Revisiting Hebeloma (Hymenogastraceae, Agaricales) in Japan: four species recombined into other genera but three new species discovered

Here, we present the results of studies of Japanese Hebeloma collections. The four species described by Imai as Hebeloma (H. fimicola, H. helvolescens, H. humosum, and H. tomoeae) are not from the genus Hebeloma, but are members of Agrocybe, Homophron, or Pholiota. Recombinations are made. Hebelomacrustuliniforme f. microspermum, described by Hongo, is a synonym of H. nanum. Three species of Hebeloma are described as new to science, all currently known only from Japan. Two of these species, H. asperosporum and H. cinnamomeum, are members of H. sect. Denudata while the third species H. citrisporum belongs to H. sect. Velutipes. Japanese records of H. cavipes, H. eburneum, H. hygrophilum, H. subtortum, and H. velutipes are validated. In total, fifteen species of Hebeloma are confirmed from Japan; this is compared with previous checklists.


Introduction
A number of species of the genus Hebeloma appear to be indigenous, or possibly even endemic, to Japan. The latter include species such as Hebeloma luchuense and H. radicosoides (H. sect. Scabrispora), H. sagarae (H. sect. Myxocybe), for which neither observations nor sequence data were found from other countries (status 16 June 2021). Hebeloma lactariolens and H. vinosophyllum (H. sect. Porphyrospora) have been described from Japan, but have also been reported from other countries (e.g., Ho et al. 2014;Cho et al. 2016;Wu et al. 2019;Eberhardt et al. 2021a). Katumoto (2010) (Rees et al. 2013). Accessing the Global Biodiversity Information Facility (www.gbif.org accessed 10 June 2021) gives 202 occurrences of Hebeloma in Japan. Of these 202 occurrences, 166 reference a preserved specimen and 131 give a species name. Additional to the names given by Katumoto and Hebeloma lactariolens are also Hebeloma leucosarx (one record, no preserved specimen), Hebeloma microsporum (one record, but not this taxon, see below under H. cinnamomeum), Hebeloma pusillum (one record, no preserved specimen), and H. sordescens (one specimen recorded as H. testaceum from 1935). iNaturalist (www.iNaturalist.org accessed 10 Section editor: Zhu-Liang Yang June 2021) gives a single record of Hebeloma for Japan, H. crustuliniforme.
One reason for this rather small number of taxa and records is that while it is normally straightforward for a competent field mycologist to identify a mushroom as being a Hebeloma, determining the species has in the past been notoriously difficult. This has been partly because of the macroscopic similarity of species, but also because of the confused interpretation of species concepts by many authors (e.g., Moser 1983;Smith et al. 1983;Arora 1986;Hansen and Knudsen 1992;Breitenbach and Kränzlin 2000;Bon 2002;Vesterholt 2005). As a result of this confusion, many mycologists ignore Hebeloma in the field, which results in few records; within herbaria, many collections determined as Hebeloma remain undetermined at species level. For example, the TNS herbarium currently holds a total of 223 collections of Hebeloma, out of which a total of 210 were collected in Japan, but more than 70 collections remain unidentified to species.
With regard to the types studied, H. luchuense and H. radicosoides were confirmed within H. sect. Scabrispora, H. sagarae was confirmed within H. sect. Myxocybe, and both H. lactariolens and H. vinosophyllum were confirmed within H. sect. Porphyrospora (Eberhardt et al. 2020b). The Japanese collections of H. spoliatum, referred to by Eberhardt et al. (2020b), were similar to or conspecific with H. danicum, and it was noted that further research was required to decide on conspecificity. This is addressed here.
Within this paper, we also present and discuss our findings with regard to additional type studies of species of Hebeloma, described from Japan, and provide a list of Hebeloma species we have found during analysis of collections sent to us by citizen scientists published on Mushroom Observer (https://mushroomobserver.org/) and herbarium collections, particularly those from the National Museum of Nature and Science (TNS). Included are three species new to science: Hebeloma asperosporum is only known from the remote island of Amami Ōshima, while H. cinnamomeum and H. citrisporum are described from the island of Honshu.

Materials and methods
All material studied was dried material from, primarily, the National Museum of Nature and Science (TNS) and the Hokkaido University Museum (SAPA), but also included a few collections sent to us directly by U. Kawasaki. This was compared with material, sequenced and discussed in Beker et al. (2016), Cripps et al. (2019), and Eberhardt et al. (2020aEberhardt et al. ( , 2020bEberhardt et al. ( , 2021a. Sequences were obtained from the dried basidiomes by direct sequencing. Internal transcribed spacer sequences were generated following methods detailed in Eberhardt (2012), and Cripps et al. (2019); RPB2 and TEF1a sequences following Eberhardt et al. (2021b); and sequences of a variable region (V6) of the mitochondrial SSU followed Gonzalez and Labarère (1998). Sequencing was carried out at LGC Genomics (Berlin). Sequences were edited using Sequencher vs. 4.9 (Gene Codes Corp., Ann Arbor, Michigan). Newly generated sequences were accessioned to Gen-Bank (MT157290, MZ725546-MZ725550, MZ724681, MZ782100-MZ782148, MZ782867-MZ782889, and MZ782921-MZ782985). Tables 1 and 2 summarize all sequences used in the analyses.
Alignments were viewed and reformatted using ALIVIEW 1.27 (Larsson 2014). Sequence alignments were done online in MAFFT using the E-INS-i option (Katoh et al. 2005(Katoh et al. , 2019 or locally with the "Mafft-globalpair" setting of MAFFT 7.471 (Katoh and Standley 2013). Maximum likelihood (ML) phylogenetic analyses were run in IQ-TREE (Nguyen et al. 2015) either locally (version 2.1.3) or online (Trifinopoulos et al. 2016). Branch support was obtained through 1000 replicates of ultrafast bootstrap (Minh et al. 2013;Hoang et al. 2018) and Shimodaira-Hasegawa (SH)-like approximate likelihood ratio tests (Guindon et al. 2010). Support values are given as SH-like approximate likelihood ratio test support [%] / ultrafast bootstrap support [%], for SH-like approximate likelihood ratio test support ≥ 80% and ultrafast bootstrap support ≥ 95%. These values were selected following the recommendations of Minh et al. (2021), who also recommend using SH-like approximate likelihood ratio test support alongside with (ultrafast) bootstrap. Model selection (Kalyaanamoorthy et al. 2017) was done using the BIC criterion, including FreeRate models and merging partitions if possible (protein coding loci were originally partitioned according to position (coding) and non-coding). The study was submitted to TREEBASE (acc. no. TB2:S28657). Trees were visualized using FIGTREE 1.4.4 (Rambaut 2006(Rambaut -2018. Results of Beker et al. (2016), Grilli et al. (2016), Frings et al. (2020, and Tian and Matheny (2021) guided the selection of the roots for the topologies. The compatibility of different loci was assessed following the principle of Kauff and Lutzoni   Beker et al. (2016). The amount of macroscopic detail available to us varied hugely from collection to collection as it was dependent on the detail provided by the collector. Where one of the authors was the collector, each specimen was photographed and observed both in the field when characters were still fresh, and later in the laboratory. Fresh basidiomes of each specimen were dried using a food dehydrator (Snackmaster Express FD-60; Nesco/American Harvest, Milwaukee, WI, USA).
All microscopic analysis was carried out on dried material, using a Leica DMRZA2 microscope with a Leica DFC495 camera connected to a computer running Leica Application Suite (LAS) V4 software. A number of photographs were taken of the spores at × 500 and × 1600, which were then measured using the LAS software. Photographs were also taken of the cheilocystidia on the lamella edge at × 500 and of individual cystidia and basidia at × 1000. The material was then examined in 5% KOH. Again, photographs were taken of the spores at × 500 and × 1600 and of the cheilocystidia (and pleurocystidia if any were present) and basidia at × 500 and × 1000.
For each Hebeloma collection, wherever possible, at least 50 spores were measured in Melzer's reagent, excluding the apiculus. The maximum length and width of each spore was measured, and its Q value (ratio of length to width) calcu-lated. Average length, width, and Q value were calculated and recorded alongside the median, standard deviation, and 5% and 95% percentiles. The degree of dextrinoidity, ornamentation, and the degree of loosening of the perispore was observed and classified. For the assessment of the degrees of ornamentation (O0, O1, O2, O3, O4), of the loosening perispore (P0, P1, P2, P3), and for the dextrinoidity (D0, D1, D2, D3, D4), we used Beker et al. (2016) and Vesterholt (2005).
The average width of the widest part of the cheilocystidium in the vicinity of the apex appears to be an important character in the separation of species within Hebeloma (Vesterholt 2005). It is also important, when determining this average width near the apex, not to be selective with regard to the cystidia chosen for measurement. To determine the average width at the apex, about 100 cheilocystidia were measured on the lamella edge. For other measurements, some 20 cheilocystidia, separated from the lamella edge, were measured from each collection. Because of the complex shapes of the cheilocystidia four measurements were made: length, width at apex (A), width at narrowest point in central region (M), and maximum width in lower half (B), see Fig. 1. The measurements were given in this order, and an average value was calculated for each of these measurements. For each cheilocystidium, the ratios A/M, A/B, and B/M were calculated and averaged across all cheilocystidia measured. Measurements were made in 5% KOH and Melzer's reagent.
For all other details with regard to our methodology, see Beker et al. (2016). Each collection studied has a database record number associated with that collection; we give these numbers as we intend to make the database publicly available.    The alignment for Agrocybe includes 24 ITS sequences and 717 positions. The result of the analysis is shown in Fig. 2. The topology is rooted with Agrocybe firma at a branch receiving 100/100 support (support not shown because branch used for rooting). The type species of Agrocybe is A. praecox, which is represented in the alignment by five sequences of different authors. The lectotype ITS of Agrocybe humosa (Hebeloma humosum) is placed in an unsupported clade together with sequences identified as A. smithii; the holotype ITS of A. imaii (H. fimicola) is placed in a clade with sequences identified as A. pediades.
The alignment for Homophron includes 35 ITS sequences and 708 positions. Homophron spadiceum is the type species of the genus. We have made an effort to sample all variation from UNITE SH1186588.08FU (which includes the epitype of Ho. spadiceum), because it was not possible to see the species limits of Ho. spadiceum. The result of the analysis is shown in Fig. 3. The topology is rooted with Lacrymaria spp. The clade indicated as "Homophron spadiceum s.l." consists of a number of subclades, some appear to be geographically restricted, among which Ho. helvolescens is in a subclade with sequences from Japan and India.
The alignment for Pholiota (P. sect. Flammuloides in Tian and Matheny 2021) includes 26 ITS sequences and 771 positions. The result of the analysis is shown in Fig. 4. The sequences of the two syntypes (one designated as lectotype below) of P. tomoeae are within the P. brunnescens clade, which includes sequences from America, Asia, and Europe.
Hebeloma asperosporum is morphologically and molecularly distinct from all other described Hebeloma species. Morphologically, it is a member of H. subsect. Clepsydroida, distinguished by its small, strongly ornamented spores, on average less than 10.5 µm long, indextrinoid and with consistently loosening perispore.
Likewise, Hebeloma cinnamomeum is morphologically and molecularly distinct from all other described Hebeloma species. Morphologically, it is a member of H. subsect. Clepsydroida, distinguished by its cinnamon-colored pileus and the spores with average Q less than 1.85 and the cheilocystidia with average width near the apex of less than 6.5 µm.
Four loci (ITS, alignment 708 positions; mitSSU V6, alignment 476 positions; RPB2, alignment 779 positions; and TEF1a, alignment 717 positions) have been employed to infer the phylogenetic relationships of taxa in Hebeloma sect. Denudata (Fig. 5). The topology is rooted with members of H. subsect. Echinospora. All loci analyzed individually resolved the two available sequences of H. asperosporum as monophyletic with high support (min. 94% SH-like approximate likelihood ratio test and 97% ultrafast bootstrap support, in most loci higher), but it was on a long branch and the position of the species within the section was unresolved. The sequences obtained for H. cinnamomeum form an independent lineage only in the ITS (support 92%/99%) and V6 (84%/- The topologies obtained for the different loci (see Treebase submission) were in conflict. This was largely due to H. asperosporum and H. matritense being placed with relevant support by one or both support criteria as sister to different species clades in the results for different loci. The result of the analysis of the concatenated alignment is shown in Fig. 5. The topology is rooted with H. echinosporum and H. populinum.
Here, H. asperosporum receives 100% support by both criteria and H. cinnamomeum 94%/97% support. The inclusion of members of H. sect. Clepsydroida in two reciprocally monophyletic groups is weakly (86% in the lesser supported clade) sup-ported by SH-like approximate likelihood ratio tests. However, conflicts between the evolutionary hypothesis of different loci concern this part of the tree.
Hebeloma citrisporum is morphologically and molecularly distinct from all other described Hebeloma species. The predominantly gently clavate cheilocystidia suggest that this     Description of the lectotype: Spores elliptical, smooth, pale brown, thick-walled, with a conspicuous germ pore, on av. 11.1 × 7.7 µm, with av. spore Q 1.44. The cheilocystidia are ventricose; the pleurocystidia are similar.
The spores of A. smithii (Watling and Bigelow 1983) and of the type of A. flexuosipes (Eberhardt et al., in press) are around the same size (11-13.5 × 6.5-8 µm respectively 12.1 × 7.7 µm) as the spores of A. humosa. Whereas the spores of the former two species are described as (indistinctly) ornamented, the spores observed in the lectotype of A. humosa are smooth. The spores of A. putaminum are described as 10-13 × 5-7 µm, thus somewhat narrower than in A. humosa, but also smooth (Maire 1913).
Both morphology (without annulus, yellowish tones in pileus, spore size, and prominent germ pore) and sequence data (Fig. 2) suggest that A. humosa is closely related with A. smithii, A. putaminum, or other related species (Nauta 2005). Another closely related species is A. flexuosipes (Eberhardt et al., in press). However, none of these species matches the type of A. humosa sufficiently closely for synonymization.  Description of the holotype: The spores are elliptical, smooth, pale brown, thick-walled, with a conspicuous germ pore, on av. 12.9 × 8.9 µm, with av. spore Q 1.44. The cheilocystidia are ventricose; pleurocystidia only near lamella edge. The pileipellis has a thin gelatinous layer, with small pieces of hypha embedded within it.

Agrocybe imaii
Japanese name: Baba-wakafusa-take (Imai 1938). Remarks: Hebeloma fimicola is an Agrocybe. Our examination of the type material was in close agreement with the description of Imai (1938). It is possible, even likely, that A. imaii is a later synonym of A. pediades s.l., and, given its habitat on dung, may be A. pediades var. fimicola (Speg.) Nauta (Agrocybe fimicola (Speg) Sing.) (Nauta 2005), but we rather leave it to experts on the genus based on multilocus sequence data to come to a conclusion. Based on ITS data, Agrocybe pediades occurs in many countries including China, India, Iraq, Pakistan, and Sri Lanka (see Fig. 2 and UNITE SH2290413.08FU).
Hebeloma asperosporum Beker & U. Eberh.,sp. nov.,Figs. 7 and 8,Mycobank MB840287 Diagnosis: The small, strongly ornamented spores, on average less than 10.5 µm long, indextrinoid and with  Branches with at least one of the support values above its threshold are thick lined. The last two letters indicate the country of origin upwards, fibrillose, at apex pruinose. Context in pileus white to cream, firm, in stipe stuffed; taste and smell not recorded. Spore deposit color not recorded.
Ecology and distribution: In mixed woodlands with Pinus, Quercus and Castanopsis. The growth habit was caespitose to scattered. To date, Hebeloma asperosporum is recorded only from the island Amami-Ŏshima.
Remarks: The mainly clavate-ventricose cheilocystidia together with distinctly to strongly ornamented spores support the placement of this taxon in H. subsect. Clepsydroida within H. sect. Denudata. The occasional presence of cystidia with thickening of the wall in the middle is further evidence of this placement. Molecularly, the position of the species within H. sect. Denudata is supported, but its assignment to subsection is not. Within H. sect. Denudata, the small spores, on average less than 10.5 µm long, indextrinoid but strongly ornamented and with consistently loosening perispore distinguishes this species from all other known members of H. section Denudata. While the description is based on just two collections, its morphological and molecular separation, from all known species of Hebeloma, leave us in no doubt that this should be regarded as a distinct species.
Of course, our description and knowledge of this taxon is limited to just these two collections from the same locality, hence our concept of this species is limited. A sequence published from China as H. alpinum (MW554385, Zouh, unpublished, submitted 26 Jan 2021; no UNITE SH to date) is, apart from obvious editing errors, identical to our H. asperosporum sequences. It is hoped that publishing this description will encourage other mycologists to search for this taxon.
Hebeloma cinnamomeum Beker & U. Eberh., sp. nov., Figs. 9 and 10, Mycobank MB840288 Diagnosis: The cinnamon-colored pileus, the spores with average Q less than 1.85 and the cheilocystidia with average width near the apex of less than 6.5 µm, are characters that separate this species from others in H. sect. Denudata.
Ecology and distribution: In deciduous or mixed woodlands apparently associated with Quercus or Pinus, but   Hiemalia). Within H. sect. Denudata, the spores with average Q less than 1.85, the cheilocystidia with average width near the apex of less than 6.5 µm and the cinnamon-colored pileus are characters that separate this species from others within this subsection. The pileus color is an eye-catching feature that may well enable field identification, or at least a reasonable suspicion of determination, of this species. While our morphological description is based on 23 collections, many (but not all) are from a relatively confined area within the Tsukuba Botanical Garden, which is planted with endogenous plants from Japan. The description provided may be too narrow, given the similar habitat for many of these collections. In addition, a sequence from a Quercus dentata root tip collected in Hokkaido (LC068995, Arai et al. unpublished, submitted27-Jul-2015)clustersinMLanalyses(notshown)withH. cinnamomeum sequences, supporting the assumptions on associates and adding evidence that it is an endogenous species in Japan. This sequence currently (9 June 2021) forms a singleton UNITE SH hypothesis (SH2292369.08FU) at the < 0.5% level, but owing to probableediting errors in this sequence this SH hasto be treated with caution.
Note that cited collection TNS-F-44863 was originally identified as Hebeloma microsporum, the name under which it is recorded on GBIF, and which was mentioned in the introduction.
Ecology and distribution: All records to date suggest Quercus as the most likely ectomycorrhizal host. The growth habit was mainly scattered, but occasionally caespitose. To date, all collections of Hebeloma citrisporum have been recorded only from the Japanese island of Honshu, in the vicinity of Tokyo.
Remarks: The predominantly gently clavate towards the apex cheilocystidia, with a few ventricose or clavateventricose, supports the placement of H. citrisporum in H. sect. Velutipes. Within this section, the citriform spores, distinctly ornamented and rather strongly dextrinoid but with the perispore only somewhat loosening in a few spores, with average width less than 7.5 µm, and the cheilocystidia with average width near the apex greater than 6.5 µm and average ratio of basal width to median width (B/M) at most 1.35, distinguish this species. While citriform spores are commonly found within spe cies of H. sect. Velutipes, rarely are they as regular and eyecatching as for this species. The species and its placement in H. sect. Velutipes are supported by molecular data (Fig. 6).
The description is based on just five collections from a relatively small geographic area around Tokyo. We would suspect that it is far more widespread, at least on Honshu Island. A Quercus dentata root tip sequence from Hokkaido (AB979728, Arai et al. 2017) might have been formed by H. citrisporum. Currently (9 June 2021), the sequence forms a singleton UNITE SH (SH1235508.08FU) at the 3% level. It differs from our H. citrisporum sequences by three indels, the "Homophron spacideum s.l." clade of Fig. 3. This clade includes a number of supported (and unsupported) subclades that appear to have restricted geographical distributions. However, the sequence variation underlying these clades is so small (see scale in Fig. 3) and the data is only from a single locus that we hesitate to recognize them as hypotheses of distinct species. Vašutová et al. (2008) stressed that ITS is not sufficient for delimiting species in the Psathyrellaceae. Thus, it is not clear whether Ho. helvolescens is indeed a later synonym of Ho. spadiceum.
Description of syntypes: The spores are smooth, ellipsoid to ovate, yellowish brown to brown and with a very small indistinct germ pore. For SAPA 1000031 the spores are on av. 7.6 × 4.9 µm, with av. spore Q 1.56, while for SAPA 1000032 they are on av. 7.9 × 4.8 µm, with av. spore Q 1.63. In both cases, the pleurocystidia are mainly ventricose, hyaline to yellowish in 5% KOH, while the cheilocystidia are more utriform in shape.
Remarks: Based on morphology and ITS sequence data (Fig. 4), the syntypes represent the same taxon. Although neither of the syntypes is in particularly good condition, of the two collections, SAPA 10000031 yielded better sequence data.
The description is in good agreement with the description given by Imai (1938). Although Imai does not mention burnt ground as the habitat, morphologically and molecularly both syntypes of H. tomoeae are in good agreement with Pholiota brunnescens, which is associated with burnt soil or wood (Matheny et al. 2018). It appears likely that P. tomoeae is a later synonym of P. brunnescens; we rather leave it to Pholiota experts to synonymize the two species, also considering that the phylograms based on several loci published by Matheny et al. (2018) and Tian and Matheny (2021) suggest that P. brunnescens sl may be rather variable and is not supported by bootstrap (Tian and Matheny 2021).

Discussion
For the newly described species as well as for the species described by Imai (1938), we had both morphological and molecular data available and the conclusions from both kinds of data supported each other. With regard to Imai's species, it is clear that, at the time he described these four species, the concepts of brown-spored genera were rather different from today. We cannot be sure that all of the new combinations are indeed non-redundant species. It is beyond the scope of this study to explore the limits of species from genera other than Hebeloma. We only have ITS data available for the types which may not be sufficient to distinguish species in Agrocybe, around Homophron spadiceum or Pholiota brunnescens. This situation leads to a dilemma between different principles of best taxonomic practices formulated by Aime et al. (2021): On the one hand, the publication of superfluous names ought to be avoided, on the other we view the recombination into the genus we consider the correct one as the most efficient means to communicate our findings to the experts for the respective genera. The experts, in our view, should make the decision on the synonymy or otherwise of A. imaii with A. pediades, A. humosa with A. smithii, Ho. helvolescens with Ho. spadiceum, and P. tomae with P. brunnescens.
Based on the almost completed study of Hebeloma types worldwide, we are confident that the new Hebeloma species, H. asperosporum, H. cinnamomeum, and H. citrisporum, are good species and all of them can, based on our current knowledge, be identified by ITS alone.
The presence of conflicts between single locus results in the genus Hebeloma was previously known. Conflicts have been discussed earlier by Grilli et al. (2016) for H. sect. Velutipes and by Eberhardt et al. (2016) for H. sect. Clepsydroida. The recent recognition of additional taxa in H. sect. Clepsydroida, i.e., H. asperosporum, H. cinnamomeum, and H. sordidulum (Eberhardt et al., submitted) appears to have increased the conflict. While more advanced analysis methods might have given a better representation than concatenation, of what the available data suggest about species relationships, it would not have changed the fact that additional loci are needed to resolve infrageneric relationships. In the case of H. sect. Denudata, additional loci are also needed for the delimitation of subsections. Currently, we do not have molecular support that the morphological characters that are used for distinguishing and delimiting H. subsects. Clepsydroida, Crustuliniformia, and Hiemalia have evolved concertedly among the members of each subsection.
Funding Open Access funding enabled and organized by Projekt DEAL. This research was funded in part by the last author and in part by JSPS KAKENHI Grants 24680085 and 20K06805 to KH and 15K16279 to TK.
Data availability The majority of the material was obtained through public collections or, in the case of the private collection of H.J.B., will be accessioned toapubliccollectionwhentheprojectis finished.Sequencedataweresubmitted to GenBank. Alignments and trees were submitted to TreeBASE.
Code availability Not applicable.

Declarations
Conflict of interest None.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.