Barrmaelia and Entosordaria in Barrmaeliaceae (fam. nov., Xylariales) and critical notes on Anthostomella-like genera based on multigene phylogenies

Phylogenetic analyses of a combined DNA data matrix containing ITS, LSU, rpb2 and tub2 sequences of representative Xylariales revealed that the genus Barrmaelia is a well-defined monophylum, as based on four of its described species (B. macrospora, B. moravica, B. oxyacanthae, B. rhamnicola) and the new species B. rappazii. The generic type of Entosordaria, E. perfidiosa, is revealed as the closest relative of Barrmaelia, being phylogenetically distant from the generic type of Clypeosphaeria, C. mamillana, which belongs to Xylariaceae sensu stricto. Entosordaria and Barrmaelia are highly supported and form a distinct lineage, which is recognised as the new family Barrmaeliaceae. The new species E. quercina is described. Barrmaelia macrospora, B. moravica and B. rhamnicola are epitypified and E. perfidiosa is lecto- and epitypified. Published sequences of Anthostomella and several Anthostomella-like species from the genera Alloanthostomella, Anthostomelloides, Neoanthostomella, Pseudoanthostomella and Pyriformiascoma are evaluated, demonstrating the necessity of critical inspection of published sequence data before inclusion in phylogenies. Verified isolates of several species from these genera should be re-sequenced to affirm their phylogenetic affinities. In addition, the generic type of Anthostomella should be sequenced before additional generic re-arrangements are proposed. Electronic supplementary material The online version of this article (doi:10.1007/s11557-017-1329-6) contains supplementary material, which is available to authorized users.


Introduction
Xylariaceae have long been treated in a conservative, morphology-based concept, and only informal subgroupings like Hypoxyloideae and Xylarioideae were accepted despite polyphyly of several genera. Recently, Wendt et al. (2017) subdivided Xylariaceae into three families based on multigene phylogeny of an ITS-LSU-rpb2-tub2 matrix. They resurrected and emended the family Hypoxylaceae, widened the Graphostromataceae to include the genera Biscogniauxia, Camillea, Obolarina and Vivantia, and restricted Xylariaceae mostly to genera with geniculosporium-like asexual morphs. This facilitates phylogenetic placement of other genera affiliated with Xylariaceae sensu lato. One example is the genus Anthostomella, which houses a number of species, whose morphological traits vary considerably and may, thus, be phylogenetically uninformative. Ascomata are usually immersed in the host tissue, covered by a clypeus or not, have amyloid or This article is part of the BSpecial Issue on ascomycete systematics in honour of Richard P. Korf who died in August 2016^.

Section Editor: Teresa Iturriaga and Marc Stadler
Electronic supplementary material The online version of this article (doi:10.1007/s11557-017-1329-6) contains supplementary material, which is available to authorized users. sometimes non-amyloid ascus apices and brown amerosporous ascospores with or without a hyaline appendage cell, with or without a gelatinous sheath. One major challenge to study them on hosts other than palms is the difficulty to spot them, as they cannot be collected regularly, and, often, they produce very limited material. Francis (1975) performed a study on the systematics of Anthostomella species on the stems and leaves of herbaceous plants and gymnosperms based on morphology alone. A similar but more voluminous study was carried out by Lu and Hyde (2000). Using a few newly collected specimens, Daranagama et al. (2015Daranagama et al. ( , 2016 determined that Anthostomella is polyphyletic within Xylariaceae and described several new genera. There is some confusion in the literature about the generic type of Anthostomella. Eriksson (1966) pointed out that lectotypification of Anthostomella with A. phaeosticta by Clements and Shear (1931) was in error and that A. limitata is the true generic type of Anthostomella. According to the ICN, this lectotypification is valid and has to be followed unless conservation with a different type is formally approved, and A. limitata is correctly listed as the generic type in Index Fungorum. In arguing that A. limitata does not exhibit several morphological characters then considered typical for the genus, Francis (1975) proposed A. tomicoides as the generic type, but this change has never been formally proposed and approved to become in effect. However, in the subsequent publications cited above, Francis (1975) was followed and A. tomicoides was accepted as the generic type. Neither A. limitata nor A. tomicoides have yet been sequenced.
Several genera have been segregated from Anthostomella or newly described, or subgenera were elevated to the generic rank. One of the latter is Lopadostoma (Jaklitsch et al. 2014) and another Entosordaria. The generic type of Entosordaria, E. perfidiosa, is characterised by non-amyloid asci and ascospores, which have a unique apical germ apparatus consisting of radiating slits (Eriksson 1966;Eriksson and Hawksworth 1986). Nonetheless, the genus was subsumed by Barr (1989) under Clypeosphaeria (see also Jaklitsch et al. 2016).
A transition to and now a member of the Diatrypaceae is the genus Anthostoma, which currently encompasses the single lignicolous species A. decipiens (Rappaz 1992;Jaklitsch et al. 2014). In a study designed to assess Anthostomella on hardwoods, Rappaz (1995) described the genus Barrmaelia, whose species, in part, also resemble Diatrypaceae, particularly in ascospore features, but, in contrast, have short-pedicellate asci and non-amyloid ascus apices. Furthermore, Barrmaelia species are typically characterised by ascomata that are immersed in the wood or bark and stromata that tend to blacken the host surface, in combination with light to dark brown, one-celled, smooth, ellipsoid to allantoid ascospores without sheath or appendages and with or without a germ slit. Rappaz (1995) combined six species in Barrmaelia (B. macrospora, B. moravica, B. oxyacanthae, B. picacea, B. pseudobombarda and B. sustenta) and described one new species, which he also selected as the generic type, B. rhamnicola. No new taxa have been added to this genus since then.
Although Rappaz (1995) only had morphology at hand, his concept withstands molecular phylogenetic analyses, as we show below. We, therefore, describe the new species B. rappazii to honour him, present the molecular systematics of five species of Barrmaelia and two of Entosordaria, including the new species E. quercina. The genera Barrmaelia and Entosordaria form a distinct lineage, which we name as the new family Barrmaeliaceae.

Isolates and specimens
All newly prepared isolates used in this study originated from ascospores of fresh specimens. The numbers of strains including NCBI GenBank accession numbers of gene sequences used to compute the phylogenetic trees are listed in Table 1. Isolates have been deposited at the Westerdijk Fungal Biodiversity Institute (CBS-KNAW), Utrecht, the Netherlands. Details of the specimens used for morphological investigations are listed in the Taxonomy section under the respective descriptions. Herbarium acronyms are according to Thiers (2017). Specimens have been deposited in the Fungarium of the Institute of Botany, University of Vienna (WU).

Culture preparation, growth rate determination and phenotype analysis
Cultures were prepared and maintained as described previously (Jaklitsch 2009). Microscopic observations were made in tap water, except where noted. Morphological analyses of microscopic characters were carried out as described earlier (Jaklitsch 2009). Methods of microscopy included stereomicroscopy using Nikon SMZ1500, Olympus SZX10 and Euromex Novex RZ 65.560, light microscopy using Euromex XHR MIC 625, Olympus BX51 and Nomarski differential interference contrast (DIC) using the compound microscopes Nikon Eclipse E600 and Zeiss Axio Imager.A1. Images and data were gathered with Nikon Coolpix 4500, Nikon DS-U2, Nikon D90, Olympus DP72 and Zeiss Axiocam 506 colour digital cameras and measured directly with the microscope, or with Olympus cellSens Dimension, NIS-Elements D v.3.0 and Zeiss ZEN Blue Edition softwares. Amyloidity of asci was assessed using Lugol or Melzer reagent. Measurements are reported as maximum and minimum in parentheses and the range representing the mean plus and minus the standard deviation of a number of measurements given in parentheses.

Analysis of sequence data
Following the phylogenetic placement of Barrmaelia macrospora within the Xylariaceae sensu lato clade in earlier analyses (Jaklitsch et al. 2014(Jaklitsch et al. , 2016, sequences of Barrmaelia and Entosordaria were analysed within the combined ITS, LSU rDNA, rpb2 and tub2 matrix of Wendt et al. (2017). As only a few tef1 sequences are available for Xylariales, this marker was not included in the matrix but the sequences were deposited at GenBank as a secondary barcode marker. To obtain a more representative taxon sampling, selected sequences were added to this matrix from Hernández-Restrepo et al.
(2016) and from Daranagama et al. (2015Daranagama et al. ( , 2016. From the latter two publications dealing with Anthostomella-like representatives, only accessions for which at least three of the four loci are available were included; before addition, it proved necessary to check these sequences with NCBI nucleotide BLAST searches for their correct gene and lineage identity, and obviously erroneous sequences as well as regions of poor sequence  Wendt et al. (2017). All alignments were produced with the server version of MAFFT (http://www.ebi.ac.uk/Tools/msa/mafft), checked and refined using BioEdit version 7.0.9.0 (Hall 1999). After exclusion of ambiguously aligned regions and long gaps, the final matrix contained 4668 nucleotide characters, i.e. 600 from the ITS, 1359 from the LSU, 1162 from rpb2 and 1547 from tub2. Maximum parsimony (MP) analysis of the combined matrix was performed using a parsimony ratchet approach. For this, a nexus file was prepared using PRAP v.2.0b3 (Müller 2004), implementing 1000 ratchet replicates with 25% of randomly chosen positions upweighted to 2, which was then run with PAUP v.4.0a151 (Swofford 2002). The resulting best trees were then loaded in PAUP and subjected to heuristic search with TBR branch swapping (MULTREES option in effect, steepest descent option not in effect). Bootstrap analysis with 1000 replicates was performed using five rounds of replicates of heuristic search with random addition of sequences and subsequent TBR branch swapping (MULTREES option in effect, steepest descent option not in effect) during each bootstrap replicate. In all MP analyses, molecular characters were unordered and given equal weight; analyses were performed with gaps treated as missing data; the COLLAPSE command was set to minbrlen.
Maximum likelihood (ML) analyses were performed with RAxML (Stamatakis 2006) as implemented in raxmlGUI 1.3 (Silvestro and Michalak 2012), using the ML + rapid bootstrap setting and the GTRGAMMA substitution model with 1000 bootstrap replicates. The matrix was partitioned for the individual gene regions, and substitution model parameters were calculated separately for them.

Assessment of published sequences
NCBI Nucleotide BLAST searches revealed serious problems for some sequences of Daranagama et al. (2015), which were, therefore, excluded from the analyses (Table 1). LSU sequences KP340547 (Anthostomella helicofissa) and KP340538 (Anthostomelloides forlicesenica) were not added to the matrix, as they did not correspond to the LSU part (ca. 540 bp) included in the ITS sequences KP297406 and KP297396 of the same accessions. Whereas LSU sequence KP340547 was revealed as xylarialean by BLAST searches but differed in 60 positions (3 gaps and 57 substitutions) from the LSU part of KP297406, a BLAST search of KP340538 revealed various Pleosporales (Kalmusia, Coniothyrium, Dendrothyrium) as the closest match (84% sequence similarity). Therefore, for these two accessions, only the LSU part of the ITS sequences was included in the LSU matrix. rpb2 sequence KP340524 (Anthostomelloides forlicesenica) was excluded as well, as a BLAST search also revealed pleosporalean affinities (80% similarity to sequence LK936413 of Leptosphaerulina chartarum, 77% similarity to sequences DQ677970 of Phaeodothis winteri and DQ677956 of Coniothyrium palmarum). tub2 sequences KP406614 (Anthostomella formosa) and KP406616 (Anthostomella obesa) were also excluded, as BLAST searches actually revealed them as rpb2 sequences. This was also confirmed in an alignment containing the rpb2 sequences included in the present study, where they were highly similar to rpb2 sequences of various Anthostomella species (not shown); however, both were different from the rpb2 sequences KP340531 and KP340533 published for the same isolates in the same publication.

Molecular phylogeny
Of the 4668 nucleotide characters of the combined matrix, 2210 are parsimony informative (338 of ITS, 422 of LSU, 638 of rpb2 and 812 of tub2). Figure 1 shows a simplified phylogram of the best ML tree (lnL = −136212.706) obtained by RAxML. Maximum parsimony analyses revealed four MP trees 32,311 steps long, which were identical except for a polytomy within the three terminal taxa of Anthostomella and an unresolved position of Hypoxylon ochraceum and H. pilgerianum relative to each other; the strict consensus tree of the four MP trees is provided in the Supplementary Information. The backbone of the MP trees was similar to the ML tree, except for a sister group relationship of Lopadostomataceae and Diatrypaceae and a slightly different position of the Calceomyces-Neoanthostomella clade; in addition, there were a few minor topological differences within the Xylariaceae and Graphostromataceae.
All families received high to maximum support in all analyses, as did the Xylariaceae sensu lato (Fig. 1). The genera Barrmaelia and Entosordaria were revealed as the closest relatives with maximum support but formed a separate lineage within Xylariaceae sensu lato, and are classified here within the new family Barrmaeliaceae. Within the Xylariaceae sensu lato, the basal position of Hypoxylaceae was highly supported, but the phylogenetic relationships between the other three families (Barrmaeliaceae, Graphostromataceae and Xylariaceae sensu stricto) remain uncertain due to the lack of significant backbone support. Clypeosphaeria mamillana is revealed as the closest relative of Anthostomelloides krabiensis with high (99% MP BS) to maximum (ML) support, and both are sister clade to the rest of the Xylariaceae sensu stricto with high support as well (Fig. 1). The second species of Anthostomelloides, A. forlicesenica, is not closely related to A. krabiensis but sister species of Brunneiperidium gracilentum within Xylariaceae sensu stricto with high (98% MP BS) to maximum (ML) support. The genera Anthostomella and Pseudoanthostomella are placed outside Xylariaceae sensu lato and form a highly supported lineage; sister group relationship to the highly supported Calceomyces-Neoanthostomella clade is revealed with medium support only in the ML analyses. Pyriformiascoma trilobatum is placed within Microdochium with maximum support in both analyses. prosenchymatous, poorly developed, without KOHextractable pigments. Ascomata (perithecia) globose, sometimes raising the substrate, singly, in small groups or gregarious. Peridium melanised, pseudoparenchymatous to prosenchymatous. Hamathecium of numerous persistent, hyaline, septate paraphyses. Asci eight-spored, cylindrical, persistent, with inamyloid or infrequently amyloid apical ascus apparatus. Ascospores yellow to dark brown; unicellular with or without germ slit (Barrmaelia), or two-celled with septum near one end, the small cell hyaline, the large cell dark brown and with an apical germ apparatus consisting of radial slits (Entosordaria); allantoid or ellipsoid, inequilateral, slightly inequilateral or nearly equilateral, with narrowly or broadly rounded ends. Asexual morph libertella-like where known (Barrmaelia; Rappaz 1995 Stromata mostly in wood, usually discolouring the wood surface grey to black, entostroma poorly developed (Rappaz 1995). Ascomata perithecial, immersed in wood or bark, rarely erumpent, often blackening the host surface, globose, ellipsoid or pyriform; ostiolar pore rounded. Peridium melanised, pseudoparenchymatous to prosenchymatous. Hamathecium of apically free, hyaline paraphyses. Asci unitunicate, cylindrical, with a short stipe, generally eight-spored; with an inamyloid apical apparatus. Ascospores light to dark brown, one-celled, smooth, asymmetrically ellipsoid to allantoid, without sheath or appendages, with or without a germ slit. Asexual morph (fide Rappaz 1995) libertella-like, only known from pure culture, conidiomata globose, more or less melanised, up to 1 mm in diam., exuding the conidia in white to pinkish droplets. Conidiophores erect, branched. Conidiogenous cells hyaline, conical or cylindrical, arranged in palisades, apex sometimes with faint annellations. Conidiogenesis holoblastic, proliferation percurrent or sympodial. Conidia hyaline, falcate, one end truncate, the other rounded or slightly acute.
Notes: As we did not observe an asexual morph in pure culture, its description is adapted from Rappaz (1995).
Colonies on CMD and MEA white; aerial hyphae abundant. No asexual morph observed.
Distribution: Europe (Czech Republic, France, Germany, Netherlands, Norway, Sweden, Switzerland, United Kingdom), possibly also the USA (fide Rappaz 1995 Notes: For synonyms, see Rappaz (1995). Concerning typification, Nitschke (1867) only cited material from Handorf on Sarothamnus in his protologue. In their list of type specimens of Nitschke deposited in B, Gerhardt and Hein (1979) mention two envelopes mounted on a sheet without a place or date on the envelopes. However, the holotype B 70 0009297 now only contains a single envelope with an asexual morph with hyaline conidia, i.e. no sexual morph is present. Therefore, epitypification became necessary. Rappaz (1995) selected B 70 0009349 as the lectotype, but that material was not cited in the protologue. It is, however, authentic material of Valsa macrospora (collected by Nitschke before publication), as both Nitschke and Rappaz considered it to be the fungus described in the protologue.
Barrmaelia macrospora is usually easy to identify due to its large and relatively narrow ascospores with one lighter coloured side. The inconspicuous germ slit was best visible in B 70 0009349 (Fig. 2t). Cannon (2015) provides a description of a slightly deviating British collection with larger, occasionally one-septate ascospores measuring (23.5-)26-29 × 7-8.5 μm, which may represent a distinct species.
Colonies on CMD and MEA white; aerial hyphae abundant. No asexual morph observed.
Habitat: In bark of thin dead branches of Salix caprea attached to the tree.
Distribution: Europe (Austria, Czech Republic, Spain). Typification. Lectotype of Eutypa moravica designated by Rappaz (1995) Notes: This species is well characterised by its light brown, allantoid and relatively small ascospores. Barrmaelia rappazii is superficially similar but differs morphologically mainly by larger and darker brown ascospores, and in having effused, black stromata with sparsely distributed perithecia. Barrmaelia rhamnicola also has allantoid ascospores but they are larger, filled with bigger oil drops and have a slightly darker colour, and it occurs on a different host.
Colonies on CMD and MEA white; aerial hyphae abundant. No asexual morph observed.
Habitat: In wood of twigs and branches of various hardwoods.
Distribution: Widespread (Africa, Asia, Europe and North America); for details, see Cannon and Minter (2013).
Notes: For synonyms, see Rappaz (1995). He found a libertella-like asexual morph in pure culture. Rappaz (1995) recognised three groups based on ascospore size within his broad concept of B. oxyacanthae. The first group with the smallest ascospores (Bmean length between 12.5 and 13^) contains the type of B. oxyacanthae and agrees very well with GZU 000317702, whose mean length of 12.6 μm (n = 30) corresponds exactly with our measurements of the type collection. The sequenced collection WU 36925 has a mean length of 13.3 μm (n = 31) and, thus, appears to be an intermediate between the first and second group, the latter of which has a Bmean length between 13.5-14^. The group with the longest ascospores (Bbetween 14.5-15^) is said to mostly contain material on Salix. Of the three collections studied on this substrate, WU 36925 belongs to either the first or second group (see above), whereas GZU 000317705 falls in the second group (mean length 13.6 μm, n = 30) and GZU 000317701 best fits in the third group (mean length 14.3 μm, n = 30). GZU 000317700 does not contain mature ascomata. GZU 000317704 and GZU 000317703 from South Dakota (USA) have mean lengths of 16.2 and 19.6 (n = 30), respectively; thus, they likely represent different, probably undescribed, species.
Barrmaelia oxyacanthae differs from other Barrmaelia species in its relatively dark brown, ellipsoid ascospores with a well-visible germ slit. It is most similar to B. pseudobombarda, which has narrower ascospores (Rappaz 1995;Mathiassen et al. 2015). Cannon and Minter (2013) give a morphological description and illustrations of B. oxyacanthae and details on its ecology and distribution.
Colonies on CMD and MEA white; aerial hyphae abundant. No asexual morph observed.
Habitat: In wood of twigs and branches of Populus tremula.
Distribution: Europe, only known from the type location in Norway.
Colonies on CMD and MEA white; aerial hyphae abundant. No asexual morph observed.
Habitat Notes: Barrmaelia rhamnicola is distinguished from other species of the genus by the often slightly curved, relatively large ascospores, which are filled with conspicuous oil drops and lack a germ slit. For comparison with the other allantoidspored species without germ slit, see notes under B. moravica and B. rappazii. Rappaz (1995) observed a libertella-like asexual morph in pure cultures.
Type species: Entosordaria perfidiosa (De Not.) Höhn. Ascomata perithecial, scattered, immersed to erumpent, depressed globose to ellipsoid, circular in transverse section. Peridium brown. Hamathecium of apically free, thin, sparsely branched paraphyses. Asci unitunicate, cylindrical, with uniseriate ascospores; apex inamyloid without distinct ring or amyloid with a discoid ring. Ascospores two-celled with septum near one end, the small cell hyaline, the large cell dark brown and with an apical germ apparatus consisting of radial slits. Asexual morph unknown.
Notes: Entosordaria was first described as a subgenus of Anthostomella (Saccardo 1882) and raised to the generic rank by Höhnel (1920), with E. perfidiosa as the generic type. Eriksson (1966) outlined the fundamental morphological differences from Anthostomella, i.e. inamyloid asci and dorsiventrally flattened ascospores with an apical germ apparatus consisting of radiating slits. He confined Entosordaria to the generic type and removed the genus from the Xylariaceae. Later, he (in Eriksson and Hawksworth 1986) argued that Stereosphaeria is the valid generic name to be used, considering Entosordaria (Sacc.) Höhn. to be a younger heterotypic homonym of Entosordaria Speg. However, Entosordaria Speg. has not been validly described according to ICN Art. 38.1, as Spegazzini (1910) neither provided a diagnosis nor a reference to a previous valid description. Therefore, Entosordaria (Sacc.) Höhn. remains available and, based on priority, is the generic name to be used. Barr (1989) classified E. perfidiosa in Clypeosphaeria, based on similarities of their ascospores, apical ascus apparatus, ascomata, clypei and peridium structure. However, molecular phylogenies do not support a close relationship, as the generic type, Clypeosphaeria mamillana, is placed in Xylariaceae s. str. with high support (Fig. 1).
With the addition of the closely related E. quercina, the genus Entosordaria also includes a species with an amyloid ascus ring, which shows that this character is not suitable for generic classification within Xylariales.
Colonies on CMD and MEA white; aerial hyphae abundant. No asexual morph observed.
Habitat: In bark of old trunks of living Acer pseudoplatanus.
Distribution: Europe. Notes: Entosordaria perfidiosa is well characterised by the ascospores with an apical germ apparatus consisting of radiating slits in combination with inamyloid asci and its growth on bark of old Acer pseudoplatanus trees. It has been classified in Clypeosphaeria by Barr (1989); however, it is only distantly related with C. mamillana, the generic type (see Fig. 1). For comparison with E. quercina, see below.
Two syntypes of Sphaeria perfidiosa are present at RO, which were studied in detail by O. Eriksson (see Eriksson and Hawksworth 1986), but he did not select a lectotype. Type specimens at RO are no longer sent out for study, but detailed photographic documentation of the two syntypes was generously provided by Mrs. A. Millozza (pers. comm.). Based on the abundance of ascomata, we select no. 413 as the lectotype. For nomenclatural stability, a recent Austrian collection for which a culture and DNA sequences are available is designated as the epitype.
Colonies on CMD white, on MEA a reddish and yellowish pigment developing; aerial hyphae abundant. No asexual morph observed.
Habitat: In bark of dead twigs of Quercus coccifera. Distribution: Only known from the type locality in Crete (Greece).
An evaluation of published sequences reveals that sequence data quality should be critically checked by BLAST searches and detailed inspection of alignments before inclusion into phylogenetic analyses. An indicator for problems in the sequence data used for phylogenetic analyses are exceptionally long branch lengths in phylograms like, for example, those seen for some clades in Daranagama et al. (2016). Marked topological differences between our analyses and those of previous publications (Daranagama et al. 2015(Daranagama et al. , 2016, but also between the latter, may, at least partly, be due to the inclusion of obviously inaccurate, dubious or erroneous sequences which have been identified and removed from our matrix. These errors may cast general doubts on the accuracy of the sequences published for these species, and all markers should be re-sequenced from verified material to corroborate their phylogenetic affinities.