Introduction

The genus Miscanthus Andersson consists of approximately 12 species of C4 perennial grasses that originate in a region ranging from East Asia to the Pacific islands [1]. Miscanthus sacchariflorus (Maxim.) Bentham and Miscanthus sinensis Andersson are the predominant species in Japan. In comparison to M. sinensis, Miscanthus sacchariflorus prefers wetter soil conditions, such as river banks, but their ranges overlap, sometimes in sympatric populations [2]. In Japan, M. sinensis is generally diploid and M. sacchariflorus is tetraploid [3], but diploid M. sacchariflorus grows in China and Korea [4, 5].

Miscanthus grasses are a popular focus of biomass production in temperate regions as a source of renewable energy on account of their high productivity, low fertilizer requirement, and relatively wide adaptability [1]. In particular, Miscanthus ×giganteus Greef & Deuter ex Hodk. & Renvoize is a promising crop in Europe and the USA [1, 6, 7]. Cytogenetic and molecular analyses has shown that M. ×giganteus is an interspecific triploid hybrid between tetraploid M. sacchariflorus and diploid M. sinensis [8, 9]. It shows remarkably high productivity due to heterosis. Its sterility, due to its triploid nature [10, 11], is advantageous for limiting the risk of spread in non-native regions. One extensively studied genotype was exported from Yokohama, Japan in the 1930s by the Danish botanist Aksel Olsen [12, 13]. To expand the genetic variation, breeders have purposely crossed M. sacchariflorus and M. sinensis [1, 14].

Naturally growing interspecific hybrids are also useful for the acquisition of breeding materials. In sympatric population areas, natural interspecific hybrids may arise if the flowering periods overlap [2]. More than half a century ago, studies reported natural putative triploid hybrids between M. sacchariflorus and M. sinensis in central to southern Japan. A putative triploid hybrid, Miscanthus ×ogiformis, was collected in Kumamoto prefecture in Kyushu (southern Japan) and taxonomically named by Honda [15]. Adati [16] found two triploid Miscanthus plants in Hyogo prefecture (western Japan), and Hirayoshi et al. [3] identified triploid plants grown from caryopses collected from M. sinensis in Gifu prefecture (central Japan).

Since the importance of Miscanthus as a biomass crop has grown and molecular techniques such as DNA fingerprinting have become available, the identification of further triploid interspecific hybrids has been reported again in East Asia: Nishiwaki and colleagues [2, 9] used cytological and molecular techniques to identify three interspecific triploid plants grown from caryopses collected from M. sacchariflorus in a sympatric population in Miyazaki prefecture, Japan. Moon et al. [4] identified a triploid Miscanthus plant with a nuclear DNA content similar to that of M. ×giganteus in Chungcheongnam-do, Korea. Ibaragi et al. [17] reexamined plants intermediate in morphology between M. sacchariflorus and M. sinensis that were previously classified as M. sinensis Andersson var. sunanensis Y. N. Lee in Korea, and concluded that this taxon should be treated as M. ×ogiformis (not M. ×giganteus, which is distinguished from M. ×ogiformis by the absence of awns in florets). In China, 11 accessions of interspecific hybrids between M. sacchariflorus and M. sinensis were identified using genome-wide genotyping techniques [18]. In addition to triploid hybrids (M. ×giganteus or M. ×ogiformis), natural diploid hybrids, previously classified as Miscanthus oligostachyus ‘Purpurascens’, were confirmed using cytological and molecular methods in China [19, 20]. Dwiyanti et al. [9] proposed the existence of a tetraploid hybrid between M. sacchariflorus and M. sinensis in Japan.

Hybridization and introgression in plants are thought to contribute to range expansion and adaptation in new environments by increasing genetic variability in hybridizing populations through new gene combinations, changes in polyploidy levels, or both [21, 22]. This hypothesis could be tested by phenotypic and genetic characterization of the natural hybrid population, in addition to the investigation of the habitats and growth environments [e.g., 23]. However, for Miscanthus hybrids, no natural spontaneous populations have been reported.

In this study, we found putative interspecific hybrids of triploid Miscanthus in the Tashiro–Fumoto area of southern Kyushu, Japan; this area lies at the southern limit of the range of M. sacchariflorus. This paper firstly aims to describe this putative hybrid population at the landscape scale using morphological, cytological, and molecular genetic methods. Secondly, we propose an explanation of the process of emergence and dispersal of this triploid Miscanthus hybrid population.

Materials and Methods

Investigation Site

We originally collected one Miscanthus plant in the Tashiro–Fumoto area (31.20°N, 130.85°E, 150–180 m a.s.l.), a small basin along the Fumoto River in Kinko-cho, Kimotsuki-gun, Kagoshima prefecture, Kyushu, Japan (Fig. 1a, b). The Fumoto River runs from east to west in the Tashiro–Fumoto area and is a tributary of Ogawa River, which further flows ca. 9 km into a sea estuary (Fig. 1b, c). The collected plant looked like M. sacchariflorus, but showed some intermediate morphology between M. sacchariflorus and M. sinensis. So we investigated the Miscanthus population in this area.

Fig. 1
figure 1

Geographical distribution of putative triploid hybrids between M. sacchariflorus and M. sinensis, in the Tashiro–Fumoto area, southern Kyushu, Japan. Symbols mark locations of plants sampled; ▲ (blue), type A; ● (red), type B1; ○, type B2; ■ (green), type C. Genotypes were characterized by using 24 simple sequence repeat markers. Dashed circle indicates samples used for seed-set rate and germination tests. Two representative plants nos. 707 and 730 used for chromosome counting are indicated by arrows

Full size image

Plant Materials and Measurements

We sampled putative hybrid plants and surveyed the distribution of the population from 114 sites mainly along the Fumoto River over a range of 2.8 km on 31 October 2012 and 3–4 October 2013. Sampling was performed at at least one site in each canopy we found. Leaf samples from 114 sites were used for flow cytometry assay and DNA analysis. One or two panicles were sampled per site. Panicles sampled from 13 out of 114 sites in 2012 were used to evaluate seed fertility (seed-set and germination rates). Flowering panicles from 39 out of 114 sites sampled in 2013 were used to evaluate pollen viability. Panicles from 29 out of 114 sites sampled in 2013 were used to count awned florets. Awns were counted in 30 florets on each of three rachis branches from one panicle. Plant length (length of a straightened shoot from the ground level to the top) and panicle length (first branch on the main axis to the top) were measured for representative shoots at nine sites. Stems and rhizomes extracted from soil were observed for representative plants. Twenty-three Miscanthus sinensis plants from this area were also analyzed. Ten Miscanthus sacchariflorus accessions collected in southern Kyushu (‘Uchinoura,’ ‘Takayamagawa,’ ‘Kimotsukigawa,’ ‘JM0126,’ ‘Hirosegawa,’ ‘Gotandagawa,’ ‘Sendaigawa’ and ‘Kumagawa’) and other regions in Japan (‘Morioka’ and ‘NARCH-OGI-62’), and one M. ×giganteus clone ‘Illinois’, were also used for each analysis.

Estimation of Nuclear DNA Content and Counting of Chromosomes

The relative nuclear DNA content was measured by flow cytometry assay. One tall fescue (Festuca arundinacea Schreb.) genotype was used as an internal standard. Nuclei prepared from chopped leaves were stained with 4′,6-diamidino-2-phenylindole (DAPI) fluorochrome by using a CyStain UV precise P reagent kit (Partec GmbH, Münster, Germany), and the filtered solution was analyzed with a Particle Analysing System (Partec). Each sample was assayed twice. Chromosome spreads were prepared by the enzymatic maceration/air-drying method [24, 25]. DAPI-stained chromosomes in root-tip squashes of two plants with two representative different genotypes revealed by DNA analysis (nos. 707 and 730, indicated in Fig. 1c) grown in pots in a glasshouse were counted under a fluorescence microscope (BX51; Olympus, Tokyo, Japan).

Genomic DNA Marker Assay and Clustering Analysis

To reveal the degree of genetic uniformity within the putative hybrid population, we examined the extent of genomic DNA polymorphisms among plants by using SSR markers. Genomic DNA was extracted from leaves by the cetyltrimethylammonium bromide (CTAB) method [26]. Table S1 lists the primer sequences of 7 intron-flanking markers used to distinguish M. sacchariflorus and M. sinensis [27] and 24 simple sequence repeat (SSR) markers used to evaluate genetic polymorphisms for phylogenetic analysis [28, 29]. Six random genotypes were used to select SSR primers showing polymorphisms in putative hybrids. In SSR marker analysis, in addition to putative hybrids, six M. sinensis, six M. sacchariflorus, and one M. ×giganteus plants were genotyped. The polymerase chain reaction (PCR) was performed as described [27]. Products amplified with intron-flanking primers were loaded into 2 % or 4 % agarose gels in TAE (Tris, acetic acid, EDTA) buffer, and those amplified with SSR primers were loaded into 6 % acrylamide gels in TBE (Tris, borate, EDTA) buffer for electrophoresis. Following staining with ethidium bromide, the banding patterns were photographed under ultraviolet light. Reproducibility of polymorphisms among genotypes of putative Miscanthus hybrids was confirmed using one representative sample from each type. Genetic distance matrixes based on SSR polymorphisms were calculated using Nei’s index [30]. Clustering was performed by the unweighted pair group method, using the arithmetic mean procedure in R software v. 3.0.2 [31].

Chloroplast DNA Analysis

Four chloroplast DNA regions previously sequenced in some Miscanthus genotypes were analyzed; namely, psbC-trnS, trnS-trnT, trnL-trnF, and rpl20-rps12 [9, 32]. DNA was extracted from 14 putative Miscanthus hybrids, six M. sinensis plants from Tashiro–Fumoto, and ten M. sacchariflorus accessions from other regions of Japan and was used as a PCR template. PCR reaction mixtures (10 μl) contained ca. 100 ng of genomic DNA, 0.2 units of AmpliTaq 360 polymerase (Applied Biosystems), 1× PCR buffer (Applied Biosystems), 1.5 mM MgCl2, 0.2 mM each dNTP, and forward and reverse primers (0.5 mM each) as described by Dwiyanti et al. [9] and Shimono et al. [32]. The cycling regime for the PCR amplification consisted of an initial denaturation step of 2 min at 95 °C; 32 cycles of 30 s at 95 °C, 30 s at 56 °C and 1 min 20 s at 72 °C; and a final extension step of 7 min at 72 °C. PCR products were cleaned up using Exo-SAP IT reagent (Affymetrix, Santa Clara, CA, USA) and directly sequenced using an ABI Prism 3130 Genetic Analyzer (Applied Biosystems). All sequences were deposited in the DDBJ database (accession nos. LC060067 to LC060146). Multiple alignment was performed using DNASIS v. 3.0 software (Hitachi Software Engineering, Tokyo, Japan).

Table 1 Relative fluorescence levels of DAPI-stained nuclei of putative hybrid Miscanthus from the Tashiro–Fumoto area and other plants
Full size table

Evaluation of Pollen Viability and Seed Fertility

Pollen viability of 39 samples of putative hybrids was investigated. Two lots of pollen grains from five to ten florets each per individual were stained with 2 % acetocarmine solution [33]. Fully stained and unstained grains were counted under a microscope.

Seed-set and germination rates of 13 putative hybrids and 3 M. sinensis plants collected in the western part of the Tashiro–Fumoto area were evaluated (Fig. 1c). To estimate the seed-set rate, we used an X-ray inspection apparatus (C-60 TV-PbO-1; Softex, Ebina, Japan) to count filled and empty seeds in three lots of about 100 seeds each per individual. Before the germination test, seeds were maintained at 4 °C over a period of 4 months. Seeds (500–5,000; including empty ones) were placed on wet filter paper and incubated for 4 weeks at 25/15 °C under a daylength of 14 h. The germination ratio (%) was estimated as:

$$ \mathrm{No}.\mathrm{germinated}\kern0.5em \mathrm{seeds}/\left(\mathrm{total}\kern0.5em \mathrm{No}.\kern0.5em \mathrm{tested}\kern0.5em \mathrm{seeds}\times \mathrm{seed}\hbox{-} \mathrm{set}\kern0.5em \mathrm{rate}\right)\times 100 $$

Statistical Analysis

Statistical analyses (ANOVA, Tukey’s and Wilcoxon’s tests) were performed in JMP software v. 9 (SAS Institute, Cary, NC, USA).

Results

Identification of a Putative Hybrid Miscanthus Population in the Tashiro–Fumoto Area of Japan

Putative hybrid Miscanthus plants were distributed mainly alongside the Fumoto River over a range of 2.8 km (Fig. 1c). As well as on river banks, some plants grew along the edges and footpaths of fields and on roadsides near the river. Some plants were found around tributaries, up to 1.4 km upstream of their confluence with the river (Fig. 1c). Miscanthus sinensis also grew naturally in this area. We could not find any putative hybrid plants in the Ogawa River basin.

The putative hybrids had elongated rhizomes, and some plants formed continuous stands up to 50 m long alongside the river or fields (Fig. 2a, b). Stems grew vertically (Fig. 2b), with buds at the nodes like M. sacchariflorus (Fig. 2c, d). The length of the plants on riverbanks ranged from 2.5 to 3.0 m at flowering stage, with panicles measuring 40 to 55 cm. Spikelets had long white callus hairs and either short or no awns on the lemmata (Fig. 2e, f).

Fig. 2
figure 2

Morphology of putative triploid hybrids between M. sacchariflorus and M. sinensis found in the Tashiro–Fumoto area. a Miscanthus plants growing along Fumoto River (hyb, putative triploid hybrid; Msi, M. sinensis). b Putative hybrid plants growing at the edge of a field. c Rhizomes (rh). d Buds at nodes. e Awnless spikelet of a putative hybrid. f Awned spikelet of a putative hybrid. g Spikelet of M. sinensis. h Spikelet of M. ×giganteus ‘Illinois’

Full size image

The mean fluorescence level of DAPI-stained nuclear DNA of the putative hybrids from 114 sites was 0.396 ± 0.008 relative to that of F. arundinacea. It was 1.30 ± 0.03 times the mean fluorescence level of 20 genotypes of M. sinensis (0.305 ± 0.011) growing in this area, almost the same as that of M. ×giganteus ‘Illinois’ (1.32 times; Table 1). The fluorescence level of DAPI-stained nuclear DNA of a tetraploid M. sacchariflorus strain ‘Uchinoura’ was 1.57 times that of M. sinensis. The chromosome number of two putative hybrids sampled was 2n = 57 (Fig. 3a,b), three times the basic number in Miscanthus (x = 19). Lastly, primer pairs, which generate specific fragments to M. sacchariflorus or M. sinensis, amplified five fragments specific to M. sacchariflorus and three specific to M. sinensis in all 114 samples (Fig. 4).

Fig. 3
figure 3

Fluorescence microscopy of DAPI-stained chromosomes of putative triploid hybrids between M. sacchariflorus and M. sinensis in Tashiro–Fumoto , plants no. 707 (type A) and no. 730 (type B1). Genotypes were characterized by using 24 simple sequence repeat markers

Full size image
Fig. 4
figure 4

Genotyping by the intron-flanking markers showing fragments specific to Miscanthus sacchariflorus (Msa) or Miscanthus sinensis (Msi) [28]. Asterisks indicate fragments detected in all M. sinensis genotypes and in some M. sacchariflorus genotypes in the study of Tamura et al. [28]. M, 100-bp ladder marker; 1–10, randomly selected putative Miscanthus plants; 11, M. ×giganteus ‘Illinois’; 12, M. sacchariflorus ‘Uchinoura’; 13, M. sacchariflorus ‘Miyakonojyo’; 14–16, M. sinensis genotypes collected in the Tashiro–Fumoto area

Full size image

All these results indicate that all tested plants are triploid hybrids between M. sacchariflorus and M. sinensis.

Genetic characterization of a putative hybrid Miscanthus population in the Tashiro–Fumoto area

Genotyping of the 114 samples using 24 polymorphic SSR markers revealed 94 polymorphic fragments among all 135 fragments. Only four genotypes were evident: A (n = 63), B1 (n = 47), B2 (n = 2) and C (n = 2). Types B1 and B2 were distinguished by only one fragment specific to each type.

Clustering analysis of the putative hybrids and other Miscanthus accessions revealed three major clusters: M. sinensis, M. sacchariflorus and putative hybrid Miscanthus from Tashiro–Fumoto (Fig. 5). Miscanthus ×giganteus was clustered with M. sacchariflorus (Fig. 5). The putative hybrid cluster was genetically closer to M. sacchariflorus than to M. sinensis. Genetic distances among A, C and B1 + B2 were similar to those among M. sinensis from Tashiro–Fumoto and among M. sacchariflorus from southern Kyushu (Fig. 5).

Fig. 5
figure 5

Phylogenetic dendrogram of putative triploid hybrids between M. sacchariflorus and M. sinensis from Tashiro–Fumoto , and other Miscanthus. Phylogenetic analysis was based on polymorphisms detected by using 24 SSR markers

Full size image

Regarding the geographical distribution of the four types of genotypes, type A plants were distributed mainly in the center of the range along a 0.8-km stretch of the Fumoto River (Fig. 1c). A cluster of type B plants was distributed mainly in the eastern part of the range along a 1-km stretch of the river. The two type B2 plants grew 60 m apart. The two plants of type C grew 12 m apart within a continuous band of vegetation at the edge of the type A area. Types A and B1 plants also grew sympatrically in the western part of the range along the river banks or the edges of fields and roads.

Six randomly selected plants of type A and six of type B1 had the same haplotype of four chloroplast DNA regions within each type. There were no differences between types A and C, or between types B1 and B2 (Table 2). There was only one difference (insertion–deletion in the trnL-trnF region) between the A + C and B groups (Table 2). All mutations identified in the putative hybrids were also present in M. sacchariflorus plants collected in Japan but not in M. sinensis plants in Tashiro–Fumoto area (Table 2). Mutations at the six loci detected in putative hybrids (Table 2) were not reported by Shimono et al. [32], who investigated chloroplast DNA variations in 636 M. sinensis individuals in Japan. The haplotype of types A + C was the same as that of M. sacchariflorus ‘Morioka’, whereas the haplotype of type B was same as those of M. sacchariflorus ‘JM0126,’ ‘Hirosegawa,’ and ‘Kumagawa’ (Table 2). Namely, two chloroplast haplotypes found in the putative hybrids were common to some M. sacchariflorus genotypes but not found in any M. sinensis genotypes.

Table 2 Polymorphic sites in four chloroplast DNA regions of putative hybrid Miscanthus, Miscanthus sinensis from the Tashiro–Fumoto area, and Miscanthus sacchariflorus from southern Kyushu and other regions of Japan
Full size table
Table 3 Proportions of florets with awns in the putative hybrid Miscanthus population in Tashiro–Fumoto
Full size table

Phenotypic Characterization of a Putative Hybrid Miscanthus Population in the Tashiro–Fumoto Area

Most florets of type B plants (B1 & B2) had awns (≥95 %) (Table 3). In contrast, only 25 % of florets of type C plants had awns (Table 3). Despite a wide variance, type A were intermediate with 36.3 % of florets with awns (Table 3).

Acetocarmine staining showed that type A plants had relatively high pollen viability, although there was variation among samples (Fig. 6a). In contrast, most pollen grains of type B1 plants were not viable (Fig. 6b). The difference between types A (73.2 %, Fig. 6c) and B1 (7.1 %, Fig. 6d) was significant (P < 0.001, Wilcoxon’s test).

Fig. 6
figure 6

Pollen viability of putative triploid hybrids between M. sacchariflorus and M. sinensis from Tashiro–Fumoto assessed by acetocarmine staining. a Type A. b Type B1. c, d Frequencies of stained pollen grains of (c) type A (n = 25) and (d) type B (B1 + B2; n = 14)

Full size image

To examine the possibility of expansion of the putative hybrid Miscanthus population by seed propagation, we investigated the seed fertility of putative hybrids and M. sinensis plants growing sympatrically. All panicles collected from putative hybrids contained filled seeds. The seed-set rate of type A plants was significantly lower than those of type B1 and M. sinensis (Table 4). Seedlings were raised from five of the nine samples of type A and all four samples of type B. The germination rate of the putative hybrid seeds (0.4 to 2.1 %) was marginally lower than that of M. sinensis seeds (Table 4).

Table 4 Seed-set rate and germination rate of the putative hybrid Miscanthus population in Tashiro–Fumoto
Full size table

The nuclear DNA content of 26 progeny obtained from eight parents of types A and B1 varied widely, ranging from 0.29 to 0.51 relative to F. arundinacea, which corresponded to 0.94 to 1.68 relative to diploid M. sinensis (Fig. S2). The mean nuclear DNA contents were not significantly different between progeny of types A (n = 14) and B1 (n = 12).

Discussion

The Hybrids Identified in the Tashiro–Fumoto Area are Triploid With a Tetraploid M. Sacchariflorus as Maternal Parent

Since the 1930s, several studies have reported naturally growing hybrids between sympatric M. sacchariflorus and M. sinensis in East Asia [4, 9, 15, 16]. However, these studies focused on individual plants, not populations. In this study, we found a large population of plants morphologically intermediate between M sacchariflorus and M. sinensis. The results of our cytological and genetic analyses indicate that the intermediate type of Miscanthus in the Tashiro–Fumoto area is a triploid hybrid between M. sacchariflorus and M. sinensis.

The genomic DNA contents of the parents are 4.5 pg per nucleus in M. sacchariflorus and 5.5 pg in M. sinensis [8]. On this basis, the DNA content of a triploid hybrid containing two M. sacchariflorus and one M. sinensis genomes is 1.32× that of diploid M. sinensis, and that of a triploid hybrid containing one M. sacchariflorus and two M. sinensis genomes is 1.41×. The mean DNA content of putative hybrids from Tashiro–Fumoto, 1.30, suggests that these plants are hybrids between tetraploid M. sacchariflorus and diploid M. sinensis, similar to M. ×giganteus [8, 9]. Similar to other triploid Miscanthus hybrids (M. ×giganteus) [18, 34], putative hybrids from Tashiro–Fumoto had M. sacchariflorus-specific haplotypes in the chloroplast genome. This suggests that maternal parents of the hybrids from Tashiro–Fumoto were tetraploid M. sacchariflorus.

Process of Emergence and Dispersal of This Triploid Miscanthus Hybrid Population

Why and how has this large (kilometer scale) hybrid population formed? In the Tashiro–Fumoto area, we found M. sinensis but no M. sacchariflorus. The native distribution of M. sacchariflorus is limited to the northern range of M. sinensis, which extends much further south to the tropics [1, 35]. The Global Biodiversity Information Facility (www.gbif.org) records the most southerly occurrence of M. sacchariflorus as 29.57°N in China, further south than the Tashiro–Fumoto area (31.2°N). However, as far as we investigated, the southern limit of M. sacchariflorus in Japan is ca. 9 km north of the Tashiro–Fumoto area (31.3°N, data not shown). The Tashiro–Fumoto area lies about 150–180 m above sea level, and the temperature is a little cooler than in the surrounding lowlands. Therefore, we suppose that the parental M. sacchariflorus had grown in this area in the not-too-distant past before dying out owing to environmental changes such as warming, while the triploid hybrids survived owing to better adaptability introduced from M. sinensis. Clark et al. [36] reported that the genome of tetraploid M. sacchariflorus in Japan has introgressions from M. sinensis (on average 7 % of a whole genome), which are more frequent in southern Japan than in the northern region. They suggested that the ancestral M. sinensis introgression into M. sacchariflorus contributed to the adaptation to a warmer environment [36]. The existence of a large hybrid Miscanthus population at the southern limit of the range of M. sacchariflorus fits this hypothesis as an extreme example of the M. sinensis genome contributing to adaptation.

Following its establishment, how did the hybrid population expand to the landscape scale? On the one hand, we found that the genetic distances among types A, C and B were comparable to those among M. sinensis from Tashiro–Fumoto and among M. sacchariflorus in southern Kyushu. Mutations in chloroplast genomes were also found among the types of putative hybrids. These results demonstrate that the hybrids in the Tashiro–Fumoto area have a polyphyletic origin; namely, they originated from progeny of multiple pairs of M. sacchariflorus and M. sinensis. On the other hand, plants with the same genotype typically formed clusters. The simple genomic and geographic structures (Fig. 1c) imply that the population propagated clonally from a few (probably F1) genotypes. In M. sacchariflorus, vegetative rhizomes are the main reproductive source for population recruitment [37–39]. Rhizome systems can be fragmented and dispersed by natural or human disturbance such as flooding or roadside maintenance [37, 39] and are recognized as a potential invasive risk in M. ×giganteus, especially in riparian ecosystems [40, 41]. According to a chronicle of the history of the area that includes Tashiro–Fumoto, the Fumoto River used to meander and often flooded following heavy rains, but 15–30 years ago it was channelized. The hybrid population area corresponds to the site of these works. Therefore, rhizomes of a few hybrid genotypes (mainly A and B1) might have been fragmented and spread in soil disturbed by floods, river improvement works, or both. Moreover, such habitat disturbances might have not only dispersed the propagules, but also led to the new conditions to which plants may become adapted following hybridization, as observed in other species [e.g., 23]. It is unlikely that the hybrid plants expanded through seed propagation, because they are triploid, and only four types were found by SSR marker analysis. Expansion through seed dispersal would result in a more heterogeneous genetic background and ploidy level, as confirmed in the plants raised from seeds of hybrid plants in this study.

Suggestions Regarding the Classification of Miscanthus Hybrids

The triploid hybrid M. ×giganteus has no awns, but the putative triploid hybrid M. ×ogiformis [11, 15, 16] has awns [15, 17]. On this basis, the Tashiro–Fumoto triploid hybrids should be classified as M. ×ogiformis rather than M. ×giganteus, although Linde-Laursen [11] insisted that M. ×giganteus should be taxonomically reclassified as M. ×ogiformis. However, the presence or absence of awns in the hybrids seems to be genetically variable, as seen in types A and B of the putative hybrids, and to be influenced by environment, as seen in the large variance among type A plants. Therefore, we think that the presence or absence of awns is not a reliable trait for classification.

Conclusion and Prospects

A landscape-scale population of triploid Miscanthus hybrids was discovered in the southern limit area of the range of M. sacchariflorus. We have proposed the hypothesis that under the environment where the growth of M. sacchariflorus was limited, hybridization between M. sacchariflorus and M. sinensis had an advantage due to its improved adaptability. Furthermore, genetic and phenotypic characterization suggest the polyphyletic origin and clonal propagation of this population. Further detailed studies, including genomic comparison with Miscanthus plants growing around this area and evaluation of the environmental adaptability of the hybrids, could test this hypothesis.

Natural Miscanthus hybrids adapted to different environments could be used as materials for breeding biomass crops. In sympatric areas where flowering times of M. sacchariflorus and M. sinensis overlap, such as in central to southern Japan, Miscanthus hybrid populations could occur; in fact, we have confirmed some other hybrid populations in Kyushu (unpublished data). As triploid hybrids can resemble M. sacchariflorus in natural habitats, previously collected M. sacchariflorus germplasms may include natural hybrids or plants with introgression from M. sinensis, which requires further inspections in the future.

Such natural hybrids, i.e., hybrids we found in a natural environment and that occurred spontaneously, set some seeds, some of which germinated. Although further testing is required to evaluate fertility accurately, these results show that the putative hybrids have partial paternal or maternal fertility. Type A plants in particular could be used as a pollen source to genetically improve hybrid Miscanthus as a biomass crop.