Skip to main content
NCBI home page
Annals of Botany logoLink to Annals of Botany
. 2016 Jul 24;118(5):941–955. doi: 10.1093/aob/mcw137

Ecological characteristics and in situ genetic associations for yield-component traits of wild Miscanthus from eastern Russia

Lindsay V Clark 1, Elena Dzyubenko 2, Nikolay Dzyubenko 2, Larisa Bagmet 2, Andrey Sabitov 2, Pavel Chebukin 2, Douglas A Johnson 3, Jens Bonderup Kjeldsen 4, Karen Koefoed Petersen 5, Uffe Jørgensen 4, Ji Hye Yoo 6, Kweon Heo 6, Chang Yeon Yu 6, Hua Zhao 7, Xiaoli Jin 8, Junhua Peng 9, Toshihiko Yamada 10, Erik J Sacks 1,*
PMCID: PMC5055818  PMID: 27451985

Abstract

Background and aims Miscanthus is a genus of perennial C4 grasses native to East Asia. It includes the emerging ligno-cellulosic biomass crop M. ×giganteus, a hybrid between M. sinensis and M. sacchariflorus. Biomass yield and cold tolerance are of particular interest in Miscanthus, given that this crop is more temperate adapted than its C4 relatives maize, sorghum and sugarcane.

Methods A plant exploration was conducted in eastern Russia, at the northern extreme of the native range for Miscanthus, with collections including 174 clonal germplasm accessions (160 M. sacchariflorus and 14 M. sinensis) from 47 sites. Accessions were genotyped by restriction site-associated DNA sequencing (RAD-seq) and plastid microsatellites.

Key Results Miscanthus sinensis was found in maritime climates near Vladivostok (43·6°N) and on southern Sakhalin Island (46·6°N). Miscanthus sacchariflorus was found inland at latitudes as high as 49·3°N, where M. sinensis was absent. Most M. sacchariflorus accessions were diploid, but approx. 2 % were tetraploids. Molecular markers revealed little population structure (Jost’s D < 0·007 among diploid groups) but high genetic diversity (expected heterozygosity = 0·14) within the collection of Russian M. sacchariflorus. Genome-wide association (GWA) analysis for traits measured at the collection sites revealed three M. sacchariflorus single nucleotide polymorphisms (SNPs) significantly associated with the number of stems per unit area, one with height and one with basal stem diameter; three were near or within previously described sorghum quantitative trait loci for related traits.

Conclusions This new Miscanthus germplasm collection from eastern Russia will be useful for breeding Miscanthus and sugarcane cultivars with improved adaptation to cold. Moreover, a strategy is proposed to facilitate the rapid utilization of new germplasm collections: by implementing low-cost SNP genotyping to conduct GWA studies of phenotypic data obtained at collection sites, plant breeders can be provided with actionable information on which accessions have desirable traits and alleles.

Keywords: Chloroplast, genome-wide association analysis (GWAS), germplasm, Miscanthus sacchariflorus, Miscanthus sinensis, population genetics, restriction site-associated DNA sequencing (RAD-seq), Russia, single nucleotide polymorphism (SNP)

INTRODUCTION

Miscanthus is a genus of perennial C4 East Asian grasses that has attracted considerable recent interest as a biomass crop for energy, heat and fibre. The most commonly grown clone for biomass has been distributed under many cultivar names, but is genetically identical to the M. ×giganteus type specimen described by Hodkinson and Renvoize (2001; Głowacka et al., 2015). We will refer to this clone as M. ×giganteus ‘1993-1780’ in reference to its accession number in the Kew Living Collection. Miscanthus is more temperate adapted (growing as far as 50°N in Russia; Hodkinson et al., 2015) than its C4 relatives sugarcane, sorghum and maize, yet M. ×giganteus ‘1993-1780’ originated from sub-tropical southern Japan (35°N, assuming origination from Yokohama; Greef et al., 1997) and is therefore unlikely to represent the maximum cold tolerance of the genus (Głowacka et al., 2014). Insufficient winter hardiness has been documented for first-year plantings of M. ×giganteus ‘1993-1780’ in parts of northern Europe, whereas better winter hardiness was observed in other Miscanthus accessions (Clifton-Brown and Lewandowski, 2000; Farrell et al., 2006). In Urbana, Illinois, we also observed severe damage to first-year plantings of M. ×giganteus ‘1993-1780’ in 2014 after an especially cold winter, whereas other, recently bred, M. ×giganteus genotypes in the same trial were not so adversely affected (unpubl. data). Several studies have also shown that, although M. ×giganteus ‘1993-1780’ has higher growth and photosynthesis at temperatures below 14 °C than other species in the tribe Andropogoneae, there is variation for these traits among and within Miscanthus species (Purdy et al., 2013; Friesen et al., 2014; Głowacka et al., 2014). Thus, a major goal of Miscanthus breeding is the production of biomass cultivars with improved cold hardiness and chilling-tolerant photosynthesis. A large public gene bank for Miscanthus does not yet exist, but several academic research groups have recently imported germplasm to Europe and North America from the native range, characterized those collections with genotyping by sequencing (GBS) or restriction site-associated DNA sequencing (RAD-seq) approaches (Slavov et al., 2013; Barth et al., 2014; Clark et al., 2014, 2015) and performed genome-wide association studies (GWAS) using phenotypic data from replicated field trials (Slavov et al., 2014).

The two most widely distributed species of Miscanthus, M. sinensis and M. sacchariflorus, are also the two parent species of the hybrid M. ×giganteus. Miscanthus sinensis and M. sacchariflorus are phylogenetically distinct from each other within Miscanthus sensu stricto (Hodkinson et al., 2002), and differ in their growth habits, with M. sinensis being caespitose (tufted), and M. sacchariflorus having a spreading habit due to long rhizomes (Chae et al., 2014). Miscanthus sacchariflorus is typically found in riparian and wetland habitats, unlike M. sinensis and other Miscanthus species, which are more commonly found on hilly sites with good drainage (Sacks et al., 2013). Although M. sinensis and M. sacchariflorus exhibit considerable overlap in their geographic distribution, the range of M. sinensis extends further south (Clifton-Brown et al., 2008; Sacks et al., 2013).

The most promising region from which to collect cold-adapted Miscanthus germplasm is expected to be at the northernmost extent of its native range, in eastern Russia. However, previous collections have been limited; Yook et al. (2014) included three Russian M. sacchariflorus accessions from 43·3–43·6°N, 131·4–132·1°E and four Russian M. sinensis accessions from 42·4–43·4°N, 130·4–132·0°E in a phylogenetic study that was focused on Miscanthus from Korea. Reports from botanical surveys, however, have indicated the presence of M. sacchariflorus as far north as Lake Bolon (49·8°N, 136·4°E) and Blagoveshchensk (50·3°N, 127·3°E), and M. sinensis primarily in maritime regions such as southern Sakhalin (to 47·4°N) and near Vladivostok (43·1°N, 131·9°E), with rare reports near Lake Khanka (45°N, 132°E) and Khabarovsk (48·5°N, 135·1°E) (Herbarium of the Komarov Botanical Institute; Harkevich, 1985).

Given the limited number of Miscanthus accessions from Russia’s Far East that have been conserved in germplasm collections previously, and the potential utility of these populations for improving cold tolerance, we conducted a plant exploration as a joint effort of the NI Vavilov Research Institute of Plant Industry, the USDA-ARS and the University of Illinois. Materials from our exploration are being maintained by the US National Plant Germplasm System and will be made available as they are propagated. To facilitate rapid use of the accessions by plant breeders, we also recorded phenotypic characteristics at the collection sites and subsequently used RAD-seq to obtain single nucleotide polymorphism (SNP) genotypes for each accession. Analyses of phenotypic, ecological and genetic data were conducted with the following objectives: (1) to identify potential geographic, climatic, ecological and anthropogenic influences on the ranges of M. sinensis and M. sacchariflorus; (2) to quantify phenotypic variation for biomass traits; (3) to understand population structure of Russian M. sinensis and M. sacchariflorus in order to identify genetic groups for germplasm conservation, association analysis and potential sources of heterosis; (4) to compare genetic diversity of Russian Miscanthus with previously characterized Miscanthus populations in order to assess its relative utility for breeding; and (5) to investigate the potential to identify quantitative trait loci (QTLs) for traits of agronomic interest via GWAS of in situ phenotypic data obtained during germplasm collection. Although it is currently unusual to perform GWAS for crop germplasm without phenotypic data from replicated field trials, there are previous examples of successful association studies using phenotypic data from natural populations, particularly in forestry (Parchman et al., 2012; Budde et al., 2014) and animal ecology (Johnston et al., 2011, 2014; Narum et al., 2013). Here we demonstrate that value can be added to a crop germplasm collection by combining in situ phenotype data and inexpensive genotyping data in a GWAS.

MATERIALS AND METHODS

Field collections and observations

From 3 to 29 September 2012, a plant exploration for Miscanthus germplasm was conducted in eastern Russia (Fig. 1). Additional details of our sampling route are provided in Supplementary Data Materials and Methods. An initial collection near Nevelsk on Sakhalin Island was followed by collections on the mainland. On the mainland, we explored the following areas: (1) the north–south corridor between Khabarovsk and Vladivostok along highway M60, which, north of Lake Khanka, was near and parallel to the Ussuri River; (2) north-east from Khabarovsk along the Amur River on its eastern side via highway P454 until the Gur River; (3) west from Khabarovsk on the M58 to Birobidzhan and from Birobidzhan south along the Bira River until near to the Amur River, which is the border with China; (4) the area south of Lake Khanka to the city of Ussuriysk; and (5) on Russky Island just south of Vladivostok. On four sections of highway, we noted the number of distinct stands of Miscanthus at least 100 m apart from each other, and calculated the average number of stands per kilometre (Fig. 1A, inset).

Fig. 1.

Fig. 1.

Open in a new tab

Collection locations, temperature data and population structure of Miscanthus in eastern Russia. (A) Miscanthus sacchariflorus collection sites are indicated with pies, with colours representing assignment (Q values) to four genetic clusters determined by Structure, using 29 260 RAD-seq SNPs across 160 individuals. Q values were highly similar among individuals within each site. Miscanthus sinensis individuals are indicated by squares, and colours indicate assignment to previously identified populations in East Asia (Clark et al., 2014) by discriminant analysis of principal components using 24 132 RAD-seq SNPs. The inset shows frequency of stands of M. sacchariflorus observed at least 100 m apart along four stretches of highway. (B) USDA plant hardiness zones, based on data from 1982–2011 from NAPPFAST (2012). (C) Mean temperatures during January based on data from 1950–2000 available at WorldClim (Hijmans et al., 2005).

Collections were made at 48 sites. At 47 sites, from one to 13 plants were measured and rhizomes collected. We endeavoured to obtain both rhizomes (live plants) and seed at each site. Seed was collected from 41 sites; the number of plants per accession that contributed seed ranged from one to 100. Material from the seed collections was not included in the genetic studies reported here. For each plant from which rhizomes were collected, we measured the height of the tallest stem (stem height), the diameter of the internode at the base of the tallest stem (stem diameter) and the number of stems in a 0·25 m2 area (stem density). We also recorded the approximate number of plants found and sampled at the collection site, the site size (in m2), the frequency of Miscanthus at the site, aspect (compass direction of slope), slope, and site physical and vegetative descriptions (Supplementary Data Dataset S1). Rhizomes were collected from 182 plants and were shipped to USDA-APHIS-PPQ in Beltsville, MD for US quarantine, and a backup set was also sent to Aarhus University in Foulum, Denmark. At the time of publication of this paper, 158 clones of M. sacchariflorus (out of 165 collected) and 14 of M. sinensis (out of 17 collected) have been released from US quarantine. Material from seed increase will be made available from USDA (http://www.ars-grin.gov; accessions W6 49502–49505; Dataset S1). Leaf tissue from each plant at Foulum was freeze-dried and shipped to the University of Illinois for DNA extraction.

Flow cytometry

Flow cytometry was performed at Aarhus University using a protocol modified from Petersen et al. (2003). Samples of M. sinensis [MS-104 (Jørgensen, 1997) or MS-110 (Petersen et al., 2003)] that were known to be diploid were used as an internal standard. Additional details are provided in Supplementary Data Materials and Methods.

Genotyping

A total of 174 plants (160 M. sacchariflorus and 14 M. sinensis) from 47 sites (44 M. sacchariflorus and three M. sinensis) were included in the genotyping analysis. Nuclear SNPs were obtained by RAD-seq (Poland et al., 2012a). Entries were also screened with ten plastid microsatellites (de Cesare et al., 2010; Jiang et al., 2012) using previously described protocols (Clark et al., 2014). Additional details are provided in Supplementary Data Materials and Methods. Sequences from RAD-seq have been deposited at the NCBI Sequence Read Archive (accession SRP063572).

Data analysis

To obtain SNP genotypes for 160 Russian M. sacchariflorus individuals, we used the UNEAK pipeline in TASSEL 3·0·162 (Lu et al., 2013). Filtering for SNPs with a minimum call rate of 0·5 and a minimum minor allele frequency of 0·01 produced 29 260 SNPs (Supplementary Data Dataset S2). To obtain SNP genotypes for M. sinensis, we ran the UNEAK pipeline under the same conditions on the 14 Russian M. sinensis individuals and 595 individuals from a previous study (576 M. sinensis individuals plus 19 M. sacchariflorus and hybrid individuals; Clark et al., 2014), and removed SNPs that appeared heterozygous in double haploid lines, yielding 24 132 SNPs.

Russian M. sacchariflorus genotypes were analysed in Structure 2·3·4 (Falush et al., 2003) to assess population structure. Three runs each at K = 1 through K = 10 were performed under default conditions with a burn-in of 10 000 reps, followed by 50 000 reps. The number of clusters (K) was selected by analysing the results in Structure Harvester (Earl and VonHoldt, 2011) using the Evanno method (Evanno et al., 2005). Mean Structure results (Q values) for each collection site were plotted using ArcGIS 10·1 (ESRI, Redlands, CA, USA).

Given the weak geographic structuring in M. sacchariflorus detected by Structure, Mantel tests were used to assess the hypothesis of genetic isolation by distance among Russian M. sacchariflorus individuals. Geographic distances between collection sites were calculated using the R (R Core Team, 2014) package geosphere (Hijmans, 2014). SNP data were imported in numeric (0 or 2 for homozygotes, 1 for heterozygotes) format, with individuals as rows and markers as columns (‘hapMap2genlight’ function; Clark et al., 2014), and genetic distances between individuals were calculated as Euclidian distances between rows using the ‘dist’ function in R. Correlation between the geographic and genetic distance matrices was then tested using the ‘mantel.rtest’ function in the R package ade4 (Chessel et al., 2004) with 999 permutations. To make a matrix of genetic distances based on plastid haplotypes, plastid microsatellite data were imported into the R package polysat (Clark and Jasieniuk, 2011) and dissimilarities were calculated with the ‘meandistance.matrix’ and ‘Lynch.distance’ functions. The genetic distance matrix based on plastid data was then used in a Mantel test, comparing it with the geographic distance matrix as was done with SNP data.

Suggested seed increase groups for M. sacchariflorus were selected based on geography and ploidy. Genetic differentiation between seed increase groups was estimated with Jost’s D across 29 260 RAD-seq SNPs using the ‘D_Jost’ function in the R package mmod (Jost, 2008; Winter, 2012).

To construct a plastid haplotype network for Russian M. sacchariflorus, the distances between haplotypes that were calculated with polysat were analysed with a modified source code from the R package pegas (Paradis, 2010; Clark et al., 2014). Additional connections were added to the network where two haplotypes only differed at one locus.

To assess the relationship between the Russian M. sinensis individuals and previously identified genetic clusters (Clark et al., 2014), and determine hybrid status, M. sinensis SNP data were evaluated by principal component analysis (PCA) and discriminant analysis of principal components (DAPC) using the R package adegenet (Jombart et al., 2010; Jombart and Ahmed, 2011). A DAPC was also performed on the M. sacchariflorus SNP dataset but was uninformative (see the Results).

For genome-wide association analysis of the phenotypic traits in M. sacchariflorus, 64-nucleotide RAD tags output by the UNEAK pipeline were aligned to the Sorghum bicolor genome version 2·0 (Paterson et al., 2009; available at http://phytozome.jgi.doe.gov) using Bowtie2 (Langmead and Salzberg, 2012) with parameters set for high sensitivity (-D 20 -R 3 -N 1 -L 18 -i S,1,0·50 –local), given that we were making cross-genus sequence alignments. A total of 16 137 pairs of RAD tags (55 % of the total) both aligned to the same location in the sorghum genome and therefore were retained at this step. Two individuals out of 160 were removed for having >50 % missing data, and three tetraploid individuals were removed, leaving 155 individuals for analysis. SNPs were then filtered to have ≤50 % missing data, minor allele frequency ≥0·05 and observed heterozygosity ≤50 %, leaving 5971 SNPs, which were analysed to detect SNP–trait associations. To reduce environmental error in traits that were significantly correlated with latitude, linear models were fit with the trait (after log transformation if applicable) as the dependent variable and latitude as the independent variable. For each individual, the predicted phenotype based on latitude was subtracted from the actual phenotype value to produce the adjusted phenotype value, which was then evaluated by GWAS. Traits evaluated were stem height (untransformed and adjusted by latitude), stem diameter (log transformed and adjusted by latitude) and number of stems per 0·25 m2 (log transformed). Of the 155 accessions that were analysed in GWAS, 91 accessions each fell into one of 30 groups of related individuals, leaving 94 unrelated individuals and groups of individuals (Supplementary Data Fig. S1; relatedness was included in the GWAS model). GWAS using the Q–K mixed model method (Yu et al. 2006), with the relationship matrix and the first four principal components included in the model, was performed both in rrBLUP (Endelman, 2011) and in TASSEL5 (Bradbury et al., 2007). In rrBLUP, the additive relationship matrix was calculated using the EM imputation method (Poland et al., 2012b), and in TASSEL5 the kinship matrix was calculated using scaled identity-by-state. In both programs, the P3D (population parameters previously determined) method was not used, given the relatively small number of SNPs being evaluated, and the K matrix was not compressed, given the small number of individuals in the study. The FDR (false discovery rate) method (Benjamini and Hochberg, 1995) was used to correct P-values for multiple testing. SNP–trait associations were considered significant if P-values were <0·05 after FDR correction. To estimate heritability based on the SNP data, the ‘kin.blup’ function of rrBLUP was used to fit each trait to the additive relationship matrix and perform REML (restricted maximum likelihood) estimations of genetic variance and error variance. The genetic variance was then divided by the sum of the genetic and error variances.

RESULTS

Ecology, ploidy and genetics of M. sinensis

We encountered M. sinensis at only three sites in eastern Russia (Fig. 1A): southern Sakhalin Island (46·58°N, 141·84°E), near highway M60 between Vladivostok and Ussuriysk (43·65°N, 132·00°E) and on Russky Island just south of Vladivostok (42·98°N, 131·91°E). On Russky Island, M. sinensis was especially abundant, and a sample was taken from a stand of > 1000 individuals (Fig. 2A). All three of the sites where M. sinensis was found have a maritime-influenced climate with relatively mild winters (USDA hardiness zones 5 and 6; average annual extreme minimum temperature of –29 to –18 °C), which probably contributed to the success of M. sinensis at these sites, in contrast to regions further inland (USDA hardiness zones 3 and 4; – 40 to –29 °C) where we only found M. sacchariflorus (Fig 1B).

Fig. 2.

Fig. 2.

Open in a new tab

Photos of Miscanthus in eastern Russia. (A) Large population of M. sinensis on Russky Island, south of Vladivostok facing the Sea of Japan. (B and C) Stands of M. sacchariflorus (white patches) growing in high frequency along roads near Lake Khanka, on land that is sloped and well drained. (D) Two rectangular hayfields near Lake Khanka with a large population of M. sacchariflorus growing in the field on the right (indicated with an arrow) and the field on the left was recently cut. (E) Near Lake Khanka, M. sacchariflorus (white patches) growing in lowlands that have recently been deforested by cutting and burning, with cows grazing in the foreground, and recently logged mountainside in the background. (F) Miscanthus sacchariflorus (indicated with an arrow) growing on an island in the Amur River near the village of Sinda, Khabarovsk Krai. (G) Miscanthus sacchariflorus growing in an open area next to a birch (Betula sp.) forest. (H) Wild soybean, Glycine max (synonym = Soja ussuriensis; indicated with an arrow) twining up a stem of M. sacchariflorus. (I) Close-up of wild soybean leaf and seed pod growing on an inflorescence of M. sacchariflorus. (J) Map of locations where photos were taken.

Using nuclear SNPs on our 14 Russian M. sinensis and 595 previously genotyped Miscanthus individuals from East Asia (Clark et al., 2014), PCA and DAPC revealed that the M. sinensis individuals from near Vladivostok on the mainland and from Russky Island were most similar to individuals from northern China (above 36°N) and Korea, and that the M. sinensis individuals from Sakhalin were most similar to individuals from northern Japan (Hokkaido and northern Honshu; Fig. 1A; Supplementary Data Fig. S2). None of the Russian M. sinensis that were collected showed evidence of hybridization with M. sacchariflorus (Fig. S2). Thus, the M. sinensis individuals that we found in eastern Russia were most genetically similar to geographically adjacent M. sinensis in neighbouring countries. Analysis of the maternally inherited plastid SSR (simple sequence repeat) alleles was consistent with the results of the nuclear SNP data. The M. sinensis individuals from Sakhalin had the most common plastid haplotype found in Hokkaido, Japan (haplotype C in Clark et al., 2014). Two out of the three M. sinensis individuals from north of Vladivostok also had haplotype C, which is also common in South Korea. However, the third M. sinensis individual from north of Vladivostok, as well as the individual from Russky Island, had haplotype G, which is uncommon in M. sinensis and previously found only in the population from northern China (Clark et al., 2014). Of the 13 M. sinensis individuals that were tested by flow cytometry, all were found to be diploid. Miscanthus sinensis plants from Sakhalin were shorter but had thicker stems than those near Vladivostok (Fig. 3A, B, G, H).

Fig. 3.

Fig. 3.

Open in a new tab

Phenotypes of Miscanthus, recorded at collection sites. Seventeen M. sinensis individuals across three sites and 165 M. sacchariflorus individuals across 44 sites were measured. Stem diameter was measured at the base of the stem. (A–F) Histograms of phenotype distributions for M. sinensis (A–C) and M. sacchariflorus (D–F). (G, H) Maps of phenotype distributions. R2 and slope values were calculated by fitting linear models relating phenotype to latitude in M. sacchariflorus only. Circles indicate M. sacchariflorus, and squares indicate M. sinensis

Ecology of M. sacchariflorus

Miscanthus sacchariflorus was found throughout most of the area that we explored. However, we did not observe any M. sacchariflorus north and east of the Anyuy River (49·33°N, 136·52°E) as we travelled along the eastern side of the Amur River on road P454 from Khabarovsk to the Gur River. Along the Amur River watershed, we found M. sacchariflorus as far inland (west) as Birobidzhan (48·67°N, 132·98°E); there may have been more M. sacchariflorus further west along the Bira River, but time constraints prevented us from travelling any further. The frequency of M. sacchariflorus also varied considerably across our study area (Fig. 1A, inset). The greatest frequency of M. sacchariflorus was observed near the south and west of Lake Khanka, with 8·79 stands km–1 (Figs 1A and 2B, C), whereas the lowest frequency was from Khabarovsk to the Gur River along the Amur River basin, with only 0·02 stands km–1 observed. West of Khabarovsk to Birobidzhan, the frequency of stands was higher (0·13 km–1) than east of Khabarovsk. However, just south of Khabarovsk, near the Ussuri River, the stand density of M. sacchariflorus was higher still at 0·39 stands km–1. Though the common name of M. sacchariflorus is Amur silvergrass, we observed that it was more common near the Ussuri River and Lake Khanka than along the Amur River. In the Lake Khanka region, we also observed rectangular hayfields that had high population densities of flowering M. sacchariflorus (Fig. 2D). Land disturbance in the Lake Khanka area, especially deforestation to make pasture and hay-lands for raising cows (Fig. 2D, E), may have contributed to the large population of M. sacchariflorus observed. Around 160 years ago, Maximowicz (1859) observed that M. sacchariflorus growing in the prairies of the Russian Far East and north-east China was used for hay and grazing; thus, there has been a longstanding interaction between livestock farmers in this region and indigenous populations of M. sacchariflorus. East of Khabarovsk in the Amur River basin, M. sacchariflorus was absent near the villages we visited; however, we saw large populations on nearby islands surrounded by the Amur River and without access by road (Fig. 2F), suggesting that human disturbance may have negatively impacted M. sacchariflorus in adjacent populated areas.

Miscanthus sacchariflorus was commonly found in relatively flat areas near surface water but not on flooded land. Less frequently, we found M. sacchariflorus on land with ≥10°slope, or not near surface water. Soil texture where M. sacchariflorus grew was typically loam or sandy loam, and less frequently clay or clay–loam. Miscanthus sacchariflorus was found in open areas, in close association with other grasses and forbs, and sometimes adjacent to forest, but not under the forest canopy (Fig. 2G). A rhizomatous grass that was found growing near M. sacchariflorus at more than half (52 %) of the collection sites was identified as Hemarthria sibirica using ITS (internal transcribed spacer) sequencing (identical to NCBI accession KF163639·1 at 622 out of 623 nucleotides; Supplementary Data Fig. S3) and morphology. Other plants commonly associated with M. sacchariflorus at the collection sites included: Allium sp., Artemisia sp., Aster sp., Betula sp., Iris sp., Juncus sp., Lespedeza sp., Phragmites sp., Quercus sp., Salix sp. and the wild soybean Glycine max (synonyms Soja ussuriensis, Glycine soja) (Fig. 2G–I). Knowledge of which species are frequently associated with M. sacchariflorus is useful for understanding their in-common environmental adaptation and the complex ecological communities to which they belong. Herbarium specimens that were collected during the expedition are described in Supplementary Data Table S1 and maintained at the Vavilov Institute.

Stem height of the M. sacchariflorus accessions at the collection sites ranged from 1·2 to 2·6 m (mean 2·0; Fig 3D, G). Most accessions had thin stems (mean diameter of 3·8 mm), but one plant (RU2012-182) had stems as thick as 8 mm (Fig. 3E, H; Dataset S1). We observed a strong correlation between latitude and phenotype of M. sacchariflorus at the collection sites. Relative to the mean, stem height decreased modestly with increasing latitude (0·24 m in 5·6°of latitude change), and stem diameter decreased substantially (1·4 mm in 5·6°of latitude change; Fig. 3G, H; Supplementary Data Fig. S4). Stem height and diameter were also significantly correlated with elevation, but more weakly than they were correlated with latitude (Fig. S4). The effect of elevation on stem height and diameter was non-significant when latitude was included as an effect in the model (data not shown), suggesting that the strong correlation between latitude and elevation (Fig. S4) drove the correlation between elevation and stem height and diameter. The number of stems per unit area, a key component of biomass yield for Miscanthus (Matumura et al., 1985, 1986, 1987), varied greatly among accessions (from 5 to 112 per 0·25 m2; Fig. 3F) but no correlation with latitude or elevation was observed for this trait (Fig. 3I). One individual in particular, RU2012-169, had above average values for height (2·42 m), stem diameter (6 mm) and stems per area (51 in 0·25 m2) and is therefore an especially good candidate for breeding biomass cultivars (Fig. S4).

Population structure, ploidy and diversity of M. sacchariflorus

Statistical analyses of the RAD-seq SNP data for 160 M. sacchariflorus individuals revealed weak population structure. The Bayesian Information Criterion produced by the ‘find.clusters’ function in the R package adegenet was at its minimum when one cluster was assumed, suggesting that it would not be meaningful to split the set into distinct groups of individuals using DAPC (Supplementary Data Fig. S5). A Neighbor–Joining tree calculated from Euclidian distances between genotypes also did not suggest clear groupings, apart from groups of closely related individuals collected at the same sites (Supplementary Data Fig. S6). Using Structure, the Evanno et al. (2005) method suggested that five clusters (K = 5) would be ideal, primarily because likelihood values were highly variable from run to run at K = 6 (Supplementary Data Fig. S7). However, we found that ancestry assignments (Q values) were not reproducible from run to run at K = 5, so we instead present the results at K = 4 (Fig. 1A) and K = 2 (Supplementary Data Fig. S8). Sets of individuals from two collection sites were placed into two respective clusters by Structure at K = 4, and Q values at K = 4 and K = 2 otherwise followed a gradient suggesting isolation by distance without clear geographic breaks (Fig. 1A; Fig. S8). Since adegenet, Structure and Neighbor–Joining indicated that population structure was absent or weak, a Mantel test was performed to determine whether a gradient of population structure was significantly associated with geography. The Mantel test confirmed that geographic distance was correlated with nuclear genetic distance at P < 0·001 (Fig. 4A).

Fig. 4.

Fig. 4.

Open in a new tab

Relationship between genetic distance and geographic distance for M. sacchariflorus individuals, shown as density plots. Significance of correlation (P) was determined by a Mantel test. Red lines were fit with linear regression. (A) Euclidian genetic distance based on 29 260 RAD-seq SNPs for 160 individuals vs. geographic distance in kilometres. (B) Proportion of plastid microsatellite alleles shared for 159 individuals vs. geographic distance.

Out of 163 M. sacchariflorus that were screened by flow cytometry, 160 were diploid and three were tetraploid. The tetraploids, which were collected at two sites in the Amur River basin and included in the SNP analyses, were genetically similar to the diploid accessions found further south in our collection range (Fig. 1A; Fig. S8).

RAD-seq SNPs indicated high genetic diversity for the Russian M. sacchariflorus (Table 1). We compared SNP diversity statistics obtained from our 160 Russian M. sacchariflorus individuals with those obtained from previous studies of Miscanthus using an identical RAD-seq protocol. In a large collection of M. sinensis that was representative of geographic diversity across most of East Asia (Clark et al., 2014), the south-east China population had the greatest number of SNPs with a minor allele frequency >0·05 (9262 SNPs), consistent with our finding that south-east China was the centre of radiation for M. sinensis after the last glacial maximum. The two most isolated populations of M. sinensis, in North Japan and the Sichuan Basin, had the lowest number of SNPs with a minor allele frequency >0·05 (6942 and 7531, respectively). In comparison, Russian M. sacchariflorus were highly diverse, with 12 265 SNPs at a minor allele frequency >0·05. However, the Japanese M. sacchariflorus–M. ×giganteus complex had 17 909 SNPs with a minor allele frequency >0·05, probably due to its interspecific nature (Clark et al., 2015). Mean expected heterozygosity across all SNPs followed a similar pattern (Table 1).

Table 1.

Genetic diversity of Russian Miscanthus sacchariflorus compared with previously studied populations (Clark et al., 2014, 2015) of Miscanthus

Population No. of individuals No. of plastid haplotypes Gini–Simpson index of plastid haplotypes No. of nuclear SNPs with minor allele frequency >0·05* Mean expected heterozygosity across all nuclear SNPs
Russia Msa† 160 13 0·82 ± 0·01 12 265 0·14
Japan Msa–Mxg‡ 78 19 0·83 ± 0·04 17 909 0·19
SE China Msi§ 125 22 0·75 ± 0·04 9262 0·13
Yangtze-Qinling Msi 114 18 0·57 ± 0·06 7902 0·13
Sichuan Msi 58 9 0·74 ± 0·04 7531 0·12
Korea, N China Msi 195 12 0·48 ± 0·04 8596 0·13
N Japan Msi 96 6 0·25 ± 0·06 6942 0·11
S Japan Msi 32 11 0·87 ± 0·04 8077 0·12
Open in a new tab

*SNPs appearing heterozygous in any one of three doubled haploid M. sinensis lines were removed from all SNP datasets for the calculation of diversity metrics.

†Msa, M. sacchariflorus.

‡Msa-Mxg, M. sacchariflorus–M. ×giganteus complex.

§Msi, M. sinensis.

Complete plastid haplotypes across ten microsatellite loci were obtained for 159 M. sacchariflorus individuals, yielding 13 unique haplotypes (Supplementary Data Table S2). Twelve of these could be connected to each other by single mutations in a haplotype network, whereas the remaining haplotype (S) only shared alleles at three out of ten loci with the most closely related haplotype (Fig. 5). Amplicon size of haplotype S was more similar to typical M. sacchariflorus haplotypes than it was to M. sinensis haplotypes. None of the Russian M. sacchariflorus individuals had plastid haplotypes from M. sinensis. Six out of the seven most common haplotypes (Z, Y, W, V, U and T) had also been found in M. sacchariflorus from China and/or Japan in our previous studies (Clark et al., 2014, 2015). Haplotype S was also observed in an individual of M. sacchariflorus ‘Robustus’ (98m0002) obtained from M. Deuter of Tinplant (Germany) and in a diploid accession of M. sacchariflorus obtained from a US nursery (UI10-00008). Two of the tetraploid individuals had haplotype U, and the third tetraploid had a unique haplotype in the dataset that differed from haplotype U by one mutation. Overall, the presence of multiple common haplotypes in the Russian M. sacchariflorus accessions resulted in a high Gini–Simpson index (probability of getting two different haplotypes if two individuals were chosen at random) of 0·82, which was also high in comparison with typical M. sinensis populations (Table 1). However, we observed considerable overlap in the geographic distributions of plastid haplotypes. A Mantel test comparing haplotype dissimilarity with geographic distance did not find significant correlation (Fig. 4B).

Fig. 5.

Fig. 5.

Open in a new tab

Plastid haplotype network of M. sacchariflorus based on ten microsatellite markers. A total of 159 individuals from eastern Russia were evaluated. Haplotypes found in more than two individuals, as well as one unusual haplotype that was dissimilar from all others, are indicated by letters. Amplicon sizes for all haplotypes are listed in Table S2.

SNP–trait associations

Using the 5971 SNPs that met our criteria for GWAS, and phenotypic data recorded at the collection sites, we identified five significant marker–trait associations. We identified three significant associations with number of stems per unit area, including two SNPs aligning to S. bicolor chromosome 2 and one SNP aligning to S. bicolor chromosome 6 (Fig. 6A). The two SNPs on chromosome 2 were both in the untranslated regions of transcribed genes, while the SNP on chromosome 6 coded for an amino acid substitution (Table 2). Intriguingly, one of the tagged genes on chromosome 2 coded for a protein kinase, and was also <1 cM from the peak of a QTL region on M. sacchariflorus linkage group 4 for the ratio of compressed circumference to basal circumference, a trait heavily dependent on stems per unit area (Hongxu Dong, unpubl. res.). Moreover, the SNP on chromosome 6 fell within a QTL for vegetative branching in sorghum that contains several homologues of rice genes also known to be involved in branching (Kong et al., 2014), and within a QTL for number of tillers per plant in sorghum (Shiringani et al., 2010; Table 2). For all three of the significant stem density SNPs in M. sacchariflorus, the minor allele was associated with fewer stems per unit area (Fig. 6B). Additionally, we identified one SNP aligning to sorghum chromosome 6 that was significantly associated with stem diameter when we corrected for the environmental influence of latitude. The stem diameter SNP was near two previously identified QTLs for stem diameter in sorghum (Shiringani et al., 2010; Phuong et al., 2013). We also identified one SNP on chromosome 3 that was significantly associated with stem height (corrected for latitude), and was near a previously identified sorghum QTL for plant height (Phuong et al., 2013). Heritability as estimated from SNP data was 0·33, 0·81 and 0·71 for stems per area, stem height and stem diameter, respectively. The latter two are similar to heritabilities of 0·88 and 0·60 found for stem height and stem diameter, respectively, in M. sinensis by Slavov et al. (2014). Individual significantly associated SNPs accounted for between 5 and 18 % of the variation in their respective traits under an additive model (Fig. 6D). The three SNPs associated with stems per area accounted for 27 % of the variation in this trait when taken together.

Fig. 6.

Fig. 6.

Open in a new tab

Results of Q–K mixed model analysis to detect SNP–trait associations in Russian Miscanthus sacchariflorus. (A–C) Manhattan plots of log-transformed P-values vs. aligned position to the Sorghum bicolor 2.0 reference genome. P-values were calculated in rrBLUP and corrected using the method of Benjamini and Hochberg (1995). The blue line indicates P = 0·05 after correction. Significant SNPs are highlighted in green. (A) Stems per area. The phenotype was measured as the number of stems in 0·25 m2 at the collection site, and was log transformed. (B) Stem height. The phenotype was adjusted for latitude. (C) Stem diameter at base. The phenotype was log transformed, then adjusted for latitude. (D) Linear models of trait vs. genotype for the five significant SNPs. A genotype of zero or two indicates a homozygote, whereas a genotype of one indicates a heterozygote.

Table 2.

SNPs significantly associated with number of stems per unit area and stem diameter in Russian Miscanthus sacchariflorus

Trait Marker Sorghum bicolor chromosome S. bicolor position* Nearest gene Position relative to gene Arabidopsis gene Gene ontology Sorghum QTL
Stems per area UIMiscanthus054659† 2 53 659 234 Sobic.002G169700 In 5′ UTR Protein kinase
UIMiscanthus038912†,‡ 2 71 469 182 Sobic.002G350900 In 3′ UTR None
UIMiscanthus015394† 6 50 139 247 Sobic.006G127200 In protein-coding region, Arg→Gln AT1G51540 Galactose oxidase/kelch repeat qM1_6·1§, tillers per plant**
Height UIMiscanthus041646‡ 3 7 815 391 Sobic.003G089400 Upstream AT5G37870 Ubiquitin protein ligase PHE
Diameter UIMiscanthus036746†,‡ 6 57 761 452 Sobic.006G221700 In 5′ UTR Protein binding SDI, stem diameter**
Open in a new tab

*Positions are in base pairs, from alignments to the S. bicolor genome version 2.0. Positions are given for the SNPs themselves, as opposed to the start of the RAD tags.

†Significant association using rrBLUP.

‡Significant association using TASSEL5.

§QTL from Kong et al. (2014).

DISCUSSION

Ecology and population genetics of M. sacchariflorus and M. sinensis

Our results further clarify the native range of M. sinensis and M. sacchariflorus in Russia, which is useful for understanding the extent of environmental adaptation in these species. We found M. sinensis in eastern Russia restricted to maritime regions with USDA hardiness zones 5 and 6, whereas M. sacchariflorus was also found further inland in hardiness zones 3–6 (Fig. 1B). Although M. sinensis has been reported further inland near Lake Khanka (Herbarium at the Komarov Botanical Institute; Harkevich, 1985), we were unable to find it in that region despite deliberate searching. Hodkinson et al. (2016) also did not find M. sinensis in the Lake Khanka region. The previous report of M. sinensis near Khabarovsk (Harkevich, 1985) was probably an error, given that this was very far outside of the range in which we observed M. sinensis. However, as the global climate continues to warm, we predict that the large population of M. sinensis on Russky Island, as well as coastal populations south-west of Vladivostok identified by Hodkinson et al. (2016), will migrate northward into the mainland of eastern Russia and interact more with the currently well-established populations of M. sacchariflorus there. In China and Japan, where M. sacchariflorus and M. sinensis are sympatric, we have observed that M. sacchariflorus is typically restricted to riparian and wetland environments, whereas M. sinensis occupies upland sites. However, in eastern Russia, we observed M. sacchariflorus in both riparian and upland environments (especially near Lake Khanka; Fig. 2B, C), suggesting that the presence or absence of interspecies competition affects the distribution of M. sacchariflorus.

One hypothesis for why M. sacchariflorus is adapted to environments with colder winters (e.g. inland eastern Russia) than M. sinensis is that the rhizomatous growth habit of M. sacchariflorus allows for better winter survival via cold avoidance of buds deep in the soil compared with the caespitose habit of M. sinensis. Rhizomatous growth habit has been associated with cold adaptation among C4 grasses in Canada due to rhizome buds being deeper underground than crown buds (Schwarz and Reaney, 1989). However, during our collecting, we observed that most of the M. sacchariflorus rhizomes were in the top 15 cm of soil, suggesting that M. sacchariflorus from eastern Russia may have physiological tolerance to cold that allows its rhizomes to overwinter in frozen soil. Future experiments in controlled environments will be needed to test this hypothesis.

We encountered an apparent north-eastern edge to the range of M. sacchariflorus at 49·3°N, 136·5°E. The absence of M. sacchariflorus north-east of 49·3°N, 136·5°E corresponds to the transition from hardiness zone 3 to 2 (Fig. 1B); thus, extreme winter temperatures may limit the range of M. sacchariflorus, although other factors such as lack of disturbed open habitat cannot be ruled out. West of Khabarovsk along the northern side of the Amur River watershed, we travelled as far inland as 48·7°N, 133·0°E without finding a western edge to the range of M. sacchariflorus. At the most extreme, we found M. sacchariflorus in areas with a mean January temperature of –24 °C (Fig. 1C). Based on our observations and the hardiness zone map (Fig. 1B), it would be worthwhile for future expeditions to explore if the western and northern boundaries of M. sacchariflorus are the Zeya and Selemdzha Rivers of eastern Russia, respectively.

In contrast to previous studies that found substantial population structure in M. sinensis (Chou et al., 2000; Iwata et al., 2005; Slavov et al., 2013; Zhang et al., 2013; Zhao et al., 2013; Clark et al., 2014, 2015; Nie et al., 2014), relatively weak population structure was observed in the present study for Russian M. sacchariflorus. Although nuclear SNPs gave a significant signal of isolation by distance (Fig. 4A), neither DAPC (Fig. S5), Structure (Fig. 1A) nor Neighbor–Joining (Fig. S6) revealed geographically distinct genetic clusters (except for groups of closely related, possibly clonal, individuals found at the same collection sites; Fig. S6), and there was no significant isolation by distance found among chloroplast haplotypes (Fig. 4B). In Japan, all or nearly all M. sacchariflorus are tetraploid (Hirayoshi et al., 1957) and many have introgressions from diploid M. sinensis (Clark et al., 2015), whereas in Russia diploidy predominates and there was no evidence of introgression with M. sinensis in our sampling range, although Hodkinson et al. (2016) found putative hybrids further south in Russia. In north-east China, Jiang et al. (2013) similarly found that diploid populations of M. sacchariflorus did not form an introgressed hybrid swarm with M. sinensis, though F1 hybrids were observed there.

It is possible that with a broader geographic sample of M. sacchariflorus, genetic patterns will emerge that explain the species’ history similarly to how the migration history of M. sinensis has been elucidated (Clark et al., 2014). On the other hand, the lack of population structure in M. sacchariflorus may reflect fundamental biological differences between M. sinensis and M. sacchariflorus. For example, M. sacchariflorus may have higher gene flow than M. sinensis, perhaps due to rhizome pieces being dispersed along waterways or by plowing (Deng et al., 2013). Moreover, being more cold-tolerant, M. sacchariflorus was probably not restricted in range to the same extent as M. sinensis during the last glacial maximum, and so the magnitude of founder effects from recolonization of Asia may not be as pronounced. Lack of recent founder effects is also consistent with the greater genetic diversity that we observed in M. sacchariflorus relative to M. sinensis (Table 1). The high levels of genetic diversity in the accessions of M. sacchariflorus from eastern Russia suggest that this germplasm collection will be especially useful for breeding.

To facilitate the long-term preservation of genes from the Russian M. sacchariflorus genotypes that we collected as clonal divisions, the best option will be to develop seed-based germplasm pools by inter-mating the clones, because seed can be kept viable and safe in cold storage for many decades, whereas loss of vegetative stock plants is a substantial risk for clonal collections. The optimal number of germplasm pools for preserving the Russian M. sacchariflorus genes should balance the desire to preserve genetic differences associated with geographic distance with the expense of maintaining and increasing seed stocks of multiple accessions for an obligate outcrossing species. At a minimum, there would need to be two seed-based germplasm pools for the Russian M. sacchariflorus accessions: one for the tetraploids and one for the diploids. More conservatively, we would recommend splitting the diploid M. sacchariflorus accessions into three germplasm pools: (1) for accessions from Vladivostok in the south to the south-west shore of Lake Khanka in the north; (2) along the Ussuri River watershed from north-east of Lake Khanka in the south to near Ussuri’s confluence with the Amur River at Khabarovsk in the north; and (3) along the Amur River watershed from near Birobidzhan and the Bira River in the west to near the village of Naykhin and the confluence of the Amur and Anyuy Rivers in the east (Supplementary Data Fig. S9). However, genetic differentiation between these Russian M. sacchariflorus groups (Supplementary Data Table S3) is an order of magnitude lower than that between the six major genetic groups of M. sinensis identified by Clark et al. (2014).

Genome-wide association analysis using field observations of M. sacchariflorus

Using an inexpensive genotyping method and phenotypic data obtained at the time of germplasm collection for the Russian M. sacchariflorus, we were able to identify three SNPs that were significantly associated with the number of stems per unit area, one with height and one with stem diameter. Four of these SNPs were in the transcribed regions of genes; one coded for an amino acid substitution, and three were near (within 10 Mb) or within sorghum QTLs for related traits (Table 2), suggesting that the associations were not spurious. The number of stems per unit area is not only important for the ecology of grasses (Harper, 1985), but is also a useful predictor of biomass yield in Miscanthus (Gauder et al., 2012). In a previous agronomic study, yield of M. sacchariflorus was negatively impacted by low number of stems per unit area, but the effect was not observed until the third year (Clifton-Brown and Lewandowski, 2002), highlighting the utility of measuring this trait during germplasm collection.

Our results indicate that value can be added to a germplasm collection if phenotyping is done at the collection site and if low-cost, high-density genotyping is performed via high-throughput sequencing. Plant breeders are unlikely to use accessions in germplasm collections until phenotypic and/or genotypic data are available to predict the value of accessions for traits of interest. If phenotypic data are not obtained by the germplasm collectors at the collection sites, then typically none will be available until the germplasm repository receives grant funding for replicated testing (McCouch et al., 2012). Although phenotypic data from collection sites are unreplicated and confounded by site to site environmental variation, there is value in collecting such data for relatively stable traits because potential users of the germplasm will find the data useful and better than no data at all. Moreover, our study and others (Parchman et al., 2012; Budde et al., 2014) have shown that although lack of replication for in situ phenotypic data would be expected to reduce the statistical power of association analyses as compared with analyses performed using replicated field trial data, it is still possible to identify useful significant SNP–trait associations with standard GWAS methodology (Q–K mixed model analysis and multiple testing correction using FDR). The high genetic diversity and absence of strong population structure within diploid Russian M. sacchariflorus also enhanced our power to detect SNP–trait associations. Overall, these results suggest that, for at least some important traits, plant phenotypic data measured in the wild can be sufficiently robust to be useful to plant breeders, particularly when adjusted for known environmental covariates, e.g. latitude in our study.

The use of in situ phenotypic data for GWAS will help to accelerate plant breeding, given that many years typically pass between when germplasm is collected and when replicated field trials are conducted and high quality phenotypic data are available. GWAS results are especially useful for the efficient utilization of germplasm collections, given that they can be used to make predictions about the performance of individual accessions, thereby aiding in the selection of accessions for further evaluation or breeding (McCouch et al., 2012). Moreover, at approx. US$20 per sample for GBS or RAD-seq, it may be less expensive to genotype an accession than to collect and import it. SNP data that are associated with DNA sequence tags can be cross-referenced to any sequence data that are generated in the future, making them much more valuable long term than older marker technologies such as RFLPs (restriction fragment length polymorphisms) and microsatellites (Kilian and Graner, 2012). Using an inexpensive sequencing-based genotyping method (RAD-seq), we were able to identify SNPs associated with traits of agronomic interest several years in advance of when we would be able to obtain data from replicated field trials. In fact, it has been possible for us to initiate replicated trials of these accessions only as early as spring of 2015, and replicated data from mature plants will not be available until the end of 2017, >5 years after the accessions were collected. We expect that many breeders will be willing to accept the risk of performing marker-assisted selection using QTLs from preliminary studies such as ours, given the potential reward of increasing yields as quickly as possible. We hope that the present study will encourage future germplasm collectors to take the time to record phenotypic data in situ and conduct genomic analyses that can rapidly provide end-users with useful guidance on how best to use the collected materials. By helping prospective end-users of the germplasm avoid a data desert, it should be possible to circumvent this common barrier to the use of germplasm collections, especially for newly collected accessions.

Applications for breeding Miscanthus and Saccharum

This collection of Miscanthus from eastern Russia, which is being curated by the US National Plant Germplasm System, along with the collection of Hodkinson et al. (2016), which is being kept at Trinity College and Teagasc, Ireland and at the Institute of Cytology and Genetics of Siberian Branch of the Russian Academy of Sciences, probably represents the maximum winter hardiness for both M. sinensis and M. sacchariflorus. Additionally, the collection includes individuals with desirable biomass traits. Prior to the establishment of this collection, the National Plant Germplasm System did not have any M. sacchariflorus accessions, and the one M. sacchariflorus genotype available from commercial horticulture nurseries in North America is short (<1·5 m) and has thin stems, which are undesirable traits for biomass production. In contrast, we collected many accessions that are tall (up to 2·6 m) and have thick stems (up to 8 mm diameter). Moreover, the three tetraploid accessions of M. sacchariflorus that we collected will be especially valuable for breeding sterile triploid M. ×giganteus (tetraploid M. sacchariflorus × diploid M. sinensis) that is more winter hardy than the currently predominant cultivar used for biomass production in North America and Europe. In addition to being useful sources of genes for improving Miscanthus, the accessions of M. sinensis and M. sacchariflorus that we collected in eastern Russia can also be crossed to Saccharum to introgress cold hardiness into sugarcane, thereby extending its range of cultivation into higher latitudes and elevations than is currently possible for this tropical crop. Previous efforts to cross Miscanthus and Saccharum have been successful, but have focused on introgressing disease resistance, rather than cold hardiness, into Saccharum (Sacks et al., 2013). Furthermore, our GWAS results may facilitate marker-assisted selection of M. sacchariflorus accessions for high numbers of stems per area, plant height and large stem diameter, which should be useful for breeding cultivars with improved biomass yield and resistance to lodging. Due to the small number of individuals included in our GWAS, it is possible that we have overestimated QTL effects (Beavis, 1998; Ioannidis, 2008); however, they represent testable hypotheses, and we intend to perform crosses to validate these QTLs so that they can potentially be used in breeding as quickly as possible. Given that both winter hardiness and stems per area of M. sacchariflorus are proabably related to the plant’s underground architecture, it will be important to determine if these two traits are correlated with each other and controlled by the same genes, because such knowledge would facilitate the breeding of cultivars that are both more winter hardy and higher yielding than current Miscanthus cultivars. With the recent interest in Miscanthus for bioenergy, we expect that this germplasm collection and its associated data will be highly requested resources.

SUPPLEMENTARY DATA

Supplementary data are available at http://www.aob.oxfordjournals.org and consist of the following. Table S1: list of herbarium specimens collected. Table S2: Miscanthus sacchariflorus plastid haplotypes found in eastern Russia. Table S3: Jost’s D statistic showing differentiation between Miscanthus sacchariflorus seed increase groups from Russia. Figure S1: relationship matrix of 155 diploid Russian M. sacchariflorus accessions used in genome-wide association analysis (GWAS). Figure S2: principal component analysis of M. sinensis SNP data. Figure S3: ITS DNA sequence used for identifying Hemarthria sibirica. Figure S4: associations between phenotypes in M. sacchariflorus. Figure S5: Bayesian Information Criterion for selecting the number of M. sacchariflorus clusters in DAPC. Figure S6: Neighbor–Joining tree of M. sacchariflorus. Figure S7: choice of number of M. sacchariflorus clusters in Structure analysis. Figure S8: Structure results for Russian M. sacchariflorus when two clusters are assumed (K = 2). Figure S9: recommendations for seed increase groups in M. sacchariflorus collected in eastern Russia. Supplementary Materials and Methods: additional details on exploration route, flow cytometry and DNA protocols. Dataset S1: 2012 Collection of Miscanthus Germplasm in Eastern Russia. Dataset S2: DNA sequences, allele frequencies, alignment positions and GWAS results for M. sacchariflorus SNPs.

Supplementary Data
supp_118_5_941__index.html (1KB, html)

ACKNOWLEDGEMENTS

We gratefully acknowledge Inga Kotlyarskaya and Andrey Kseudz for their expert guiding and driving during the collection expedition, and Melina Salgado for assisting with DNA extraction and PCR. Karen Williams, Melanie Harrison, the editor Susanne Barth and two anonymous reviewers provided valuable feedback on earlier drafts of this manuscript. This work was supported by the USDA-ARS NPGS Plant Exploration Program [Cooperative Agreement #59-1275-1-338] and the DOE Office of Science, Office of Biological and Environmental Research (BER) [grant nos DE-SC0006634 and DE-SC0012379].

LITERATURE CITED

  1. Barth S, Cormican P, Michel T, Velmurugan J, Klaas M. 2014. Genotyping by sequencing of a collection of Miscanthus spp. accessions. In: Plant and Animal Genome XXII Conference. San Diego, CA.
  2. Beavis WD. 1998. QTL analyses: power, precision, and accuracy In: Paterson AH, ed. Molecular dissection of complex traits. Boca Raton, FL: CRC Press, 145–162. [Google Scholar]
  3. Benjamini Y, Hochberg Y. 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society B 57: 289–300. [Google Scholar]
  4. Bradbury PJ, Zhang Z, Kroon DE, Casstevens TM, Ramdoss Y, Buckler ES. 2007. TASSEL: software for association mapping of complex traits in diverse samples. Bioinformatics 23: 2633–2635. [DOI] [PubMed] [Google Scholar]
  5. Budde KB, Heuertz M, Hernández-Serrano A, et al. 2014. In situ genetic association for serotiny, a fire-related trait, in Mediterranean maritime pine (Pinus pinaster). New Phytologist 201: 230–241. [DOI] [PubMed] [Google Scholar]
  6. de Cesare M, Hodkinson TR, Barth S. 2010. Chloroplast DNA markers (cpSSRs, SNPs) for Miscanthus, Saccharum and related grasses (Panicoideae, Poaceae). Molecular Breeding 26: 539–544. [Google Scholar]
  7. Chae WB, Hong SJ, Gifford JM, Lane Rayburn A, Sacks EJ, Juvik JA. 2014. Plant morphology, genome size, and SSR markers differentiate five distinct taxonomic groups among accessions in the genus Miscanthus. GCB Bioenergy 6: 646–660. [Google Scholar]
  8. Chessel D, Dufour AB, Thioulouse J. 2004. The ade4 package – I: one-table methods. R News 4: 5–10. [Google Scholar]
  9. Chou C-H, Chiang Y-C, Chiang T-Y. 2000. Genetic variability and phytogeography of Miscanthus sinensis var. condensatus, an apomictic grass, based on RAPD fingerprints. Canadian Journal of Botany 78: 1262–1268. [Google Scholar]
  10. Clark LV, Jasieniuk M. 2011. POLYSAT: an R package for polyploid microsatellite analysis. Molecular Ecology Resources 11: 562–566. [DOI] [PubMed] [Google Scholar]
  11. Clark LV, Brummer JE, Głowacka K, et al. 2014. A footprint of past climate change on the diversity and population structure of Miscanthus sinensis. Annals of Botany 114: 97–107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Clark LV, Stewart JR, Nishiwaki A, et al. 2015. Genetic structure of Miscanthus sinensis and Miscanthus sacchariflorus in Japan indicates a gradient of bidirectional but asymmetric introgression. Journal of Experimental Botany 66: 4213–4225. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Clifton-Brown JC, Lewandowski I. 2000. Overwintering problems of newly established Miscanthus plantations can be overcome by identifying genotypes with improved rhizome cold tolerance. New Phytologist 148: 287–294. [Google Scholar]
  14. Clifton-Brown JC, Lewandowski I. 2002. Screening Miscanthus genotypes in field trials to optimise biomass yield and quality in Southern Germany. European Journal of Agronomy 16: 97–110. [Google Scholar]
  15. Clifton-Brown J, Chiang Y-C, Hodkinson TR. 2008. Miscanthus: genetic resources and breeding potential to enhance bioenergy production In: Vermerris W, ed. Genetic improvement of bioenergy crops. Berlin: Springer, 273–294. [Google Scholar]
  16. Deng Z, Chen X, Xie Y, Li X, Pan Y, Li F. 2013. Effects of size and vertical distribution of buds on sprouting and plant growth of the clonal emergent macrophyte Miscanthus sacchariflorus (Poaceae). Aquatic Botany 104: 121–126. [Google Scholar]
  17. Earl DA, VonHoldt BM. 2011. STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genetics Resources 4: 359–361. [Google Scholar]
  18. Endelman JB. 2011. Ridge regression and other kernels for genomic selection with R package rrBLUP. Plant Genome Journal 4: 250–255. [Google Scholar]
  19. Evanno G, Regnaut S, Goudet J. 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology 14: 2611–2620. [DOI] [PubMed] [Google Scholar]
  20. Falush D, Stephens M, Pritchard JK. 2003. Inference of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics 164: 1567–1587. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Farrell AD, Clifton-Brown JC, Lewandowski I, Jones MB. 2006. Genotypic variation in cold tolerance influences the yield of Miscanthus. Annals of Applied Biology 149: 337–345. [Google Scholar]
  22. Friesen PC, Peixoto MM, Busch FA, Johnson DC, Sage RF. 2014. Chilling and frost tolerance in Miscanthus and Saccharum genotypes bred for cool temperate climates. Journal of Experimental Botany 65: 3749–3758. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Gauder M, Graeff-Hönninger S, Lewandowski I, Claupein W. 2012. Long-term yield and performance of 15 different Miscanthus genotypes in southwest Germany. Annals of Applied Biology 160: 126–136. [Google Scholar]
  24. Głowacka K, Adhikari S, Peng J, et al. 2014. Variation in chilling tolerance for photosynthesis and leaf extension growth among genotypes related to the C4 grass Miscanthus×giganteus. Journal of Experimental Botany 65: 5267–5278. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Głowacka K, Clark LV, Adhikari S, et al. 2015. Genetic variation in Miscanthus × giganteus and the importance of estimating genetic distance thresholds for differentiating clones. Global Change Biology Bioenergy 7: 386–404. [Google Scholar]
  26. Greef J, Deuter M, Jung C, Schondelmaier J. 1997. Genetic diversity of European Miscanthus species revealed by AFLP fingerprinting. Genetic Resources and Crop Evolution 44: 185–195. [Google Scholar]
  27. Harkevich SS. 1985. The vascular plants of Soviet Far East. Nauka Publisher. [Google Scholar]
  28. Harper JL. 1985. Modules, branches, and the capture of resources In: Jackson JBC, Buss LW, Cook RE, eds. Population biology and evolution of clonal organisms. New Haven, CT: Yale University Press, 1–34. [Google Scholar]
  29. Hijmans RJ. 2014. geosphere: spherical trigonometry. R package version 1.3-11.
  30. Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A. 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25: 1965–1978. [Google Scholar]
  31. Hirayoshi I, Nishikawa K, Kubono M, Murase T. 1957. Cyto-genetical studies on forage plants (VI): On the chromosome number of Ogi (Miscanthus sacchariflorus). Research Bulletin of the Faculty of Agriculture, Gifu University 8: 8–13. [Google Scholar]
  32. Hodkinson TR, Renvoize S. 2001. Nomenclature of Miscanthus×giganteus (Poaceae). Kew Bulletin 56: 759–760. [Google Scholar]
  33. Hodkinson T, Chase M, Lledó D, Salamin N, Renvoize SA. 2002. Phylogenetics of Miscanthus, Saccharum and related genera (Saccharinae, Andropogoneae, Poaceae) based on DNA sequences from ITS nuclear ribosomal DNA and plastid trnL intron and trnL-F intergeneic spacers. Journal of Plant Research 115: 381–392. [DOI] [PubMed] [Google Scholar]
  34. Hodkinson TR, Klaas M, Jones MB, Prickett R, Barth S. 2015. Miscanthus: a case study for the utilization of natural genetic variation. Plant Genetic Resources 13: 219–237. [Google Scholar]
  35. Hodkinson TR, Petrunenko E, Klaas M. et al. 2016. New breeding collections of Miscanthus sinensis, M. sacchariflorus and hybrids from Primorsky Krai, Far Eastern Russia. In: Proceedings of the Biomass 2015: Perennial biomass crops for a resource constrained world. Springer.
  36. Ioannidis JPA. 2008. Why most discovered true associations are inflated. Epidemiology 19: 640–648. [DOI] [PubMed] [Google Scholar]
  37. Iwata H, Kamijo T, Tsumura Y. 2005. Genetic structure of Miscanthus sinensis ssp. condensatus (Poaceae) on Miyake Island: implications for revegetation of volcanically devastated sites. Ecological Research 20: 233–238. [Google Scholar]
  38. Jiang J-X, Wang Z-H, Tang B-R, Xiao L, Ai X, Yi Z-L. 2012. Development of novel chloroplast microsatellite markers for Miscanthus species (Poaceae). American Journal of Botany 99: e230–e233. [DOI] [PubMed] [Google Scholar]
  39. Jiang J, Zhu M, Ai X, Xiao L, Deng G, Yi Z. 2013. Molecular evidence for a natural diploid hybrid between Miscanthus sinensis (Poaceae) and M. sacchariflorus. Plant Systematics and Evolution 299: 1367–1377. [Google Scholar]
  40. Johnston SE, McEwan JC, Pickering NK, et al. 2011. Genome-wide association mapping identifies the genetic basis of discrete and quantitative variation in sexual weaponry in a wild sheep population. Molecular Ecology 20: 2555–2566. [DOI] [PubMed] [Google Scholar]
  41. Johnston SE, Orell P, Pritchard VL, et al. 2014. Genome-wide SNP analysis reveals a genetic basis for sea-age variation in a wild population of Atlantic salmon (Salmo salar). Molecular Ecology 23: 3452–3468. [DOI] [PubMed] [Google Scholar]
  42. Jombart T, Ahmed I. 2011. adegenet 1.3-1: new tools for the analysis of genome-wide SNP data. Bioinformatics 27: 3070–3071. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Jombart T, Devillard S, Balloux F. 2010. Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. BMC Genetics 11: 94. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Jørgensen U. 1997. Genotypic variation in dry matter accumulation and content of N, K and Cl in Miscanthus in Denmark. Biomass and Bioenergy 12: 155–169. [Google Scholar]
  45. Jost L. 2008. GST and its relatives do not measure differentiation. Molecular Ecology 17: 4015–4026. [DOI] [PubMed] [Google Scholar]
  46. Kilian B, Graner A. 2012. NGS technologies for analyzing germplasm diversity in genebanks. Briefings in Functional Genomics 11: 38–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Kong W, Guo H, Goff VH, Lee T-H, Kim C, Paterson AH. 2014. Genetic analysis of vegetative branching in sorghum. Theoretical and Applied Genetics 127: 2387–2403. [DOI] [PubMed] [Google Scholar]
  48. Langmead B, Salzberg SL. 2012. Fast gapped-read alignment with Bowtie 2. Nature Methods 9: 357–359. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Lu F, Lipka AE, Glaubitz J, et al. 2013. Switchgrass genomic diversity, ploidy, and evolution: novel insights from a network-based SNP discovery protocol. PLoS Genetics 9: e1003215. doi:10.1371/journal.pgen.1003215. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Matumura M, Hasegawa T, Saijoh Y. 1985. Ecological aspects of Miscanthus sinensis var. condensatus, M. sacchariflorus, and their 3×-, 4×-hybrids (1) Process of vegetative spread. Research Bulletin of the Faculty of Agriculture, Gifu University 50: 423–433. [Google Scholar]
  51. Matumura M, Hasegawa T, Saijoh Y. 1986. Ecological aspects of Miscanthus sinensis var. condensatus, M. sacchariflorus and their 3×-, 4×-hybrids (2) Growth behaviour of the current year’s rhizomes. Research Bulletin of the Faculty of Agriculture, Gifu University 51: 347–362. [Google Scholar]
  52. Matumura M, Hasegawa T, Saijoh Y. 1987. Ecological aspects of Miscanthus sinensis var. condensatus, M. sacchariflorus and their 3×-, 4×-hybrids. (3) Above ground standing crop and response cutting. Research Bulletin of the Faculty of Agriculture, Gifu University 52: 315–324. [Google Scholar]
  53. Maximowicz CJ. 1859. Primitiae Florae Amurensis. St. Petersburg: Buchdruckerei der Kaiserlichen Akademie der Wissenschaften.
  54. McCouch SR, McNally KL, Wang W, Hamilton RS. 2012. Genomics of gene banks: a case study in rice. American Journal of Botany 99: 407–423. [DOI] [PubMed] [Google Scholar]
  55. NAPPFAST . 2012. NAPPFAST Global Plant Hardiness Zones based on CFSR base data. Last accessed February 17, 2015.
  56. Narum SR, Buerkle CA, Davey JW, Miller MR, Hohenlohe PA. 2013. Genotyping-by-sequencing in ecological and conservation genomics. Molecular Ecology 22: 2841–2847. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Nie G, Zhang X-Q, Huang L-K, et al. 2014. Genetic variability and population structure of the potential bioenergy crop Miscanthus sinensis (Poaceae) in southwest China based on SRAP markers. Molecules 19: 12881–12897. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Paradis E. 2010. pegas: an R package for population genetics with an integrated-modular approach. Bioinformatics 26: 419–420. [DOI] [PubMed] [Google Scholar]
  59. Parchman TL, Gompert Z, Mudge J, Schilkey FD, Benkman CW, Buerkle CA. 2012. Genome-wide association genetics of an adaptive trait in lodgepole pine. Molecular Ecology 21: 2991–3005. [DOI] [PubMed] [Google Scholar]
  60. Paterson AH, Bowers JE, Bruggmann R, et al. 2009. The Sorghum bicolor genome and the diversification of grasses. Nature 457: 551–556. [DOI] [PubMed] [Google Scholar]
  61. Petersen KK, Hagberg P, Kristiansen K. 2003. Colchicine and oryzalin mediated chromosome doubling in different genotypes of Miscanthus sinensis. Plant Cell, Tissue and Organ Culture 73: 137–146. [Google Scholar]
  62. Phuong N, Stützel H, Uptmoor R. 2013. Quantitative trait loci associated to agronomic traits and yield components in a Sorghum bicolor L. Moench RIL population cultivated under pre-flowering drought and well-watered conditions. Agricultural Sciences 4: 781–791. [Google Scholar]
  63. Poland JA, Brown PJ, Sorrells ME, Jannink J-L. 2012a. Development of high-density genetic maps for barley and wheat using a novel two-enzyme genotyping-by-sequencing approach. PLoS One 7: e32253. doi:10.1371/journal.pone.0032253. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Poland J, Endelman J, Dawson J, et al. 2012b.. Genomic selection in wheat breeding using genotyping-by-sequencing. Plant Genome Journal 5: 103. [Google Scholar]
  65. Purdy SJ, Maddison AL, Jones LE, et al. 2013. Characterization of chilling-shock responses in four genotypes of Miscanthus reveals the superior tolerance of M. ×giganteus compared with M. sinensis and M. sacchariflorus. Annals of Botany 111: 999–1013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. R Core Team . 2014. R: a language and environment for statistical computing R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.
  67. Sacks EJ, Juvik JA, Lin Q, Stewart JR, Yamada T. 2013. The gene pool of Miscanthus species and its improvement In: Paterson AH, ed. Genomics of the Saccharinae. New York: Springer, 73–101. [Google Scholar]
  68. Schwarz AG, Reaney MJT. 1989. Perennating structures and freezing tolerance of northern and southern populations of C4 grasses. Botanical Gazette 150: 239–246. [Google Scholar]
  69. Shiringani AL, Frisch M, Friedt W. 2010. Genetic mapping of QTLs for sugar-related traits in a RIL population of Sorghum bicolor L. Moench. Theoretical and Applied Genetics 121: 323–336. [DOI] [PubMed] [Google Scholar]
  70. Slavov G, Robson P, Jensen E, et al. 2013. Contrasting geographic patterns of genetic variation for molecular markers vs. phenotypic traits in the energy grass Miscanthus sinensis. GCB Bioenergy 5: 562–571. [Google Scholar]
  71. Slavov GT, Nipper R, Robson P, et al. 2014. Genome-wide association studies and prediction of 17 traits related to phenology, biomass and cell wall composition in the energy grass Miscanthus sinensis. New Phytologist 201: 1227–1239. [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Winter D. 2012. MMOD: an R library for the calculation of population differentiation statistics. Molecular Ecology Resources 12: 1158–1160. [DOI] [PubMed] [Google Scholar]
  73. Yook MJ, Lim S-H, Song J-S, et al. 2014. Assessment of genetic diversity of Korean Miscanthus using morphological traits and SSR markers. Biomass and Bioenergy 66: 81–92. [Google Scholar]
  74. Yu J, Pressoir G, Briggs WH, et al. 2006. A unified mixed-model method for association mapping that accounts for multiple levels of relatedness. Nature Genetics 38: 203–208. [DOI] [PubMed] [Google Scholar]
  75. Zhang QX, Shen YK, Shao RX, et al. 2013. Genetic diversity of natural Miscanthus sinensis populations in China revealed by ISSR markers. Biochemical Systematics and Ecology 48: 248–256. [Google Scholar]
  76. Zhao H, Wang B, He J, et al. 2013. Genetic diversity and population structure of Miscanthus sinensis germplasm in China. PLoS One 8: e75672. doi: 10.1371/journal.pone.0075672. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Data
supp_118_5_941__index.html (1KB, html)
supp_mcw137_aob-16101-s01.docx (3.1MB, docx)
supp_mcw137_aob-16101-s02.xlsx (5.5MB, xlsx)
supp_mcw137_aob-16101-s03.xlsx (2.5MB, xlsx)

Articles from Annals of Botany are provided here courtesy of Oxford University Press

RESOURCES