Introduction
Miscanthus is a perennial rhizomatous grass genus that is currently under intense development as a bioenergy crop. It has, since the late 1970s, come to the attention of the plant breeding community for energy and fibre (1970s onwards; Jones and Walsh, Reference Jones and Walsh2001) and is hence considered undomesticated (Yan et al., Reference Yan, Chen, Luo, Ma, Meng, Li, Zhu, Li, Zhou, Zhu, Han, Ge, Li and Sang2012; Slavov et al., Reference Slavov, Nipper, Robson, Farrar, Allison, Bosch, Clifton-Brown, Donnison and Jensen2014). Cultivars of Miscanthus used as crops or tested in trials are largely clonally propagated (single-genotype), wild sourced material or clonally propagated F1 hybrids (Hodkinson et al., Reference Hodkinson, Chase, Takahashi, Leitch, Bennett and Renvoize2002c; Głowacka et al., Reference Głowacka, Adhikari, Peng, Gifford, Juvik, Long and Sacks2014a, Reference Głowacka, Clark, Adhikari, Peng, Stewart, Nishiwaki, Yamada, Jørgensen, Hodkinson, Gifford, Juvik and Sacksb). There is a need to generate new broadly adapted genotypes suitable for a range of environments including both agricultural and marginal lands (Clifton-Brown et al., Reference Clifton-Brown, Chiang, Hodkinson and Vermerris2008; Chou, Reference Chou2009; Jørgensen, Reference Jørgensen2011; Qin et al., Reference Qin, Zhuang, Zhu, Cai and Zhang2011; Jing et al., Reference Jing, Conijn, Jongschaap and Bindraban2012; Nijsen et al., Reference Nijsen, Smeets, Stehfest and Vuuren2012). There is a movement towards developing crops suited for marginal land so that fertile land is not taken away from food production (Cai et al., Reference Cai, Zhang and Wang2011; Donnelly et al., Reference Donnelly, Styles, Fitzgerald and Finnan2011; Gopalakrishnan et al., Reference Gopalakrishnan, Cristina Negri and Snyder2013). For example, the EU FP7 project GrassMargins aims to develop genotypes suitable for growth on European marginal land (http://www.grassmargins.com). Furthermore, China possesses 100 million hectares of marginal and degraded land, especially in the northern and western regions, that has the potential to produce approximately 1 billion tons of Miscanthus feedstock (Sang, Reference Sang2011; Sang and Zhu, Reference Sang and Zhu2011). To achieve this potential, many plant traits will need to be optimized including yield, flowering, drought tolerance, frost and cold tolerance, and biomass chemical composition (reviewed in Jones et al. (Reference Jones, Finnan and Hodkinson2014)).
Miscanthus breeding is a case study of genetic resource utilization for novel crop development. The genus has a wealth of genetic resources, and the progress made in characterizing and utilizing this diversity is outlined in this review. We do not consider reverse-genetic studies and genetic engineering approaches for crop development here. Such details can be found elsewhere (Wang et al., Reference Wang, Yamada, Kong, Abe, Hoshino, Sato, Takamizo, Kanazawa and Toshihiko2011; Xie and Peng, Reference Xie and Peng2011; Feltus and Vandenbrink (Reference Feltus and Vandenbrink2012); Perera et al., Reference Perera, Baldwin and Reichert2013). We instead focus on the problems and prospects of using natural genetic variation for Miscanthus crop production. Much progress has been made on the fundamental characterization of Miscanthus species such as on their taxonomy and phylogenetics. Furthermore, several studies have outlined population genetic variation and examined adaptive variation of a range of genotypes. The ongoing challenge is to combine the genotypic and phenotypic knowledge for crop development and to better incorporate natural genetic diversity into breeding programmes. Next-generation sequencing and breeding technologies utilizing association studies and genomic selection (GS) offer considerable potential in this respect. Although many genetic resource collections of Miscanthus exist in Europe and the Americas outside of Asia, there is neither a directory of Miscanthus collections nor a coordinated programme for the conservation of its genetic resources.
Taxonomy, phylogeny and distribution
Miscanthus sensu lato (s.l., in the broad sense) includes about 20 species depending on the author (Clayton and Renvoize, Reference Clayton and Renvoize1986; Scally et al., Reference Scally, Hodkinson, Jones, Jones and Walsh2001a, Reference Scally, Waldren, Hodkinson and Jonesb; Clayton et al., Reference Clayton, Vorontsova, Harman and Williamson2006 onwards). However, its generic limits have been revised based on molecular phylogenetics (Hodkinson et al., Reference Hodkinson, Chase and Renvoize1997, Reference Hodkinson, Renvoize and Chase2002a; Swaminathan et al., Reference Swaminathan, Alabady, Varala, De Paoli, Ho, Rokhsar Dan, Arumuganathan, Ming, Green, Meyers, Moose and Hudson2010). DNA sequences and fingerprinting data reported by Hodkinson et al. (Reference Hodkinson, Renvoize and Chase2002a, Reference Hodkinson, Chase, Lledó, Salamin and Renvoizeb) showed that some species included in Miscanthus s.l. are more closely related to other genera than to Miscanthus. Miscanthus sensu stricto (s.s., in the strict sense) includes only those species with a basic chromosome number of 19. Its taxonomic type species is M. floridulus (Labil.) Warb. ( = M. japonicus Anderss; basionym Saccharum floridulum Labillardière described in 1824).
Synonymy is high in the genus. The International Plant Names Index (IPNI, 2014) lists over 60 species, but only 11–12 species can be recognized in Miscanthus s.s. (Table 1). Although hybridization is known to occur within the genus, few hybrids have been identified and named despite the lack of breeding barriers and the sympatry of several taxa. Miscanthus× giganteus Greef et Deuter ex Hodkinson and Renvoize was described by Hodkinson and Renvoize (Reference Hodkinson and Renvoize2001). They showed that the name M. × giganteus Greef et Deuter is illegitimate because neither the type was specified nor a Latin description was provided. They chose to keep the species epithet × giganteus to prevent confusion in the literature, but updated the authority names accordingly. New records of natural hybridization between M. sacchariflorus and M. sinensis have been reported (Nishiwaki et al., Reference Nishiwaki, Mizuguti, Kuwabara, Toma, Ishigaki, Miyashita, Yamada, Matuura, Yamaguchi, Rayburn, Akashi and Stewart2011). The name Miscanthus ogiformis is not correctly applied to Miscanthus× giganteus as it does not recognize the hybrid nature of the taxon and cannot be linked to the type (Ibaragi et al., Reference Ibaragi, Lim, Yook, Chang and Kim2013). Miscanthus floridulus and M. sinensis also have sympatric distributions and similar morphology. Phenotypic evaluation of these show that the two species intergrade in their morphology and that hybrids are potentially common (Scally et al., Reference Scally, Hodkinson, Jones, Jones and Walsh2001a). There is clearly a need for more research on natural hybrids and hybrid zones in Miscanthus and the taxonomic treatment of these taxa.
List of Saccharinae genera and species belonging to Miscanthus s.s.
spp., species.
a Likely to be an infraspecific taxon of M. sacchariflorus.
Miscanthus is classified in the predominantly tropical grass tribe Andropogoneae and subtribe Saccharinae (Clayton and Renvoize, Reference Clayton and Renvoize1986; Clayton et al., Reference Clayton, Vorontsova, Harman and Williamson2006 onwards; Bouchenak-Khelladi et al., Reference Bouchenak-Khelladi, Salamin, Savolainen, Forest, Bank, Van Der, Chase and Hodkinson2008; Teerawatananon et al., Reference Teerawatananon, Jacobs and Hodkinson2011; Kellogg, Reference Kellogg2013). Saccharinae includes the sugarcane genus Saccharum L. s.l. and several less well-known genera (Table 1). The term ‘Saccharum complex’ has been used to describe a taxonomically difficult subset of Saccharinae (Erianthus, Miscanthus, Narenga, Saccharum and Sclerostachya) implicated in the origin of sugarcane (Daniels and Roach, Reference Daniels, Roach and Heinz1987). Miscanthus species are unusual among Andropogoneae because they have bisexual paired spikelets, both with hermaphrodite flowers (Fig. 1). Other Andropogoneae have paired spikelets, but with the exception of a few genera such as Ischaemum L. and Schizachyrium Nees, one of these is usually male or sterile (Clayton and Renvoize, Reference Clayton and Renvoize1986).
Line drawings of (a) Miscanthus sinensis and (b) M. sacchariflorus (from Sun et al. (Reference Sun, Lin, Yi, Yang Z- and Zhou2010), with permission). (a) A, Panicle and leaf; B, paired spikelets; C, back of a lower glume; D, ventral side of an upper glume; E, ventral side of an upper lemma with awn; and F, stamens and gynoecium. (b) A, Rhizome and culm; B, panicle and leaf; C, paired spikelets; D, back and ventral sides of a lower glume; E, ventral side of an upper glume; F, back of a lower lemma; G, back of an upper lemma without awn; and H, lodicule.
Morphological descriptions of Miscanthus are included in several floras including Chen and Renvoize (Reference Chen and Renvoize2006) for China, Koyama (Reference Koyama1987) and Osada (Reference Osada1993) for Japan, Cope (Reference Cope, Nasir and Ali1982) for Pakistan, Gilliand (Reference Gilliand1971) for Malaya and Hodkinson (submitted) for Thailand. Miscanthus species are perennial and rhizomatous (Fig. 1) with erect cane-like stems growing up to 7 m tall (in M. lutarioriparius = M. sacchariflorus). They are sometimes tufted with short rhizomes. The inflorescence is terminal and bears plumose racemes. Its spikelets are pedicellate and paired (one with a short pedicel and another with a long pedicel). The inflorescence axis may be long and have relatively short racemes as in M. floridulus or may be short with long racemes (subdigitate inflorescence, as in most M. sinensis and M. sacchariflorus; Fig. 1).
Some comparative morphological and anatomical studies have been published on Miscanthus (Lee Reference Lee1964a, Reference Leeb, Reference Leec, Reference Leed; Scally et al., Reference Scally, Hodkinson, Jones, Jones and Walsh2001a, Reference Scally, Waldren, Hodkinson and Jonesb; Sun et al., Reference Sun, Lin, Yi, Yang Z- and Zhou2010). These studies helped define species boundaries, improved infrageneric classification and quantified morphological variation. Important diagnostic characteristics are found in inflorescence axis length, raceme length and number, spikelet size, spikelet callus hair length, glume and lemma size, nerves on glumes, dorsal hairs of glume, and presence or absence of awns (Lee Reference Lee1964a, Reference Leeb, Reference Leec, Reference Leed; Scally et al., Reference Scally, Hodkinson, Jones, Jones and Walsh2001a, Reference Scally, Waldren, Hodkinson and Jonesb; Chen and Renvoize, Reference Chen and Renvoize2006). For example, Scally et al. (Reference Scally, Hodkinson, Jones, Jones and Walsh2001a, Reference Scally, Waldren, Hodkinson and Jonesb) used 31 morphological characteristics predominantly from spikelets and the inflorescence to study variation in Miscanthus species using principal component analysis and detrended correspondence analysis. Miscanthus sacchariflorus and M. sinensis were clearly differentiated with these methods, but the other species clustered with the M. sinensis group. There is huge morphological variation present in M. sinensis. A standardized list of morphological descriptors has not yet been published, but would be of high value for phenotyping studies (Scally, Reference Scally2001; De Cesare, Reference De Cesare2012).
Groups of species at sectional rank within Miscanthus have been described and keys to Miscanthus species provided by Hodkinson et al. (Reference Hodkinson, Chase and Renvoize1997) and Chen and Renvoize (Reference Chen and Renvoize2006). The most comprehensive effort to taxonomically subdivide the genus was made by Lee (Reference Lee1964a, Reference Leeb, Reference Leec, Reference Leed), who separated the genus into four sections. Three can be assigned to Miscanthus s.s. (sections Kariyasua, Miscanthus and Triarrhena) and one (section Diandra) is not part of Miscanthus s.s. because of DNA sequence evidence (Hodkinson et al., Reference Hodkinson, Renvoize and Chase2002a) and chromosome number (Fig. 4; Table 2). Section Diandra species also have two anthers compared with three anthers in Miscanthus s.s. Other species assigned to Miscanthus s.l. are better included in Miscanthidium (an African taxon; M. ecklonii, M. junceus and M. sorghum, M. violaceus), Sclerostachya fusca and Diandranthus (various combinations including M. nepalensis and M. nudipes) (Hodkinson et al., Reference Hodkinson, Renvoize and Chase2002a).
List of chromosome studies carried out on the x=19 Miscanthus s.s. taxaa
a Several other taxa classified as Miscanthus s.l. do not share a basic chromosome number of 19; they are more commonly based on 10 or 15 such as M. fuscus, n= 15 (Li, Reference Li, Loh and Lee1959); M. nepalensis, n= 20 (Mehra et al., Reference Mehra, Khosla, Kohli and Koonar1968); M. nudipes, n= 20 (Mehra et al., 1968); Miscanthidium violaceum, n= 14 (Brett, Reference Brett1954); and Narenga porphyrocoma, n= 15 (Burner et al., Reference Burner1991).
b Linde-Laursen recorded mostly 58 chromosomes with some at 57.
c Based on flow cytometry.
Miscanthus s.s. is native to Eastern Asia, Southeastern Asia and the South Pacific (Fig. 2), with the highest species diversity being recorded in Eastern Asia, especially in China and Japan (Chen and Renvoize, Reference Chen and Renvoize2006; Sun et al., Reference Sun, Lin, Yi, Yang Z- and Zhou2010). Its native latitudinal range extends from temperate Southeast Russia at 50°N to tropical Polynesia at 22°S. Its native longitudinal distribution extends from Burma and Andaman and Nicobar Islands at 92°E to Fiji at 179°W. Its species have radiated to occupy a wide range of biomes and climatic zones. Some species such as M. floridulus generally grow at sea level or in warm tropical climates, but others such as M. paniculatus can tolerate high altitudes of up to 3100 m on dry mountain slopes of Guizhou, Sichuan and Yunnan in China (Chen and Renvoize, Reference Chen and Renvoize2006).
Geographical distribution of Miscanthus s.s. species (from Clifton-Brown et al. (Reference Clifton-Brown, Chiang, Hodkinson and Vermerris2008), with permission). The range of M. × giganteus is not known, but can potentially occur in sympatric regions of M. sinensis and M. sacchariflorus. The distribution of M. intermedius (Japan) is not shown.
Given such a wide native distribution, it is not surprising that Miscanthus has also become naturalized following human introduction in many regions of the world including Eurasia, North and South America, and New Zealand (Meyer et al., Reference Meyer, Paul and Anderson2010; Quinn et al., Reference Quinn, Allen and Stewart2010, Reference Quinn, Matlaga, Stewart and Davis2011, Reference Quinn, Culley and Stewart2012; Barney et al., Reference Barney, Mann, Kyser and DiTomaso2012; Matlaga et al., Reference Matlaga, Quinn, Davis and Stewart2012; Clark et al., Reference Clark, Brummer, Głowacka, Hall, Heo, Long, Peng, Yamada, Yoo, Yu, Zhao and Sacks2014). Clark et al. (Reference Clark, Brummer, Głowacka, Hall, Heo, Long, Peng, Yamada, Yoo, Yu, Zhao and Sacks2014) used high-density single-nucleotide polymorphism (SNP) markers to show that naturalized populations of M. sinensis were derived from a subset of ornamental cultivars that were themselves derived from Southern Japan.
Chromosome variation
The chromosomes of Miscanthus s.s. are relatively small, generally 25 μm in metaphase of mitosis (Adati, Reference Adati1958; Burner, Reference Burner1991; Linde-Laursen, Reference Linde-Laursen1993; Hodkinson et al., Reference Hodkinson, Chase and Renvoize2001; Chramiec-Głąbik et al., Reference Chramiec-Głąbik, Grabowska-Joachimiak, Sliwinska, Legutko and Kula2012), compared with those of some grasses, but are not unusual in Panicoideae (Celarier and Paliwal, Reference Celarier and Paliwal1957; Sede et al., Reference Sede, Escobar, Morrone and Zuloaga2010). Early studies on M. floridulus and M. sinensis failed to reach a consensus on the basic (monoploid) chromosome number of the genus (Avdulov, Reference Avdulov1928, Reference Avdulov1931; Church, Reference Church1929; Hunter, Reference Hunter1930). However, subsequent meiotic and mitotic counts of M. floridulus, M. × giganteus, M. intermedius, M. oligostachyus, M. sacchariflorus, M. sinensis and M. tinctorius (Table 2 and Fig. 3) established the basic number at x= 19 (Bremer, Reference Bremer1934; Li et al., Reference Li, Shang, Hsiao and Yong1948; Li and Ma, Reference Li, Ma and Hughes1951; Adati and Mitsuishi, Reference Adati and Mitsuishi1956; Adati, Reference Adati1958). Regular meiotic behaviour with 19 bivalents has been observed in all Miscanthus s.s. with 2n= 38 chromosomes. Further evidence for x= 19 comes from the examination of chromosome numbers in polyploids, ranging from diploids to hexaploids (Table 2) that are represented by multiples of 19 (Adati and Shiotani, Reference Adati and Shiotani1962). Karyotypes have been described by Adati (Reference Adati1958) for M. floridulus, M. intermedius, M. oligostachyus, M. sacchariflorus, M. sinensis and M. tinctorius, by Lafferty and Lelley (Reference Lafferty and Lelley1994) for M. × giganteus, and by Chramiec-Głąbik et al. (Reference Chramiec-Głąbik, Grabowska-Joachimiak, Sliwinska, Legutko and Kula2012) for M. × giganteus, M. sacchariflorus and M. sinensis.
Chromosomes of Miscanthus in metaphase of mitosis (from Hodkinson et al. (Reference Hodkinson, Chase and Renvoize2001), with permission). (a, b) M. floridulus (2n= 2x= 38), (c) M. × giganteus (2n= 3x= 57).
Adati and Shiotani (Reference Adati and Shiotani1962) proposed that the x= 19 basic chromosome number of Miscanthus is of allopolyploid origin from two parental lineages with x= 10 and x= 9, but this hypothesis remains to be rigorously tested. Recent mapping studies have shown a high similarity of the Miscanthus genome to the Sorghum genome and indicated whole-genome duplication in Miscanthus relative to Sorghum (Kim et al., Reference Kim, Zhang, Auckland, Rainville, Jakob, Kronmiller, Sacks, Deuter and Paterson2012; Ma et al., Reference Ma, Jensen, Alexandrov, Troukhan, Zhang, Thomas-Jones, Farrar, Clifton-Brown, Donnison, Swaller and Flavell2012; Swaminathan et al., Reference Swaminathan, Chae, Mitros, Varala, Xie, Barling, Glowacka, Hall, Jezowski, Ming, Hudson, Juvik, Rokhsar Daniel and Moose2012). Ma et al. (Reference Ma, Jensen, Alexandrov, Troukhan, Zhang, Thomas-Jones, Farrar, Clifton-Brown, Donnison, Swaller and Flavell2012) used genotyping by sequencing (GBS) of diploid x= 19 M. sinensis to demonstrate that Miscanthus is an ancient polyploid relative to Sorghum bicolor consisting of two subgenomes. Each pair of the 19 M. sinensis linkages aligned to one sorghum chromosome, except one that mapped to two sorghum chromosomes. Swaminathan et al. (Reference Swaminathan, Chae, Mitros, Varala, Xie, Barling, Glowacka, Hall, Jezowski, Ming, Hudson, Juvik, Rokhsar Daniel and Moose2012) used RNA sequencing (RNA-Seq)-based markers to also determine 19 linkage groups and showed the genome-wide duplication in Miscanthus relative to Sorghum with subsequent insertional fusion of a pair of chromosomes. Whether this ancient duplication in the Miscanthus genome involved allopolyploidy or autopolyploidy remains to be determined (Ma et al., Reference Ma, Jensen, Alexandrov, Troukhan, Zhang, Thomas-Jones, Farrar, Clifton-Brown, Donnison, Swaller and Flavell2012; Swaminathan et al., Reference Swaminathan, Alabady, Varala, De Paoli, Ho, Rokhsar Dan, Arumuganathan, Ming, Green, Meyers, Moose and Hudson2010, Reference Swaminathan, Chae, Mitros, Varala, Xie, Barling, Glowacka, Hall, Jezowski, Ming, Hudson, Juvik, Rokhsar Daniel and Moose2012).
The basic chromosome number of 19 in Miscanthus s.s. does not correspond to some other Miscanthus (sensu Clayton and Renvoize, Reference Clayton and Renvoize1986) species including Asian M. fuscus, M. nepalensis and M. nudipes and African M. ecklonii, M. junceus, M. sorghum and M. violaceus that generally have a basic chromosome number of 10 or 15 (Table 2; footnote). These taxa are better treated in genera separate from Miscanthus (Hodkinson et al., Reference Hodkinson, Renvoize and Chase2002a; as described above).
Genome size variation
Genome size has been studied by flow cytometry in Miscanthus and found to exhibit considerable variation among species (Table S1, available online). Rayburn et al. (Reference Rayburn, Crawford, Rayburn and Juvik2009), using three accessions of each species, showed that diploid M. sinensis had a 1C nuclear DNA content of 2.75 pg and diploid M. sacchariflorus 2.25 pg. Therefore, they estimated the genome size of diploid M. sinensis to be approximately 20% greater than that of diploid M. sacchariflorus.
Li et al. (Reference Li, Hu, Luo, Zhu, Li, Luo, Li and Yan2013) examined nuclear DNA content variation in M. lutarioriparius, M. sacchariflorus and M. sinensis collected from a range of habitats, altitudes and latitudes in China. They found little variation among the species at the diploid level, suggesting that genome size was stable within the species (among populations). However, in accordance with the results reported by Rayburn et al. (Reference Rayburn, Crawford, Rayburn and Juvik2009) and De Cesare (Reference De Cesare2012), their results indicated a large difference among diploid species (1C = 2.69 pg in M. sinensis compared with 2.19 pg in M. sacchariflorus and M. lutarioriparius). Li et al. (Reference Li, Hu, Luo, Zhu, Li, Luo, Li and Yan2013) also estimated the genome sizes of tetraploid accessions of M. sacchariflorus and M. lutarioriparius and found that they had smaller genomes than expected when compared with the genome sizes of their diploid progenitors (1C = 4.27 pg and 4.28 pg compared with the expected value of 4.37 pg). This could indicate genome downsizing after polyploidization (Leitch et al., Reference Leitch, Hanson, Lim, Kovarik, Chase, Clarkson and Leitch2008; Bento et al., Reference Bento, Gustafson, Viegas and Silva2011).
Li et al. (Reference Li, Hu, Luo, Zhu, Li, Luo, Li and Yan2013) did not include M. × giganteus in their studies, but Rayburn et al. (Reference Rayburn, Crawford, Rayburn and Juvik2009) showed that triploid M. × giganteus had a total nuclear content of 7.0 pg, diploid M. sacchariflorus had a 1C content of 2.25 pg and diploid M. sinensis had a 1C content of 2.75 pg (Table S1, available online). Rayburn et al. (Reference Rayburn, Crawford, Rayburn and Juvik2009) therefore, by simple deduction from predicted genome sizes, provided evidence that M. × giganteus is more likely the result of a combination of a 2 × M. sacchariflorus gamete and a 1 × M. sinensis gamete (sum 4.5+2.75 = 7.25 pg) than that of a 2 × M. sinensis gamete and a 1 × M. sacchariflorus gamete (5.5+2.25 = 7.75 pg).
From these values, it is possible to estimate genome size in base pairs (bp) for the following three species: M. × giganteus, M. sacchariflorus and M. sinensis (diploids to tetraploids; higher-ploidy plants excluded). The genomes (Table S1, available online), ranging in estimated size from 2.1 Gbp (diploids) to 5.62 Gbp (tetraploids), are large in comparison with those of Arabidopsis (125 Mbp), similar in size to those of maize (2.3 Gbp), small in comparison with those of bread wheat (17 Gbp) and tiny in comparison with the largest genome measured thus far, Paris japonica (Pellicer et al., Reference Pellicer, Fay and Leitch2010), of 150 Gbp.
Swaminathan et al. (Reference Swaminathan, Alabady, Varala, De Paoli, Ho, Rokhsar Dan, Arumuganathan, Ming, Green, Meyers, Moose and Hudson2010) used genomic and small RNA-Seq to characterize the genome of M. × giganteus. Coding regions were found to show a high sequence similarity to those in other grasses, but 95% of the genome was found to fall within 12 repeat classes of DNA related to transposons or centromeric DNA. The major repeats actively produce small RNAs. Most small RNAs (sRNAs) in grasses are in the 24-nucleotide size range (probably small interfering RNA (siRNAs)). Retrotransposons (class 1 transposons) are the most common sRNA (32%), followed by DNA transposons (class 2 transposons). Thus, siRNAs were suggested to represent a large component of the small-RNA transcriptome of Miscanthus (Swaminathan et al., Reference Swaminathan, Alabady, Varala, De Paoli, Ho, Rokhsar Dan, Arumuganathan, Ming, Green, Meyers, Moose and Hudson2010).
Polyploidy
Ploidy estimation in Miscanthus has been achieved by flow cytometry and counting techniques (Table 2). Miscanthus is a polyploid complex with diploids, triploids, tetraploids, pentaploids and hexaploids. Hodkinson et al. (Reference Hodkinson, Chase, Takahashi, Leitch, Bennett and Renvoize2002c) used amplified fragment length polymorphism (AFLP) fingerprinting in combination with chromosome counting to show that many plants labelled as M. sacchariflorus are in fact M. × giganteus. Morphologically these taxa are hard to separate even with flowering specimens (Hodkinson and Renvoize, Reference Hodkinson and Renvoize2001; Hodkinson et al., Reference Hodkinson, Renvoize and Chase2002a), and a combination of methods including ploidy determination is often required to correctly assign a name to specimens. Many of the M. sacchariflorus polyploids have been assigned infraspecific status including M. sacchariflorus var. brevibarbis (triploid), M. sacchariflorus var. glaber (triploid), M. ogiformis (triploid) and M. sacchariflorus f. latifolius (pentaploid).
Evidence for autopolyploidy has been provided for some genotypes or taxa including autotriploid M. sinensis ‘Goliath’ (De Cesare, Reference De Cesare2012), autotriploid M. sinensis var. condensatus (Adati and Mitsuishi, Reference Adati and Mitsuishi1956; Adati, Reference Adati1958), M. sinensis ‘Autumn Light’ (Swaminathan et al., Reference Swaminathan, Alabady, Varala, De Paoli, Ho, Rokhsar Dan, Arumuganathan, Ming, Green, Meyers, Moose and Hudson2010), autotetraploid M. sacchariflorus (Adati and Mitsuishi, Reference Adati and Mitsuishi1956; Adati, Reference Adati1958) and autotriploid M. sacchariflorus. However, it is possible that some of these taxa are the result of hybridization and hence allopolyploidy.
Evidence for allopolyploidy has been provided for several Miscanthus taxa (Adati and Shiotani, Reference Adati and Shiotani1962), but most notably for M. × giganteus (Linde-Laursen, Reference Linde-Laursen1993; Hodkinson et al., Reference Hodkinson, Chase, Lledó, Salamin and Renvoize2002b; Nishiwaki et al., Reference Nishiwaki, Mizuguti, Kuwabara, Toma, Ishigaki, Miyashita, Yamada, Matuura, Yamaguchi, Rayburn, Akashi and Stewart2011) and some taxa in the M. sacchariflorus complex including allotriploid M. sacchariflorus var. brevibarbis and M. sacchariflorus var. glaber (Adati and Shiotani, Reference Adati and Shiotani1962). Adati and Shiotani (Reference Adati and Shiotani1962) used karyotype analysis and observations of chromosome pairing in meiosis to show that some tetraploid M. sacchariflorus are of allopolyploid origin. These tetraploids were composed of two different chromosomal sets, one with a satellite chromosome and another without a satellite chromosome. Two sets are homologous to M. sinensis and two partially homologous. They also argued, on the basis of meiotic and morphological studies, that pentaploid M. sacchariflorus var. latifolius is an allopolyploid combining genomes of M. sacchariflorus and M. sinensis and that M. intermedius is an allopolyploid combining genomes of M. oligostachyus and M. tinctorius.
Origin of Miscanthus ×giganteus and polyploid M. sacchariflorus taxa
The allopolyploid origin of M. × giganteus has been established via morphological, geographical, cytogenetic, molecular genetic and pollen fertility/seed viability studies. Linde-Laursen (Reference Linde-Laursen1993) examined meiotic pairing in M. × giganteus and found few trivalents and nearly equal numbers of bivalents and univalents, which indicates that two of the three genomes have high homology and one has low homology to the other two. All pollen grains were sterile with two to five apertures (compared with single-aperture grains in fertile Miscanthus). Meiotic pairing in M. × giganteus contrasts with that in autotriploid M. sinensis ssp. condensatus, which was shown to have a high number of trivalents in pollen mother cells at metaphase 1 (Adati, Reference Adati1958).
Hodkinson et al. (Reference Hodkinson, Chase, Lledó, Salamin and Renvoize2002b) used nuclear ribosomal DNA sequences from the internal transcribed spacer (ITS) region to show that both M. sinensis and M. sacchariflorus were the parental genome donors of M. × giganteus. One ITS repeat type in M. × giganteus matched M. sinensis and the other M. sacchariflorus (Fig. 4). AFLP and inter-simple-sequence repeat (ISSR) fingerprinting also confirmed this observation. The molecular cytogenetic techniques such as fluorescent in situ hybridization and genomic in situ hybridization were unable to differentiate among the different parental genomes present in M. × giganteus, indicating that the parental genomes of the triploid are extremely similar at the repetitive DNA level.
Summary of phylogenetic relationships in Miscanthus and related taxa based on nuclear ribosomal sequences (adapted from Hodkinson et al. (Reference Hodkinson, Renvoize and Chase2002a, Reference Hodkinson, Chase, Lledó, Salamin and Renvoizeb, Reference Hodkinson, Chase, Takahashi, Leitch, Bennett and Renvoizec) and Swaminathan et al. (Reference Swaminathan, Alabady, Varala, De Paoli, Ho, Rokhsar Dan, Arumuganathan, Ming, Green, Meyers, Moose and Hudson2010)). 
, Note M. sacchariflorus and M. × giganteus accessions share plastid haplotypes, indicating that M. sacchariflorus is the ovule donor and M. sinensis the pollen donor of all M. × giganteus hybrids studied thus far by the current authors. 
, well-supported clades.
Plants classified as M. sacchariflorus also have complex ancestry and are difficult to classify and name because chromosome complements range from diploid to pentaploid (Adati, Reference Adati1958; Adati and Shiotani, Reference Adati and Shiotani1962; Fedorov, Reference Fedorov1969). Miscanthus sinensis and M. sacchariflorus hybridize and introgression is expected among these taxa to produce monoploid and polyploid taxa (Adati and Shiotani, Reference Adati and Shiotani1962). The morphological characteristics that differentiate the two species, such as the absence/presence of an awn, length of the callus hairs and culm buds, are insufficient to separate interspecific hybrids. More work is required to fully understand the M. sacchariflorus ploidy complex (Lledó et al., Reference Lledó, Renvoize and Chase2001).
Plastid genome variation has been studied in Miscanthus using gene sequencing (Hodkinson et al., Reference Hodkinson, Renvoize and Chase2002a, Reference Hodkinson, Chase, Lledó, Salamin and Renvoizeb; Feng et al., Reference Feng, Lourgant, Castric, Saumitou-Laprade, Zheng, Jiang D and Brancourt-Hulmel2014) and microsatellite markers (De Cesare et al., Reference De Cesare, Hodkinson and Barth2010; De Cesare, Reference De Cesare2012; Głowacka et al., Reference Głowacka, Adhikari, Peng, Gifford, Juvik, Long and Sacks2014a, Reference Głowacka, Clark, Adhikari, Peng, Stewart, Nishiwaki, Yamada, Jørgensen, Hodkinson, Gifford, Juvik and Sacksb). Different, and species-specific, plastid haplotypes were detected by Hodkinson et al. (Reference Hodkinson, Renvoize and Chase2002a, Reference Hodkinson, Chase, Lledó, Salamin and Renvoizeb) and De Cesare (Reference De Cesare2012), and these were used to assess the maternal origin of M. × giganteus and also the phylogeny of Miscanthus species in combination with nuclear ribosomal DNA. Plastid DNA is generally maternally inherited in grasses, and M. × giganteus was shown to have the plastid type of M. sacchariflorus in all samples studied. Therefore, the allotriploid M. × giganteus inherited its plastid (and by extrapolation mitochondrial DNA) from a M. sacchariflorus lineage (Fig. 4).
Some artificial crosses of M. sinensis and M. sacchariflorus were included in the study carried out by De Cesare (Reference De Cesare2012). In several of these, the hybrid had the plastid genome of M. sinensis, showing that hybridization is possible in both directions (with both species as maternal parent). This is supported by Clark et al. (Reference Clark, Brummer, Głowacka, Hall, Heo, Long, Peng, Yamada, Yoo, Yu, Zhao and Sacks2014), who determined, in a major SNP study, many US ornamentals labelled as M. sinensis to be in fact BC1 or BC2 hybrids of M. sacchariflorus and M. oligostachyus with M. sinensis as the recurrent female parent. There is no reason to believe that the formation of M. × giganteus in the wild is unidirectional, but the plastid studies carried out by Hodkinson et al. (Reference Hodkinson, Renvoize and Chase2002a) and De Cesare (Reference De Cesare2012) suggest that this could be nearly the case as all putatively wild sourced M. × giganteus accessions have M. sacchariflorus plastid DNA. Clark et al. (Reference Clark, Brummer, Głowacka, Hall, Heo, Long, Peng, Yamada, Yoo, Yu, Zhao and Sacks2014) found the M. sacchariflorus plastome in nine of the 11 Chinese interspecific sacchariflorus× sinensis hybrids collected from the wild. Triploid seeds have also been found on M. sacchariflorus inflorescences in a sympatric zone with M. sinensis in Japan (Nishiwaki et al., Reference Nishiwaki, Mizuguti, Kuwabara, Toma, Ishigaki, Miyashita, Yamada, Matuura, Yamaguchi, Rayburn, Akashi and Stewart2011). Unidirectional hybridization can be caused by several factors including nuclear cytoplasmic DNA incompatibility effects (Anderson and Maan, Reference Anderson and Maan1995) or by population factors. For example, if M. sinensis was rare and M. sacchariflorus common (or if phenological differences created such a pattern), the vast number of seeds set would be from M. sacchariflorus ovule donors. However, a small number of M. sinensis plants can potentially father a large number of M. × giganteus seeds.
Nishiwaki et al. (Reference Nishiwaki, Mizuguti, Kuwabara, Toma, Ishigaki, Miyashita, Yamada, Matuura, Yamaguchi, Rayburn, Akashi and Stewart2011) investigated natural occurrences of triploidy in sympatric populations of tetraploid M. sacchariflorus and diploid M. sinensis in Japan. The interspecific hybrid, now known as Miscanthus× giganteus, was first collected in Yokohama, Japan, by a Danish plant collector (Nielsen, Reference Nielsen1990) and subsequently introduced around the world. Japan is therefore a likely source of new natural allotriploid M. × giganteus. Nishiwaki et al. (Reference Nishiwaki, Mizuguti, Kuwabara, Toma, Ishigaki, Miyashita, Yamada, Matuura, Yamaguchi, Rayburn, Akashi and Stewart2011) measured seed set of sympatric M. sinensis and M. sacchariflorus and assessed their DNA content with flow cytometry. Triploid seeds were found on the inflorescences of M. sacchariflorus. These plants have great potential as new sources of variation in breeding programmes. However, they originate from the warm moist regions of Southern Japan. The authors speculate that more cold-tolerant M. × giganteus would be expected from more northerly and cooler regions of Japan (Nishiwaki et al., Reference Nishiwaki, Mizuguti, Kuwabara, Toma, Ishigaki, Miyashita, Yamada, Matuura, Yamaguchi, Rayburn, Akashi and Stewart2011).
Aneuploids and B chromosomes
Linde-Laursen (Reference Linde-Laursen1993) reported a hyperploid chromosome number of 58 in M. × giganteus (trisomic). Aneuploidy has not otherwise been confirmed in many other cytological studies. However, the occurrence of accessory (B) chromosomes has been reported in some but not all Miscanthus species (Li and Ma, Reference Li, Ma and Hughes1951; Price, Reference Price1963a, Reference Priceb; Linde-Laursen, Reference Linde-Laursen1993). Price (Reference Price1963a) recorded between 0 and 11 B chromosomes in six clones of M. floridulus, and Linde-Laursen (Reference Linde-Laursen1993) reported between 0 and 4 B chromosomes approximately 0.7 μm in length in two clones of M. × giganteus. Chramiec-Głąbik et al. (Reference Chramiec-Głąbik, Grabowska-Joachimiak, Sliwinska, Legutko and Kula2012) reported one to four B chromosomes in M. × giganteus, two in M. sinensis and four in M. sacchariflorus.
Artificial polyploids and haploids
Chromosome doubling has been used to generate artificial polyploids in Miscanthus and has potential to introduce new genetic diversity into breeding programmes especially for M. × giganteus types by manipulating the ploidy of the parental species, restoring fertility or disrupting the self-incompatibility system (Petersen et al., Reference Petersen, Hagberg and Kristiansen2002; Głowacka et al., Reference Głowacka, Jeżowski and Kaczmarek2009, Reference Głowacka, Jeżowski and Kaczmarek2010a, Reference Głowacka, Jeżowski and Kaczmarekb; Yu et al., Reference Yu, Kim, Rayburn, Widholm and Juvik2009). Another stimulus for artificial polyploid formation has been the desire to generate novel sterile genotypes that lower the risk of invasiveness following introduction as a crop (Petersen et al., Reference Petersen, Hagberg and Kristiansen2003; Barney and Ditomaso 2008; Jørgensen, Reference Jørgensen2011). Petersen et al. (Reference Petersen, Hagberg and Kristiansen2002, Reference Petersen, Hagberg and Kristiansen2003) generated tetraploid M. sinensis from diploid source plants using colchicine or oryzalin treatments during callus induction, during callus proliferation, or on in vitro shoot apices and leaf explants. These tetraploids can be used as parental species in triploid Miscanthus production with M. sacchariflorus. Treatment of shoot apices with colchicine was shown to be the most efficient method for the four genotypes tested.
Triploid M. × giganteus is sterile in a post-zygotic barrier that results from abnormal male and female gametophyte production (Słomka et al., Reference Słomka, Kuta, Płażek, Dubert, Żur, Dubas, Kopeć and Żurek2012). Hexaploid M. × giganteus has been generated from triploid source material in an attempt to restore its fertility. For example, Yu et al. (Reference Yu, Kim, Rayburn, Widholm and Juvik2009) treated triploid callus, obtained from immature panicles, with colchicine and oryzalin to generate hexaploids. These were also found to have an increased stomata size (30 μm in the hexaploids compared with 24.3 μm in the triploids), but they did not report any findings for the fertility of the hexaploids. Touchell and Ranney (Reference Touchell and Ranney2012) also used oryzalin for in vitro chromosome doubling of M. × giganteus. Fertility of the resulting hexaploids was shown using pollen viability staining and crossing of the hexaploids with diploid M. sinensis, but in vitro embryo culture was required to obtain viable plantlets.
Haploid plants and double-haploid plants have also been reported (Głowacka et al., 2009; Głowacka et al., Reference Głowacka, Kaczmarek and Jeżowski2012) and used in the gene expression studies of Miscanthus (Barling et al., Reference Barling, Swaminathan, Mitros, James, Morris, Ngamboma, Hall, Kirkpatrick, Alabady, Spence, Hudson, Rokhsar and Moose2013). Głowacka et al. (Reference Głowacka, Kaczmarek and Jeżowski2012) developed a methodology for haploid formation by anther culture in M. sinensis. Androgenesis has also been attempted in M. × giganteus (Zur et al., Reference Zur, Dubas, Słomka, Dubert, Kuta and Płazek2013), but its efficiency is very low due to cytological chromosome imbalance.
Genotyping: genetic variation and phylogeography
Several multi-locus marker systems have been applied to Miscanthus such as restriction fragment length polymorphism (RFLP; Hernández et al., Reference Hernández, Dorado, Laurie, Martín and Snape2001), randomly amplified polymorphic DNA (RAPD; Chiang et al., Reference Chiang, Chou, Huang and Chiang2003), ISSR polymerase chain reaction (ISSR-PCR; Hodkinson et al., Reference Hodkinson, Chase, Takahashi, Leitch, Bennett and Renvoize2002c; Zhang et al., Reference Zhang, Guo, Kim, Lee, Li, Robertson, Wang, Wang and Paterson2013a, Reference Zhang, Shen, Shao, Fang, He, Ren, Zheng and Chenb) and AFLP (Greef et al., Reference Greef, Deuter, Jung and Schondelmaier1997; Hodkinson et al., Reference Hodkinson, Chase, Takahashi, Leitch, Bennett and Renvoize2002c). Single-locus co-dominant markers have also been applied including isozymes (Chou et al., Reference Chou, Hwang and Chang1987; Chou and Chang, Reference Chou and Chang1988; Chou and Ueng, Reference Chou and Ueng1992; Von Wühlisch et al., Reference Von Wühlisch, Deuter and Muhs1994). Many simple-sequence repeat (SSR) markers have been developed for the nuclear genome (Hernández et al., Reference Hernández, Dorado, Laurie, Martín and Snape2001; Hung et al., Reference Hung, Chiang, Chiu, Hsu and Ho2009; Ho et al., Reference Ho, Wu, Hsu, Huang, Huang and Chiang2011; Zhou et al., Reference Zhou, Li and Ge2011; Hu et al., Reference Hu, Diao, Zheng, Qu, Zhou and Hu2012; Kim et al., Reference Kim, Zhang, Auckland, Rainville, Jakob, Kronmiller, Sacks, Deuter and Paterson2012; Yu et al., Reference Yu, Zhao, Zhu, Chen and Peng2013), but fewer have been developed for the plastid/chloroplast genome (De Cesare et al., Reference De Cesare, Hodkinson and Barth2010; Jiang et al., Reference Jiang, Wang, Tang, Xiao, Ai and Yi2012). Recently, comprehensive SNP surveys have been conducted using next-generation sequencing approaches (Slavov et al., Reference Slavov, Nipper, Robson, Farrar, Allison, Bosch, Clifton-Brown, Donnison and Jensen2014; Clark et al., Reference Clark, Brummer, Głowacka, Hall, Heo, Long, Peng, Yamada, Yoo, Yu, Zhao and Sacks2014; Głowacka et al., Reference Głowacka, Adhikari, Peng, Gifford, Juvik, Long and Sacks2014a, Reference Głowacka, Clark, Adhikari, Peng, Stewart, Nishiwaki, Yamada, Jørgensen, Hodkinson, Gifford, Juvik and Sacksb). A more detailed history of molecular marker development has been given elsewhere (Głowacka, Reference Głowacka2011; Ma et al., Reference Ma, Jensen, Alexandrov, Troukhan, Zhang, Thomas-Jones, Farrar, Clifton-Brown, Donnison, Swaller and Flavell2012; Hodkinson et al., Reference Hodkinson, De Cesare and Barth2013).
Studies have demonstrated considerable genetic diversity in breeding collections and wild populations of Miscanthus at the infraspecific level (Greef et al., Reference Greef, Deuter, Jung and Schondelmaier1997; Hodkinson et al., Reference Hodkinson, Chase, Takahashi, Leitch, Bennett and Renvoize2002c; Głowacka et al., Reference Głowacka, Adhikari, Peng, Gifford, Juvik, Long and Sacks2014a, Reference Głowacka, Clark, Adhikari, Peng, Stewart, Nishiwaki, Yamada, Jørgensen, Hodkinson, Gifford, Juvik and Sacksb). Greef et al. (Reference Greef, Deuter, Jung and Schondelmaier1997) and Hodkinson et al. Reference Hodkinson, Chase, Takahashi, Leitch, Bennett and Renvoize(2002c) showed that AFLP markers could easily differentiate cultivars and infraspecific taxa of Miscanthus. However, they detected very little variation among the accessions of M.× giganteus collections and used the markers to help identify clonal material.
Diversity in M. × giganteus collections is a major cause for concern. Głowacka et al. (Reference Głowacka, Adhikari, Peng, Gifford, Juvik, Long and Sacks2014a, Reference Głowacka, Clark, Adhikari, Peng, Stewart, Nishiwaki, Yamada, Jørgensen, Hodkinson, Gifford, Juvik and Sacksb) used nuclear and chloroplast SSRs in combination with restriction site-associated DNA sequencing to estimate genetic similarity in over 30 M. × giganteus accessions of unknown provenance (legacy cultivars) from collections in North America and Europe and some newly bred M. × giganteus genotypes grown from seed and found that genetic variation in the legacy cultivars was extremely low. A total of 27 of these legacy cultivars were inferred as clones matching the M. × giganteus type specimen.
Population genetics and genetic diversity
Population genetic and adaptive variation data are required to determine gene pools for Miscanthus breeding and to understand physiological adaptations to abiotic stress such as temperature, drought and salinity. These limiting factors are crucial obstacles to overcome for developing crops that are suitable for growth in a wide range of climates and environments including marginal land (Jones et al., Reference Jones, Finnan and Hodkinson2014). Population genetic information is also important to develop knowledge about the evolution of Miscanthus and the impact of past and future climate on its distribution (Hodkinson (Reference Hodkinson, Hodkinson, Jones, Waldren and Parnell2011); De Souza et al., Reference De Souza, Arundale, Dohleman, Long and Buckeridge2013; Clark et al., Reference Clark, Brummer, Głowacka, Hall, Heo, Long, Peng, Yamada, Yoo, Yu, Zhao and Sacks2014).
Several studies have been carried out on genetic variation in Miscanthus, especially in M. sinensis, and the geographical centres of diversity including China, Korea and Japan. For example, Slavov et al. (Reference Slavov, Nipper, Robson, Farrar, Allison, Bosch, Clifton-Brown, Donnison and Jensen2014) used SNP and SSR markers to study putatively neutral geneticdiversity in a large breeding collection of Miscanthus. They also included 17 phenotypic traits related to biomass, phenology, cell-wall composition and morphology. They used the resulting data to delineate a reduced population of 145 M. sinensis genotypes to be used for association mapping and GS. Their data revealed considerable population genetic differentiation/structure in M. sinensis over the geographical space from Korea to Japan with a longitudinal cline (from 124° to 142° E) accounting for a high proportion of the molecular variation. In contrast, they found that latitude and altitudinal variation best explained variation in the phenotypic traits.
A genetic diversity study was conducted by Zhao et al. (Reference Zhao, Wang, He, Yang, Pan, Sun and Peng2013a, Reference Zhao, Li, He, Yu, Yang, Liu and Pengb) in over 450 M. sinensis accessions collected from a representative range across China using 23 SSR markers. High genetic diversity was detected and clustering of individuals was consistent with geographical distribution. However, within-subpopulation variation was substantially greater (83%) than among-subpopulation variation (17%), which is not unusual given the outbreeding and perennial nature of the species. Miscanthus sinensis also has good dispersal ability via its light feathery spikelets (Fig. 1) that facilitate gene flow.
Mating system has also been shown to contribute to patterns of population diversity and differentiation using RAPD markers and DNA sequence variation in outcrossing M. sinensis (from Japan, China and Taiwan) and inbreeding M. condensatus from Taiwan (Chou et al., Reference Chou, Chiang and Chiang2000; Chiang et al., Reference Chiang, Chou, Huang and Chiang2003). Chiang et al. (Reference Chiang, Chou, Huang and Chiang2003) studied sequence variation at the nuclear ADH1 locus and plastid trnL-F spacer regions. Low levels of genetic diversity were detected in M. condensatus that could be explained by bottlenecks caused by selfing in all populations. The ADH1 locus was under positive selection in lineages of M. condensatus that could be explained by pressure to evolve in response to different ecological conditions in saline habitats in which it is distributed (Chiang et al., Reference Chiang, Chou, Huang and Chiang2003).
A recent study carried out by Clark et al. (Reference Clark, Brummer, Głowacka, Hall, Heo, Long, Peng, Yamada, Yoo, Yu, Zhao and Sacks2014) examined a sample of over 600 M. sinensis accessions covering a large proportion of its native range in China, South Korea and Japan using a high-density set of SNP markers and ten plastid microsatellites. The markers detected six genetic clusters from geographically distinct regions. Four clusters were from mainland Asia (Southeast China, Yangtze-Qinling, Sichuan Basin and Korea/North China) and two were from Japan (Southern and Northern). They also included some M. floridulus in their analyses and found them to cluster with M. sinensis, demonstrating their close relationship and questioning their species status. All plastid haplotypes observed in M. floridulus were also common in M. sinensis. This was consistent with the results of the study carried out by Hodkinson et al. Reference Hodkinson, Renvoize and Chase(2002a) in which M. floridulus accessions were found to be embedded in a M. sinensis clade and with morphological intergradation of these species (Scally et al., Reference Scally, Hodkinson, Jones, Jones and Walsh2001a, Reference Scally, Waldren, Hodkinson and Jonesb). Only four M. floridulus accessions were included in the study carried out by Clark et al. (Reference Clark, Brummer, Głowacka, Hall, Heo, Long, Peng, Yamada, Yoo, Yu, Zhao and Sacks2014), and further studies are required to confirm these early observations.
Clark et al. (Reference Clark, Brummer, Głowacka, Hall, Heo, Long, Peng, Yamada, Yoo, Yu, Zhao and Sacks2014) also provided evidence that Southeast China was the centre of origin for the M. sinensis accessions found in temperate Eastern Asia. Their data were consistent with the hypothesis that Southeast China acted as a refugium during the last glacial maximum. They did not include other more southerly populations of M. sinensis, so it is not clear how important this refugium was in comparison with others that could have existed in former Indo-China, the Philippines, Indonesia and the South Pacific.
Genetic structure has also been detected on finer geographical scales. For example, Iwata et al. (Reference Iwata, Kamijo and Tsumura2004) used AFLP fingerprinting and PCR-RFLP to detect three regional subgroups of M. sinensis ssp. condensatus in Miyake Island, Japan. They also detected a rare haplotype most probably transmitted from outside the island. Shimono et al. (Reference Shimono, Kurokawa, Nishida, Ikeda and Futagami2013) investigated variation in Miscanthus sinensis in Japan using chloroplast DNA and detected nine haplotypes from over 600 individuals sampled from 30 populations. Two putative ancestral lineages were detected in the Ryukyu Islands, suggesting that they might have migrated from China via Taiwan or possibly the Korean Peninsula.
Adaptive variation
Field trials and laboratory-based controlled experiments, using a broad range of genotypes, have revealed variation in agronomic traits such as yield (Jeżowski et al., Reference Jeżowski, Głowacka and Kaczmarek2011; Gauder et al., Reference Gauder, Graeff-Hönninger, Lewandowski and Claupein2012), drought tolerance (Clifton-Brown and Lewandowski, Reference Clifton-Brown, Lewandowski, Bangerth and Jones2002), temperature control of leaf growth (Farrell et al., Reference Farrell, Clifton-Brown, Lewandowski and Jones2006), frost and cold tolerance (Clifton-Brown and Jones, Reference Clifton-Brown and Jones1997; Weng and Ueng 1997; Zub et al., Reference Zub, Arnoult, Younous, Lejeune-Hénaut and Brancourt-Hulmel2012; Głowacka et al., Reference Głowacka, Adhikari, Peng, Gifford, Juvik, Long and Sacks2014a, Reference Głowacka, Clark, Adhikari, Peng, Stewart, Nishiwaki, Yamada, Jørgensen, Hodkinson, Gifford, Juvik and Sacksb), flowering time (Clifton-Brown et al., Reference Clifton-Brown, Chiang, Hodkinson and Vermerris2008; Jensen, Reference Jensen2009; Jensen et al., Reference Jensen, Farrar, Thomas-Jones, Hastings, Donnison and Clifton-Brown2011; Zhang et al., Reference Zhang, Wyman, Jakob and Yang2012), senescence (Robson et al., Reference Robson, Mos, Clifton-Brown and Donnison2011), chemical composition and morphology (Jørgensen, Reference Jørgensen1997; Kaack et al., Reference Kaack, Schwarz and Brander2003; Hodgson et al., Reference Hodgson, Lister, Bridgwater, Clifton-Brown and Donnison2010, Reference Hodgson, Nowakowski, Shield, Riche, Bridgwater, Clifton-Brown and Donnison2011; Allison et al., Reference Allison, Morris, Clifton-Brown, Lister and Donnison2011; Zhao et al., Reference Zhao, Wang, He, Yang, Pan, Sun and Peng2013a, Reference Zhao, Li, He, Yu, Yang, Liu and Pengb, Reference Zhao, Huai, Xiao, Wang, Yu, Ding and Peng2014), and seed germination (Dwiyanti et al., Reference Dwiyanti, Stewart, Nishiwaki and Yamada2014). These studies have demonstrated huge phenotypic variation in and among Miscanthus species (Zub and Brancourt-Hulmel, Reference Zub and Brancourt-Hulmel2010; Jones et al., Reference Jones, Finnan and Hodkinson2014) that can be utilized in breeding.
Other researchers have set up common garden experiments with different genotypes grown at multiple locations to provide insights into the natural levels of adaptive variation (Clifton-Brown et al., Reference Clifton-Brown, Lewandowski, Andersson, Basch, Christian, Kjeldsen, Jørgensen, Mortensen, Riche, Schwarz, Tayebi and Teixeira1999; Clifton-Brown and Lewandowski, Reference Clifton-Brown and Lewandowski2000; Yan et al., Reference Yan, Chen, Luo, Ma, Meng, Li, Zhu, Li, Zhou, Zhu, Han, Ge, Li and Sang2012). Clifton-Brown and Lewandowski (Reference Clifton-Brown and Lewandowski2000) used field trials to examine the overwintering success of newly established Miscanthus genotypes from different sources in Asia. They planted these at four sites across a temperature gradient in Europe (Sweden, Denmark, Germany and England) and found considerable variation among the limited number of genotypes that they tested. Yan et al. (Reference Yan, Chen, Luo, Ma, Meng, Li, Zhu, Li, Zhou, Zhu, Han, Ge, Li and Sang2012) also used common garden experiments, but for a much larger sample of Miscanthus (93 genotypes) collected across their natural geographical range in China. They grew these in three locations representing temperate grassland with cold winter, semi-arid Loess Plateau and relatively warm and wet Central China and detected high variation in growth traits and significant levels of site × population interactions for most traits. Genotypes with high levels of plasticity that can produce good yields, in a broad range of habitats, were identified. These physiological experiments, field trials and common garden studies are helping to delineate populations of Miscanthus genotypes suitable for association mapping and GS (Slavov et al., Reference Slavov, Nipper, Robson, Farrar, Allison, Bosch, Clifton-Brown, Donnison and Jensen2014).
Linking genotype to phenotype
Some recent studies have used gene expression analysis to understand phenotypic variation in Miscanthus using methods such as RNA-Seq. Chouvarine et al. (Reference Chouvarine, Cooksey, McCarthy, Ray, Baldwin, Burgess and Peterson2012) used transcriptome sequencing of rhizome samples to generate an exome sequence database for Miscanthus complete with gene ontology functional annotations. Their data were used to differentiate closely related Miscanthus cultivars. Barling et al. (Reference Barling, Swaminathan, Mitros, James, Morris, Ngamboma, Hall, Kirkpatrick, Alabady, Spence, Hudson, Rokhsar and Moose2013) also generated a comprehensive expressed sequence tag (EST) catalogue using RNA-Seq that was predicted to represent a high proportion of the Miscanthus transcriptome using comparisons with sorghum gene models. They compared gene expression profiles in different tissues and a range of developmental stages. They also analysed expression profiles in rhizomes characterized in the spring compared with those characterized in the autumn to reveal biological pathways that exhibit altered regulation. Some candidate gene work has also been undertaken to understand variation in important lignin-related genes. For example, Suman et al. (Reference Suman, Ali, Arro, Parco, Kimbeng and Baisakh2011) studied variation in caffeic acid O-methyltransferase (COMT), cinnamyl alcohol dehydrogenase (CAD), cinnamoyl-CoA reductase (CCR) and ferulate 5-hydroxylase (F5H) genes with target region amplification polymorphism markers and detected sufficient variation to distinguish species of the Saccharum complex. However, they did not include sufficient numbers of genotypes to assess variation within and among the Miscanthus species.
Another study has focused on generating genetic linkage maps of Miscanthus that are needed for several applications such as quantitative trait locus (QTL) analysis and marker-assisted selection (MAS). High-resolution maps based on sequence markers allow the use of QTLs accessible from other grass species through alignment based on syntenic relationships (Ma et al., Reference Ma, Jensen, Alexandrov, Troukhan, Zhang, Thomas-Jones, Farrar, Clifton-Brown, Donnison, Swaller and Flavell2012). However, such maps have been produced only recently.
Mapping
Some studies have used markers for genetic mapping, but progress has been slow because of the large and heterozygous genome of Miscanthus. Mapping projects have therefore focused on diploid M. sinensis to facilitate genetic inheritance studies. The first published linkage map for Miscanthus (Atienza et al., Reference Atienza, Satovic, Petersen, Dolstra and Martín2002) was a breakthrough in the field. This map was generated using 257 PCR fingerprinting markers (RAPD) for offspring cross-mapping using an outbred population of 89 M. sinensis individuals (both parents full sibs). The markers were spread over 28 linkage fragments that spanned a total map length of 1074.5 cM with an average density of 4.2 cM per marker (but half of the fragments contained only two to four markers). Maps based on non-sequence-based markers (RAPD, AFLP and diversity array technology markers) do not provide alignable information for cross-utilization studies (Zhang et al., Reference Zhang, Guo, Kim, Lee, Li, Robertson, Wang, Wang and Paterson2013a, Reference Zhang, Shen, Shao, Fang, He, Ren, Zheng and Chenb).
Higher-resolution genetic maps of Miscanthus species based on DNA sequence markers have recently been generated using next-generation sequencing technology (Ma et al., Reference Ma, Jensen, Alexandrov, Troukhan, Zhang, Thomas-Jones, Farrar, Clifton-Brown, Donnison, Swaller and Flavell2012; Swaninathan et al., Reference Swaminathan, Chae, Mitros, Varala, Xie, Barling, Glowacka, Hall, Jezowski, Ming, Hudson, Juvik, Rokhsar Daniel and Moose2012). This has allowed for data transferability and several comparative genomic analyses. The map of M. sinensis developed by Swaminathan et al. (Reference Swaminathan, Chae, Mitros, Varala, Xie, Barling, Glowacka, Hall, Jezowski, Ming, Hudson, Juvik, Rokhsar Daniel and Moose2012) was based on a full-sib (F1) population produced by reciprocally crossing two ornamental clonally propagated M. sinensis accessions (Grosse Fontaine × Undine). Their analysis, including 868 segregating SNP and SSR markers, detected 19 linkage groups (consistent with the basic chromosome number x= 19). The total length on the new max likelihood map was 1782 cM (estimated total length of 1884 cM accounting for telomeric ends). In an integrated map of Grosse Fontaine and Undine, 97% of the mapped markers lie within 10 cM of another marker.
In the same year, Ma et al. (Reference Ma, Jensen, Alexandrov, Troukhan, Zhang, Thomas-Jones, Farrar, Clifton-Brown, Donnison, Swaller and Flavell2012) used an alternative sequencing approach known as GBS to identify the 19 linkage groups and produced a higher-resolution genetic map. It was based on an outcrossing full-sib F1 mapping population (called M × 2). Their composite linkage map combining markers from both parental linkage maps included 3745 SNP markers spanning 2396 cM with an average resolution of 0.64 cM. The mapping population of Ma et al. (Reference Ma, Jensen, Alexandrov, Troukhan, Zhang, Thomas-Jones, Farrar, Clifton-Brown, Donnison, Swaller and Flavell2012) segregates for important agronomic traits such as flowering time, biomass yield, stem number, senescence and spring emergence and can be applied for QTL studies and MAS.
QTLs
Despite their comparatively low resolution, the early maps (Atienza et al., Reference Atienza, Satovic, Petersen, Dolstra and Martín2002) were applied to QTL analysis of agronomic and combustion traits (Atienza et al., Reference Atienza, Satovic, Petersen, Dolstra and Martín2003a, Reference Atienza, Satovic, Petersen, Dolstra and Martínb, Reference Atienza, Satovic, Petersen, Dolstra and Martínc, Reference Atienza, Satovic, Petersen, Dolstra and Martínd). Atienza et al. Reference Atienza, Satovic, Petersen, Dolstra and Martín(2003a) used their genetic map (Atienza et al., Reference Atienza, Satovic, Petersen, Dolstra and Martín2002) to localize QTLs in M. sinensis controlling total height, flag leaf height and basal culm diameter. Field data were collected over two years to investigate developmental and environmental effects. Of the potential 11 reported QTLs, three were considered to be significant including total height, basal culm diameter and flag leaf height. Atienza et al. Reference Atienza, Satovic, Petersen, Dolstra and Martín(2003b) almost simultaneously published a paper using a similar methodology to investigate QTLs of yield components in M. sinensis. They detected 20 potential QTLs: six associated with yield, eight with stem yield, two with leaf yield and four with top yield. Atienza et al. (Reference Atienza, Satovic, Petersen, Dolstra and Martín2003c, d) also applied the same mapping population and RAPD markers to investigate QTLs influencing combustion quality traits. Atienza et al. Reference Atienza, Satovic, Petersen, Dolstra and Martín(2003c) detected nine putative QTLs: two for calcium, two for sulphur and five for phosphorus, and Atienza et al. Reference Atienza, Satovic, Petersen, Dolstra and Martín(2003d) detected four for chlorine and two for potassium.
These studies represent significant first steps in QTL detection, but it is not known how stable they are over time (years of trial and age of the plants) and how much they are influenced by the environment (Atienza et al., Reference Atienza, Satovic, Petersen, Dolstra and Martín2003d). We are currently in a period of considerable progress in QTL mapping in Miscanthus with the application of high-density/resolution genetic maps (Armstead et al., Reference Armstead, Huang, Ravagnani, Robson and Ougham2009). Because of the advances in DNA sequencing technology, it is likely that the limiting step will be high-quality phenotyping (Myles et al., Reference Myles, Peiffer, Brown, Ersoz, Zhanga, Costicha and Buckler2009).
MAS programmes in Miscanthus are underway at several institutions, for example, the University of Illinois, USA, on traits such as yield, stability, flowering time, overwintering ability, low-temperature photosynthesis, leaf extension and drought tolerance (Sacks, pers. commun.). An introgression programme of Saccharum into Miscanthus is ongoing at the same research institute (http://www.energybiosciencesinstitute.org/directory/sacks-erik). Furthermore, a significant MAS Miscanthus breeding programme is being carried out at the Institute of Biological, Environmental and Rural Sciences (IBERS), Wales (http://www.aber.ac.uk/en/ibers/). MAS for salt tolerance is being investigated at Wageningen University, the Netherlands (http://edepot.wur.nl/155120).
Association mapping and GS
Association mapping (linkage disequilibrium (LD) mapping) is a method of mapping QTLs that takes advantage of historical LD to link phenotypes to genotypes (Myles et al., Reference Myles, Peiffer, Brown, Ersoz, Zhanga, Costicha and Buckler2009). The genome is sampled for markers (such as SNPs) and associations are statistically detected between markers and a particular phenotype. Associations are independently verified to show that they (1) directly contribute to the trait of interest or (2) are linked to (in LD with) a QTL that contributes to the trait of interest. For example, Zhao et al. (Reference Zhao, Wang, He, Yang, Pan, Sun and Peng2013a, Reference Zhao, Li, He, Yu, Yang, Liu and Pengb) found nine SSRs associated with heading date and biomass yield in M. sinensis using association analysis between measured traits and 115 SSR marker alleles.
Association mapping in the form of a genome-wide association study (GWAS) is an advance on standard association mapping and has been most widely applied to the study of human diseases and cattle breeding and more recently to plants including Miscanthus (Slavov et al., Reference Slavov, Nipper, Robson, Farrar, Allison, Bosch, Clifton-Brown, Donnison and Jensen2014). Slavov et al. (Reference Slavov, Nipper, Robson, Farrar, Allison, Bosch, Clifton-Brown, Donnison and Jensen2014) used GWAS to study 17 traits related to phenology, biomass and cell-wall composition using a sample of 138 Miscanthus sinensis genotypes and over 100,000 single-nucleotide variants.
In crops, GS has successfully been implemented first in model crop species such as rice and maize. In rice, GS has been carried out for eight traits (yield, tiller number, grain number, 1000-grain weight, grain length, grain width, heading date and apicule colour; Xu, Reference Xu2013). In maize, GS for kernel spacing has been reported (Crossa et al., Reference Crossa, Beyene, Kassa, Pérez, Hickey, Chen, Burgueño, Windhausen, Buckler, Jannink, Lopez Cruz and Babu2013). It remains to be seen whether the high-density marker association approaches can prove suitable for GS in Miscanthus for advances in biomass-related traits such as stem diameter, stem-to-leaf ratio, cell-wall composition, or improved hardiness under adverse climatic or soil conditions.
Comparative genomics
Currently, there are few genomic resources available to Miscanthus breeders, except for some genomic and EST data (Kim et al., Reference Kim, Lee, Guo, Chung, Paterson, Kim and Lee2014), compared with rich QTL knowledge and physical data aligned with a high-quality reference genome of Sorghum (Zhang et al., Reference Zhang, Guo, Kim, Lee, Li, Robertson, Wang, Wang and Paterson2013a, Reference Zhang, Shen, Shao, Fang, He, Ren, Zheng and Chenb). However, the genomic resources available to breeders are likely to increase enormously over the next decade and will be utilized together with the resources of other well-characterized grass species such as sorghum, wheat, rice and maize. These resources of other Saccharinae and Sorghinae will prove particularly useful. Comparative genomic resources such as the CSGRqtl database (http://helos.pgml.uga.edu/qtl/) will facilitate the cross-utilization of information among Saccharinae taxa and complement Gramene (http://www.gramene.org), which includes mapping data from a broad diversity of grass taxa. The CSGRqtl database uses sorghum genome sequence as its central reference. It helps facilitate QTL mapping and characterize the function of genes that underlie QTLs. It can facilitate the investigation of genetic control of traits across genomes of divergent taxa and paleoduplicated subgenomes, as is the case in Miscanthus. These resources will combine genome data when they become available for Miscanthus species.
Conclusions
Natural genetic diversity is high in the Miscanthus polyploid complex and much progress has already been made in the characterization, evaluation and utilization of these resources so that artificial selection is not restricted by a lack of variation. The natural genetic diversity in Miscanthus has been characterized to define gene pools and used to help direct novel crossing work, manipulate ploidy, undertake QTL and association mapping studies, and develop GS selection programmes. Miscanthus therefore serves as a model for the use of genetic resources for new crop development. Advances in genetics underlying agronomic traits and the manipulation of these characteristics in breeding programmes will depend on the efficient utilization of existing collections and also on future collections aimed at targeting a maximum natural genetic diversity. There is a need for detailed phenotyping descriptor lists, a network of genetic resource collections and better seed/field bank coordination at the international level.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S147926211400094X
Acknowledgements
This work was supported by FP7 KBBE.2011.3.1-02 grant number 289461 (GrassMargins).
