Multi-locus phylogenies reveal three new true-like taxa and the traces of interspecic hybridization in Octaviania (Boletales)

Among many convergently evolved sequestrate fungal genera in Boletaceae (Boletales, Basidiomycota), the genus Octaviania is the most diverse. We recently collected many specimens of Octaviania subgenus Octaviania, including several undescribed taxa, from Japan and the Americas. Here we describe two new species in subgenus Octaviania, O. tenuipes and O. tomentosa, from temperate to subtropical evergreen Fagaceae forests in Japan based on morphological observation and robust multi-locus phylogenetic analyses (nrDNA ITS and partial large subunit [LSU], translation elongation factor 1-α gene [TEF1] and the largest subunit of RNA polymerase II gene [RPB1]). Based on specimens from the Americas as well as studies of the holotype, we also taxonomically re-evaluate the taxon O. asterosperma var. potteri. Our analysis suggests that O. asterosperma var. potteri is a distinct taxon within the subgenus Octaviania so we recognize this as O. potteri. We unexpectedly collected O. potteri specimens from geographically widespread sites in the USA, Japan and Colombia. This is the rst veried report of Octaviania from the South American continent. Our molecular analyses also revealed that the RPB1 sequence of one O. tenuipes specimen was identical to that of a closely related species, O. japonimontana, and that one O. potteri specimen from Minnesota had an RPB1 sequence of an unknown species of O. subg. Octaviania. Additionally, one O. japonimontana specimen had an unusually divergent TEF1 sequence. Gene-tree comparison and phylogenetic network analysis of the multi-locus dataset suggest that these heterogenous sequences are most likely the result of previous inter- and intra-specic hybridization. We hypothesize that frequent hybridization events in Octaviania may have promoted the high genetic and species diversity within the genus. sensu stricto from South America. Our primary research objective is to clarify the phylogenetic and systematic positions of those taxonomically unsettled specimens in a robust phylogenetic framework. Here we propose two new species and one new taxonomic status based on morphological observations and nuclear multi-locus phylogenies. Furthermore, we found strong topological conicts in some species of O. subg. Octaviania among gene trees. We therefore examined the cause of these conicts based on gene-tree comparisons and phylogenetic network analyses and we discuss the possibility of inter- and intra-specic hybridization within the subgenus based on their ecology and phylogeography.


Introduction
The Boletaceae (Boletales, Basidiomycota) is generally recognized as a large family that consists of epigeous, mushroom-forming fungi. However, recent systematic studies have revealed a considerable number of sequestrate (tru e-like and secotioid) fungal lineages in the family that have evolved independently from boletoid mushrooms (e.g., Castellano  The genus Octaviania is comprised of sequestrate, tru e-like species that have a marbled gleba and dextrinoid or non-amyloid basidiospores with coarse, conical to pyramidal ornamentation ). Historically, the generic concept of Octaviania was unsettled and the genus was previously considered by some authors as a synonym of Arcangeliella Cavara (Russulaceae), Hydnangium Wallr (Hydnangiaceae). or Melanogaster Corda (Paxillaceae). Pegler and Young (1979) provided evidence that Octaviania is distinct from those morphologically similar sequestrate genera and Orihara et al (2012a) rede ned the current generic concept of the genus. Orihara et al. (2012a) further divided the genus into three subgenera, subg. Octaviania, Fulvoglobus Orihara and Parcaea Orihara, based on multigene phylogenies and the morphology of basidiomata. Paz et al. (2016) reviewed the European species of Octaviania and critically examined the type species, O. asterosperma. They found that O. asterosperma senus stricto has a pseudoparenchymatous peridium, which is one of the major characteristics of subg. Fulvoglobus. Accordingly, they concluded that the subg. Fulvoglobus that was introduced by Orihara  Octaviania subg. Octaviania sensu Paz et al. (2016) (hereafter referred to simply as Octaviania subg. Octaviania) is characterized by cavities in the gleba lled with slightly viscid to dry, brown to blackish brown spore masses, and a peridium composed of in ated hyphae and isodiametric, pseudoparenchymatous cells. So far, the subgenus accommodates eight described species that are known only from the northern hemisphere ; Paz et al. 2016). Orihara et al. (2012a) further suggested that there were at least two additional, taxonomically unsettled species (Octaviania sp. "E" from Japan and Octaviania sp. from North America). Since the publication of Orihara et al. (2012a), we have collected a number of additional specimens of O. subg. Octaviania from Japan, including the two species mentioned above. In addition, we collected fruitbodies of O. subg. Octaviania from a Quercus humboldtii forest in Colombia, which constitutes the rst known record of Octaviania sensu stricto from South America. Our primary research objective is to clarify the phylogenetic and systematic positions of those taxonomically unsettled specimens in a robust phylogenetic framework. Here we propose two new species and one new taxonomic status based on morphological observations and nuclear multi-locus phylogenies. Furthermore, we found strong topological con icts in some species of O. subg. Octaviania among gene trees. We therefore examined the cause of these con icts based on gene-tree comparisons and phylogenetic network analyses and we discuss the possibility of inter-and intra-speci c hybridization within the subgenus based on their ecology and phylogeography.

Methods
Taxon sampling and morphological observation For microscopy, hand-cut sections of fresh or dried specimens were mounted in water, 3% KOH or lactoglycerol. To determine amyloidity of basidiospores, dried material was stained with Melzer's reagent. Basidiospore dimensions (range of spore length, from the hilar appendage to the spore tip × spore width), their standard deviations (SD) and the length to width ratio (Q) were determined based on 50 random measurements unless otherwise mentioned. The 95% prediction intervals of basidiospore diameter are shown without parentheses in taxonomic descriptions. Both endpoints of the spore dimensions are shown in parentheses, but when the value is the same as the 95% prediction interval, only the latter is shown. Measurements include the hilar appendage but not the spore ornamentation or pedicel. Basidium sizes are presented as the range of the lengths × the range of the widths. Scanning electron microscopy (SEM) was performed with the HITACHI TM-4000Plus Tabletop Microscope (Hitachi High-Technologies Co. Ltd., Japan). Small fragments of a dried gleba were excised and immersed in 8% ionic liquid (1-ethyl-3-methyl-imidazolium tetra uoroborate) for conductive treatment (Yanaga et al., 2012) and were observed at 10-15 kV.
DNA extraction, PCR ampli cation and sequencing DNA was extracted from fresh or dried fruitbodies using Indicating FTA Cards (Whatman International Ltd, Maidstone, England) based on the protocol by Orihara et al. (2012a, b). We also extracted genomic DNA from some fruitbodies using the protocol of Izumitsu et al. (2012). PCR ampli cation of the ITS and the large subunit (LSU; 28S) of the nuclear ribosomal DNA (nrDNA), and TEF1 followed Orihara et al. (2012a). For RPB1 ampli cation, we used a newly designed primer set based on sequences of Boletaceae deposited in the International Nucleotide Sequence Databases (INSD). The new primers include forward primer RPB1-TO-Bf (5'-AAGGCYGATATYGTGAGTC − 3'), which is located in the intron A between domains A and B of RPB1, reverse primer RPB1-TO-Br (5'-GCTTTGATGATRTCYCC − 3'), and reverse primer RPB1-TO-Br2 (5'-ARGCYTTGATRATRTCYCC-3'). Both of the reverse primers are located in the conserved (exon) domain C. These primer pairs target an 850-1100 bp amplicon which spans the region between primer RPB1-Bf (Nuhn et al, 2013) and primer RPB1-Cr (Matheny et al., 2002). The PCR ampli cation of RPB1 was performed using the following procedure: initial incubation at 95 °C for 10 min; subsequent step of 30 cycles at 94 °C for 30 s, 53 °C for 60 s, and 72 °C for 90 s, followed by 13 cycles at 94 °C for 30 s, 52 °C for 60 s, and 72 °C for 90 s; a nal elongation step at 72 °C for 7 min. Unidirectional sequencing of the PCR products in the forward and reverse directions were completed according to Orihara et al. (2012a). Sequences were edited and assembled with Sequence Scanner v. 1.0 (Applied Biosystems, Foster City, California, USA), BioEdit version 7.0.9 (Hall 1999) and SeaView version 4 (Galtier et al. 1996). A total of 178 newly obtained sequences were deposited in INSD (Table 1).

Phylogenetic analyses
For the combined ITS-nLSU-TEF1-RPB1 dataset, we retrieved 170 sequences from INSD ( Table 1). The sequences were carefully selected so that the dataset could represent all genera and subgenera in the subfamily Leccinoideae, which includes the genera Chamonixia, Leccinellum, Leccinum, Octaviania, Rossbeevera and Turmalinea (Orihara et al. 2016a). We selected Spongispora temasekensis, Spongiforma thailandica, Borofutus dhakanus, Tylocinum griseolum, Binderoboletus segoi and Retiboletus spp. in subfamily Leccinoideae as outgroup taxa based on Henkel et al. (2016) and Wu et al. (2016Wu et al. ( , 2018. We only included specimens in our analysis whose nucleotide sequences covered more than 50% of the total length of the aligned, full ITS-nLSU-TEF1-RPB1 dataset to reduce a negative effect caused by the lack of sequences in the dataset (i.e., no less than 1802 bp in length in the concatenated 4gene dataset). Accordingly, we could not include sequences of Ionosporus, Rhodactina, Pseudoaustroboletus and two of the polyphyletic clades of Leccinum in the Leccinoideae previously shown in Kuo and Ortiz-Santana (2020) (i.e., the Leccinum talamancae and L. longicurvipes lineages). Similarly, several species of Octaviania, including O. asterosperma s. str. and O. arbucalensis, which belong to subg. Octaviania (Yang et al. 2006;Vadthanarat et al. 2017), were not included in the analyses due to insu cient number of DNA loci available from INSD. Instead, we prepared an additional singlegene nLSU dataset that included as many Octaviania species as possible available from INSD, including the type species O. asterosperma s. str. ( Table 1). The ML analyses were conducted with RAxML 8. SeaView (Castresana 2000) to exclude ambiguously aligned sites. Accordingly, the longest part of the insert within the ITS2 region found in all known species of O. subg. Octaviania ) was excluded from our analyses. Prior to the multigene analyses, we compared the BioNJ tree topologies among the ITS, nLSU, TEF1 and RPB1 datasets to see if there were any topological con icts among the gene trees. Sequences that caused considerable topological con icts (BioNJ bootstrap values ≥ 75%; 1000 replicates) among the four single-locus phylogenies were excluded from the multi-locus analyses. Accordingly, RPB1 sequences of "Octaviania tenuipes" nom. prov. KPM-NC 27968 (INSD acc. no.: MT868858) and "Octaviania potteri" nom. prov. KPM-NC 17828 (MT868837 & MT868838), that are proposed as new taxa in this study were omitted from the combined multi-locus dataset. We subsequently concatenated the four datasets for the multi-locus analyses. The ITS rDNA region was partitioned by ITS1 + ITS2 and 5.8S, and the TEF1 and RPB1 regions were partitioned by codons and introns, and best-t likelihood models were estimated for each partition and nLSU with MrModeltest 2.3 (Nylander 2004).
Bayesian analyses were conducted with MrBayes 3.2 (Ronquist and Huelsenbeck 2003). Nucleotide substitution models for maximum likelihood (ML) analyses were selected by the Akaike Information Criterion (AIC) in jModeltest2 (Darriba et al., 2012;Guindon and Gascuel 2003). The GTR + I + G model was applied to ITS1 + ITS2, nLSU, the second codon of RPB1 and the rst and third codons of TEF1; SYM + I for 5.8S rDNA; F81 + I for the second codon of TEF1; HKY + I for the rst codon of RPB1; HKY + G for the third codon of RPB1; and HKY + I + G for the introns of TEF1 and RPB1. Bayesian posterior probabilities (PP) were estimated by the Metropolis-coupled Markov chain Monte Carlo method (Geyer 1991). In the multi-gene (ITS + nLSU + TEF1 + RPB1) analysis, two parallel runs were conducted with one cold and seven heated chains each for 4M generations. The parameter for temperature of the seven heated chains in both runs was set to 0.10. The 0.10 heating scheme was used instead of the default 0.20 setting, because in previous phylogenetic studies on the Leccinoideae, the Markov chains with the 0.10 heating setting converged more smoothly and were less likely to become trapped at local optima (Orihara et al. 2016a;Orihara and Smith 2017). Trees were saved to a le every 1000th generation. We determined that the two runs reached convergence when the average standard deviation of split frequencies (ASDSF) was continuously lower than 0.01. The ASDSF was monitored every 5000 generations. We also veri ed the convergence by checking that the effective sample size (ESS) of each resulting statistic was su ciently large (> 200). Trees obtained before reaching convergence were discarded as the burn-in, and the remaining trees were used to calculate a 50% majority consensus topology and to determine PP values for individual branches.
Maximum likelihood (ML) analyses were conducted with RAxML 8.2.10. The same partitioned datasets as those for the Bayesian analyses were used so that different α-shape parameters, GTR rates, and empirical base frequencies could be assigned to each partition. The best-t ML tree was estimated under the GTRCAT + I model. The rapid bootstrap (BS) analysis was implemented with 1000 replicates.
The single-gene nLSU phylogenies that included all the representative species of Octaviania available from INSD were estimated using the ML and BioNJ methods. The ML analysis was conducted using RAxML 8.2.10, setting the substitution model to GTRCAT + I and the number of rapid BS replicates to 1000. The BioNJ analysis was done by SeaView version 4 with the number of BS replicates set to 1000.
To compare tree topologies and examine precise phylogenetic placement of our three target taxa in Octaviania subg. Octaviania, we further inferred ML gene trees from individual ITS, nLSU, RPB1, and TEF1 datasets of the subgenus using RAxML 8.2.10. The datasets were partitioned by genes for ITS (i.e., ITS1 + ITS2 and 5.8) and by codons for RPB1 and TEF1. The best-t ML tree was estimated under the GTRCAT + I model. The rapid BS analysis was implemented with 1000 replicates.
Since the comparison of the four gene trees of Octaviania subg. Octaviania detected several heterogenous sequences in the RPB1 and TEF1 regions, we further conducted phylogenetic network analysis based on a smaller multi-locus dataset to nd the traces of reticulate evolution among infrageneric taxa in the subgenus Octaviania. The dataset for this analysis included 1-2 representative specimens for each species of the subgenus Octaviania. We selected specimens for which molecular data were available from all four DNA regions (i.e., ITS, nLSU, RPB1 and TEF1). The RPB1 sequences of "Octaviania tenuipes" nom. prov. (MT868858 [KPM-NC 27968]) and "Octaviania potteri" nom. prov. (seq1: MT868837 [KPM-NC 17828; RH1181]), which were omitted in the multi-gene Bayesian and ML analyses discussed above, were included in the combined dataset for this analysis. The analysis was executed with SplitsTree 4 (Huson and Bryant 2006). Networks were constructed by the NeighborNet method using the "distance estimation to uncorrected P value" setting. The resultant networks were displayed with the EqualAngle algorithm (Dress and Huson 2004). Bootstrap analysis was then conducted with 1000 replicates.

Results
Morphological evaluation of the North American Octaviania species in subgenus Octaviania.
The phylogenetic analyses in Orihara et al. (2012a) explicitly showed that three specimens of Octaviania (KPM-NC 17827, KPM-NC 17828 and OSC 13925) from eastern North America (i.e., Iowa, Minnesota and Florida) formed a distinct clade within Octaviania subg. Octaviania but there was no molecular taxonomic treatment of the unidenti ed taxon. We critically examined the morphology and habitat of the taxon and we compared it with the previously published literature on North American Octaviania species.
The overall macro-morphology, peridial structure and the basidiospore and basidia dimensions matched the original description of Octaviania asterosperma var. potteri Singer and Smith (Singer and Smith 1960), which was reported from Michigan, USA (see description of O. potteri below). We studied the holotype of O. asterosperma var. potteri (MICH 12376 [Potter 8898]) from the University of Michigan Herbarium (MICH). The specimen was well-preserved, but the cells of the peridium were collapsed. The basidiospore morphology matched those of the three North American specimens of Octaviania sp. (Table 2). We therefore identify the North American Octaviania species as O. asterosperma var. potteri. Below we propose a new status for this species as Octaviania potteri stat. nov.

Phylogenetic placement of new taxa inferred from the multi-locus phylogeny
The multi-locus dataset comprised of ITS and LSU nrDNA, TEF1 and RPB1 sequences of the Leccinoideae consisted of 94 specimens and 3603 aligned nucleotide positions. The Bayesian inference reached convergence after ca. 1.38M generations. Accordingly, we discarded the rst 1400 trees in each parallel run, and the remaining 2601 trees in each run were summarized to approximate Bayesian posterior probabilities (PPs). ESS of all the model parameters were su ciently large (> 200). The total arithmetic and harmonic mean of estimated marginal log likelihoods (lnL) for runs were − 27576.51 and − 27653.13, respectively. In the RAxML analysis, the nal ML optimization of log likelihood was − 27424.085894. The overall topologies between the Bayesian and ML trees were nearly identical. Specimens of O. tenuipes sp. nov. exhibited minimal infraspeci c genetic divergence. In contrast, both O. potteri stat. nov. and O. tomentosa sp. nov. showed considerable genetic divergence among specimens. In the O. tomentosa clade, a specimen from Amami-oshima Island in the Ryukyu island chain, was genetically divergent from the other specimens from mainland Japan. In the O. potteri stat. nov. clade, the geographically isolated specimens from Hokkaido, Japan (KPM-NC 18032) and Colombia (HUA 222100) were nested among the North American specimens.
Although the generic type species, O. asterosperma, was not included in the multi-locus phylogenies, the nLSU gene tree indicate that O. potteri stat. nov. is genetically distant from O. asterosperma var. asterosperma and it should be treated as a distinct taxon ( Fig. S1; the lnL of the ML tree = -3239.444877).
Comparison of single-gene tree topologies within Octaviania subgenus Octaviania The four ML gene trees based on ITS nrDNA (ITS1-5.8S-ITS2), LSU nrDNA, RPB1 and TEF1 datasets were estimated with the nal ML optimization of lnL of -1806.362004, -1975.613722, -2139.994105 and − 2417.247141, respectively (Fig. 2). All of the species-level clades in subg. Octaviania except O. potteri stat. nov. were recovered in each tree with high bootstrap values.
Interestingly, one O. tenuipes specimen collected from a Castanopsis sieboldii forest in Mt. Tohken, Kagoshima Prefecture, Japan (KPM-NC 27968) had an identical RPB1 sequence to O. japonimontana, which was inferred to be sister to O. tenuipes in the RPB1 tree with moderate BS support. This result was con rmed by sequencing the RPB1 region of the specimen twice using different primer pairs. Furthermore, one O. potteri specimen from Minnesota, USA (RH1181; KPM-NC 17828) had at least two heterogeneous RPB1 sequences, one of which was apparently derived from O. potteri but another was remarkably divergent from the other O. potteri sequences. The ML phylogeny showed that the divergent sequence from the RH1181 specimen forms its own clade and is an unknown species-level lineage that is sister to O. kobayasii (Fig. 2). It should also be noted that the TEF1 sequence of a O. japonimontana specimen from Mts. Tanzawa

Taxonomy
Based on our morphological studies and phylogenetic results (Fig. 1), we describe two new species, O. tenuipes and O. tomentosa, from Japan. The multi-locus phylogenetic analyses also strongly support O. potteri stat. nov. as sister to O. hesperi in O. subg. Octaviania (Fig. 1). Furthermore, our nLSU gene tree shows that O. potteri stat. nov. is phylogenetically distant from O. asterosperma var. asterosperma (Fig.  S1). We conclude that the taxon previously considered as O. asterosperma var. potteri is a distinct species from O. asterosperma s. str., and we propose a new status, Octaviania potteri, for the former taxon.
We are aware of the work by Kuo and Ortiz-Santana (2020) that proposed the synonymy of all the sequestrate Leccinoideae within a broadly circumscribed genus Leccinum s.l.. While the phylogenetic analyses of Kuo and Ortiz-Santana (2020) resolved monophyletic clades for all of the sequestrate genera (e.g., Octaviania, Chamonixia, Turmalinea and Rossbeevera), several other major clades of epigeous Leccinoideae taxa were otherwise poorly resolved. Based on the lack of resolution in their phylogenetic trees and the uncertainty that remains about the evolutionary relationships within the Leccinoideae, we opt to retain the sequestrate genus names Octaviania, Chamonixia, Rossbeevera and Turmalinea. We acknowledge that these sequestrate taxa belong to the Leccinoideae but feel that synonymy with Leccinum is premature, results in the loss of information, and does little to clarify the taxonomy and phylogeny of this group. In our multi-locus phylogeny, some relationships within Leccinoideae clades that were unresolved in Kuo and Ortiz-Santana (2020) were resolved with high statistical support (i.e., ML BS ≥ 75% and PP ≥ 0.97; Fig. 1). For example, phylogenetic placement of the core clade of Leccinum (= Leccinum s.str.) was not resolved in Kuo and Ortiz-Santana (2020), whereas our phylogeny supported that Leccinum s.str. was sister to the clade comprised of Leccinellum s. str., two unde ned leccinoid clades, Rossbeevera and Turmalinea (Fig. 1). Our phylogeny also supported monophyly of the Leccinellum s. str. clade, which was not strongly supported in Kuo and Ortiz-Santana (2020). However, we could not include the two independent clades, Leccinum talamancae and L. longicurvipes, whose phylogenetic positions within the Leccinoideae remained uncertain. Our results suggest that we need to further address the phylogeny and systematics of Leccinoideae before lumping together all the wellde ned epigeous and sequestrate genera into one large and broadly circumscribed genus. We will wait to determine a nal taxonomic scheme for the sequestrate genera until a more highly resolved phylogeny becomes available that provides appropriate insight on this group.
Etymology referring to the slender stipe (sterile base) of the species, which is unique in the genus. Description Basidiomata sequestrate, up to 21 mm in diam., soft, depressed-globose or reniform, surface smooth or slightly occose, white at rst, becoming dirty white to light yellowish brown at maturity, turning immediately wine red when immature specimens are touched, immediately blue when mature specimens are touched, gradually oxidizing further to black, with a more or less developed stipe that sometimes exceeds 10 mm long, becoming conspicuously slender at the bottom. Peridium thin, mostly less than 0.5 mm thick. Gleba whitish in youth, then becoming brown and nally blackish brown, somewhat watery, composed of darker-colored locules lled with basidiospores and whitish mycelial veins. Stipe (sterile base) well developed compared to other typical species of the genus, often with some subhyaline spots inside, occasionally turning red (in immature basidiomata) or blue (in mature basidiomata) when cut. Odor pleasant, sweet.

Remarks
Orihara et al. (2012a) tentatively described O. tenuipes as "Octaviania sp. E" because, at that time only one collection of an immature basidioma had been examined and the morphology of this new species was not su ciently known. This species has now been recorded from subtropical to temperate regions in Japan, associated with Castanopsis and evergreen Quercus tree species. Morphologically, O. tenuipes tends to have a rather slender and well-developed stipe compared to the other species of Octaviania. Octaviania japonimontana, which is phylogenetically close to O. tenuipes, is somewhat similar morphologically, but O. japonimontana occurs in deciduous Fagaceae forests (with Q. crispula and Fagus spp.) and tends to have basidiomata with thicker peridia and a more rubbery texture. However, these differences are sometimes inconspicuous so molecular methods are sometimes necessary to con rm the species identi cation. Another closely related species, O. durianelloides Orihara, also resembles O. tenuipes when the basidiomata are immature. However, at maturity the basidiomata of O. durianelloides have conspicuous brown scales or warts on the surface, which is unique in the genus.
Octaviania tomentosa Orihara, sp. nov. MycoBank MB 836875 Description Basidiomata sequestrate, up to 15 mm in diam., soft, subglobose, depressed-globose or reniform, surface felty to tomentose, white to dirty white, not becoming yellowish or brownish with age, turning blue or bright red when touched or injured, immature basidiomata tending to turn red rather than blue where touched, after exposure gradually turning black, stipe short, not exceeding 3 mm long, with white rhizomorphs. Peridium usually very thin (less than 0.25 mm) but sometimes variable in thickness, in some cases up to 0.4 mm, context white, showing the same pattern of discoloration as the peridial surface. Gleba beige in youth, becoming brown at maturity, somewhat watery, particularly in young basidiomata, composed of locules lled with yellowish brown to brown basidiospores and whitish mycelial veins, typical of the genus. Stipe (sterile base) context white, sometimes with some subhyaline spots inside. Odor sweet at maturity.

Remarks
This rare species has only been found from four sites in and around the Kanto region in Honshu and from one site in Amami-oshima Island in the Ryukyu island chain despite extensive long-term collecting of Octaviania spp. throughout Japan. These two disjunct areas are about 1200 km apart and the climates and vegetation are also quite different between the two areas (temperate evergreen forests on mainland Japan vs. subtropical forests in the Ryukyu Islands). The multi-locus tree (Fig. 1) as well as the singlegene trees (Fig. 2) clearly show generic divergence between the two disjunct lineages. The specimen from Amami-oshima Island had a thicker peridium than the specimens from the Kanto region (ca. 150-400 µm thick in the Amami-oshima specimen vs. 70-250 um thick in specimens from Kanto). However, we treat these two lineages as infraspeci c variations because of the lesser degree of genetic divergence compared to other species-level divergence in both the species tree and individual gene trees (Figs. 1-2). For instance, the sequence similarity of nLSU between the Amami-oshima specimen and the holotype from Honshu is 99.15% (935 bp / 943 bp), whereas the nLSU similarity between holotypes of O. yaeyamaensis and O. etchuensis, which are genetically the least divergent species within subgenus Octaviania, is 98.31% (875 bp / 890 bp). In addition, we cannot infer the potential mating incompatibility between these geographically isolated, but unculturable specimens.  Habitat, known distribution and seasonality future. Morphologically, it is di cult to characterize this species because most characters of the basidiomata are typical of other taxa in the subgenus Octaviania. However, the distinctive dextrinoid reaction of the basidiospores and the very thin peridiopellis (i.e., an outermost lamentous layer of the peridium) that is sometimes absent in patches are two features that are distinct compared to any of the closely related species. Notably, we also found that the distribution of O. potteri extends to a montane dry forest in Colombia dominated by the ectomycorrhizal host tree Quercus humboldtii. This is the rst record of a true Octaviania from South America. Horak (1964) described Octaviania chilensis Horak from Chile, but this species was later transferred to Stephanospora Pat. in the Agaricales (Vidal 2004). Species of Octaviania subg. Octaviania are always associated with Fagaceae trees, and Colombian Q. humboldtii is considered to have migrated from Central America via the Isthmus of Panama in the Middle to Late Pleistocene (van der Hammen 1974). Thus, it is most likely that O. potteri migrated from North and Central America along with Q. humboldtii, the only oak species that is native to South America.

Discussion
Generic relationships in the Leccinoideae, particularly among the genera Leccinum, Leccinellum, Chamonixia, Octaviania, Rossbeevera and Turmalinea, have never been fully resolved with con dence in previous phylogenetic and systematic studies (e.g., Kuo and Ortiz-Santana 2020; Orihara et al. 2012aOrihara et al. , 2016aWu et al. 2014Wu et al. , 2016. Orihara et al. (2016a) conducted phylogenetic analyses based on three nuclear loci (i.e., ITS, nLSU and TEF1), but only the sister relationship between Rossbeevera and Turmalinea was resolved with strong statistical support (PP ≥ 0.98 and ML BS ≥ 70%). Kuo and Ortiz-Santana (2020) provided a large-scale, multi-locus phylogeny that focused on epigeous Leccinum and Leccinellum species based on the nLSU, TEF1 and RPB2 regions. The resulting phylogenetic tree showed many polyphyletic clades of Leccinum and Leccinellum s.l. within Leccinoideae and most of their phylogenetic relationships were unresolved. Accordingly, Kuo and Ortiz-Santana (2020) treated all the epigeous bolete and hypogeous sequestrate taxa in Leccinoideae as a broadly circumscribed genus Leccinum s.l.. In our study, we incorporated the RPB1 region as well as ITS, nLSU and TEF1 into our multilocus dataset. The resulting trees provided robust phylogenetic support for most of the generic relationships in the Leccinoideae. The two exceptions were the lack of resolution in the branching pattern between Chamonixia and the other genera as well as the phylogenetic placement of some generic-level clades of epigeous Leccinoideae excluded from Leccinum and Leccinellum s.str. (i.e., Leccinellum albellum, L. quercophilum and Leccinum violaceotinctum; Fig. 1). This multi-locus phylogeny also resolved most of the species-level relationships in Octaviania. This exempli es the usefulness of RPB1 for phylogenetic studies on Leccinoideae and highlights the fact that our multi-locus phylogeny shows promise for resolving the genus-level relationships within the Leccinoideae in the near future.
Interestingly, we found that some of the Octaviania specimens had heterogeneous RPB1 sequences compared to the other specimens of the same species. Comparison of the four single-gene tree topologies unexpectedly revealed that one O. tenuipes specimen (KPM-NC 27968) had an RPB1 sequence identical to O. japonimontana (Fig. 2). Furthermore, the RPB1 phylogeny showed that one O. potteri specimen from Minnesota (KPM-NC 17828; RH1181) was not placed within any of the known specieslevel clades in O. subgenus Octaviania. Instead this specimen formed a unique branch that was distant from any species-level clades in the subgenus (Fig. 2). These topological inconsistencies would be most reasonably explained by interspeci c hybridization between two closely related species.
Octaviania tenuipes consistently occurs in evergreen Quercus and Castanopsis forests in subtropical to temperate regions of Japan, whereas O. japonimontana occurs in deciduous Quercus and Fagus crenata forests in mountainous, temperate regions. The possible "hybrid" specimen between O. tenuipes and O. japonimontana (KPM-NC 27968) was collected in the Takakuma mountain range of Kyushu, Japan, which is known as a southern border of the distribution of F. crenata, the potential ectomycorrhizal host of O. japonimontana. Although we have not been able to nd O. japonimontana in that mountain range, it is possible that the two closely related species incidentally inhabited the two adjacent vegetations and hybridized due to minimal putative reproductive barriers between the two species. This potential interspeci c hybridization likely occurred very recently because the "hybrid" O. tenuipes specimen (KPM-NC 27968) has a conserved RPB1 sequence of O. japonimontana (i.e., 100% identical to the other O. japonimontana sequences). The hybrid nature of this specimen was also supported by the network analysis based on the combined dataset of the ITS, nLSU, RPB1 and TEF1 regions (Fig. 3). Stukenbrock (2016) summarized that when two allopatric, fungal species come into contact by chance they more readily hybridized than sympatric species, referring to the case of Neurospora species (Turner et al. 2011). Leducq et al. (2016) clari ed that one North American lineage in Saccharomyces paradoxus is an incipient, hybrid species resulting from secondary contact of two geographically isolated, allopatric lineages after the last glaciation. Similarly, Gladieux et al. (2011) showed that hybridization between two closely related European Microbotryum species tends to be induced by secondary contact following initial divergence in allopatry. Although genome-level genetic comparisons will be needed to verify a potential hybridization between O. tenuipes and O. japonimontana, our hypothesis of hybridization is supported by other cases of hybridization in fungi such as those discussed above.
Similarly, the heterogenous RPB1 sequence in the O. potteri specimen from Minnesota (KPM-NC 17828 [RH1181]) is the result of another interspeci c hybridization between O. potteri and an unknown North American species within subgenus Octaviania. So far, no such species closely related to O. kobayasii has been described, but our results strongly suggest that there is another undescribed species in this lineage from North America. We assume that this unknown Octaviania sp. is sympatric with O. potteri or at least they have bordering distributions somewhere in the upper Midwest of eastern North America. Further collecting surveys for tru e-like fungi in this region may con rm the existence of this unknown Octaviania species in the future.
In addition, the distinct intraspeci c divergence between the O. japonimontana KPM-NC 26723 specimen and the other specimens in the TEF1 phylogeny could be an additional signature of past intraspeci c hybridization (Figs. 2 and 4 g). This topological pattern is similar to that of O. potteri in the RPB1 phylogeny, although the genetic distance between the two lineages is less in the case of O. japonimontana. Therefore, it is likely that an unknown intraspeci c lineage genetically distant to the core O. japonimontana clade exists or existed in the recent past around the habitat of the KPM-NC 26723 specimen. Another possible cause of the topological differences is unusually accelerated molecular evolution of the KPM-NC 26723 specimen, but no such clear divergence was recognized in the other three DNA regions and the rate of divergence in the TEF1 tree seems stable in other species-level clades (Fig. 2). Therefore, accelerated evolutionary rates seems less likely than hybridization. Geographically, the site where this specimen was collected was only about 250 m away from where another O. japonimontana specimen (KPM-NC 26723) was collected on the same day, and we did not nd any clear morphological differences between these specimens. We will refrain from assigning a new taxonomic status of this intraspeci c lineage until specimens that do not show any trace of recent hybridization are collected.
Intra-and inter-speci c hybridization in fungi has been frequently documented in plant pathogens (e.g., (2016a) illustrated that the frequent topological incongruences among gene trees of the sequestrate bolete genus, Rossbeevera, which is closely related to Octaviania, were likely to be derived from intraspeci c gene introgression as well as incomplete lineage sorting (ILS). As far as we know, however, the present study is the rst case that demonstrated interspeci c hybridization in sequestrate basidiomycetes based on molecular evidence. Given the multiple traces of introgression among intraspeci c lineages shown by Orihara et al. (2016a) and the interspeci c hybridization within Octaviania discussed here, it appears that the sequestrate genera in the Leccinoideae may be less reproductively isolated when the two lineages have been ecologically isolated (e.g. there is therefore no need for reproductive isolation to reinforce species boundaries). So far, the precise mechanism of these interspeci c hybridizations has not been discovered and the ploidy of putative hybrid specimens should be examined based on genome-level comparisons and analyses of chromosomes. We nevertheless suppose that interspeci c hybridization may have promoted high genetic diversity within the sequestrate genera of the subfamily Leccinoideae.

Conclusion
The multi-locus phylogeny comprised of ITS and LSU nrDNA, TEF1 and RPB1 sequences provided a robust phylogenetic framework of our study and revealed the phylogenetic placement of two new Octaviania species, O. tenuipes and O. tomentosa collected from Japan. In addition to the phylogenetic analyses, critical morphological examination based on recently collected North American specimens and the holotype concluded that O. asterosperma var. potteri should be treated as an independent species, O. potteri stat. nov. Interestingly, O. potteri exhibited an unusually broad range of distribution (i.e., North America, Japan and Colombia), and this is the rst record of Octaviania from South America. Comparison of the four single-gene tree topologies revealed that one O. tenuipes specimen had an RPB1 sequence identical to O. japonimontana, and one O. potteri specimen from Minnesota had heterogenous RPB1 sequences, one of which was proved to be that of an unknown species of O. subg. Octaviania. The comparison also showed that one O. japonimontana specimen had remarkably divergent TEF1 sequence from the other O. japonimontana specimens. Based on these results and potential phylogeography of these species, we consider that these topological inconsistencies are due to inter-and intra-speci c hybridization between two phylogenetically closely related lineages. Thus, the high genetic and species diversity of Octaviania may partly be due to frequent inter-and intra-speci c hybridization. Further genomic comparison among closely related species and precise population genetics will enlighten the speciation and diversi cation mechanisms within Octaviania and other sequestrate genera in the Leccinoideae.  Comparison of four ML trees of O. subg. Octaviania based on ITS and LSU nrDNA, RPB1 and TEF1 datasets. Branches supported by BS ≥ 75% are depicted as thickened black lines. BS value below 50% are not shown. Specimens that have heterogeneous RPB1 or TEF1 sequences are depicted in coloured fonts. Octaviania potteri KPM-NC 17828 (RH1181) specimen had two highly heterogenous RPB1 sequences (i.e., "seq1" and "seq2") as shown in the red font.