Monotropastrum kirishimense (Ericaceae), a new mycoheterotrophic plant from Japan based on multifaceted evidence

Due to their reduced morphology, non-photosynthetic plants have been one of the most challenging groups to delimit to species level. The mycoheterotrophic genus Monotropastrum, with the monotypic species M. humile, has been a particularly taxonomically challenging group, owing to its highly reduced vegetative and root morphology. Using integrative species delimitation, we have focused on Japanese Monotropastrum, with a special focus on an unknown taxon with rosy pink petals and sepals. We investigated its flowering phenology, morphology, molecular identity, and associated fungi. Detailed morphological investigation has indicated that it can be distinguished from M. humile by its rosy pink tepals and sepals that are generally more numerous, elliptic, and constantly appressed to the petals throughout its flowering period, and by its obscure root balls that are unified with the surrounding soil, with root tips that hardly protrude. Based on genome-wide single-nucleotide polymorphisms, molecular data has provided clear genetic differentiation between this unknown taxon and M. humile. Monotropastrum humile and this taxon are associated with different Russula lineages, even when they are sympatric. Based on this multifaceted evidence, we describe this unknown taxon as the new species M. kirishimense. Assortative mating resulting from phenological differences has likely contributed to the persistent sympatry between these two species, with distinct mycorrhizal specificity. Supplementary Information The online version contains supplementary material available at 10.1007/s10265-022-01422-8.

Here, we have applied an integrative taxonomic approach to test whether this taxon should be considered a distinct species from Monotropastrum humile. We investigated discontinuities in flowering phenology and floral morphology that could uniquely diagnose this taxon. We then reconstructed the phylogenetic relationships to examine whether the taxon represents distinct evolutionary lineages, based on MIG-seq [multiplexed inter-simple sequence repeat (ISSR) genotyping by sequencing] data. Finally, given that fungal-associate identity may be relevant in delimiting mycoheterotrophic taxa (Barrett and Freudenstein 2011;Barrett et al. 2022;Freudenstein and Barrett 2014), we investigated the mycorrhizal fungal communities of this taxon and M. humile, including at a sympatric site, using high-throughput DNA sequencing. Our multifaceted evidence leads us to conclude that this taxon is morphologically, phenologically, phylogenetically, and ecologically distinct, and should, therefore, be recognized as a separate species. Consequently, we have described it as a new species, M. kirishimense Suetsugu. Our data are consistent with a study that has shown that fungal host utilization enhances species delimitation in mycoheterotrophic orchids (Freudenstein and Barrett 2014). Our study presents the exciting possibility that a host shift in M. kirishimense, toward a specific Russula lineage, triggered ecological speciation.

Specimen collection and preservation
We collected 50 Monotropastrum kirishimense plants encompassing ten Japanese populations. A total of 38 individuals of M. humile, including five M. humile f. roseum plants, were collected throughout Japan and Taiwan from a total of eight populations, as shown in Table S1. For comparative study, we also collected one specimen from Vietnam [Hsu 10691 (STG00764), hereafter referred to as Monotropastrum sp. 1]. This specimen differs morphologically from typical M. humile, and has glabrous flowers and broad, somewhat ridged fruits (Fig. S1). To minimize disturbance to the local populations, the minimum number of samples required for molecular analysis were collected. However, at least one voucher specimen encompassing the entire plant was deposited in KYO, MZ, TAIF, TI and TNS representing each population. Scale leaves for DNA analysis were immediately dried using silica gel and stored at room temperature until DNA extraction, while 1-3 root fragments (ca. 1 mm in diameter and 3-5 mm in length) were collected from each specimen for molecular barcoding of mycorrhizal fungi. Each root sample was transferred to a 1.5 mL tube containing 99.5% ethanol, and stored at − 20 °C. The herbarium acronyms follow Index Herbariorum (Thiers 2021).

Morphological observation
We compared the morphological characters of Monotropastrum kirishimense, M. humile, and Monotropastrum sp. 1, using the samples listed in Table S1. The morphological variation in M. kirishimense and M. humile was also investigated by reviewing the literature and herbarium specimens (at KYO, TAIF and TI) from other localities. The morphological characters were visually observed under a stereomicroscope and measured using a digital caliper. We note that M. kirishimense is somewhat similar to M. humile f. roseum described from Sadogashima, Niigata Prefecture, Japan, and has rosy pink flowers. Therefore, we examined the M. humile f. roseum type specimen at TI (TI00205063) in detail, and additional M. humile f. roseum specimens from other localities, to identify consistent morphological differences between M. kirishimense and M. humile.

Flowering phenology analysis
Field observations of the flowering phenology from a sympatric site were used to determine whether differences in the flowering phenology play a role in maintaining reproductive integrity between Monotropastrum kirishimense and M. humile. For quantitative comparison, we counted the scapes of M. kirishimense and M. humile from the Onami population (31° 55′ N 130° 50′ E), where both species occur sympatrically, between April 26 and July 22, 2003, and between May 6 and July 17, 2004. They were classified into four developmental stages: (A) emerging (the aboveground organs becoming visible through the leaf litter), (B) flowering (anthers and stigma becoming visible from the perianth tube), (C) wilting (with blackened tepals), and (D) fruiting (ovary becoming large and protruding from the dried out tepals). These stages were counted manually while walking along a fixed route of ca. 500 m, at intervals of approximately 3 weeks.

High-throughput plant phylogenetic analysis
A phylogenetic tree of Monotropastrum plants was constructed based on MIG-seq, which encompasses microsatellite-associated reduced-representation DNA sequencing with restriction site-associated DNA sequencing (RAD-seq) (Suyama and Matsuki 2015). After extracting the genomic DNA from the silica-dried samples using the cetyltrimethylammonium bromide (CTAB) method, we prepared an MIG-seq library, as per Suetsugu et al. (2021b) and Suyama et al. (2022). This included 32 M. kirishimense samples from seven populations, 19 M. humile samples from eight populations, including two M. humile f. roseum individuals, and one Monotropastrum sp. 1 sample (Table S1). The library was sequenced using an Illumina MiSeq Sequencer (Illumina, San Diego, CA, USA) with a MiSeq Reagent Kit v. 3 (150 cycle, Illumina). The raw MIG-seq data were deposited in the DDBJ Sequence Read Archive (DRA, accession number DRA014598).
After removing the primer regions and low-quality sequencing reads (Suetsugu et al. 2021b), 7,014,511 reads (137,539 ± 7234 reads per sample) were obtained from the original 7,804,056 raw reads (153,021 ± 7863 per sample). The Stacks v. 2.60 pipeline was used for de novo singlenucleotide polymorphism (SNP) discovery (Rochette et al. 2019). The following parameters were used: minimum depth of coverage required to create a stack (m) = 3, maximum distance allowed between the stacks (M) = 2, number of mismatches allowed between the sample loci when building the catalog (n) = 2. Only SNPs retained by 26 or more samples were extracted, and SNPs with high heterozygosity (Ho ≥ 0.6) were removed. Moreover, SNP sites with fewer than three minor alleles were filtered out. Finally, 1000 SNPs from 543 loci were provided for the subsequent analyses. SNP-based maximum likelihood (ML) phylogeny was inferred using RAxML v. 8.2.10 (Stamatakis 2014), with a GTR substitution model with Lewis' ascertainment bias correction and 1000 bootstrap replicates.

Molecular analysis of the mycorrhizal fungi
Genomic DNA was extracted from the root tips of 10 Monotropastrum kirishimense plants from four populations (Table S1), and 23 M. humile plants (including two M. humile f. roseum plants) from five populations, using CTAB methods. We amplified the ITS region of the mycorrhizal fungi using the primer set ITS86F/ITS4 (Waud et al. 2014) fused with 3-6-mer Ns and with the Illumina forward/reverse sequencing primer. To add the Illumina sequencing adapters, supplemental PCR was also performed as described in Suetsugu et al. (2021a, b). Equal volumes of each PCR amplicon were pooled and purified using the AMPure XP Kit (Beckman Coulter, CA, USA). The sequencing libraries were processed in an Illumina MiSeq sequencer, with the MiSeq Reagent Micro Kit v. 2 (300 cycles, Illumina, USA). The sequence data were deposited in the DRA (accession number DRA013047).
After sequencing, we performed bioinformatic analysis using Claident v. 0.2.2019.05.10 (Tanabe and Toju 2013), as described in Suetsugu and Matsubayashi (2021). Erroneous sequence reads were removed based on the CD-HIT-OTU method (Li et al. 2012), using the clcleanseqv command in Claident. The remaining sequencing reads were clustered into operational taxonomic units (OTUs) at a 97% threshold similarity, using VSEARCH v. 2.8.0 (Rognes et al. 2016). The OTUs were subjected to de novo and reference-based chimera removal, based on the UCHIME algorithm (Nilsson et al. 2019). The OTU taxonomic assignment was performed based on the query-centric auto-k-nearest-neighbor (QCauto) and the lowest common ancestor (LCA) algorithms (Tanabe and Toju 2013). The functional guild for each fungal OTU was estimated using the FUNGuild database (Nguyen et al. 2016). In subsequent analyses, we used the taxa designated as ectomycorrhizal fungi by FUNGuild, because all monotropoid species obtain their carbohydrates from mycorrhizal fungi that link them to the surrounding trees, on which the fungi form ectomycorrhizae Bruns 2001, 2002;Matsuda et al. 2011;Yokoyama et al. 2005).
Because all the Monotropastrum kirishimense and M. humile plants were predominantly colonized by OTUs assigned to Russula, we downloaded several Russula sequences closely related to the OTUs detected here, based on BLAST searches, from the International Nucleotide Sequence Database Collaboration (INSDC) database. The sequences obtained were aligned using ClustalW in MEGA X (Kumar et al. 2018). The aligned sequences were then used to reconstruct phylogenetic relationships using MEGA X (Kumar et al. 2018) with ML analysis, with a GTR + I + G model and 1000 bootstrap replicates (lnL = − 3108.04). Our infrageneric classification of Russula follows Shimono et al.

Morphological characters
Review and analysis of herbarium specimens, protologues, and living plants revealed few morphological characters that consistently differed between Monotropastrum kirishimense and M. humile (Figs. 1,2,3,4,5). Although the shape of the floral disc is recognized as a diagnostic character between M. humile and its closely related species (Tsukaya et al. 2008), the floral disc of M. kirishimense has thin protrusions elongated to bend backward (Fig. 4d), similar to that of M. humile (Tsukaya et al. 2008).
However, Monotropastrum kirishimense can be distinguished from M. humile by its rosy pink tepals and several other features. First, the flowers of M. kirishimense usually bear 4-9 (up to 11) generally elliptic sepals that are constantly appressed to the petals throughout its flowering period, while the flowers of M. humile usually bear 2-3 (up to 5) generally oblong sepals that are usually spreading during peak anthesis (Figs. 1,2,3,4,5). Second, M. kirishimense flowers and ovaries are more rounded, with a lower aspect ratio than those of M. humile. Although the M. humile ovary also expands and becomes rounded after pollination, the flowers and ovaries of M. kirishimense are stout even before pollination (Figs. 1,2,3,4,5). Third, M. kirishimense is always shorter above ground (typically < 5 cm, vs. > 5 cm in M. humile; Figs. 2a, 3b), whereas the underground stalk is longer in M. kirishimense (typically > 10 cm, vs. < 5 cm in M, humile; Figs. 2c, 3a). More importantly, as suggested by Imamura and Kurogi (2003), the root ball morphology differs completely between these two species. The M. kirishimense root ball is obscure and unified with the surrounding soil, with little protrusion of root tips, and a white mantle indicating mycorrhizal formation (Figs. 3f,4b,5a). Meanwhile, that of M. humile can easily be distinguished from the soil and has relatively distinct root tips (Fig. 2e).
Although Monotropastrum humile f. roseum is also distinguished from M. humile f. humile by its red flowers, Honda (1957) did not describe its characteristics other than floral coloration. Its holotype comprises only the aboveground parts. Nonetheless, due to the spreading oblong-elliptic to obovateelliptic sepals, we can conclude that M. kirishimense is not humile, other than in coloration. Additional sampling has also shown that M. humile f. roseum has a reddish ovary (Fig. 1f), and that M. kirishimense has rosy pink tapels (Fig. 1a-c).

Plant phylogeny
The Monotropastrum ML phylogenetic tree separated M. kirishimense from the remaining taxa, and the monophyly of each clade was highly supported (a 100% bootstrap value; Fig. 7). Although Monotropastrum sp. 1 is not morphologically identical to M. humile, because of its glabrous flower organs and broad, somewhat ridged fruits, it was embedded  Our fungal ITS sequence ML phylogenetic analysis has shown that the Monotropastrum kirishimense mycobionts were clustered nearest to sequences of Russula aff. alboareolata (AB509955), belonging to the subsection Virescentinae (Fig. 9). This subsection has not been reported as a mycobiont of the 93 M. humile individuals collected in Japan, Taiwan, and China across 15 populations (Bidartondo and Bruns 2001;Matsuda et al. 2011;Min et al. 2012;Yokoyama et al. 2005). In contrast, all the Russula OTUs associated with the M. humile individuals investigated in the present study formed a clade with mycobionts previously detected in M. humile (Fig. 9).

Discussion
Our morphological investigation has indicated that Monotropastrum kirishimense can be distinguished from M. humile by its rosy pink tepals, more numerous elliptic sepals constantly appressed to the petals throughout its flowering period, and obscure root balls unified with the surrounding soil, with little protrusion of the root tips. Our MIG-seq-based phylogenetic tree has also shown that M. kirishimense and the other M. humile species complex can be separated into two monophyletic clades, with a 100% bootstrap value. We have therefore determined that M. kirishimense should be treated as an independent species, based on its morphological and phylogenetic distinctness.
Floral trait differentiation between taxa plays a key role in reducing interspecific pollen transfer, either through phenological isolation (a pre-mating barrier caused by differences in flowering time) or floral isolation (a pre-mating barrier caused by differences in morphological, visual, or olfactory traits) (Chapurlat et al. 2020). Given their overall similarity in floral features, the two species are unlikely to experience pollinator-mediated isolation. Indeed, we have observed that Monotropastrum kirishimense and M. humile are mainly pollinated by the bumblebee Bombus diversus (K. Suetsugu, unpublished data). Reproductive isolation between M. kirishimense and M. humile is also unrelated to the spatial separation of these species, given that they grow adjacent to each other. Phenological isolation between plant taxa has substantial potential to lead to reproductive isolation. Monotropastrum humile flowers more than a month before M. kirishimense, with only a brief period (if any) of overlap at the end of M. humile flowering. This divergence in flowering time could be directly selected as a reinforcement mechanism to reduce hybridization before complete speciation (Osborne et al. 2020). Therefore, speciation between M. kirishimense and M. humile may be reinforced by differences in the timing of floral maturation. However, it is also possible that flowering-time divergence could be selected after speciation is complete, as a mechanism to avoid wasting reproductive effort on unfit hybrids (Hopkins 2013). Reproductive asynchrony in flowering time can reduce heterospecific pollen Metabarcoding-based community profiling revealed that Monotropastrum kirishimense and M. humile are predominantly associated with different Russula lineages. Monotropastrum kirishimense is consistently associated with Russula OTU2 (in subsection Virescentinae), even in the Onami population, where M. humile, associated with different OTUs, grows sympatrically within a few meters. Therefore, we can conclude that their genetic characteristics drive the differences in mycorrhizal interaction between the two species. For M. kirishimense, the association with Russula OTU2 encompasses four sampling localities spanning the geographic distribution of the species. Despite extensive studies on its mycorrhizal communities based on 113 individuals collected in Japan, Taiwan, and China across 20 populations (Bidartondo and Bruns 2001;Matsuda et al. 2011;Min et al. 2012;Yokoyama et al. 2005; present study), the subsection Virescentinae has never been reported as a mycobiont of M. humile.
In contrast, all the Russula OTUs associated with M. humile individuals collected from all the five populations in our studies were closely related or identical to OTUs previously reported as mycobionts of M. humile (Bidartondo and Bruns 2001;Matsuda et al. 2011;Min et al. 2012;Yokoyama et al. 2005). Consequently, we conclude that the two species differ in mycorrhizal specificity, and that M. kirishimense has a specialized association with Russula OTU2, although more extensive assessments may reveal that some M. humile individuals are associated with Russula OTU2.  1 cm (b, c), 5 mm (d-i), and 2 mm (j). Drawn by Kumi Hamasaki Although M. kirishimense is widely distributed in Japan, it is specialized on Russula OTU2, and is much rarer locally than M. humile, which is associated with diverse members of the Russulaceae. Although the high host specificity of M. kirishimense may contribute to its rarity, the ecology of its fungal OTU, including its preferred habitat, soil requirements, or fidelity to a specific ectomycorrhizal host tree, remains unexplored. Notably, M. kirishimense occurs only in coniferous forests (dominated primarily by Pinus densiflora), while M. humile commonly occurs in not only coniferous forests, but also other ectomycorrhizal forests, such as fagaceous forests. Russula OTU2 may be preferentially associated with P. densiflora. Further studies are needed to investigate the distribution patterns and abundance of Russula species associated with M. kirishimense and M. humile.
Our findings imply that the distinct mycorrhizal communities play a crucial role in the niche partitioning and coexistence of M. kirishimense and M. humile. Because (i) classical theoretical ecology predicts that two species competing for the same resources cannot stably coexist (Gause 1934) and (ii) mycoheterotrophic plants depend on mycorrhizal fungi for their carbon demands (Merckx 2013), divergent mycorrhizal associations can play a vital role in reducing resource competition. Previous studies have also showed that sympatric (at least initially) mycoheterotrophic plants often have distinct mycorrhizal communities and display strong spatial segregation, even if they share some fungal OTUs (Bidartondo and Bruns 2005;Jacquemyn et al. 2014;McCormick and Jacquemyn 2014;Taylor and Bruns 1999). Our findings may therefore indicate that niche differentiation via segregation of mycorrhizal fungi represents an important mechanism contributing to sympatry. Given that vertical depth partitioning among closely related ectomycorrhizal fungi is a common phenomenon (Mujic et al. 2016;Taylor et al. 2014), the different root depths of M. kirishimense and M. humile may be an adaptation to effectively exploit vertically separated fungal partners.
Furthermore, speciation between Monotropastrum kirishimense and M. humile may be partially due to resource partitioning, with specialization on different fungal hosts leading to reproductive isolation (see also Barrett and Freudenstein 2011;Barrett et al. 2022). Speciation via host shift is one of the most plausible modes of ecological speciation (Calcagno et al. 2007;Fry 2003). This scenario begins with the formation of host races comprising host-affiliated, genetically differentiated groups within the parental species (Drès and Mallet 2002;Jacquemyn et al. 2018). Disruptive selection on host-specificity, indicated by trade-offs in performance between hosts, can lead to further specialization, and promote the formation of two daughter species (Jacquemyn et al. 2018;Rundle and Nosil 2005).
Although little is known about the genetic basis of mycoheterotroph-mycorrhizal associations, the high mycorrhizal specificity observed in many mycoheterotrophs is thought to be the result of physiological fine-tuning to adapt to particular fungi (Hynson and Bruns 2009). Breakdown of coadapted gene complexes controlling host specificity may, therefore, be responsible for postzygotic isolation, in the form of reduced hybrid fitness. Hybridization between ecotypes within a single mycoheterotrophic species with different host specificity can considerably reduce progeny fitness due to a lower probability of mycoheterotrophic growth (Jacquemyn et al. 2016(Jacquemyn et al. , 2018. Consequently, differences in mycorrhizal communities have been suggested to contribute to reproductive isolation among mycoheterotrophic plants ( Barrett and Freudenstein 2011;Barrett et al. 2022;Jacquemyn et al. 2018). Future investigations, including artificial interspecific cross-pollination experiments and in-situ seed baiting, are required to determine whether mycorrhizal associations can prevent hybrid seeds from establishing successful seedlings, thus acting as a post-mating barrier in these two Monotropastrum species. We also note that genotypically distinct M. humile individuals in the different populations tended to be predominantly colonized by different Russula OTUs, highlighting potential race formation within M. humile. However, it is impossible to exclude the possibility that the local availability of Russula species is the primary determinant, because there was no geographic mixing of different genotypes. More extensive sampling across a much broader geographic range would facilitate a more robust understanding of the evolutionary dynamics of mycorrhizal specificity within M. humile.
In summary, we have shown that Monotropastrum kirishimense is distinct from M. humile based on morphology, flowering phenology, and the molecular identity of itself and its fungal partners. Phenological differences (a pre-mating reproductive barrier) and distinct mycorrhizal specificity (a post-mating reproductive barrier) are likely to contribute to the ongoing sympatry of M. kirishimense and M. humile. Mycoheterotrophic plants are often susceptible to environmental destruction, because they are highly dependent on the fungi and the trees that sustain them (Suetsugu et al. 2020). Therefore, many members of the Monotropoideae are restricted to old-growth forests and are now in danger of extinction (Min et al. 2012). The rare and previously unrecognized M. kirishimense can now receive conservation recognition for the first time. This study highlights the importance of integrative taxonomy to avoid under-assessing biodiversity. Suetsugu,sp. nov. (Figs. 1ac,3,4,5) Type. JAPAN. Kagoshima Pref, Kirishima-shi, Makizonocho, Ohnami-ike, 25 June 2019, Kenji Suetsugu KS424 (holotype: KYO!, dried plant on an herbarium sheet and liquidpreserved material in a bottle labeled as the same specimen; isotypes: TI!, TNS!, dried plant on an herbarium sheet).

Diagnosis.
Monotropastrum kirishimense is similar to M. humile but differs in its rosy pink tepals, more numerous (4-11) elliptic sepals constantly appressed to the petals throughout its flowering period, and obscure root balls unified with the surrounding soil, with little protrusion of the root tips.
Terrestrial, mycoheterotrophic herb. Root ball unified with the surrounding soil, with little protrusion of the root tips, 4.7-6.3 cm in diam; roots 0.7-0.9 mm in diam. Stems erect,   Etymology. The species is named after the type locality, Kirishima. To distinguish it from Monotropastrum humile f. roseum (beni-bana-gin-ryo-so, in Japanese) described by Honda (1957), we use Kirishima-gin-ryo-so as a Japanese name, after the type locality.