• Open Access

Invasiveness potential of Miscanthus sinensis: implications for bioenergy production in the United States

Authors


L. D. Quinn, e-mail: ldquinn@illinois.edu

Abstract

Miscanthus sinensis (Anderss.) is a perennial grass species that has been grown widely as an ornamental since the late 1800s and is now being considered for bioenergy production in the United States. With its ability to be grown from seed and tolerate cold climates, this species offers practical advantages over current cultivars of the higher-yielding hybrid species, M.×giganteus. Yet a large-scale release of M. sinensis for bioenergy production in colder northern regions could result in new invasions into natural areas. We show, with reference to historical records and data collected in six wild US populations of M. sinensis in 2009, that ornamental varieties of this species have a long history of localized escape in the Eastern United States, primarily within the Appalachian region. To prevent further escape and gene flow, we recommend the development of sterile or functionally sterile varieties of M. sinensis or the restriction of its usage as a donor of genetic material to new sterile cultivars of M. ×giganteus. Other appropriate precautions for new biomass varieties include experimental demonstration of low invasiveness in the target region ahead of commercial production, along with postintroduction stewardship programs.

Introduction

Miscanthus species show promise in ameliorating carbon emissions and sequestering carbon in plant tissues (Clifton-Brown et al., 2007) without replacing food crops (Heaton et al., 2008). As a candidate for bioenergy production, Miscanthus sinensis Andersson (Japanese silvergrass/Chinese silvergrass/Maiden grass/Eulalia) shows notable potential (Stewart et al., 2009). This perennial caespitose grass, which is native to Japan, Korea, China, Taiwan, and parts of Russia, attains aboveground yields ranging from 1.8 to 13 t ha−1 in native grasslands in Japan (Stewart et al., 2009 and references therein). Yield data has not been published for M. sinensis in the United States, but test plots in Europe showed yields of 9.5–14.8 t ha−1 for wild accessions and 13.5 and 19.4 t ha−1 for two breeding lines (Clifton-Brown et al., 2001). These values are small in comparison with yields seen in M.×giganteus, a sterile triploid hybrid of M. sinensis and M. sacchariflorus, which has produced 29.6 t ha−1 in Illinois, USA test plots (Heaton et al., 2008). However, M. sinensis has practical agronomic advantages compared with its sterile relative (Gutterson & Zhang, 2009; Jakob et al., 2009; Zub & Brancourt-Hulmel, 2010). Because it produces viable seed, M. sinensis can be easily and economically sown using traditional farm equipment (Christian et al., 2005). Current varieties of M. ×giganteus must be established from rooted plantlets from tissue culture or as rhizome fragments (Venturi et al., 1998). In addition, selections of M. sinensis from northern latitudes appear to tolerate cold climates to a greater degree than M. ×giganteus (Jorgensen, 1997; Clifton-Brown et al., 2001; Farrell et al., 2006). In European test plots, M. ×giganteus attained greater yields than M. sinensis accessions in southern latitudes, but only M. sinensis and its hybrids survived the first winter in Sweden or Denmark (Clifton-Brown et al., 2001). M. sinensis may also be better adapted to drought stress than current varieties of M. ×giganteus (Clifton-Brown & Lewandowski, 2000; Clifton-Brown et al., 2002). For these reasons, M. sinensis is being promoted as a genetic resource for new hybrids and evaluated as a potential bioenergy crop (Stewart et al., 2009; Zub & Brancourt-Hulmel, 2010).

Production of M. sinensis could be a boon for the bioenergy industry in the United States, particularly in colder northern climates, but several sources have expressed concern that this and other potential bioenergy crops could escape production to become invasive species (Raghu et al., 2006; Barney & DiTomaso, 2008; Buddenhagen et al., 2009). Warnings about invasiveness of bioenergy crops are based on widely accepted weed risk assessment (WRA) protocols (Barney & DiTomaso, 2008;Cousens, 2008; Buddenhagen et al., 2009), and the reasoning that crop species with the capacity to thrive on marginal land with little irrigation or fertilizer may have much in common with disturbance-adapted invasive species (Raghu et al., 2006). While it may be capable of high productivity on marginal soils, M. ×giganteus received low (noninvasive) WRA scores for the United States, primarily due to its inability to produce fertile seed (Barney & DiTomaso, 2008). Conversely, M. sinensis not only produces viable seed in the United States (Meyer & Tchida, 1999), it also acts as a pioneer species in its native range, colonizing and eventually dominating heavily disturbed volcanic sites (Tsuyuzaki & Hase, 2005) and clear-cuts (Ohtsuka et al., 1993), especially where management (e.g. burning) prevents transition to forest (Stewart et al., 2009). Genotypes of M. sinensis tolerate a number of stressful conditions, including low-fertility soils, cold temperatures, heavy metals, low pH, and frequent burning (Stewart et al., 2009). In addition, it has been shown to tolerate shade in the United States (Meyer, 2003; Horton et al., 2010). These attributes suggest that an evaluation of the invasive potential of M. sinensis is warranted before development of new varieties of Miscanthus for biofuel production.

This potential may already be realized in some parts of the United States. One review notes that M. sinensis has a history of naturalization in the United States (Barney & DiTomaso, 2008), and there is a growing categorization of M. sinensis as an invasive species (Kaufman & Kaufman, 2007) by horticulturists (Meyer & Tchida, 1999; Peters et al., 2006; Wilson & Knox, 2006), floristic databases (EDDMaps, 2010; USDA NRCS, 2010), state and regional invasive plant councils (SE-EPPC, 2010), and the United States Forest Service (Miller et al., 2004; USDA Forest Service, 2006). In contrast, in a recent assessment of the invasive potential of ornamental M. sinensis in North Carolina, overall recommendations ranged from ‘insignificant/moderate invasiveness’ to ‘noninvasive’ classifications (Trueblood, 2009). However, this recommendation was made without quantification of the extent of invasion in North Carolina natural areas. Similar contradictions are likely to continue as a result of the absence of peer-reviewed analysis of M. sinensis invasion ecology in the United States. In this paper, we discuss the history of early introduction and first escapes, present the current distribution of naturalized populations, and summarize new data from surveys of six Eastern US populations. Acknowledging substantial potential benefits of bioenergy production, we also provide practical solutions for breeding M. sinensis varieties that will have a low likelihood of invasiveness.

A note on terminology: defining invasiveness

Several terms exist to describe undesirable plants (Radosevich et al., 2007). These include, but are not limited to: weeds, aliens, pests, colonists, exotics, escapes, naturalized species, and invaders. Even if a common term is used, it can carry different meanings depending on the user or the context (Richardson et al., 2000). Thus, in recent years, several attempts have been made to standardize the definition of ‘invasive’. Although some debate remains, it has been generally agreed upon that invasive species are those taxa that pass through a number of geographic and environmental filters to become established in sites of introduction and subsequently spread via dispersal of seeds or vegetative propagules to establish new populations (Rejmanek, 2000; Richardson et al., 2000; Colautti & MacIsaac, 2004; Radosevich et al., 2007; Valery et al., 2008). In our continued discussion of M. sinensis as a potential invader, we will refer to this general definition. In some cases, we refer to the species as having naturalized in a particular area. By this, we mean that M. sinensis has established self-sustaining populations outside of cultivation and without anthropogenic assistance aside from the initial introduction event.

Introduction history and current distribution of M. sinensis in the United States

M. sinensis was first introduced as an ornamental species to the United States from Japan (Beal, 1896) in the late 1800s (Favretti & Favretti, 1997). It was recorded in the list of plantings on the Biltmore Estate in Asheville, NC between 1893 and 1895 (P. Andes, personal communication), and at least four varieties were developed at the Biltmore Nursery and sold via mail order starting in 1907 (Alexander, 2007). It was noted in Washington, DC gardens in 1894 (Anon, 1894), and by 1913, it was documented that the species had escaped from cultivation in Washington, DC and was also present in Florida and New York (Britton & Brown, 1913). By the early 1940s, it had established in West Virginia natural areas (Core, 1941), and was considered ‘abundantly naturalized’ in Pennsylvania, New Jersey, and Washington, DC (Moldenke, 1942). Later sources confirmed M. sinensis can escape (Pohl, 1978) to become a ‘nuisance’ in natural areas (Hitchcock, 1950). More specifically, it can displace native grasses and dominate roadsides and pastures (Kaufman & Kaufman, 2007).

While the original location of introduction is unknown, several factors may have acted to increase the distribution of M. sinensis throughout the eastern US over time. Strong genetic variation between introduced populations suggests multiple independent introduction events (L.D. Quinn, T.M. Culley & J.R. Stewart, unpublished data) to various locations in the United States. In addition to the original introduction in the late 1800s, several ornamental varieties of M. sinensis were introduced from Japan into the United States in the 1970s (Grounds, 1998), and it continues to be one of the most popular garden grass plantings (Wilson, 2008; Maynard, 2010). For example, sales of M. sinensis exceed $39 million in North Carolina (Trueblood, 2009). Today, its naturalized distribution is concentrated in the eastern US (EDDMaps, 2010; USDA NRCS, 2010), but has been documented as far west as California (Hickman, 1993) (Fig. 1). The Southeast Exotic Pest Plant Council (SE-EPPC) database lists 139 observations of naturalized M. sinensis, of which 44 have an infested area described (EDDMaps, 2010). These average 1165 m2 in area; 41% of the infestations are ≤1 m2 (i.e., ∼1 plant), and the largest described infestation is 1.5 ha (EDDMaps, 2010). The observations west of the Appalachian region generally appear not to be escapes, but rather reflect ornamental plantings in current or abandoned gardens (Fig. 2; Steve Baluch, personal communication).

Figure 1.

 Current distribution of Miscanthus sinensis in the United States (redrawn from EDDMaps, 2010). Counties with reported naturalized populations are shaded.

Figure 2.

 Examples of Miscanthus sinensis observations in the Southeast Exotic Pest Plant Council (SE-EPPC) database (EDDmaps, 2010) west of the Appalachian region which appear to be (a) current or (b) abandoned ornamental plantings. Photographed 7 November 2009 by Steve Baluch in (a) Poplar Bluff, MO (36.8695, −90.3151) and (b) Carbondale, IL (37.7171984, −89.1881489).

Characterizing largest extant M. sinensis populations

One of the most reliable predictors of invasion success in new locations is evidence of invasiveness elsewhere (Radosevich et al., 2007; Barney & DiTomaso, 2008). Therefore, we chose to focus on large naturalized populations in less-disturbed habitats (away from major roadways) to characterize a ‘worst case scenario’ for invasion of this species. While not all populations will necessarily reach this scale, we show that it is possible for some to become quite large and dense, justifying our recommendations for the breeding techniques outlined in the next sections.

Several extensive naturalized M. sinensis populations have been identified in the eastern US (Meyer, 2003; Horton et al., 2010). The largest known naturalized populations are concentrated along roadways, particularly the Pennsylvania Turnpike (Interstate 76) near Valley Forge National Monument, and Interstate 40, where the population extends for approximately 160 km east of Asheville, NC (Meyer, 2003). Because roads commonly act as conduits of invasive plant spread into new locations (Christen & Matlack, 2009), these populations should not be disregarded. Several M. sinensis populations that are currently established in less-disturbed habitats are thought to have originated from roadside populations (Meyer, 2003; A. Strassman, D. Taylor, M. Ortt & J. Horton, personal communication).

We sampled six naturalized M. sinensis populations in late June and July 2009. These populations were located in five eastern states: Kentucky, North Carolina, New Jersey, Ohio, and Pennsylvania (henceforth referred to by standard State abbreviations) (see Table 1 for additional location information). Two populations were sampled in NJ because their habitats differed: NJ1 was in an open meadow and NJ2, approximately 1.5 km to the northeast, was in a shaded mixed hardwood understory. Detailed or partial historical records were available for the NC, KY, NJ, and PA populations, but not for the OH population (see Table 1). The perimeter of each population was measured using a handheld GPS device (Garmin eTrex Vista, Olathe, KS, USA), except in PA where the perimeter was estimated from an aerial photo on Google Maps. Population density was estimated for each site (except PA, where access was limited) by counting all individuals in eight randomly placed 10 × 10 m plots. Population size was determined by extrapolating average counts to the total area over which the population existed. This population size was then extrapolated or interpolated to a standard area (1 ha) to give density per hectare (Table 1). Seedheads were collected from 20 individuals each in the OH, NC, and KY populations in autumn 2009. After storage at 4 °C for 5 months, 100 seeds from each individual were placed in rolls of moistened germination paper and allowed to germinate over 4 weeks at 25/15 °C (12 h day/12 h night). We were not able to obtain seeds from the NJ and PA populations, as local collectors were not available at the time of seed production.

Table 1.   Site data and population density for six invasive Miscanthus sinensis populations in the eastern United States
Sampled populations
PopulationLatitudeLongitudeDateArea
sampled
(ha)
Density
(indiv./ha)
Germination
Min–max,
average (%)
Site detailsLocation of original
plantings
Original
planting
date
Distance from
sampled
population (m)
  • Population density was not estimated in the Pennsylvania population. Location of original plantings and original planting dates were derived from the following sources:

  • *

    T. Almendinger, Ecologist, Duke Farms Foundation, estate planting records.

  • †P. Andes, Director of Horticulture, Biltmore Estate, estate planting records.

  • ‡D. Taylor, Botanist, Daniel Boone National Forest, historical records from former caretaker of Natural Bridge State Resort Park, Slade, KY.

  • §

    §M. Meyer, Professor of Horticulture, University of Minnesota (2003).

Ohio39.4760−81.30456/26/20090.762820–51, 22.4Private property; grazed pastureUnknownUnknownUnknown
New Jersey 140.5463−74.63466/30/200911650NAConservation trust land; open meadow40.5572, −74.6308*1990s*630
New Jersey 240.5543−74.62387/2/20091.53300NAConservation trust land; mixed hardwood understory40.5572, −74.6308*1990s*1215
North Carolina35.5341−82.54687/15/20092.959000–23, 4.0Biltmore Estate; recent (2001) clearcut in pine forest35.5385, −82.55211890s570
Kentucky37.8034−83.66287/20/20090.615 0758–67, 37.9National Forest; old clearcut in mixed oak forest37.7771, −83.68071920s3000
Pennsylvania40.0769−75.49696/29/20091NANAPrivate property; residential areaUnknown1950s§Unknown

The populations we sampled ranged from 0.6 to 2.9 ha in area, and densities ranged from 1650 to over 15 000 plants ha−1 (Table 1). Seed germination rates were generally low (4–38%, on average) (Table 1), but viability may have been affected by storage time (approximately 5 months) or timing of collection. Note that seeds from the NC population were collected and subjected to germination tests in 2003 (Meyer, 2003). In that study, a much greater germination percentage was observed (approximately 50%, as opposed to our average value of 4%) (Meyer, 2003). In all of the populations in our germination tests, however, at least some individuals produced viable seed.

Recalling that invasive plants are those that pass through geographic and environmental filters to become established and reproductive beyond sites of introduction (Rejmanek, 2000; Richardson et al., 2000; Colautti & MacIsaac, 2004; Radosevich et al., 2007), it seems appropriate to designate the populations we sampled as invasive. In the sites where introduction history is known, it is clear that M. sinensis has established populations well away (570–3000 m) from original ornamental plantings (Table 1). In NJ, large M. sinensis populations have become established over 1 km from source plantings within just 20 years. One definition states that invasive species are those that establish reproductive populations more than 100 m from sources within 50 years (Richardson et al., 2000). While we were unable to obtain seeds from the NJ or PA populations, we did find that at least some individuals produced viable seed in OH, KY, and NC. Other definitions of invasiveness invoke measures of dominance and spatial extent (Colautti & MacIsaac, 2004). Several populations exceeded 1 ha (Table 1) in area, and other naturalized M. sinensis populations are known to be much larger (e.g. roadside populations stretching 3–160 km) (Meyer, 2003). Also, most of the populations we sampled displayed high densities. Population density in Kentucky was 15 075 individuals ha−1 (Table 1), which appeared to exclude most other vegetation (Fig. 3) and exceeds the recommended planting density for fields of M. giganteus for bioenergy production (10 000/ha) (Bullard, 2001). The OH and NC populations were also quite dense at more than 5000 plants/ ha−1 (Table 1).

Figure 3.

 Dense population of Miscanthus sinensis in Red River Gorge Geological Area (Daniel Boone National Forest), near Slade, KY. Photographed 20 July 2009 by Lauren Quinn.

Mechanisms of potential escape from cultivation

As large plantations of fertile Miscanthus cultivars are envisioned for bioenergy production, breeders and potential growers should be aware of the mechanisms of escape to natural areas. Seed dispersal is the most obvious means of escape, and many invasive species have become established as a result of seed dispersal from gardens (Reichard & White, 2001) and agricultural fields (Gressel, 2005). Since ornamental M. sinensis varieties produce viable seeds in a wide geographic gradient across the United States (Meyer & Tchida, 1999), and because seed dispersal distances of several hundred meters have been observed in this species (L.D. Quinn & D. Matlaga, unpublished data), we can assume that invasive populations established as a result of seed dispersal from ornamental plantings. However, sexual reproduction is not a requirement for escape. Plants can also escape cultivation through dispersal of vegetative propagules. Fragmentation and movement of rhizomes can result in large-scale invasions of genetically identical clones (e.g. Arundo donax; Khudamrongsawat et al., 2004; Ahmad et al., 2008). Prevention of seed and rhizome dispersal through careful breeding and management practices will be discussed in the next section.

Cultivated species can also escape in more subtle ways. Gene flow (i.e. pollen transfer) between domesticated species, including crops and ornamentals, and wild relatives can result in hybrids of variable invasiveness (Gressel, 2005; Kowarik, 2005). For example, barley (Hordeum vulgare subsp vulgare), occasionally hybridizes with wild Hordeum species, but these tend not to become problematic invasive species in most areas (Andersson & de Vicente, 2010). Alternatively, one of the world's most invasive species, Johnsongrass (Sorghum halepense) (Holm et al., 1977), is thought to be derived from hybridization between cultivated sorghum (S. bicolor subsp. bicolor) and wild S. propinquum (Ejeta & Grenier, 2005; Andersson & de Vicente, 2010). Moreover, it appears that spontaneous natural hybridization between domesticated species and wild relatives is the rule, not the exception, for the majority of important crop species (Ellstrand, 2003). In the United States, 11 of the 20 most important crop species are known to hybridize spontaneously with wild relatives, and 18 of those species are known to hybridize with wild relatives in other parts of the world (Ellstrand, 2003). In addition, the examples of fertile interspecific hybrids given here are not rare exceptions; in fact, many hybrid plants are at least partially fertile (Ellstrand, 2003; National Research Council, 2004). As we have shown, fertile Miscanthus populations are already present in the United States. The potential outcome of hybridization between cultivated Miscanthus and wild populations cannot be predicted, but steps can be taken to prevent the potential for gene flow.

Breeding noninvasive cultivars of Miscanthus

It is possible to breed for noninvasiveness while retaining valuable agronomic properties of crops (Anderson et al., 2006). In the case of perennial Miscanthus species grown for cellulosic ethanol and power production, the most important agronomic traits are high growth rates and large yields from aboveground biomass. Therefore, it is not necessary to retain sexual reproduction in Miscanthus cultivars for bioenergy production. Complete sterility or multiple mechanisms of functional sterility, in addition to careful monitoring, could dramatically reduce the likelihood of escape or hybridization with extant wild M. sinensis populations (see Table 2).

Table 2.   Hierarchy of invasive-related traits which could be targeted in a Miscanthus breeding and management program to minimize invasive potential, along with the commercial implications Thumbnail image of

The most effective method of containment is complete (male and female) sterility (Li et al., 2004; National Research Council, 2004; Gressel & Al-Ahmad, 2005; Anderson et al., 2006). This would prevent seed set in biomass production fields as well as potential gene flow into wild populations. Sterility can be achieved in breeding programs by inducing triploidy (National Research Council, 2004; Anderson et al., 2006), a trait that has been linked to low potential invasiveness in M.×giganteus (Barney & DiTomaso, 2008). In the case of M.×giganteus, triploidy has occurred in Japan as a result of natural hybridization between diploid M. sinensis and tetraploid M. sacchariflorus (Greef & Deuter, 1993). A Japanese accession of this hybrid, collected in 1935, is the variety in commercial use for biomass production in Europe (Clifton-Brown et al., 2001). While this accession is highly productive, its sterility prevents improvement through traditional breeding programs. In addition, vegetative propagation requires higher planting costs. In order to circumvent these issues, seed production fields planted with M. sinensis and M. sacchariflorus could generate F1 seed of triploid M. ×giganteus which could be planted by farmers for biomass production without a risk of further seed production. Seed production fields should be located strategically to avoid planting fertile varieties where invasion risk is known to be high (e.g. Appalachia). In addition, a buffer zone of female-sterile Miscanthus species could be planted around seed production fields to ‘catch’ wayward pollen (Gressel & Al-Ahmad, 2005). This would be analogous to the tetraploid by diploid cross used in the commercial seed production system for seedless watermelon (Compton et al., 1996). In contrast to watermelon, however, this is an interspecific cross and thus low rates of seed set in F1 progeny may be anticipated (Sleper & Poehlman, 2006).

Sterility can also be achieved through manipulating expression of plant hormones or cytotoxin genes in reproductive tissues (Li et al., 2004). In addition, development of variety genetic use restriction technology (V-GURT or terminator technology) (National Research Council, 2004) in Miscanthus could allow growers to sow fields with seed, but would result in inviable seeds on adult plants. However, there are many technical, commercial, and regulatory hurdles to overcome to apply such transgenic solutions to Miscanthus.

Functional sterility is an additional measure to control the potential for dispersal and gene flow. Choosing late-flowering genotypes for breeding programs is one potential solution. With the native distribution of M. sinensis reaching from southern China up into Russia (Stewart et al., 2009), it should be expected that there is large variation in flowering time within different genotypes. Many Miscanthus genotypes in a breeding program in Wales (Aberystwyth, UK; 52°N) flowered very late in the season or not at all (Clifton-Brown et al., 2008), allowing little to no time for mature seeds to develop before the first killing frost. To obtain seed from most varieties in the Mendel/Tinplant Miscanthus breeding program in Magdeburg, Germany (52°N), plants typically need to be transferred to a greenhouse from the field (Deuter, 2000). Given the sensitivity of flowering of M. sinensis to the accumulation of heat units, the late flowering observed in Northern Europe cannot be assumed to apply to the United States. However, a genotype derived from M. sinensis and grown from seed has been identified which flowers too late to produce seed even as far south as Auburn, AL (33°N) (Fig. 4). This demonstrates functional sterility in Miscanthus, where localized seed production fields in the long-season subtropics could generate seeds that could be planted by farmers in temperate biomass production regions where the season is too short for seed production. The combination of genotype by environment interactions for both flowering time and biomass yield require multiyear testing of the integrity of this system in target biomass production markets to confirm safety ahead of release. Importantly, climate change may destabilize this system in future generations. Therefore, climate projections should be considered in testing this system, and additional climate-neutral safeguards should be taken. Other traits can be manipulated to minimize seed-related invasiveness in Miscanthus (Table 2). For example, minimizing pollen viability/longevity, elimination of seed shattering and dormancy are common domestication traits limiting spread and establishment outside of the cropping area (Li et al., 2004; Anderson et al., 2006). In addition, seed dispersal distance could potentially be decreased if glabrous or ‘flightless’ seeds were produced.

Figure 4.

 Functional sterility demonstrated in a genotype derived from Miscanthus sinensis at Mendel's Auburn, AL facility (33°N), through identification of a non-flowering and high-biomass breeding line that was grown from seed (foreground plot) in comparison to flowering genotypes (background). Non-flowering M. sinensis derived germplasm provided by Dr. Jorge da Silva, Texas AgriLife Research. Photographed 5 November 2009 by Gregory Van Dubay, Jr.

Controlling for seed sterility or encoding functional sterility; however, may not be sufficient to ensure containment. Another bioenergy candidate, giant reed (A. donax), is highly invasive in coastal riparian systems despite its sterility (Bell, 1997; Dudley, 2000). The success of this species has been correlated with vegetative reproduction from rhizome and stem fragments (Boose & Holt, 1999; Decruyenaere & Holt, 2001; Wijte et al., 2005; Boland, 2006; Quinn & Holt, 2008), which is a trait thought to increase the likelihood of invasiveness across a large number of plant taxa (Baker, 1974; Rejmanek, 2000; Kolar & Lodge, 2001; Lloret et al., 2005). While there is not much published information on vegetative spread in M. sinensis, it may be less aggressive than M.×giganteus in these regards, due to its caespitose habit and shorter rhizomes (Greef & Deuter, 1993). Nonetheless, tracking traits such as rhizome length, basal radial growth, and stem axillary bud viability within breeding programs appears appropriate.

Finally, not all genotypes of M. sinensis are expected to have the same invasive potential (Meyer & Tchida, 1999; Fig. 3). Analysis of the genetic, morphological, and physiological properties of the escaped genotypes, in comparison with the existing ornamental and developing biomass varieties is of critical importance to enable the avoidance of invasive characteristics in commercial varieties. However, as has been observed in sorghum, if weedy relatives exist in the cropping region, outcrossing of noninvasive domesticated cultivars can significantly contribute to their invasive risk (Andersson & de Vicente, 2010).

Minimizing invasion risk through management practices

In addition to the efforts of breeders, there are management strategies which can further minimize invasive risk. The localization of most known US M. sinensis escapes to the Appalachian and Eastern seaboard regions (Figs 1 and 2), despite country-wide use as an ornamental, suggests that climate and geography play a role in invasion risk. Additional studies are required to uncover the relative roles of genetics, climate, and land management in M. sinensis invasiveness, but this raises the possibility that there are specific genotypes and regions with low risk of invasion.

It is also important for growers to be aware of the potential for vegetative fragments to establish away from cultivation. Rhizome and stem fragments should be carefully contained during transport to and from production fields, contained on field margins (e.g. by tillage), and by maximizing distances between biomass production fields and suspected natural transport vectors like rivers and streams. As Miscanthus production scales up in a new region or with a new variety, stewardship programs, which include active monitoring protocols to identify unwanted volunteers and development of eradication methods to eliminate escapes, may be warranted.

US regulatory issues

In the United States, government restrictions on growing known invasive and noxious weeds are in effect. Noxious weeds are defined as those which invade cropland with a significant economic impact, and therefore, prohibited noxious weeds are typically identified by the agriculture departments at both the state and federal levels (National Plant Board: Laws and Regulations, 2010). While noxious weeds are commonly defined at the species level, weedy genotypes of crop species can also be included. For example, in Indiana (National Plant Board: Laws and Regulations, 2010) all genotypes of the species S. halepense (Johnsongrass) and S. almum (Columbus grass) are defined as noxious weeds, yet within the species S. bicolor, one genotype (shattercane) is considered noxious, but another (grain/forage sorghum) is a commercially important crop planted on 2.7 million hectares (as of 2009–2010) in the United States (USDA ERS, 2010). No Miscanthus species are included among the 105 US federal noxious weeds (USDA APHIS, 2010). M. sinensis is not banned in any US state, although it is listed as potentially invasive on the noxious species list in Connecticut (CT Invasive Plants Council, 2009). Regulatory restrictions on sale and production exist for M. floridulus in Hawaii and M. sacchariflorus in Massachusetts. In addition, Florida requires a permit for moving any relative of sugarcane, including Miscanthus (National Plant Board: Laws and Regulations, 2010), but these regulations are directed at minimizing pests and diseases of sugar production, not invasive risks.

In contrast to legally defined noxious weeds invading agricultural areas, species that invade nonagricultural areas are not regulated at the federal or state level in a consistent manner. US Executive Order 13112 (1999) directs federal agencies to prevent the introduction of invasive species. Lists of species that have been identified as potentially invasive in different regions can be found, but these are typically not government sanctioned. State or regional exotic or invasive plant councils maintain lists of important invaders in natural areas, but unlike noxious weeds lists, these carry no legislative mandate for control. However, these lists are often much more relevant to ecologists and managers of natural landscapes than noxious weeds lists based on agricultural systems (Quinn et al., 2008). A list of such databases can be found at ‘About the weeds of the US’ (USDA, 2010). M.×giganteus is not found in any of these lists, but M. sinensis was included on the list maintained by the Southeast Exotic Pest Plant Council (SE-EPPC, 2010) and others.

Conclusions

After a century of widely distributed ornamental use, the behavior of M. sinensis in its introduced range is currently under-studied, but existing information supports the conclusion that although it is currently not in the worst threat category, at least some genotypes are invasive in some locations. This situation could change if new high-biomass varieties of M. sinensis become a significant component of large tracts of dedicated bioenergy crops. Because the small amount of information available points towards invasive potential, those interested in production of M. sinensis should choose varieties with a combination of traits that minimize invasive potential in the region of interest. Breeding programs should select for reduced invasive potential, including sterility or functional sterility. In addition, growers must be made aware of the potential for escape via rhizome fragmentation and any breakdown of sterility, and should take practical precautions to prevent it. Miscanthus has shown the potential to mitigate carbon emissions and provide a clean and renewable source of fuel (Clifton-Brown et al., 2007; Davis et al., 2010). If we proceed carefully, it should be possible to take advantage of these potential benefits.

Acknowledgements

We wish to thank Parker Andes, Jonathan Horton, Marilyn Ortt, David Taylor, Andrew Strassman, and Thom Almendinger for helping to coordinate M. sinensis data collection sites in eastern USA natural areas. We thank Steve Baluch and Gregory Van Dubay, Jr for providing photographs and Neal Gutterson for suggestions improving drafts of this manuscript. We are grateful to Drs Jorge da Silva and Mike Gould, Texas AgriLife Research, for providing seed. We acknowledge the Energy Biosciences Institute and Mendel Biotechnology Inc. for funding this research.

Ancillary