Corrections to recent changes in the taxonomy of the Sordariales

The classification of taxa belonging to the Sordariales has been problematic over the years. With the beginning of the DNA era, ascospore morphology, which was the main criterium for the delimitation of taxa in the Sordariales, was demonstrated to not be useful for inferring taxonomic relationships especially at the genus level. In the past decades, the combination of both morphological and molecular data allowed the reclassification of these taxa. Recently, a study of some often overlooked Diaporthomycetidae and Sordariomycetidae included a new taxonomic classification for members of the Sordariales, many of which were based on nomenclatural errors or which lacked sufficient data to support their hypotheses. The authors did not contribute any new DNA sequences, but instead relied on datasets generated by previous authors in their published phylogenetic studies. Surprisingly, different results were obtained contradicting these previous studies and, in an act of taxonomic vandalism, five new families were introduced without performing further molecular analyses to verify the incongruencies with these previous studies. Three of these new families, which we consider doubtful, are Bombardiaceae, Lasiosphaeridaceae and Zygospermellaceae. The family Strattoniaceae is here considered superfluous since it was introduced to accommodate only a single genus and delimited based on a species that is not the type species of Strattonia. The Neoschizotheciaceae was erected based on the new genus Neoschizothecium, which was introduced to accommodate members of Schizothecium since Huang et al. (2021) considered Schizothecium as a synonym of Podospora after misinterpreting their type species as the same. However, Schizothecium and Podospora have been two independent genera based on two different type species for half a century, making Neoschizothecium and Neoschizotheciaceae superfluous. Moreover, they proposed 32 new combinations, 16 of which are now superfluous or doubtful. Most of these taxonomic errors could have been avoided if a proper literature review had been performed. Two examples are the new superfluous combinations of Triangularia tarvisina and Cladorrhinum olerum, because the former is considered conspecific with Triangularia setosa, and the latter conspecific with Cladorrhinum foecundissimum, the anamorph of Arnium olerum. The focus of the current review is to provide a scientifically responsible alternative to the erroneous novelties proposed at the family, genus and species level in the recent classification of Sordariales.


Introduction
The order Sordariales is one of the most diverse groups in the Sordariomycetes and the classification of their members based solely on morphology has been problematic (Hawksworth and Eriksson 1986;Miller andHuhndorf 2004a, 2005;Huhndorf et al. 2004;Kruys et al. 2015). The taxonomic classification of taxa belonging to this order was traditionally based on ascospore morphology; however, Miller and Huhndorf (2005) demonstrated that this character is extremely homoplastic and not useful for inferring taxonomic relationships. Instead, the ascomatal wall morphology appears to be more phylogenetically informative, even though it has its limitation since not all taxa exhibit distinctive features in their ascomatal walls, and/or similar ascomatal wall characteristics are found in different monophyletic lineages.
For nearly two decades, extensive molecular studies have been combined with morphological data to delimit sordarialean taxa at the family, genus and species level (Miller and Huhndorf 2004a;Cai et al. 2005;Kruys et al. 2015;Wang et al. 2019a;Marin-Felix et al. 2020). Historically, the Sordariales has contained 7 to 14 families, depending on the authors' concepts (Hawksworth and Eriksson 1986;Eriksson et al. 2001). Huhndorf et al. (2004) restricted the order to 3 families, i.e. Chaetomiaceae, Lasiosphaeriaceae and Sordariaceae, the first two of which were considered polyphyletic based on molecular data. The Chaetomiaceae has been extensively studied in recent years based on a polyphasic approach combining morphological and molecular data, resulting in its delimitation as a monophyletic lineage (Wang et al. 2019a, b). However, the polyphyly of Lasiosphaeriaceae, which was the largest family of the order, remained problematic until it was recently partly resolved . Kruys et al. (2015) demonstrated the separation of the family into four different clades. Subsequently, Wang et al. (2019a) introduced the new family Podosporaceae to accommodate taxa belonging to clade IV in Kruys et al. (2015). Marin-Felix et al. (2020) delimited Lasiosphaeriaceae to clade III, which contained the type genus Lasiosphaeria. Moreover, the new families Diplogelasinosporaceae, Naviculisporaceae and Schizotheciaceae were introduced to accommodate three monophyletic lineages containing taxa resembling those in the Lasiosphaeriaceae. Once again, the morphological characters used to delimit these families turned out to be homoplastic and appeared independently in multiple lineages. Other lasiosphaeriaceous taxa were nested in an unsupported lineage (Lasiosphaeriaceae s. lat. I, Fig. 1), and therefore a new family was not introduced for this clade. Finally, another monophyletic lineage with two genera was shown as an unsupported sister group to Schizotheciaceae (Lasiosphaeriaceae s. lat. II, Fig. 1). However, a new family was not introduced until further studies including additional taxa and molecular data could be performed. Figure 1 shows the phylogenetic tree obtained from the study performed by Marin-Felix et al. (2020).
The genera belonging to this order have also been taxonomically challenging to delineate over the years. For example, the genera Gelasinospora and Neurospora, both in the Sordariaceae, were established based on different patterns of ascospore ornamentation. García et al. (2004) demonstrated that this character was not phylogenetically inform a t i v e a n d s y n o n y m i z e d G e l a s i n o s p o r a u n d e r Neurospora. This observation agreed with Miller and Huhndorf (2005), who found that ascospore morphology was a homoplastic character throughout the order. The largest genera in the Chaetomiaceae have also been recently refined, i.e. Chaetomium, Humicola and Thielavia, resulting in the introduction of 17 new genera and more than 70 new combinations to accommodate taxa not included in the molecular-based monophyletic lineages of these recircumscribed genera (Wang et al. 2016a(Wang et al. , 2019a. Genera traditionally placed in Lasiosphaeriaceae are currently being studied. For example, Wang et al. (2019a) delimitated Cladorrhinum, Podospora and Triangularia, whereas Marin-Felix et al. (2020) established 5 new genera, i.e. Areotheca, Lundqvistomyces, Pseudoechria, Pseudoschizothecium and Rhypophila. These taxonomic changes were done to accommodate taxa previously placed in Cercophora, Podospora and Triangularia based on morphology, but that no longer occurred in the more narrowly defined, DNA-based monophyletic lineages representing these genera.
Recently, Huang et al. (2021) performed a phylogenetic study based on ITS, LSU, RPB2 and TUB2 sequences already available in GenBank, without generating a single new sequence. As a result, they introduced 5 new families, 1 new genus and 32 new combinations in the order Sordariales. However, most of these taxonomic novelties are based on errors of interpretation or lack sufficient data for their creation (Table 1). In order to compare the contradictory results between the phylogenetic study of Huang et al. (2021) and previous studies (Kruys et al. 2015;Marin-Felix et al. 2020), we tried to reanalyze the original molecular alignments generated by Huang et al. (2021). However, these alignments are not available in TreeBase even though they stated in their publication that final alignments and trees were deposited in TreeBase. Therefore, we had to download all sequences from GenBank to perform a phylogenetic analysis that included all sequences previously used in Marin-Felix et al. (2020) and Huang et al. (2021). No sequences in Chaetomiaceae were included since no novelties were proposed in this family ( Table 2). The maximum likelihood (ML) and Bayesian inference (BI) analyses including the four loci were performed as described by Harms et al. (2021). The lengths of the individual alignments used in the combined dataset were 556 bp (ITS), 916 bp (LSU), 978 bp (RPB2) and 618 bp (TUB2), and the final combined alignment was 3068 bp. A poorly aligned ambiguous region was manually deleted from the ITS1 region. Figure 2 shows the most likely tree obtained from the RAxML analysis of the combined dataset generated in our study. It is identical in the topology of the 95% majority-rule consensus tree generated by the Bayesian analysis. It is unfortunate that Huang et al. (2021) did not indicate the GenBank accession numbers of the sequences they used to produce their trees, making it very difficult for *These 15 taxa are nomenclaturally invalid # These six taxa are not accepted by the present authors due to insufficient data the scientific community to reproduce their results. It should have been mandatory to indicate which sequences were incorporated in their phylogenetic study, as well as to cite the underlying papers giving proper attribution to the authors that generated those data, as recommended in Aime et al. (2020), a recent ICTF-sponsored paper on best practices in taxonomy. Surprisingly, the second author of Huang et al. (2021) is also a co-author of the Aime et al. New families superfluously introduced or without sufficient evidence The complex nomenclatural histories surrounding the type species of Podospora and Schizothecium have been discussed in detail (Lundqvist 1972;Ament-Velásquez et al. 2020;Vogan et al. 2021 fig. 105, Jul 1838). This segregation, which was based on morphological data and discussed by Lundqvist (1972), was later supported by molecular data (Cai et al. 2005). Wang et al. (2019a) overlooked the conserved type of Podospora (i.e. Sordaria fimiseda Ces. & De Not.) and incorrectly cited S. fimicola Corda as type species. To further complicate matters, Wang et al. (2019a) designated an epitype specimen (CBS H-24048) for S. fimicola that was induced to produce fertile ascomata in culture (CBS 482.64), but this epitype represents P. fimiseda, not S. fimicola! We have examined the ex-epitype culture (CBS 482.64) and could not reproduce the fruiting of ascomata in culture. The epitype specimen was also examined, but no ascomata were found to study their morphology in detail, despite images of mature ascomata fruiting in culture provided in Wang et al. (2019a).
According to the pictures of the epitype shown by Wang et al. (2019a), this specimen produces hyphal-like ascomatal hairs surrounding the entire ascoma, which match P. fimiseda (Miller 2003). Schizothecium fimicola has swollen agglutinated ascomatal hairs only in the upper part of the ascomata. Thus, the type species of Schizothecium should be conserved as S. fimicola (typ. cons. pending). The ascomal wall of CBS 482.64 needs to be studied to verify that it is pseudobombardioid and matches the description of P. fimiseda (Miller 2003) and, therefore, it is a suitable epitype for this later species. If the conservation proposal to change the type species of Podospora from P. fimiseda to P. anserina (Vogan et al. 2021) is accepted, this would also bring additional resolution to the complicated and often confused nomenclatural history of these two genera. Since Podospora and Schizothecium are clearly not synonyms as Wang et al. (2019a) and Huang et al. (2021) believed, there was no reason to create a new family (i.e. Neoschizotheciaceae) for Schizotheciaceae. Thus, Neoschizotheciaceae is nomenclaturally illegitimate (Art. 14.3, Shenzhen Code) and a superfluous synonym of Schizotheciaceae.
It is worthwhile to mention that one should always attempt to make decisions that lead to as few taxonomic name changes and reduce the possibility of publishing taxonomically superfluous names, as recommended in the recent guidelines for publishing new fungal species or names, version 3.0 . This is especially important in this group of fungi, in which many names are not represented by DNA sequences of type material and many new combinations are possible. Therefore, the proposal to introduce a new genus with numerous subsequent new combinations for an established lineage should have never been accepted and was easily avoidable if a review of the literature was performed.
The new family Strattoniaceae was erected to accommodate the genus Strattonia. Surprisingly, the genus was redefined without the study of any material of the type species (St. tetraspora), but based only on the already available sequences of the type strain of another species, St. oblecythiformis. The redefinition was made only on morphological characters even though it has already been clearly demonstrated that taxonomists must be careful proposing any taxonomic reclassification of members of the Sordariales and that a polyphasic approach based on both morphological and molecular data is required (Miller and Huhndorf 2004a;Kruys et al. 2015;Marin-Felix et al. 2020). It is of note that no cultures of St. tetraspora are available in any recognized culture collection. Therefore, it is of upmost importance to recollect this species and include it in phylogenetic analyses to confirm the monophyly of Strattonia. Huang et al. (2021) argued that the characteristics of Strattonia should be reduced to those species producing ascospores with an upper cell surrounded by a gelatinous sheath. On the other hand, taxa that have no sheath have been demonstrated to belong to different genera, i.e. Jugulospora     -Cai et al. (2005) and Lundqvistomyces . Therefore, the presence or absence of an ascospore sheath is polyphyletic and one cannot assume that all species with ascospores surrounded by a sheath are going to belong to the same monophyletic lineage. In fact, other species characterized by ascospores surrounded by a gelatinous sheath are placed in different genera, such as Echria (Schizotheciaceae), Fimetariella (Lasiosphaeriaceae s. lato) and Sordaria (Sordariaceae). This family is erected based on a clade containing only two species, S. oblecythiformis and the new combination S. petrogale, but both species only contain sequences from the internal transcribed spacer region (ITS) and the nuclear rDNA large subunit (LSU) from the study of Vu et al. (2019). For these reasons, we reject the family Strattoniaceae until additional taxa, including the type species, and additional genes (i.e. RPB2 and TUB2) have been studied. The families Lasiosphaeridaceae and Zygospermellaceae, erected for only one and two genera, respectively, are also points of contention since they were proposed based on limited taxon sampling and poor taxonomic practices. In the recent phylogenetic studies from Kruys et al. (2015) and Marin-Felix et al. (2020), both clades now considered independent families by Huang et al. (2021) were clustering together in a well-supported clade. Huang et al. (2021) prematurely introduced both families since they formed two independent lineages in their phylogenetic study. However, both families are included in the Schizotheciaceae clade in our Bayesian phylogenetic study (0.97 pp) (Fig. 2). More taxa and sequences should be incorporated in further phylogenetic analyses to verify if the Lasiosphaeridaceae and Zygospermellaceae are two independent lineages or belong to the Schizotheciaceae. For these reasons, we reject these families until more data can be analysed and we refer to both lineages as Lasiosphaeriaceae s. lato.
In our phylogenetic study, the only family not supported is Lasiosphaeriaceae s. str. However, the conflicts may be caused by the high number of taxa in which only the ITS and LSU sequences are available, which most likely influences the backbone relationships and support for these phylogenetic relationships resulting in misleading higher-level classifications. Although most strains included in Huang et al. (2021) are available in the CBS collection, and therefore additional loci could have been sequenced, they chose not to generate these critical sequence data for their phylogenetic analyses.
Finally, the family Bombardiaceae was introduced by Huang et al. (2021), even though the ML bootstrap support was only 76% and there was no Bayesian inference support for this clade. Huang et al. (2021) explained that Marin-Felix et al. (2020) discussed the low support of this lineage, and for that reason no family was introduced to accommodate it. Nonetheless, they considered this lineage to be wellsupported in their phylogenetic study, and therefore the new family Bombardiaceae was introduced to represent it. In our phylogenetic study (Fig. 2), which included the same sequences that Huang et al. (2021) used, the clade representing this family is not supported (42% bs/0.89 pp), so we do not accept the Bombardiaceae as a justifiable family. We take this opportunity to clarify that most of the outgroups used in Huang et al. (2021) do not belong to the family Microascales as they erroneously indicated, but rather to the Coronophorales (syn. Melanosporales) as in the case of all the Microthecium spp. included, as well as Thielavia basicola.
New genera superfluously introduced or redefined based on poor taxonomic practice As mentioned above, Neoschizothecium is a superfluous genus based on misinterpretation of the type species of Podospora and Schizothecium so it is nomenclaturally illegitimate (Art. 14.3, Shenzhen Code) and thus, a superfluous synonym under Schizothecium. The delimitation of Strattonia based on a single species is a poor taxonomic decision since the type species of the genus was not included in the study of Huang et al. (2021). Finally, the delimitation of Cercophora is also based on poor taxonomic practice since Huang et al. (2021) chose a dubious candidate to represent the type species of C. mirabilis in their analyses but thankfully, they did not designate an epitype. It is impossible to confirm the morphological identification of this strain (CBS 120402) because it does not sporulate in culture as mentioned by Marin-Felix et al. (2020). Therefore, it is not a suitable representative of the type species of Cercophora and a suitable epitype must be studied and designated before this genus can be properly delimited. The placement of Cercophora spp. in different lineages and families support the necessity of a detailed study of this genus and the careful designation of Fig. 2 (continued) an epitype for the type species in order to determine the proper taxonomic placement of Cercophora within the order.

New combinations superfluously introduced or failing to follow good taxonomic practice
The new combination of Cladorrhinum olerum proposed to accommodate Arnium olerum is superfluous because Marin-Felix et al. (2020) already synonymized this latter name under Cl. foecundissimum. The nucleotide similarity between both species is 99.81% for ITS, 100% for LSU, 98.60% for RPB2 and 100% for TUB2, suggesting that these represent the asexual and sexual morphs of the same taxon.
The new combination T. tarvisina is superfluous since P. tarvisina has been considered a synonym of T. setosa for 50 years based on their morphological similarity (Lundqvist 1972). This is corroborated based on the high nucleotide similarity greater than 99.5% of the four loci (ITS, LSU, RPB2 and TUB2) between both species according to a BLAST comparison. Therefore, this new combination was made in error due to the failure of a proper literature review by Huang et al. (2021). Likewise, the new combination T. praecox is also superfluous since this species shows morphological and molecular similarity also with T. setosa, and therefore P. praecox is here considered a synonym to this later species. The description of T. setosa is here emended to incorporate the sizes observed in P. praecox when it was introduced, which are similar to T. setosa except for the ascus dimensions (up to 190 μm wide in P. praecox vs. up to 60 μm in T. setosa) (Cailleux 1969;Lundqvist 1972;Wang et al. 2019a Triangularia tarvisina (Sacc.) S.K. Huang & K.D. Hyde, Fungal Divers. 111: 515. 2021. Podospora praecox Cailleux, Cahiers de La Maboké 7: 102. 1969. Triangularia praecox (Cailleux) S.K. Huang & K.D. Hyde, Fungal Divers. 111: 515. 2021.
As mentioned above, the genus Strattonia is delimited based on the non-type species S. oblecythiformis, which should not be accepted due to the difficulty of delimiting genera in the Sordariales based on only morphological data (for further details, see section "New families superfluously introduced or without sufficient evidence"). For this reason, the new combination S. petrogale should not be taken up until further studies that include the type species confirm the correct taxonomic placement of the lineage representing Strattonia.
The nine new combinations in the newly erected genus Neoschizothecium proposed for accommodating the species previously belonging to Schizothecium are also here considered superfluous for the same reasons the family Neoschizotheciaceae is invalid. Moreover, N. minicauda was introduced to accommodate P. minicauda. However, the strain included in the phylogenetic study does not represent type material and its morphology was not studied. Therefore, the transfer of this species to Schizothecium should await further studies.
Cladorrhinum brunnescens is transferred to Podospora as P. brunnescens. However, in our phylogenetic study, this species is not placed in the well-supported clade (99% bs/1 pp) representing Podospora. Only ITS and LSU sequence data of the type strain are available. Therefore, the other two loci should be sequenced and additional analyses conducted before this combination is accepted.
Finally, two additional new combinations that followed poor taxonomic practices are Cl. leucotrichum and T. nannopodalis, which were proposed based on strains that do not represent type material and whose morphology was not studied.
Limitation in the morphological study Huang et al. (2021) performed morphological studies of type material for several taxa belonging to the Sordariales, although they mostly repeated what can already be found in the literature without providing any new information while at the same time destroying irreplaceable type material. Examination of type material is important, but generating sequence data from types or designating epitypes with molecular data is a higher priority. No molecular data were generated in Huang et al. (2021). Due to the difficulty of delimiting species and genera of this order based only on morphology, it is much more important to generate sequences from these types of materials to incorporate in phylogenetic analyses instead of re-examining material that has previously been studied and well documented. This is the case of Triangularia horridula, a new combination proposed after a morphological and molecular study of the holotype from the fungarium of Saccardo (Forin et al. 2021). Using next-generation sequencing techniques, the generation of molecular data from type material is possible and essential to improve taxonomic classifications.

Conclusions and final remarks
In conclusion, it is clear that the authors of Huang et al. (2021) were not able to fully appreciate or understand the problems surrounding the reclassification of the order Sordariales, which has been studied using a combination of meticulous morphological and molecular analyses by various workers for more than a decade. Even in the last few years, chemotaxonomic data have been combined with morphological and molecular data to verify some taxonomic relationships. For example, Shao et al. (2020) demonstrated that different specimens and strains of Jugulospora produced the same xanthoquinodins compounds, corroborating the redefinition of the genus performed by Marin-Felix et al. (2020). Moreover, the new genus Pseudorhypophila was introduced based on the production of the same class of compounds, i.e. zopfinol and derivatives, by members of the same monophyletic lineage (Harms et al. 2021). In this later work, they observed that producers of sordarins, which are a class of natural antifungal agents, were all located in the Naviculisporaceae, supporting the introduction of this family based on molecular data by Marin-Felix et al. (2020). The taxonomy of the Sordariales should not be rearranged following the results of a single phylogenetic study and numerous new family names should not be introduced without careful consideration. The study of Huang et al. (2021) is reminiscent of those of Raymond Hoser (Kaiser et al. 2013;Rhodin et al. 2015;Wüster et al. 2021) or Alexander Doweld (2001), who frequently practice taxonomic vandalism by formally naming clades based on the work of previous workers. The introduction of unnecessary families, genera and species for unstable and unsupported clades that will eventually change justifies the criticisms of other mycological disciplines. These claim fungal taxonomists only want to constantly change names and make systematic mycology a more difficult field. New scientific names will remain in databases forever even if they are proven to be invalid or superfluous and have the potential to be used by future careless workers and further populated in the literature making them appear as valid and accepted names. Therefore, the introduction of new taxa should be done responsibly by following the best taxonomic practices available, such as a thorough review of past literature, conducting polyphasic studies to corroborate initial hypotheses and collaborating with experts in the group of fungi under study Lücking et al. 2020).
Furthermore, it is possible that DNA-based phylogenies can be misleading in the recognition of species just like morphological studies alone. Polymorphisms can interfere in species identification based only on morphology, but also the lack of loci or the use of loci inappropriate for the fungal taxa under study can also lead to the wrong conclusions (Lücking et al. 2020). This is the case for some of the new combinations proposed by Huang et al. (2021) based only on ITS and LSU sequences. Therefore, both phenotypic and molecular approaches should be combined for a more robust taxonomic classification.
It is worthy to mention that according to the recent guidelines for publishing a new fungal species or name provided by Aime et al. (2020), it is strongly recommended to examine the types of the basionyms before making changes and include notes on how to differentiate the new recombined species from other similar species in that genus. Even though one of the co-authors of this guideline was also one of the authors of all new combinations proposed, none of these recommendations were followed in any of the new combinations performed by Huang et al. (2021).