi (Ertz et al. 2009) that produce bitunicate asci in ascohymenial apothecia and was proposed as an intermediate group or “Zwischengruppe” (Henssen & Thor 1994). This placement raises intriguing questions regarding the origins of ascolocular development and further illustrates the importance of including lichen-forming fungi in dothideomycete phylogenies.
While considerable progress has been made in defining these fungi the placement of Dothideomycetes in relation to the majority of other Ascomycota classes remains unresolved. Here, greater clarity would likely require a huge increase of characters from genome projects. In this regard, the first phylogenomic studies have shown low resolution for this relationship (Fitzpatrick et al. 2006, Kuramae et al. 2006, Robbertse et al. 2006). This could indicate a rapid radiation event, but more likely suggests taxon sampling bias. This latter view is supported by the fact that none of these studies has included lichenised species that represent about 25 % of the number of species in Ascomycota.
The authors of this volume have focused on two primary goals. These are to considerably expand the taxon sampling of existing orders by including saprobes, asexual species and other poorly sampled groups. Secondly we aim to sample widely within specific environmental niches and present a multigene phylogeny that exposes the highly diverse nature of Dothideomycetes.
The majority of fungal cultures were obtained from the CBS culture collection and additional sources mentioned in other papers of this volume. DNA was also provided by authors of several papers presented in this volume and the reader is referred to Boehm et al. (2009a), Crous et al. (2009a), Suetrong et al. (2009) and Zhang et al. (2009). For additional details see Table 1 - see online Supplementary Information. Fungal genomic DNA was obtained by scraping mycelium from PDA plates. Samples were subsequently pulverised and the DNA was extracted using the FastDNA® kit and the FastPrep® instrument from MPI Biochemicals (Irvine, CA, U.S.A.). DNA amplifications were completed using Taq polymerase (GenScript, Piscataway, NJ, U.S.A.), with FailSafe™ PCR 2× PreMix E (Epicentre, San Diego, CA, U.S.A.). Primers were used as noted in the Assembling the Fungal Tree of Life project (AFTOL; Schoch et al. 2009a). This resulted in DNA sequence data obtained from the small and large subunits of the nuclear ribosomal RNA genes (SSU, LSU) and three protein coding genes, namely the translation elongation factor-1 alpha (TEF1) and the largest and second largest subunits of RNA polymerase II (RPB1, RPB2). Primer sets used for these genes were as follows: SSU: NS1/NS4; LSU: LR0R/LR5; TEF1 983/2218R (initially obtained from S. Rehner: ocid.nacse.org/research/deephyphae/EF1primer.pdf); RPB2: fRPB2-SF/fRPB2-7cR; RPB1: RPB1-Ac/RPB1-Cr (obtained from V. Hofstetter). Primer sequences are available at the WASABI database at the AFTOL website (aftol.org). PCRs for these genes were performed in various laboratories of the coauthors mentioned but the majority of reactions were run under conditions described previously (Lutzoni et al. 2004, Schoch et al. 2009a). Two duplicate sets of sequences were inadvertently included in the analysis (indicated in Table 1).
Sequences were obtained from WASABI (Kauff et al. 2007) as well as from previous publications (e.g. Lutzoni et al. 2004, Schoch et al. 2009a). Introns were removed and an initial core set of 171 taxa were aligned by using default options for a simultaneous method of estimating alignments and tree phylogenies, SATé (Liu et al. 2009). In order to consider codons without the insertion of unwanted gaps, protein coding fragments were translated in BioEdit v. 7.0.1 (Hall 2004) and aligned within SATé as amino acids. These were then realigned with their respective DNA sequences using the RevTrans 1.4 Server (Wernersson & Pedersen 2003). After the removal of intron sequences the alignment was examined manually in BioEdit with a shade threshold of 40 % and regions with high amounts of gap characters were excluded. This resulted in a reduction of 99 columns in the LSU data set, 118 in RPB1 and 162 in RPB2, for a total of 379. Nothing was removed for TEF1. In order to allow for the extension of our alignment as newly generated sequences became available from other studies in this volume, these were subsequently added to this core alignment with MAFFT v. 6.713 (Katoh et al. 2009). The E-INS-i setting, focused on high accuracy with a high percentage of unalignable regions such as introns, was applied and the SATé alignment was used as a seed. This resulted in a supermatrix of five genes (LSU, SSU TEF1, RPB1, RPB2) consisting of 52 % gaps and undetermined characters out of a total of 6 582 characters. GenBank accession numbers are shown in Table 1.
Conflict tests on the initial core set of 204 taxa were conducted by selecting single gene data sets and doing comparisons on a gene by gene basis. This was done using the “bootstopping” criterion in RAxML v. 7.0.4 (Stamatakis et al. 2008) under the CIPRES v. 2.1 webportal to produce trees of comparative gene sets where all taxa have the gene present. Comparisons between all potential sets of gene trees with no missing taxa were done using a script (Kauff & Lutzoni 2002) obtained through the Lutzoni lab website and to detect present or absent taxa within clades with a cut-off bootstrap value of 70 %. This is described in more detail elsewhere (Miadlikowska et al. 2006, Schoch et al. 2009a).
A phylogenetic analysis was performed using RAxML v. 7.0.4 (Stamatakis 2006) applying unique model parameters for each gene and codon. The dataset was divided in 11 partitions as previously described in Schoch et al. (2009a). A general time reversible model (GTR) was applied with a discrete gamma distribution and four rate classes following procedures laid out in Schoch et al. (2009). Ten thorough maximum likelihood (ML) tree searches were done in RAxML v. 7.0.4 under the same model, each one starting from a randomised tree. Bootstrap pseudo replicates were performed 2000 times using the fast bootstrapping option and the best scoring tree form 10 separate runs were selected. The resulting trees were printed with TreeDyn v. 198.3 (Chevenet et al. 2006). All alignments are deposited in TreeBASE. Additionally, the data sets were analyzed in GARLI v. 0.96 (Zwickl 2006) using the GTR-gamma-invariant model. In this case 200 bootstraps were run under default conditions.
Ancestral reconstructions were performed in Mesquite v. 2.6 with character states traced over 2000 bootstrapped trees obtained with RAxML-MPI v. 7.0.4 (Stamatakis 2006). Following the phylogeny presented (Fig. 2) this reconstruction was performed with a maximum-likelihood criterion using the single parameter Mk1 model. Ancestral states were assigned to a node if the raw likelihood was higher by at least 2 log units than the likelihood value of the other ancestral state(s) according to default settings. Character states were also mapped using TreeDyn v. 198.3 (Chevenet et al. 2006), shown in Fig. 3. This is presented as a clockwise circular tree, starting with outgroup taxa. Only clades with more than two taxa of the same state are shown and bootstrap recovery was not considered in assigning character states. In applying the character states of saprobes (including rock heterotrophs), plant associated fungi (including pathogens, endophytes and mycorrhizae) and lichenised fungi the broad concepts presented were followed as laid out in Schoch et al. (2009a). Some character assessments were taken from Zhang et al. (2009; this volume). Ecological characters of sampling sources, terrestrial, fresh water and marine were assessed based on papers elsewhere in this volume (Suetrong et al. 2009, Shearer et al. 2009).
A MCL (Markov Cluster Algorithm) protein analysis of 52 fungi and one metazoan (Drosophila melanogaster) (Table 2 - see online Supplementary Information) and the phylogenetic placement of these species was used to characterise the phylogenetic profile of each cluster. Chytridiomycota and Mucoromycotina each were represented by one and two species, respectively. In Dikarya, Basidiomycota and Ascomycota were represented by 8 and 40 species respectively. The Pezizomycotina (filamentous ascomycetes) was presented by 26 species in four classes [Sordariomycetes (12), Leotiomycetes (2), Dothideomycetes (6) and Eurotiomycetes (6)].
The phylogram presented in Fig. 2 represents the largest ever phylogenetic assessment of Dothideomycetes to date. Here the focus has been on expanding taxon diversity in the class while specifically avoiding a small number of taxa that other analyses suggest reside on long unstable branches. This still allowed for an extensive sweep of dothideomycete taxon diversity; in doing so we followed the premise of allowing for missing data in our supermatrix (Wiens 2006). An effort was made to intersperse taxa with poor character sampling amongst those having better sampling throughout the tree, but the inclusion of missing characters could still have unanticipated effects on phylogenetic assessments (Lemmon et al. 2009). While recognising this caveat, a recent expansive data set covering all of Ascomycota noted very little changes in major nodes even after the removal of taxa with high proportions of missing characters (Schoch et al. 2009a). The phylogeny presented here agrees well with broad phylogenies in this volume and elsewhere (Schoch et al. 2006, Crous et al. 2007a, Zhang et al. 2008, Crous et al. 2009b). After all introns and 379 ambiguous character positions were removed, the matrix consisted of 52 % missing and indeterminate characters. This maximum-likelihood analysis had 5 069 distinct alignment patterns and produced a best known likely tree with a log likelihood of -207247.761117.
The ancestral reconstructions in Fig. 3 indicate that phytopathogenicity can be confined to a number of terminal clades throughout the tree and that these always reside within saprobic lineages. A maximum of seven transitions likely occurred in several lineages of the orders Pleosporales, Capnodiales and singular lineages in Myriangiales, Botryosphaeriales and Venturiaceae (also see in this volume; Crous et al. 2009a, Zhang et al. 2009). Several transitions to lichenisation have also occurred, although phylogenetic uncertainty may limit this to a minimum of two. Due to the use of lichenised Arthoniomycetes as outgroup a broader assessment is required to determine whether the Dothideomycetes evolved from a lichenised ancestor. Previous studies suggested that the saprobic habit is an ancestral trait but only with marginal support (Schoch et al. 2009a). Similar conclusions can be reached for the aquatic ecological characters – the majority of fresh water and marine clades reside within terrestrial clades as has been shown previously e.g. (Spatafora et al. 1998, Vijaykrishna et al. 2006). Transitions from a terrestrial life style to fresh water likely occurred at least three times and transitions to marine environments up to six times. Phylogenetic uncertainty for the placement of some marine clades can limit this to a minimum of four times (Fig. 2). Reversions from aquatic to terrestrial environments are rare, with one possible exception in the Lentitheciaceae where bambusicolous saprobes reside, nested within several fungi occurring in freshwater habitats (for additional details see Zhang et al. 2009; this volume). Phylogenetic resolution will have to improve to test this further.
An analysis of recently released genomes was compared to consider whether genome composition reinforces phylogenetic support for Dothideomycetes (Fig. 4). Relative to a clustering analysis of proteins from 52 sequenced fungi and Drosophila melanogaster, about 5 515 protein coding genes from Dothideomycetes shared protein clusters with proteins from other dothideomycete fungi only. This comprises roughly 8–11 % of the protein coding genes in each of six sequenced Dothideomycetes. The species profile of each protein cluster was used to assign a phylogenetically informed designation. The profiles most frequently seen were those of the most conserved proteins, namely clusters designated as having a shared Ophistokont phylogenetic profile. Among the more derived nodes of the Dothideomycetes, protein clusters were observed that had a species composition that could reflect the result of selection pressure on more distantly related fungi that share the same niche.
A phylogenomic profile (Fig. 4) of the proteins from six Dothideomycetes from the two largest orders seen in Fig. 1 is presented (Mycosphaerella graminicola, Mycosphaerella fijiensis, Phaeosphaeria nodorum, Alternaria brassicicola, Pyrenophora tritici-repentis, Cochliobolus heterostrophus). The highest percentage of proteins (excluding species specific proteins) were conserved outside kingdom Fungi (Ophistokont node, 23 %), followed by proteins specific for the Dikarya (14 %) and the Pezizomycotina (13 %). This breakdown was also prevalent within other Pezizomycotina classes. Approximately 8 % of the proteins from the six Dothideomycetes were conserved across and within derived nodes in this class. Relative to this analysis 28 % of the proteins were specific to the Dothideomycetes (including species specific proteins). The other class containing loculoascomyetes, Eurotiomycetes, had 19.5 % proteins characterised as class specific. This means the percentage dothideomycete specific proteins were about 8.5 % more. Eurotiomycetes in the analysis were mostly human pathogens, with most having no known sexual state whereas the Dothideomycetes in the analysis were all plant pathogens and mostly with known sexual states. This breakdown of nutritional modes, although not comprehensive for these two classes, is somewhat representative. In Eurotiomycetes human pathogens are more diverse and plant pathogens uncommon, with the converse being true for Dothideomycetes. Both classes contain melanised species with similar morphologies and more comprehensive comparative studies need to expand sampling to incorporate species from the different nutritional modes for both classes.
In the phylogram presented (Fig. 2) the two dothideomycete subclasses previously described based on presence or absence of pseudoparaphyses (Schoch et al. 2006) could be recovered with varying levels of bootstrap representation. Subclass Pleosporomycetidae previously included Pleosporales plus a single species, representing Mytilinidiaceae, namely Lophium mytilinum (Schoch et al. 2006). Taxon sampling for the Mytilinidiaceae was considerably expanded by Boehm et al. (2009b), with the addition of a number of new taxa, leading to the establishment of the Mytilinidiales. Likewise, extensive taxon sampling for the family Hysteriaceae led to a newly redefined Hysteriales also included in this subclass (Boehm et al. 2009a; this volume). It appears that persistent, hysteriaceous carbonaceous ascomata that dehisce via a longitudinal slit (e.g., hysterothecia) have evolved multiple times within Pleosporomycetidae (Mugambi & Huhndorf 2009,Mugambi & Huhndorf 2009). Pleosporomycetidae can be expanded to tentatively include Jahnulales (Fig. 2B) based on strong bootstrap support from RAxML analyses and morphology. Perithecioid ascomata and a hamathecium of wide cellular pseudoparaphyses are characteristic of Jahnulales (Inderbitzin et al. 2001, Pang et al. 2002; Shearer et al. 2009; this volume) and agree with diagnostic features for Pleosporomycetidae. We also recommend that the definition of the subclass be reassessed with more inclusive character sets. Also, Leptosphaerulina species characterised by the absence of pseudoparaphyses reside within the pseudoparaphysate Pleosporales (Fig. 2C; Silva-Hanlin & Hanlin 1999, Kodsueb et al. 2006), indicating that pseudoparaphyses could have been lost multiple times. It should be noted that the maturity of ascomata may play an important role in these assessments. Immature specimens may contain pseudoparaphyses that dehisce when mature and these characteristics need to be evaluated with more complete sampling of the numerous aparaphysate taxa still listed as incertae sedis. The second subclass, Dothideomycetidae, previously circumscribed based on the absence of pseudoparaphyses remains well supported (Fig. 2C).
The results of this study provided continued support for ten orders within class Dothideomycetes, namely Pleosporales, Hysteriales, Mytilinidiales, Patellariales, Botryosphaeriales, Jahnulales, Dothideales, Capnodiales, Myriangiales and Trypetheliales. The latter order was recently proposed (Aptroot et al. 2008) and represents the largest lichen forming clade in Dothideomycetes. Another recently proposed order, Botryosphaeriales includes only the single family, Botryosphaeriaceae. The analysis (Fig. 2B), however, shows strong support for a narrower interpretation of the Botryosphaeriaceae, typified by Botryosphaeria dothidea and related genera, excluding a separate clade of species residing in Guignardia (with Phyllosticta anamorphs). Bagnisiella examinens and Saccharata protea did not reside in either of the above clades, placed on early diverging branches. A more extensive taxon sampling is required to address the diversity in this order, which most likely will validate the separation of additional families. Another currently accepted order, Microthyriales, consisting of species occurring as saprobes or epiphytes on stems and leaves is represented in this study by only a single sample, Microthyrium microscopicum (Fig. 2C). Members of this order are poorly represented in culture and have unusual thyrothecial ascomata that have a scutate covering comprising a thin layer of radiating cells. This structure is generally lacking a basal layer and is quite unlike any morphologies in other orders. This positioning adjacent to the plant parasitic Venturiaceae and coprophilic Phaeotrichaceae, is unexpected but since the single representative of the Microthyriales is on a long branch this is a relationship that will require more intensive taxon sampling.
Additional families that could not be placed in an order are Tubeufiaceae and Gloniaceae (Fig. 2B). Species in Tubeufiaceae have superficial clustered ascomata and characteristic bitunicate asci with relatively long ascospores, often with helicosporous anamorphs (Kodsueb et al. 2008). Members of Tubeufiaceae, which frequently occur in freshwater habitats include anamorph genera, such as Helicoon and Helicodendron, and are ecologically classified as aeroaquatic species. A few teleomorph taxa such as Tubeufia asiana occur on submerged wood (Tsui et al. 2007), and Tubeufia paludosa occur on herbaceous substrates in wet habitats (Webster 1951). The Gloniaceae are saprobic, have dichotomously branched, laterally anastomosed pseudothecia that form radiating pseudo-stellate composites and dehisce by an inconspicuous, longitudinal, but evaginated slit. They reside sister to the saprobic Mytilinidiales but due to conspicuous morphological differences and moderate statistical support they are placed in Pleosporomycetidae incertae sedis (Boehm et al. 2009a, this volume).
Several other well supported clades representing families were evident in this study (Fig. 2). These include several families in Pleosporales, treated elsewhere (Zhang et al. 2009; this volume). Other clades have lower levels of support. For example Leptosphaeriaceae (Fig. 2A) have moderate bootstrap support and it is treated in the very broad sense here. There was also support for several newly described families treated in different papers within this volume. In Pleosporales these include Amniculicolaceae and Lentitheciaceae (Zhang et al. 2009; this volume). The Lindgomycetaceae (Shearer et al. 2009; this volume, Hirayama et al. 2010) encompassing a majority of species isolated from fresh water habitats. Two other novel families, Aigialaceae and Morosphaeriaceae include mainly marine species (Suetrong et al. 2009; this volume). In addition to these, the sampling of a wide diversity of fungi on bamboo yielded the description of Tetraplosphaeriaceae (Tanaka et al. 2009; this volume). Another novel family, Dissoconiaceae, is proposed by Crous et al. 2009 (this volume) for foliicolous commensalists on Eucalyptus leaves, some of which are putative hyper parasites and reside in Capnodiales.
Results of this study suggest that sampling within existing families also requires continued expansion as familial definitions in Dothideomycetes remains problematic. A paper focused on two families, with poor representation in molecular data sets, Melanommataceae and Lophiostomataceae addresses this in more detail (Mugambi & Huhndorf 2009,Mugambi & Huhndorf 2009; this volume). Numerous other clades in our tree remain without familial placement. This includes a diverse group in Capnodiales (Fig. 2C, clade C) a newly described group of hysteriaceous fungi in Pleosporales (Fig. 2A, clade G) and additional marine lineages (clades H, L, Fig. 2A). An interesting clade tentatively circumdescribed by Zhang et al. (2009; this volume) as Massariaceae contains bambusicolous fungi and appears related to the lichenised Arthopyreniaceae (Fig. 2A).
Finally, a clade including Corynespora anamorphs (clade K, Fig. 2A) is placed for the first time, but without clear relationship to any other currently defined families. The genus Corynespora includes anamorphic fungi with tretic, percurrent, and acropetal conidiogenesis. The melanised, pseudoseptate conidia have a pronounced hilum from which the conidial germ tube emerges and are borne apically from solitary, melanised conidiophores. Though nearly 100 species are described based on differences in morphology, considerable phenotypic plasticity within individual isolates complicates species recognition, and molecular analyses that may result in taxonomic clarification have not been done. Corynespora species fill a diversity of roles as saprobes, pathogens, and endophytes on and in woody and herbaceous plants, other fungi, nematodes, and human skin (Dixon et al. 2009). One of the species represented here, C. cassiicola is an important pathogen of rubber. The teleomorphic fungi Pleomassaria swidae (Pleomassariaceae; Tanaka et al. 2005) and Corynesporasca caryotae (Corynesporascaceae; Sivanesan 1996) have unnamed Corynespora species as anamorphs. In this study, species currently placed in Corynespora are not monophyletic and are positioned in at least two families: Massarinaceae and Clade K (Fig. 2A).
The previously mentioned Dissoconiaceae relies on taxonomic descriptions based on anamorph characters. This is a theme that is expected to continue for mitosporic taxa in Dothideomycetes as molecular data accelerates their integration. The artificial nature of the “higher” taxa of anamorphs e.g., deuteromycetes (Kirk et al. 2001) is now well recognised, but the integration of anamorphs into the phylogenetic classification of teleomorphs remains a significant challenge in fungal systematics (Shenoy et al. 2007). The correlation of teleomorphs and anamorphs (Seifert et al. 2000) is not always predictive but it has been applied in some genera within Dothideomycetes, e.g. Botryosphaeria and Mycosphaerella (Crous et al. 2006, 2009b). However, numerous examples underscoring anamorph convergence can be found throughout the class e.g. Dictyosporium (Tsui et al. 2006, Kodsueb et al. 2008), Sporidesmium (Shenoy et al. 2006), Cladosporium (Crous et al. 2007b) and Phoma (Fig. 2A; Aveskamp et al. 2009, de Gruyter et al. 2009, Woudenberg et al. 2009) as well as Fusicoccum and Diplodia (Crous et al. 2006, Phillips et al. 2008). The use of large multigene phylogenies will be essential to bring taxonomic order to cryptic anamorph lineages.
Besides the unclassified diversity found in anamorphic genera, numerous ecological niches contain diverse lineages of fungi lacking systematically sampled molecular characters. Several examples of this knowledge gap can be found in papers in this volume. In this regard, the rock inhabiting fungi are amongst the least understood. These fungi exist ubiquitously as melanised, slow growing colonies and that usually do not produce generative structures. They subsist on bare rock surfaces and are consequently highly tolerant of the environmental stresses induced by lack of nutrients, water and extremes in radiation and temperature (Palmer et al. 1990, Sterflinger 1998, Ruibal 2004, Gorbushina et al. 2008). Members of this ecological guild are diverse and occur in two classes – Eurotiomycetes and Dothideomycetes. Ruibal et al. 2009 (this volume) present the results of an expanded sampling of rock-inhabiting fungi that include lineages residing within Dothideomycetes and sister class Arthoniomycetes. These rock inhabiting fungi can be placed in Capnodiales, Pleosporales, Dothideales and Myriangiales, as well as some unclassified lineages of Dothideomycetes. Interestingly, some associated lineages were without clear placement within either Arthoniomycetes or Dothideomycetes. The rock isolates included in Fig. 2C illustrate a subsection of genetic diversity seen in these extremophiles, in particular for the Capnodiales, with two rock isolates-rich lineages Teratosphaeriaceae and Clade C (Fig. 2C). A more detailed analysis (Ruibal et al. 2009; this volume) allows for the presentation of hypotheses related to evolution of pathogenicity and lichenisation because these modes of nutrition are often found in close proximity of rock inhabiting fungal lineages.
The lichenised fungi allied with the Dothideomycetes represent another poorly sampled group of fungi. Several lichenised species remain enigmatically placed after they were confirmed as members of Dothideomycetes based on DNA sequence data (Lumbsch et al. 2005, Del Prado et al. 2006). Although the number of species is comparatively small, their placement can play an important link in determining how transitions to and from lichenisation influenced dothideomycete evolution. Trypetheliaceae known for its anastomosing, branched pseudoparaphyses was until very recently still placed within Pyrenulales, an ascohymenial order in Eurotiomycetes, based on bitunicate asci and lense-shaped lumina in the ascospores (Del Prado et al. 2006). Attempts to resolve members of this family remain challenging as they tend to occur on long, rapidly evolving branches in our phylogenetic analyses, which often lead to artifacts. Nelsen et al. 2009 (this volume) demonstrate the occurrence of two additional lichen-forming lineages within Dothideomycetes representing the families Strigulaceae and Monoblastiaceae. The delineation of lichenised family Arthopyreniaceae should continue to be assessed given their placement with a clade containing bambusicolous fungi (Tanaka et al. 2009; this volume) and their non monophyly is also confirmed elsewhere (Nelsen et al. 2009; this volume). The relationship between the lichenised groups and bambusicolous genera Roussoella and Roussoellopsis (Didymosphaeriaceae; Ju et al. 1996, Lumbsch & Huhndorf 2007) is strongly supported, but their affinity is not fully understood due to their considerable morphological differences.
The fungi collected from marine and freshwater habitats contain yet more varied species that have not been assessed well within a molecular based framework. Their diversity is supported by the fact that whole orders (Jahnulales) and several families, already mentioned, almost exclusively consist of species collected from these environments. A recent assessment of marine fungi tallied a number of more than 500 species with more than a fifth of these suggested to reside in Dothideomycetes (Jones et al. 2009). The number for fungi from fresh water habitats is somewhat lower (about 170 taxa).
Despite similarities in their preferred medium for spore dispersal (water) an examination of phylogenetic diversity within Dothideomycetes indicates that these groups of fungi tend to reside in divergent parts of the tree (Figs (Figs2, 2, ,3).3). However, some exceptions may occur: For example, members of Aigialaceae are weakly supported to share ancestry with members of freshwater clade Lindgomycetaceae (Raja et al. 2010). The Jahnulales represents another recently delineated aquatic lineage with an interesting mixture of fresh water and marine taxa. It was delineated based on molecular and morphological data (Inderbitzin et al. 2001, Pang et al. 2002) and now contains four genera and several species (Campbell et al. 2007). Previously, two anamorphic species in the Jahnulales, Xylomyces rhizophorae (described from mangrove wood of Rhizophora) and X. chlamydosporus have been reported from mangroves and thus saline habitats (Kohlmeyer & Volkmann-Kohlmeyer 1998). It has further been documented that X. chlamydosporus is the anamorph of Jahnula aquatica, a freshwater species (Sivichai, pers. comm.).
Marine Dothideomycetes generally exist in association with algae and plants in marine and brackish environments, usually with intertidal or secondary marine plants (e.g., mangroves). The majority of these fungi have been classified in families and genera that comprise mostly terrestrial species (e.g., Pleospora) and no definitive clades of marine Dothideomycetes have been identified. Here we find support for diverse aquatic lineages similar to the situation in Sordariomycetes. Papers by Suetrong et al. 2009 (this volume) and Shearer et al. 2009 (this volume) continue to address this disparity by using multigene phylogenies to describe several lineages within a class wide context. In contrast, many marine members of the Dothideomycetes await interrogation at the DNA sequence level, especially the genera Belizeana, Thalassoascus, Lautospora and Loratospora, all exclusively marine taxa.
The final environmentally defined group sampled in this volume is the bambusicolous fungi. More than 1 100 fungal species have been described or recorded worldwide from bamboo (Hyde et al. 2002). Furthermore, their ecological specialisation as pathogens, saprophytes, and endophytes has been relatively well documented (e.g. Hino 1961). However, relatively few studies based on DNA sequence comparisons have been undertaken for many bambusicolous fungi. Several unique lineages, e.g. the Katumotoa bambusicola-Ophiosphaerella sasicola clade in a freshwater lineage (Lentitheciaceae) and the Roussoella-Roussoellopsis clade close to lichen-forming families could be found (Fig. 2). Particularly, a new family Tetraplosphaeriaceae including five new genera characterised by a Tetraploa anamorph s. l. is introduced as a lineage of fungi with bamboo habitat (Tanaka et al. 2009; this volume). It is clear that much additional diversity within this group of fungi remains to be sampled using DNA sequence data
A number of other niches remain poorly discussed in this volume. Coprophilous fungi occur in three families Delitschiaceae, Phaeotrichaceae, and Sporormiaceae (Figs 2A, C). These families are not closely related and it is clear that the fimicolous life style has arisen more than once in the Dothideomycetes. Also, many species from these groups are not strictly dung-inhabiting, but can be found on other substrates like soil, wood, and plant-debris. Interestingly, some are human pathogens, plant endophytes and lichenicolous fungi. As is true throughout the Ascomycota, a change in substrate is apparently not a substantial evolutionary step in these taxa (Kruys & Wedin 2009).
Several orders e.g. Dothideales, Myriangiales and Microthyriales have not been treated using the extensive systematic sampling that is true for studies treated in this volume. However, individual smaller studies continue to provide interesting and surprising results. One such example is the first described meristematic and endoconidial species residing in Myriangiales (Fig. 2C) reported by Tsuneda et al. (2008). These Endosporium species were isolated from very different substrates such as: poplar twigs and a dead bird. They also have a close relationship to a single lineage of rock inhabiting fungi. The nutritional shifts represented by these closely related species correlate well with scenarios described by Ruibal et al. (2009; this volume) for rock inhabiting fungi. Another melanised meristematic fungus, Sarcinomyces crustaceus, isolated from pine trees appears in a similar position in a phylogeny presented in the aforementioned paper (Ruibal et al. 2009; this volume).
Another unusual species, Catinella olivacea is included in Fig. 2C, but without any clearly resolved position, diverging early to Dothideomycetidae. This species was initially placed in Leotiomycetes, due to their flattened apothecia, found on the underside of moist, well-decayed logs of hardwood. Asci are unitunicate but they appear to form after ascolocular development. As in the previous analysis, it was not possible to identify relationships between this species and any known order, although there are indications of a close relationship with the Dothideomycetidae (Greif et al. 2007).
The placement of the single asexual mycorrhizal lineage representing Cenococcum geophilum in the Dothideomycetes (LoBuglio et al. 1996), allied to members of the saprobic Gloniaceae is intriguing (Fig. 2B; Boehm et al. 2009a; this volume). No resolved placement for this species in Dothideomycetes has been possible in the past. The results of this study were also unexpected because no biological data suggest a connection to the family. Cenococcum is a fungus that is intensively used in environmental studies and this could suggest a very interesting biology for members of the ostensibly saprobic Gloniaceae. Results of this study advocate a more expansive sampling of Cenococcum in order to confirm this intriguing result.
One of the major obstacles in dothideomycete systematics remains the lack of a clear understanding of what species are members of the class based on morphology alone. Throughout most of the 20th Century, comparative morphological studies have been the only character on which to base phylogenetic relationships. The advent of large DNA-sequence data sets should allow for a substantially improved interpretation of morphological characters for this class of fungi. Studies in this volume and elsewhere have provided a clear understanding that many of the characters classically used in taxonomy and systematics of the group are homoplastic and not helpful for reconstructing phylogenetic relationships. Dothideomycete taxonomy also needs to keep pace with the rapid advances being made in phylogenetics, genomics and related fields. The important principle here is that our classification should communicate diversity accurately and allow dothideomycete biologists from disparate fields to have access to an agreed upon set of taxonomic names to aid communication. In addition, it should allow for a focus on under-sampled groups and clades (i.e. poorly sampled saprobes and others). A major task ahead will be to add asexual genera to present phylogenetic schemes, and integrate these into the existing familial and ordinal classification. As most of these asexual genera are in fact poly- and paraphyletic, their type species will need to be recollected to clarify their phylogenetic position. In addition to this, it appears that even some concepts of teleomorphic taxa will require extensive reconsideration. Finally, we should attempt to incorporate valuable biological information from past workers, such as the three mycologists to which this volume is dedicated, by reliably assessing culture and sequence identity. It is hoped that the papers in this volume will make a meaningful contribution towards these goals.
Authors from individual papers in this volume contributed to this work and specific acknowledgements to that regard can be found in individual papers. Work performed for this paper by the first author after 2008 was supported in part by the Intramural Research Program of the NIH, National Library of Medicine. Part of this work was also funded by grants from NSF (DEB-0717476) to J. W. Spatafora (and C.L. Schoch until 2008) and (DEB-0732993) to J.W. Spatafora and B. Robbertse.