Growth and fecundity of fertile Miscanthus × giganteus (“PowerCane”) compared to feral and ornamental Miscanthus sinensis in a common garden experiment: Implications for invasion

Abstract Perennial grasses are promising candidates for bioenergy crops, but species that can escape cultivation and establish self‐sustaining naturalized populations (feral) may have the potential to become invasive. Fertile Miscanthus × giganteus, known as “PowerCane,” is a new potential biofuel crop. Its parent species are ornamental, non‐native Miscanthus species that establish feral populations and are sometimes invasive in the USA. As a first step toward assessing the potential for “PowerCane” to become invasive, we documented its growth and fecundity relative to one of its parent species (Miscanthus sinensis) in competition with native and invasive grasses in common garden experiments located in Columbus, Ohio and Ames, Iowa, within the targeted range of biofuel cultivation. We conducted a 2‐year experiment to compare growth and reproduction among three Miscanthus biotypes—”PowerCane,” ornamental M. sinensis, and feral M. sinensis—at two locations. Single Miscanthus plants were subjected to competition with a native grass (Panicum virgatum), a weedy grass (Bromus inermis), or no competition. Response variables were aboveground biomass, number of shoots, basal area, and seed set. In Iowa, all Miscanthus plants died after the first winter, which was unusually cold, so no further results are reported from the Iowa site. In Ohio, we found significant differences among biotypes in growth and fecundity, as well as significant effects of competition. Interactions between these treatments were not significant. “PowerCane” performed as well or better than ornamental or feral M. sinensis in vegetative traits, but had much lower seed production, perhaps due to pollen limitation. In general, ornamental M. sinensis performed somewhat better than feral M. sinensis. Our findings suggest that feral populations of “PowerCane” could become established adjacent to biofuel production areas. Fertile Miscanthus × giganteus should be studied further to assess its potential to spread via seed production in large, sexually compatible populations.


| INTRODUCTION
Perennial, non-food plants that are bred to achieve higher growth rates with minimal chemical inputs are prime candidates for biofuel development (Somerville, Youngs, Taylor, Davis, & Long, 2010;Tilman, Hill, & Lehman, 2006). These feedstock candidates may also possess stress-tolerant traits such as high nutrient-and water-use efficiencies, allowing them to grow on marginal land (Quinn, Allen, & Stewart, 2010;Smith & Barney, 2014). An ongoing ecological concern, however, is that these traits may promote weediness of biofuel cultivars, causing unintended disturbance within neighboring natural areas (Clark et al., 2014;Owens et al., 2013;Somerville et al., 2010). Presumed economic and ecological benefits of new bioenergy crops must be balanced against environmental risk (Raghu, Spencer, Davis, & Wiedenmann, 2011;Raghu et al., 2006). Research on possible risks is ongoing (reviewed in Barney, 2014) with the aim of minimizing unintended consequences of biofuel development such as promoting species' invasions.
Grass species in the Asian genus Miscanthus present both opportunities and challenges for biofuel cultivation. Due to high rates of biomass accumulation, cold tolerance, and stable performance across broad climate gradients, Miscanthus sinensis, Miscanthus sacchariflorus, and M. × giganteus (a hybrid between tetraploid M. sacchariflorus and diploid M. sinensis) are promising candidates for extensive biofuel production in North America, Europe, and China (Arundale et al., 2015;Friessen, Peixoto, Busch, Johnson, & Sage, 2014;Heaton, Clifton-Brown, Voigt, Jones, & Long, 2004). The hybrid M. × giganteus can be sterile (triploid) or fertile (e.g., tetraploid) (Sacks, Jakob, & Gutterson, 2013;Sacks, Juvik, Lin, Stewart, & Yamada, 2013). A sterile triploid hybrid of M. × giganteus has been tested as a biofuel in Europe and the USA and produces greater amounts of annual biomass compared to its parent species and other potential perennial grass biofuel candidates (Heaton, Dohleman, & Long, 2008). In this study, we focus on seed-producing M. × giganteus, described below.
Quantitative ecological studies are needed to investigate the potential of biofuel feedstock biotypes such as fertile M. × giganteus to escape cultivation and establish invasive populations (Barney, 2014). By identifying the conditions that favor germination, survival and growth, these studies complement screening assessments that are designed to prevent or mitigate unintended invasions of candidate biofuel species (Flory, Lorentz, Gordon, & Sollenberger, 2012).
Mechanisms for establishment of feral populations of Miscanthus are species dependent, with M. sacchariflorus spreading largely from rhizomes rather than seeds (Bonin et al., 2014;Hager, Rupert, et al., 2015;Mutegi et al., 2016). In contrast, seed dispersal is the primary mechanism for population growth in M. sinensis (e.g., Quinn et al., 2010Quinn et al., , 2011. Triploid M. × giganteus is sterile and therefore less likely to establish feral populations than other Miscanthus biotypes; this cultivar is propagated vegetatively, largely from rhizomes (Heaton et al., 2010). The fertile, seed-producing tetraploid, "PowerCane," has been developed to reduce growers' propagation costs associated with planting rhizomes. Smith and Barney (2014) reported high seedling mortality and low emergence for "PowerCane," which could suggest a low risk of escape from cultivation. However, we conducted seed addition experiments in Iowa and Ohio and found that "PowerCane" had higher establishment and produced more biomass per plot than ornamental or feral M. sinensis at both locations (Bonin et al., 2017).
Considering this ability of "PowerCane" to escape cultivation, it is important to assess its performance relative to neighboring vegetation.
Perennial species that are strong competitors may possess a stronger ability to persist and spread compared to those that show reduced growth in the presence of competitors. To further assess the potential for "PowerCane" to persist and become invasive outside of cultivation, we used common garden experiments to examine responses of "PowerCane" to three competition treatments and compared these responses to those of ornamental and feral M. sinensis (below, we refer to these three taxa as biotypes). Experimental plots were located at two sites within the targeted range of cultivation for Miscanthus: in central Ohio and central Iowa. Response variables of the three biotypes were biomass, basal area, number of shoots, and number of seeds per plant, measured 2 years after planting. The first three traits were used to characterize vegetative growth responses, and seed number was used to assess reproductive output. We used contrast analysis to evaluate if the relative performance of "PowerCane" (compared to ornamental and feral M. sinensis) was altered by competition treatments, and to assess the presence of a statistical interaction between biotype and competition.

| Study sites
The two study sites were located at The Ohio State University-
F I G U R E 1 Weather data for Boone County (Iowa) and Franklin County (Ohio) for the (a) monthly low temperatures, and (b) monthly total precipitation during the study period

| Competitor species
Panicum virgatum "Cave-in-Rock" (switchgrass) and Bromus inermis (bromegrass) were the competitor species in this experiment. Panicum virgatum, a warm-season C 4 perennial grass native to North America, is typically found in prairie grasslands. Similar to M. sinensis, P. virgatum is being considered for biofuel feedstock. In contrast, B. inermis is a cool season C 3 perennial grass native to Hungary and Russia. Bromus inermis, first introduced to North America in the late 1800s as a forage plant, is considered invasive due to its ability to rapidly divide and spread through rhizomes and form sod patches, reaching heights of >1 m. This species establishes in agricultural fields, forests, pastures and has caused drastic ecological alterations by establishing large populations in native prairies (Dillemuth, Rietschier, & Cronin, 2009).
These two species have overlapping ranges through most of the continental United States, including our study sites and represent contrasting competitors with which to evaluate the success of escaped Miscanthus. Seeds for both competitor species were purchased from Millborn Seeds Inc., SD, USA.

| Experimental design
Two common garden experiments were established: one in Ohio and the other in Iowa. Each used a randomized block, factorial design to reduce bias due to underlying habitat heterogeneity. Treatment variables were biotype and competition. The focal Miscanthus biotypes were planted in three competition treatments: no competition, competition with P. virgatum, or competition with B. inermis, with 15 replicates per biotype in each level of competition. Initially, we initially considered each feral population of M. sinensis as a separate biotype, with 15 replicates for each population. Later, these were grouped together for data analyses because differences among the feral populations were not significant. At each site, the experimental area was divided into 15 rows, with row representing a single block, and one plot per row was randomly assigned to one of the 15 different treatment x biotype combinations resulting in a total of 225 plots.
An individual plot measured 1.5 m × 1.5 m, with a 1.5-m buffer zone. Each plot had one focal Miscanthus plant at its center.

| Planting methods
Miscanthus seeds were cleaned in a 5% bleach solution for 5 min to kill fungal pathogens, and were subsequently rinsed twice using distilled water. In Ohio, seeds were germinated in Petri dishes with 1% agar solution and stored inside an incubation chamber at 35°C. In Iowa, seeds were started in germination boxes. Once radicles protruded through their seed coat, the seedlings were transplanted into either 2.54-cm peat pots with Fafard ® 2 Mix (Ohio) or 2.54-cm plastic pots with

| Data collection
At the end of the second growing season, we recorded total dry biomass, basal area, total number of shoots (reproductive and T A B L E 1 Description of Miscanthus biotypes used in this experiment, their classification, origin, and GPS coordinates nonreproductive), and seed production (estimated based on three panicles per plant). We measured the basal diameter for every focal plant and used it to calculate basal area (Area = ¼πd 2 ). The total number of reproductive and nonreproductive shoots were counted and combined to provide the total number of shoots. A subsample (½, ¼, or ⅛, with a larger fraction for smaller plants) of the fresh biomass was collected and weighed for each plot to limit the amount of time spent harvesting. To estimate dry weights, ten plots of each focal biotype were randomly subsampled for fresh biomass (~200 g) and dried at 37.8°C, until the samples reached a constant weight (~2 weeks). The dry weight from each subsample was divided by its original weight to get a fresh to dry weight ratio. The average fresh to dry weight ratio for each focal biotype was used as a conversion factor to convert fresh weight to dry weight, and finally scaled up depending on the fraction that was sampled to get the total dry weight per plot. To estimate seed production, a subsample of three panicles were collected and processed, using rubber blocks and blower techniques to extract seeds. The seeds were then weighed in parcels of three subsamples of 50 seeds. This weight was used to extrapolate the total number of seeds from the three collected panicles. The number of seeds per plant was calculated as the number of flowering panicles x the mean number of seeds per panicle.

| Data analysis
A general linear mixed-effect model was used to measure the effects of biotype, competition, and their interaction on biomass, basal area, number of shoots, and estimated seed production. Block was included as a random factor. With the exception of number of shoots, variables were log-transformed to meet homoscedasticity requirements.
Tukey's HSD contrasts were used to determine significant differences within treatments. All analyses were completed using JMP (v12, SAS Institute Inc., Cary, NC, 1989-2007). As noted above, because there were no significant differences in the responses among feral populations, these were combined into a single category.

| RESULTS
During the first winter of the experiment, heavy mortality occurred in Iowa, most likely due to colder and drier weather conditions (National Climatic Data Center Oct-Dec 2013, Figure 1) For all dependent variables, the effects of competition and biotype were significant and there were no significant interactions between treatments (Table 2). Thus, the effects of competition were similar within biotypes (Figure 3). Competition from P. virgatum vs. competition from B. inermis generally the growth of Miscanthus; although not significant, the intensity of this inhibition differed among the biotypes ( Figure 4). Across competition treatments, "PowerCane" had greater biomass and basal area than ornamental or feral Miscanthus, but its seed production was an order of magnitude lower relative to the other biotypes ( Figure 4). Results are summarized below.
F I G U R E 3 The performance of each Miscanthus biotype at each competition treatment (mean ± standard error).
The legend for all symbols is located in the upper right corner of the Basal area figure (a). All biotypes performed better in the absence of competition (closed circles). Although there was no significant interaction between treatments, Jelitto showed more variable responses to competitors than feral plants or "PowerCane." In contrast, seed production for feral plants was more strongly suppressed by competition than for Jelitto or "PowerCane" . For all plots, means ± standard error are depicted. Lowercase letters represent significant differences using Tukey's HSD with P. virgatum produced fewer seeds than those without competition, but not for those in competition with B. inermis ( Figure 6, Tukey's HSD α = 0.05). Among the biotypes, feral and M. sinensis "Jelitto" plants produced significantly more seeds than "PowerCane" (Figure 5, Tukey's HSD α = 0.05).

| DISCUSSION
All Miscanthus biotypes suffered a reduction in biomass, basal area, total number of shoots, and seed production in response to competition, as expected. Our results show that competition did not modify the effects of biotype; therefore, our discussion focuses on the overall differences among biotypes. A major conclusion from our study is that "PowerCane" generally performed as well or better than ornamental or feral biotypes in terms of vegetative growth, but its seed production was an order of magnitude lower than the other biotypes. We discuss this low seed production below. This result suggests that if "PowerCane" establishes feral populations, they should persist at least as well as feral M. sinensis.
The relative invasiveness of species introduced for agriculture or horticulture has been associated with factors such as high productivity, disease resistance, flowering phenology, and seed production (Knight, Havens, & Vitt, 2011;Simberloff, 2008). Although "PowerCane" did not produce as many seeds as the other Miscanthus biotypes, Knight et al. (2011) concluded that large differences in fecundity may have a relatively small effect on the population growth rates of long-lived plants that have the potential to be invasive because population growth models of invasive plants, in particular perennial species, showed that population growth rate responded little to reductions in fecundity. The superior growth and potentially large planting area could contribute to an invasion risk of "PowerCane," in particular if pollen limitation (reviewed in Knight et al., 2005) explains the low seed production in our experiment. "PowerCane" was intended to be grown in managed settings, with annual harvest of inflorescences prior to seed maturation, but the volatile nature of bioenergy markets has made commercial fields a potential liability if abandoned.
Regarding competition, our findings are consistent with those of Barney, Mann, Kyser, and DiTomaso (2012), who documented better performance of M. × giganteus in noncompetitive lowland sites compared to competitive upland sites in California. Although we did not examine the competitive effect (sensu Goldberg & Fleetwood, 1987) of Miscanthus, it has been reported that naturalized M. sinensis suppressed P. virgatum in a glasshouse setting, suggesting that it may be a stronger competitor than P. virgatum in the field (Meyer et al., 2010).
There was greater variance in the response of "PowerCane" for basal area and biomass compared to the M. sinensis biotypes (Figure 3), but overall, the absence of an interaction between biotype and competition showed that relative differences among the biotypes diminished in the presence of competition. Further, we found that the competitor effects were the same across all biotypes independent of the competitor. Taken together, these results suggest that the greater growth of "PowerCane" may not translate to stronger invasive ability compared to its parent species, M. sinensis, at least when feral populations are established in relatively undisturbed, vegetated areas.
Low seed set of "PowerCane" warrants further examination because (1) as the only biotype with a 4x ploidy, "PowerCane" seed set may have been pollen limited, (2) life-history trade-offs may contribute to a negative correlation between allocation to biomass and reproduction (Stearns, 1992;Weiner, 2004), and (3) being derived from a half-sib cross (Sacks, Juvik, et al., 2013) "PowerCane" may suffer from inbreeding depression. If pollen limitation explains low seed set, then extensive plantings of "PowerCane" would be expected to have greater levels of seeds set per plant than observed in our study. On the other hand, because resource limitations to reproduction can respond to local conditions (Obeso, 2004;Pulido et al., 2014), examination of trade-offs between allocation to growth or reproduction across a broad geographic range is necessary to understand potential propagule pressure of "PowerCane." Low seed production alone does not eliminate potential invasion of "PowerCane" because Smith and Barney (2014) found that sites with available bare ground and low resident plant competition were invaded by M. × giganteus. Our parallel seed addition study (Bonin et al., 2017) also suggests that low seed set may not be as limiting to "PowerCane" as it might be for less vigorous Miscanthus varieties. Finally, inbreeding depression across generations of "PowerCane" may reduce seed viability or the competitive ability of resulting plants. Because inbreeding depression in outcrossing plants varies widely, empirical studies are needed to assess such consequences for "PowerCane." Our separate seed addition study further indicates that "PowerCane" established from seed produces large, competitive, flowering culms after 2 years (Bonin et al., 2016). Even if seed viability is low, these plants may become invasive asexually as M. sacchariflorus, a congener with low seed set (Mutegi et al., 2016) has already done in the United States (Bonin et al., 2014). The general finding is that once established, "PowerCane" plants persist.
Our ability to examine geographic variation in competitive responses was not possible due to major mortality after the first winter at the Iowa site, which experienced colder temperatures and lower precipitation prior to the second growing season than in Ohio ( Figure 1). This result has relevance because dispersal ability and cold tolerance are important controls for initial establishment of M. sinensis (e.g., Quinn et al., 2010Quinn et al., , 2011 emphasizing the importance of the first year of establishment for long-term persistence. Our experiment contributes to the ongoing risk assessment of Miscanthus cultivation. International recognition of human-induced climate change has stimulated an enormous amount of research and development toward producing plant-based renewable energy to reduce reliance on fossil fuels and emissions of glasshouse gases (e.g., Heaton et al., 2008;Powlson, Riche, & Shield, 2005). This multifaceted research includes evaluation of potential plant species in terms of their economic costs and benefits affecting land use (Heaton et al., 2008;Somerville et al., 2010), yield (Jeżowski, Głowacka, & Kaczmarek, 2011;Kim, Kim, Jeong, Jang, & Chung, 2012;Powlson et al., 2005), and ecological impact (Field, Campbell, & Lobell, 2008;Mack, 2008;Raghu et al., 2006Raghu et al., , 2011Wiens, Fargione, & Hill, 2011). Our findings suggest that development of seeded varieties of M. × giganteus may lead to feral populations with invasive potential, especially if larger populations produce more seeds than observed in our small-scale field experiment. Lack of seed production is the critical trait that allows sterile M. × giganteus to be white-listed (Quinn, Gordon, Glaser, Lieurance, & Flory, 2015). We have shown that with the exception of seed production, seeded "PowerCane" produces larger plants and shows comparable performance under competition compared to feral accessions. Taken together with our seed addition experiment (Bonin et al., 2016), our results support the potential of seeded Miscanthus to establish feral populations. Further research on limitations to seed production and spread of feral populations will be useful for assessing the invasive potential of "PowerCane."