SEARCH

SEARCH BY CITATION

Keywords:

  • bioenergy crop;
  • integrodifference equation models;
  • large statured invasive grasses;
  • matrix model;
  • risk assessment;
  • spatial population dynamics

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  1. Many species prioritized for bioenergy crop development possess traits associated with invasiveness, necessitating a priori efforts by ecologists to identify species or cultivars with minimal invasive potential. The grass Miscanthus × giganteus Greef et Deu ex Hodkinson et Renvoize is a candidate for biomass production in the northern US maize belt, with both sterile and fertile varieties commercially available in the near future. Prior to widespread deployment, the invasive potential of both varieties must be quantified.
  2. Using M. × giganteus demographic and seed dispersal data, we parameterized an age-/stage-structured integrodifference equation model to estimate potential spread rates of sterile and fertile M. × giganteus. We identified thresholds for reproductive parameters, above which population numbers and space occupied are likely to increase. Our simulations considered lateral spread of M. × giganteus but not dispersal of rhizome fragments.
  3. When clonal recruitment is absent, population growth rate for sterile M. × giganteus is projected to be slightly <1 (λ = 0·979), indicating gradual population decline over the long term. A sterile M. × giganteus population may increase in numbers and space under certain conditions: annually rhizome sprouting must be >20% and rhizome production must be ≥1 per plant. The relatively slow spread rates (0–0·09 m year−1) estimated for sterile M. × giganteus would not apply in scenarios where rhizomes were dispersed long distance. For a fertile M. × giganteus genotype, even low rates of seed viability and survival, seedling survival and seed germination support rapidly expanding populations.
  4. Synthesis and applications. Spatial demographic models offer a powerful tool for quantifying risk of invasive spread by bioenergy crops. Our results suggest that sterile and fertile cultivars of M. × giganteus have markedly different invasive potential and therefore should be considered separately in management and policy decisions. Feral populations of sterile M. × giganteus would need to experience frequent and severe disturbance to pose a significant invasion risk, indicating that they should be grown well away from riparian areas prone to streambank scouring. In contrast, cultivars of M. × giganteus bearing fertile seed may be very difficult, if not impossible, to contain.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Government policies in the United States and Europe encourage an increasing proportion of energy to be derived from biofuels (Robertson et al. 2008). Dedicated bioenergy crops are hence being developed and deployed on a large scale for fuel production. In the United States, the Biomass Crop Assistance Program (BCAP) instituted by the United States Department of Agriculture Farm Service Agency (USDA-FSA) has initiated projects in bioenergy production focussing on extensive herbaceous plantings (50 000 acres per state by 2014) in Ohio, Pennsylvania, Arkansas, and Missouri (USDA-FSA 2011). There is pressure for bioenergy crops to yield large quantities of biomass while avoiding pitfalls associated with grain-based biofuels. By cultivating bioenergy crops on marginal lands unfit for food crops, it may be possible to avoid the ‘food vs. fuel debate’ (Tilman et al. 2009). The significant environmental and economic costs associated with irrigation and pesticide application may be mitigated by bioenergy crops that are drought tolerant, competitive with weeds and pest resistant (Heaton, Voigt & Long 2004). Such vigour is not without its costs, however: a downside of these traits is the potential for these plants to thrive outside of cultivation (Raghu et al. 2006).

Sheltering plants from stresses through cultivation can reduce environmental stochasticity, allowing for repeated sampling of recruitment sites and improving the likelihood of population establishment outside of cultivation (Mack 2000). The cultivation of bioenergy crops poses several risks, in terms of facilitating invasions. Propagule pressure is correlated with invasion success (Lockwood, Cassey & Blackburn 2005), and considering the large spatial scale of planned bioenergy plantings (USDA-FSA 2011), it is likely that propagule pressure may be high and sustained. Traits typifying an ideal bioenergy crop (e.g. fast growth, highly efficient nutrient/water/light use) overlap with those of weedy and invasive species (Heaton, Voigt & Long 2004; Raghu et al. 2006). Additionally, many bioenergy candidates have a history of invasion (Buddenhagen, Chimera & Clifford 2009), which is troublesome considering that prior invasion is one of the most reliable predictors of future invasion success (Hayes & Barry 2008).

Many leading bioenergy candidates are perennial grasses (Lewandowski et al. 2003), a group of plants also recognized for their world-wide success as invaders (D'Antonio & Vitousek 1992). Grasses may regenerate by both wind-dispersed seeds and rhizome fragments (Briske & Derner 1998). In the Midwest United States, Miscanthus × giganteus has been identified as a promising candidate bioenergy crop, yielding on average 29·6 Mg ha−1 (Heaton, Dohleman & Long 2008), and a focal species in the USDA-FSA BCAP program (USDA-FSA 2011). Miscanthus × giganteus is a naturally occurring triploid hybrid between the diploid Miscanthus sinensis and the tetraploid Miscanthus sacchariflorus (Stewart et al. 2009). Both parent species of M. × giganteus produce fertile seeds and after being introduced to the United States as ornamentals have escaped cultivation to form naturalized populations (Quinn, Allen & Stewart 2010). In Australia, naturalized populations of M. sinensis have been found in the Blue Mountains (Jorgensen 2011). In Europe, M. sinensis seedlings have been observed outside of cultivation (Pysek, Sadlo & Mandak 2002; M. Deuter, personal communication). Clonal production systems for sterile M. × giganteus have been designed where ‘mother fields’ are a source of clonal material, and biomass production fields are planted with rhizome fragments or rooted plantlets using specialized machinery (Jorgensen 2011). In the agronomic literature, M. × giganteus is typically indicated to only produce sterile spikelets due to its triploid genome; however, an anecdotal account of fertile seed production has been reported (Nielsen 1987). A greenhouse study conducted in Illinois screening over 7 million spikelets did not yield a seedling, supporting the assumption of sexual sterility (Matlaga, Schutte & Davis 2012). The Illinois clone variety of M. × giganteus has been the focus of the majority of production trials and the USDA-FSA BCAP program in Ohio, Pennsylvania, Arkansas, and Missouri (USDA-FSA 2011) and represents c. 90% of the commercial market (E. Heaton, personal communication). The Illinois clone variety of M. × giganteus has been found to have lateral vegetative expansion rates of 0·15 m year−1 in unmanaged arable lands in Illinois (Matlaga, Schutte & Davis 2012).

Concern about the invasive potential of this species is supported by invasive behaviour of species that combine strictly clonal reproduction with aggressive spread outside of their native range. For example, giant reed Arundo donax is a sexually sterile (Mariani et al. 2010) bioenergy candidate, which is highly invasive in California where it spreads via recruitment from rhizome mats and fragments dispersed through riparian systems (Khudamrongsawat, Tayyar & Holt 2004). Hence, an a priori assumption that sterile M. × giganteus has low invasive potential is not supported by current scientific data.

Drawbacks of a clonal production system, namely the high costs of vegetative planting and limited opportunity to improve yields through traditional breeding techniques (Quinn, Allen & Stewart 2010), have spurred efforts to engineer fertile varieties of M. × giganteus (Yu et al. 2009). A report in the agricultural media indicates that commercialization of a fertile-seeded M. × giganteus variety is imminent (Ross 2011). Several production scenarios have been proposed to prevent or limit escape of fertile M. × giganteus seed. Seed production fields, strategically placed in areas with low invasion risk, planted with M. sinensis Andress. and M. sacchariflorus (Maxim.) Hack. could generate predominantly triploid F1 M. × giganteus seed for planting in biomass production fields (Quinn, Allen & Stewart 2010). Additionally, functional sterility could be used where late flowering genotypes are planted in northern latitudes allowing little or no time for seed maturation prior to the onset of winter (Quinn, Allen & Stewart 2010). It is crucial to understand what threshold of seed viability must be avoided to prevent escape and spread of M. × giganteus.

Conducting comprehensive risk analysis of bioenergy crop invasion is a major challenge. The negative impact of an invasion is dependent on several factors, including population dynamics of the invader, expressed as rates of population growth and spread, as well as impacts on native communities and the cost of eradication (Parker et al. 1999; Neubert & Parker 2004). To minimize and mitigate the potential of invasive bioenergy crop cultivars, a nested-sieve framework for cultivar evaluation has been proposed. Under this framework, candidate bioenergy crops must pass through pre-entry sieves, including trait-based screening and demographic climate matching analysis, and post-entry sieves, including estimation of population growth in potential range, prior to widespread release (Davis et al. 2010). The invasion risk posed by M. × giganteus in Florida (Gordon et al. 2011) and California (Barney & Ditomaso 2008) has been deemed acceptable using Weed Risk Assessment (WRA) systems methods, primarily based on the species sterility. However, M. × giganteus was evaluated for introduction into the North Central region of the USA by the same method and received a score of ‘evaluate further’ (S. Raghu and A. Davis, unpublished observations). Considering that establishment of large-scale M. × giganteus plantings are underway in the United States (USDA-FSA 2011), the time for pre-entry screening methods like WRAs has likely passed.

An important next step in the safe deployment of bioenergy crops is to guide risk analysis and management with demographic models parameterized with vital rates estimated empirically in the proposed production area. To date, this has been accomplished on a limited spatiotemporal scale for a single oilseed crop (Camelina sativa) in the intermountain West of the USA (Davis et al. 2011). Here, we expand upon this approach using a spatial demographic model describing the population growth and spread of the herbaceous perennial bioenergy crop M. × giganteus. To evaluate the invasive potential of both a sterile cultivar currently in production and a fertile cultivar under development, we simulate feral populations with specific sexual and clonal recruitment scenarios to understand approximately where within M. × giganteus reproductive parameter space thresholds for increasing population size (population growth rate, λ, >1) and spatial extent (wavespeed, c*, >1) are located. The comparison of model projections for fertile and sterile lines of M. × giganteus not only provides information on invasion risk for this particular bioenergy crop, but also represents an opportunity to understand the role of plant reproductive strategy in biological invasions. Finally, the quantitative approach described here may serve as a model system for evaluating bioenergy crops and aiding in the development of non-invasive ideotypes. Here, we report the results of a quantitative risk analysis study, in which empirically derived parameter estimates were used to implement a model of M. × giganteus spread and establishment. Our modelling analysis was framed by the hypothesis that variation in sexual reproduction in this species makes a larger contribution to population growth rate than does variation in clonal reproduction.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

General Model

We use Neubert & Caswell's (2000) integrodifference equation approach for spatial demographic analysis of age- or stage-structured populations to estimate population growth rate (λ) and wavespeed (c*) of M. × giganteus. The general form of the model is:

  • display math(eqn 1)

where population growth, given by the population projection matrix (Bn), is coupled with population spread over space, represented by a matrix of dispersal kernels (K). The population projection matrix is composed of survival, growth and reproduction for all life stages. The dominant eigenvalue of the population projection matrix gives the in situ population growth rate (λ) at low density (Neubert & Caswell 2000). The matrix of dispersal kernels is stage-structured as well and describes how each stage moves in space when undergoing life cycle transitions. During one time step, at each point y, the population grows and some individuals disperse from point y to location x. Each age or stage class can have a distinct dispersal kernel and stages without dispersal have a kernel in which the entire probability is concentrated at the origin. The integration of this process over all points y gives the population at x after a single time step. The model estimates spatial population dynamics at the leading edge of an invasion into an unoccupied area. Therefore, it is assumed that there is no Allee effect and that vital rates and dispersal kernels are at low densities.

A population conforming to (eqn (eqn 1)) will eventually spread into a homogenous and unoccupied area in the form of an invasion wave moving at a constant speed. To estimate the speed of the invasion wave, Neubert & Caswell (2000) combined the population projection matrix Bn with a matrix of moment generating functions by element-wise multiplication (Hadamard product) to produce a new matrix that combines demography and dispersal (p). Neubert & Caswell (2000) demonstrated that the asymptotic rate at which a population spreads over space during a time step is represented by:

  • display math(eqn 2)

where s is the waveshape parameter.

Miscanthus × giganteus Model Development

We used published estimates of M. × giganteus' vital rates (Table 1; Matlaga, Schutte & Davis 2012) and dispersal (Quinn et al. 2011) to construct an integrodifference equation model in the matlab technical computing environment (Mathworks 2011).

Table 1. Lower-level demographic parameters, their abbreviations and value(s) used in the Miscanthus × giganteus matrix projection model. Numbered subscripts indicate age-class
Lower-level parameterAbbreviationValue(s) used in model
Rhizome fragment production rp 0–5
Rhizome sprouting p 0–1
Rhizome fragment survival s r 0, 0·35, 0·7
Survival of newly sprouted rhizomes s t 0·248
Seed survival in soil s s 0·000001, 0·00001, 0·0001, 0·001
Seed germination g 0–1
Seed viability v 0–1
Seedling survival s 0 0·001,0·01, 0·1, 1·0
1st-year seed production sd 1 10 497
2nd-year seed production sd 2 85 179
3rd-year seed production sd 3 164 527
4th-year seed production sd 4 187 853
1st-year survival s 1 0·248
2nd-year survival s 2 0·916
3rd-year survival s 3 0·974
4th-year survival s 4 0·979

We constructed and parameterized the population projection matrix (Fig. 1), which is comprised of survival, growth and reproduction, primarily from vital rates of M. × giganteus empirically estimated in the Midwest United States. The demographic fate of M. × giganteus individuals is strongly dependent on their age (Matlaga, Schutte & Davis 2012), and therefore, we constructed a population model that is primarily age-structured with a 1-year time interval. Because the empirically estimated demographic rates of plants remain relatively constant after the 4th year in our model, the demographic fate of plants does not change after their 4th year. In our model, M. × giganteus has the ability to produce two propagule types: sexual (seeds) and clonal propagules (rhizome fragments). Propagules can transition to 1st year plants within the same time interval in which they are produced (a3,3–6 in Fig. 1) or transition to a ‘seed bank’ (s in Fig. 1) or ‘rhizome bank’(r in Fig. 1). Individuals can persist in the seed bank (a2,2 in Fig. 1), but the rhizome bank is transient, lasting a single time step, and therefore can be considered delayed sprouting.

image

Figure 1. Schematic life cycle of Miscanthus × giganteus and its expression in A matrix format representing annual transitions between dormant seeds (s), rhizome fragments (r), and adult plants 1–4 years of age (1–4). Transitions are displayed as aij values, representing the value in the ith row and jth column of the population projection matrix. Shaded aijs represent empirically estimated transitions and non-shaded aijs represent transitions for which data are unavailable and are the subject of simulations.

Download figure to PowerPoint

To calculate survival rates for 1–4-year-old plants, we used the fitted asymptotic exponential relationship (abecx, where = 0·979, = 0·248, = 0·894) between plant age and survival estimated by Matlaga, Schutte & Davis (2012). Survival of newly sprouted rhizome fragments is assumed to be the same as 1st-year plants (Table S1, Supporting information). Fecundity of 1–4-year-old plants was reported by Matlaga, Schutte & Davis (2012). Table S1 (Supporting information) demonstrates the calculation of matrix elements (aij) from the lower-level parameters.

Both sexual (dispersal of spikelets which contain seeds) and clonal (lateral vegetative expansion) dispersal kernels of M. × giganteus were previously estimated in field settings in Illinois. The dispersal kernel for M. × giganteus spikelets (referred to as caryopses in publication) was estimated in an experimental release and capture experiment (Quinn et al. 2011). In this experiment, a point source of 600 panicles was exposed to wind, which approximates the conditions of a small feral population. During the experiment, the majority of spikelets (73%) were captured within 20 m of the release point but a minority of spikelets travelled greater distances, including 400 m, the maximum distance sampled (Quinn et al. 2011). The probability–density function describing the distance spikelets travelled was best supported by a lognormal distribution, with μ = 2·53 and σ = 1·98. Our model utilized the raw data of the number of spikelets captured per distance. Matlaga, Schutte & Davis (2012) recorded the lateral extent of individually marked M. × giganteus during three successive years and used those data to calculate annual clonal expansion for plants, which ranged from −0·160 to 0·41 m year−1. The distribution of annual clonal expansion was best fit using a Gaussian distribution with μ = 0·158 m year−1 and σ = 0·113 m year−1 (Matlaga, Schutte & Davis 2012). Dispersal of rhizome fragments has not been investigated and was not considered in our study.

Approach

Our approach was to identify approximate thresholds within the demographic parameter space for increasing M. × giganteus population size and spatial extent. We used this approach to explore two varieties of M. × giganteus with contrasting reproductive abilities: the current sterile variety of M. × giganteus known as the ‘Illinois clone’ (from here on ‘sterile variety’) and a hypothetical variety with fertile seed production to represent the impending commercialization of a fertile M. × giganteus line (from here on ‘fertile variety’). We constructed a matrix population model for M. × giganteus that allowed us to perform simulations for the sterile and fertile varieties in which we varied reproductive parameters and calculated the effect on the population growth rate (λ) and population spread rate (c*).

Initially, we performed exploratory simulations varying reproductive parameters for the sterile (sr, rp, p Fig. 1) and fertile (s0, g, ss, v Fig. 1) models across their full range (0–1) to determine the approximate threshold for a growing (λ > 1) and spreading (c* > 0) population. We then narrowed our simulations to explore parameter space around the critical thresholds for growth and population spread. In the sterile variety model, we varied the number of rhizome fragments produced from 0 to 5 at 0·25 intervals (numerical spacing) and the proportion of rhizome fragments that sprout from 0 to 1 at 0·05 intervals, under three scenarios of rhizome fragment survival in the soil (0%, 35% and 70%). In the fertile variety model, we varied seed germination from 0 to 1 at 0·01 intervals and seed survival in the soil from 0 to 1 at 0·01 intervals, under 16 scenarios where seedling survival (0·001, 0·0001, 0·00001, 0·000001) and seed viability (0·001. 0·01. 0·1, 1·0) were varied factorially.

Simulation results were graphed, displaying zero-growth isoclines for population growth rate (λ) and population spread rate (c*). Area beneath the λ = 1 isocline indicates parameter space yielding a population which is declining in numbers and in spatial extent, conversely area above the λ = 1 isocline indicates parameter space yielding a population growing in numbers and spatial extent.

We used elasticity analysis to understand how variation in demographic parameters impacts population growth rate (λ), calculating the elasticity of λ to changes in each aij of the A matrix as

  • display math(eqn 3)

where Eij is the elasticity of λ to proportional perturbations of aij, and the partial derivative of λ with respect to aij, λ/∂aij, is referred to as the sensitivity of λ to additive changes in aij (Caswell 2001). By making use of the chain rule for differentiation, eqn (eqn 3) can be extended allowing for calculation of elasticities of λ to lower-level demographic parameters (individual life stage transitions making up each aij), such that

  • display math(eqn 4)

where the elasticity of λ to proportional perturbations in lower-level demographic parameter x is represented by Ex (Caswell 2001).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Miscanthus × giganteus Sterile Variety Model

Projections of population growth and spread from our sterile variety model indicate that in the absence of rhizome fragment production (rp = 0), M. × giganteus populations slowly decline over time (λ = 0·979, Figs 1, S1 and S2, Supporting information). When rhizome fragment production exceeds 0, even slightly, positive population growth can occur given specific combinations of rhizome fragment sprouting (p) and rhizome fragment survival (sr). However, even in scenarios that yield rapid population growth (i.e. λ = 1·5), only relatively slow population spread rates (i.e. c* = 0·09 m year−1) are achieved (Figs S1 and S2, Supporting information).

The threshold for positive population growth is lowered with increasing rhizome fragment production (rp), rhizome sprouting (p) and rhizome fragment survival (sr; Fig. 2). In scenarios where fewer than one rhizome fragment is produced per individual, but >0, proportional sprouting (p) must exceed 0·2 to yield positive population growth (Fig. 1). If several rhizome fragments are produced per individual (rp = 2–5), positive population growth can be achieved at proportional sprouting rates (p) as low as 0·10 (Fig. 2). Rapid population growth rates occur (λ = 1·2–1·5) in scenarios where rhizome fragment production (rp) exceeds 3 per plant and proportional sprouting rates (p) exceed 0·60; however, rates of population spread remain modest (c* = 0·06–0·09 m year−1) (Figs S1 and S2, Supporting information). Increasing survival of rhizome fragments in the soil alters isocline topography, as well as the maximum rates achieved for population growth and spread (Figs 1, S2 and S3, Supporting information). As rhizome fragment survival in the soil increases, the threshold for positive population growth (λ = 1) is exceeded at lower proportional sprouting rates (p) and rhizome fragment production (rp). For example, in cases where rhizome fragment production (rp) is 1 per plant, proportional sprouting rates (p) >0·4 and 0·18 are needed to exceed the λ = 1 threshold when survival in the soil is 0% and 75%, respectively. Additionally, maximum population growth and spread rates occurred at lower values of rhizome fragment production with increasing rhizome fragment survival in the soil (Fig. 1).

image

Figure 2. Simulation results showing population growth rate (λ) of sterile Miscanthus × giganteus where shaded areas indicate parameter space yielding a growing population (λ > 1) and non-shaded areas indicate parameter space yielding a declining population (λ < 1).

Download figure to PowerPoint

Elasticity analysis revealed that population growth rate is most sensitive to perturbations of survival of 4th-year plants (a6,6, Table S2, Supporting information). When rhizome fragment survival in the soil (sr) increased, population growth rate is more sensitive to perturbations of this parameter (Table S2, Supporting information).

Miscanthus × giganteus Fertile Variety Model

Results from our fertile variety model indicate that all lower-level sexual demographic parameters [e.g. seedling survival (s0), seed survival (ss), germination (g) and seed viability (v)] influence the location of the population growth threshold of λ = 1 (Figs 3, S3 and S4, Supporting information). Population spread rates, c*, are over an order of magnitude faster for the fertile variety compared with the sterile variety model (Figs S1 and S3, Supporting information). The topography of the population growth rate isocline, determined by varying seed survival (ss) and germination (g), changes depending on the context of seedling survival (s0) and seed viability (v; Fig. S3, Supporting information). At high rates of seed viability (= 1·0) and seedling survival (s0 = 0·001), population growth rate is much more sensitive to germination (g) than to seed survival (ss; Table S3, Supporting information).

image

Figure 3. Simulation results showing population growth rate (λ) of fertile Miscanthus × giganteus where shaded areas indicate parameter space yielding a growing population (λ > 1) and non-shaded areas indicate parameter space yielding a declining population (λ < 1).

Download figure to PowerPoint

Elasticity analysis reveals that population growth rate is most sensitive to perturbations of survival of 4th-year plants except at high rates of seed viability (Table S2, Supporting information).

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Our analyses of the spatial population dynamics of sterile and fertile cultivars of M. × giganteus support the hypothesis that variation in sexual reproduction of this species influences population growth to a greater extent than does variation in clonal reproduction. The threshold for positive population growth (λ = 1) was exceeded at lower demographic rates for seed/seedling establishment compared with rates associated with rhizome fragment establishment. As a matter of concern for producers, under similar rates of population growth (λ), the fertile variety model displayed markedly faster rates of population spread (c*) compared with the sterile variety. These fundamental differences in potential population dynamics between sterile and fertile varieties imply that the invasive potential of these varieties, and management guidelines for safe containment, should be addressed separately.

Analyses of the demographic behaviour of the sterile cultivar indicated that the survival and sprouting of rhizome fragments in the soil may strongly influence population dynamics of M. × giganteus. Matlaga, Schutte & Davis (2012) experimentally investigated the within-year establishment success of 1–20 g M. × giganteus rhizome fragments in a field experiment in Illinois. Overall, 9·5% of fragments established and fragment size was a significant predictor of establishment success; small and large fragments had 0% and 42% establishment probabilities, respectively (Matlaga, Schutte & Davis 2012). Considering the overall rhizome sprouting rate of 9·5% (therefore = 0·095), our model results imply that adult plants must produce several rhizome fragments per individual (rp = 2–4) to achieve a growing population under these conditions. At the upper end of rhizome sprouting rate observed in the field (= 0·42), we observed an expanding population if one rhizome fragment was produced per every two individuals (rp = 0·5). Because the rhizome system of M. × giganteus is extremely tough and requires significant force to fragment, rhizome production rates of rp = 0·5–4 are possible only under cases of severe and frequent disturbance.

Based on these model results, we identify two types of growth environments that require special attention by producers of sterile M. × giganteus cultivars: field margins and riparian areas. It is possible that, on the margins of biomass production fields, machinery could provide regular and severe disturbance resulting in the fragmentation of M. × giganteus rhizomes and creating the potential for positive population growth. For production fields abutting waterways, flooding could also drive invasive spread if flow rates are sufficient to scour and destabilize banks, leading to rhizome fragmentation or even movement of entire rhizome mats, as has been observed for Phragmites australis and A. donax (Keller 2000; Khudamrongsawat, Tayyar & Holt 2004). We advise producers to create setbacks of at least 5 m from high watermarks, plant such buffer areas to a mowed perennial sward, and to monitor these field margins annually.

Observations of λ > 1 for fertile M. × giganteus cultivars, even in model scenarios with low seed viability, were consistent with recent work showing that reductions in fecundity of 95–100% are needed to slow the population growth rate of long-lived species below λ = 1·0 (Crone et al. 2011; Knight, Havens & Vitt 2011). A growing number of reports suggest that population growth of long-lived species is sensitive to adult survival but not fecundity (Ramula et al. 2008; Knight, Havens & Vitt 2011).

These results offer concrete guidance for the development of bioenergy crop cultivars with lower invasive potential. Breeding programmes aimed at developing fertile M. × giganteus cultivars should explicitly estimate the viability of seed produced in biomass production fields as well as seedling survival, seed survival and germination success within nearby habitats. We recommend that agronomists do not exceed the demographic rates associated with the lower left corner of Fig. 3, seedling survival 0·000001 and seed viability of 0·001, when developing a non-invasive fertile M. × giganteus crop. This is not, however, a guarantee that any specific rates will prevent invasive population dynamics over the long term.

We acknowledge that results from this model, like all population models, are most valuable for the heuristic insights they provide (Crone et al. 2011), identifying specific demographic parameters, and approximate magnitudes, that if targeted could reduce the invasive potential of M. × giganteus. A specific limitation of our study is omission of rhizome fragment dispersal. Rhizome systems in large grasses can be fragmented by disturbance and dispersed potentially long distances. For example, fragmentation and dispersal of A. donax rhizomes in California riparian habitats can occur by machinery and flooding (Boland 2008). We suggest that the importance of rhizome dispersal for the spatial population dynamics be assed in three steps. A dialogue between agronomist and invasion biologist could identify the most likely and important (greatest distance) rhizome dispersal pathways of M. × giganteus (e.g. small watercourses, large streams/rivers, machinery movement between production field, machinery movement between production regions). Empirical field and laboratory studies could estimate dispersal kernel parameters in each pathway, although there is considerable difficulty in determining long-distance dispersal in waterways. Lastly, a modelling simulation approach could be used that focussed on determining the influence that varying the portion of rhizome fragments entering each dispersal pathway has on rates of population spread.

Our approach considers only the demography and population dynamics of hypothetical feral populations of M. × giganteus outside of cultivation. The strength and spatial scale of propagule pressure originating from bioenergy production was not considered in our modelling scenarios. Metapopulation modelling efforts should be conducted in the future with the goal of understanding source (bioenergy production fields) and sink (adjacent unmanaged lands) dynamics. The relatively slow spread rates we estimated for sterile M. × giganteus would not apply in scenarios where rhizomes were dispersed long distance by machinery or through watercourses.

Future efforts in the United States are needed to evaluate the invasive potential of the most common M. × giganteus varieties both in terms of empirical field estimates demographic performance across a range of environments and modelling simulations to project population dynamics over time. To our knowledge, information on the demography of M. × giganteus outside of cultivation is limited in time-scale, habitats considered, and variety of M. × giganteus. Only two published studies have evaluated the demography of M. × giganteus outside of cultivation and both considered the ‘Illinois clone’ variety (Barney et al. 2012; Matlaga, Schutte & Davis 2012). The demographic performance of the ‘Freedom clone’ variety of M. × giganteus, which constitutes a minority of the overall commercial market in the United States but is specifically targeted for the Southeast United States (B. Baldwin, personal communication), has not been considered outside of cultivation. Rhizome fragment establishment of the Illinois clone has been studied in riparian and dryland habitats of Central California over 2 years (Barney et al. 2012). The demography of adult plants and rhizome establishment has been studied in early successional habits of Illinois over 3 years (Matlaga, Schutte & Davis 2012). Future studies are needed that consider a wider range of both managed and unmanaged habitats, longer time-scales and additional varieties of M. × giganteus. Perhaps most crucial, estimates are needed of sexual reproductive parameters (e.g. seed survival, seed germination, seedling survival) for fertile M. × giganteus varieties within habitats adjacent to production fields. Additionally, studies are needed to determine whether non-reproductive demographic rates (survival, growth and lateral expansion) are identical between fertile and sterile varieties of M. × giganteus. Collaborations are needed between agronomists developing fertile varieties of bioenergy crops and ecologist tasked with evaluating invasive potential. When deciding which cultivars should receive approval for large-scale production, economic benefits of crop production must be weighed against invasion scenarios based on local empirical data.

While our study focussed on the Midwest United States, it must be recognized that efforts are underway to jumpstart bioenergy crop development and deployment on a global scale (Dornburg et al. 2010). Due to intense international interest in growing perennial grasses (i.e. Panicum virgatum, A. donax, Phalaris arundinaceae and M. × giganteus) for bioenergy (Lewandowski et al. 2003; Jorgensen 2011), a need exists to assess the risk of invasion for these species across many countries. Despite this global need, we argue that because demographic rates are inherently site and region specific (e.g. Shea et al. 2005) and play an important role in determining invasion success risk should be evaluated on a local scale. Our perspective is in contrast to the WRA approach that has been carried out to evaluate invasion risk of bioenergy crops for locations, such as U.S. state of Florida (Gordon et al. 2011). In Europe, a mixture of varieties of M. × giganteus have been considered for biomass production, the most common being the Hornum variety (Heaton et al. 2010), as well as the fertile seed producing M. sinensis (Lewandowski et al. 2003). To our knowledge no attempts have been made to evaluate the invasive potential of these varieties in Europe or elsewhere. However, observations that single M. × giganteus plants have established in Germany from garden rubbish (Brennenstuhl 2008) suggest that rhizome establishment is a viable recruitment pathway in this region. Additionally, seedlings of M. sinensis have been found outside of cultivation in Europe (see 'Introduction').

Our study considers the potential population trajectory of M. × giganteus outside of cultivation in Illinois, United States but does not provide insight into the potential impact of an expanding population. It is known that invasive plants can alter population, community and ecosystem processes through various mechanisms (Levine et al. 2003). It is not known whether traits that typify an ideal bioenergy crop [e.g. rapid growth, competitive with weeds, drought tolerant (Heaton, Voigt & Long 2004)] would predispose these species to have a significant impact in the case of escape and invasion of natural areas. However, it remains difficult to assess the risk associated with bioenergy crops without data on their potential impacts on natural ecosystems.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank the National Center for Ecological Analysis and Synthesis working group ‘New Synthesis of Demography and Dispersal’ for developing Matlab code, which was provided to us by C.C. Horvitz. This manuscript was improved by helpful comments from Nick Jordan, Sheri Huerd and Jim Eckberg.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Barney, J.N. & Ditomaso, J.M. (2008) Nonnative species and bioenergy: are we cultivating the next invader? BioScience, 58, 6470.
  • Barney, J.N., Mann, J.J., Kyser, G.B. & DiTomaso, J.M. (2012) Assessing habitat susceptibility and resistance to invasion by the bioenergy crops switchgrass and Miscanthus × giganteus in California. Biomass and Bioenergy, 40, 143154.
  • Boland, J.M. (2008) The roles of floods and bulldozers in the break-up and dispersal of Arundo donax (Giant reed). Madrono, 55, 216222.
  • Brennenstuhl, G. (2008) Zur einbu¨ rgerung von Vinca- und Miscanthus-Taxa - Beobachtungen im gebiet um Salzwedel. Mitteilungen zur floristischen Kartierung in Sachsen-Anhalt, 13, 7784.
  • Briske, D.D. & Derner, J.D. (1998) Clonal biology of caespitose grasses. Population Biology of Grasses (ed. G.P. Cheplick), pp. 106135. Cambridge University Press, Cambridge.
  • Buddenhagen, C., Chimera, C. & Clifford, P. (2009) Assessing biofuel crop invasiveness: a case study. PLoS ONE, 4, e5261.
  • Caswell, H. (2001) Matrix Population Models: Construction, Analysis and Interpretation, 2nd edn. Sinauer Associates, Sunderland, MA.
  • Crone, E.E., Menges, E.S., Ellis, M.M., Bell, T., Bierzychudek, P., Ehrlen, J., Kaye, T.N., Knight, T.M., Lesica, P. & Morris, W.F. (2011) How do plant ecologists use matrix population models? Ecology Letters, 14, 18.
  • D'Antonio, C.M. & Vitousek, P.M. (1992) Biological invasions by exotic grasses, the grass fire cycle, and global change. Annual Review of Ecology and Systematics, 23, 6387.
  • Davis, A.S., Cousens, R.D., Hill, J., Mack, R.N., Simberloff, D. & Raghu, S. (2010) Screening bioenergy feedstock crops to mitigate invasion risk. Frontiers in Ecology and the Environment, 8, 533539.
  • Davis, P.B., Menalled, F.D., Peterson, R.K. & Maxwell, B.D. (2011) Refinement of weed risk assessment for biofuels using Camalina sativa as a model species. Journal of Applied Ecology, 48, 989997.
  • Dornburg, V., van Vuuren, D., van de Ven, G., Langeveld, H., Meeusen, M., Banse, M., van Oorschot, M., Ros, J., van den Born, G.J. & Aiking, H. (2010) Bioenergy revisited: key factors in global potentials of bioenergy. Energy and Environmental Science, 3, 258267.
  • Gordon, D.R., Tancig, K.J., Onderdonk, D.A. & Gantz, C.A. (2011) Assessing the invasive potential of biofuel species proposed for Florida and the United States using the Australian weed risk assessment. Biomass and Bioenergy, 35, 7479.
  • Hayes, K.R. & Barry, S.C. (2008) Are there any consistent predictors of invasion success? Biological Invasions, 4, 483506.
  • Heaton, E.A., Dohleman, F.G. & Long, S.P. (2008) Meeting US biofuel goals with less land: the potential of Miscanthus. Global Change Biology, 14, 20002014.
  • Heaton, E., Voigt, T. & Long, S.P. (2004) A quantitative review comparing the yields of two candidate C-4 perennial biomass crops in relation to nitrogen, temperature and water. Biomass and Bioenergy, 27, 2130.
  • Heaton, E.A., Dohleman, F.G., Miguez, A.F., Juvik, J.A., Lozovaya, V., Widholm, J., Zabotina, O.A., Mcisaac, G.F., David, M.B., Voigt, T., Boersma, N.N. & Long, S.P. (2010) Miscanthus: a promising biomass crop. Advances in Botanical Research, 65, 76124.
  • Jorgensen, U. (2011) Benefits versus risks of growing biofuel crops: the case of Miscanthus. Current Opinion in Environmental Sustainability, 3, 2430.
  • Keller, B.M. (2000) Genetic variation among and within populations of Phragmites australis in the Charles River watershed. Aquatic Botany, 66, 195208.
  • Khudamrongsawat, J., Tayyar, R. & Holt, J.S. (2004) Genetic diversity of giant reed (Arundo donax) in the Santa Ana River, California. Weed Science, 52, 395405.
  • Knight, T.M., Havens, K. & Vitt, P. (2011) Will the use of less fecund cultivars reduce the invasiveness of perennial plants? BioScience, 61, 816822.
  • Levine, J.M., Vila, M., Antonio, C.M.D., Dukes, J.S., Grigulis, K. & Lavorel, S. (2003) Mechanisms underlying the impacts of exotic plant invasions. Proceedings of the Royal Society of London B: Biological Sciences, 1517, 775781.
  • Lewandowski, I., Scurlock, J.M.O., Lindvall, E. & Christou, M. (2003) The development and current status of perennial rhizomatous grasses as energy crops in the US and Europe. Biomass and Bioenergy, 25, 335361.
  • Lockwood, J.L., Cassey, P. & Blackburn, T. (2005) The role of propagule pressure in explaining species invasions. Trends in Ecology & Evolution, 20, 223228.
  • Mack, R.N. (2000) Cultivation fosters plant naturalization by reducing environmental stochasticity. Biological Invasions, 2, 111122.
  • Mariani, C., Cabrini, R., Danin, A., Piffanelli, P., Fricano, A., Gomarasca, S., Dicandilo, M., Grassi, F. & Soave, C. (2010) Origin, diffusion and reproduction of the giant reed (Arundo donax L.): a promising weedy energy crop. The Annals of Applied Biology, 157, 191202.
  • Mathworks (2011) MATLAB. Version 7.12.0. Mathworks, Natick, MA, USA.
  • Matlaga, D., Schutte, B. & Davis, A. (2012) Age-dependent demographic rates of the bioenergy crop Miscanthus × giganteus in Illinois. Invasive Plant Science and Management, 5, 238248.
  • Neubert, M.G. & Caswell, H. (2000) Demography and dispersal: calculation and sensitivity analysis of invasion speed for structured populations. Ecology, 81, 16131628.
  • Neubert, M.G. & Parker, I.M. (2004) Projecting rates of spread for invasive species. Risk Analysis, 24, 817831.
  • Nielsen, P.N. (1987) Vegetativ formering af elefantgraes, Miscanthus sinensis ‘Giganteus’. Tidsskrift for Planteavl, 91, 275281.
  • Parker, I., Simberloff, D., Lonsdale, W., Goodell, K., Wonham, M., Kareiva, P., Williamson, M., Von Holle, B., Moyle, P. & Byers, J. (1999) Impact: toward a framework for understanding the ecological effects of invaders. Biological Invasions, 1, 319.
  • Pysek, J., Sadlo, J. & Mandak, B. (2002) Catalogue of alien plants of the Czech Republic. Preslia, 74, 97186.
  • Quinn, L.D., Allen, D.J. & Stewart, J.R. (2010) Invasiveness potential of Miscanthus sinensis: implications for bioenergy production in the United States. Global Change Biology Bioenergy, 2, 310320.
  • Quinn, L.D., Matlaga, D.P., Stewart, J.R. & Davis, A.S. (2011) Empirical evidence of long-distance dispersal in Miscanthus sinensis and Miscanthus × giganteus. Invasive Plant Science and Management, 4, 142150.
  • Raghu, S., Anderson, R.C., Daehler, C.C., Davis, A.S., Wiedenmann, R.N., Simberloff, D. & Mack, R.N. (2006) Adding biofuels to the invasive species fire? Science, 313, 1742.
  • Ramula, S., Knight, T.M., Burns, J.H. & Buckley, Y.M. (2008) General guidelines for invasive plant management based on comparative demography of invasive and native plant populations. Journal of Applied Ecology, 45, 11241133.
  • Robertson, G.P., Dale, V.H., Doering, O.C., Hamburg, S.P., Melillo, J.M., Wander, M.M., Parton, W.J., Adler, P.R., Barney, J.N., Cruse, R.M., Duke, C.S., Fearnside, P.M., Follett, R.F., Gibbs, H.K., Goldemberg, J., Mladenoff, D.J., Ojima, D., Palmer, M.W., Sharpley, A., Wallace, L., Weathers, K.C., Wiens, J.A. & Wilhelm, W.W. (2008) Agriculture – Sustainable biofuels Redux. Science, 322, 4950.
  • Ross, M. (2011) New miscanthus development possible biomass game-changer? FarmWeek, May 16, 2011, pp. 9.
  • Shea, K., Kelly, D., Sheppard, A.W. & Woodburn, T.L. (2005) Contextdependent biological control of an invasive thistle. Ecology, 86, 31743181.
  • Stewart, J.R., Toma, Y., Fernandez, F.G., Nishiwaki, A., Yamada, T. & Bollero, G. (2009) The ecology and agronomy of Miscanthus sinensis, a species important to bioenergy crop development, in its native range in Japan: a review. Global Change Biology Bioenergy, 1, 126153.
  • Tilman, D., Socolow, R., Foley, J.A., Hill, J., Larson, E., Lynd, L., Pacala, S., Reilly, J., Searchinger, T., Somerville, C. & Williams, R. (2009) Beneficial biofuels-the food, energy, and environment Trilemma. Science, 325, 270271.
  • USDA-FSA (2011) Environmental assessment: proposed BCAP Giant Miscanthus (Miscanthus × giganteus) establishment and production in Arkansas, Missouri, Ohio, and Pennsylvania. Sponsored by Aloterra Energy LLC and MFA Oil Biomass LLC.
  • Yu, C.Y., Kim, H.S., Rayburn, A.L., Widholm, J.M. & Juvik, J.A. (2009) Chromosome doubling of the bioenergy crop, Miscanthus × giganteus. Global Change Biology Bioenergy, 1, 404412.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
jpe12057-sup-0001-FigS1-S4.pptxapplication/197K

Fig. S1. Simulation results showing population growth rate (λ) of sterile Miscanthus × giganteus.

Fig. S2. Simulation results showing population spread rate (c*) of sterile Miscanthus × giganteus.

Fig. S3. Simulation results showing population growth rate (λ) of fertile Miscanthus × giganteus.

Fig. S4. Simulation results showing population spread rate (c*) of fertile Miscanthus × giganteus.

jpe12057-sup-0002-TableS1-S3.docxWord document22K

Table S1. Demographic transitions (aij) in Miscanthus × giganteus matrix projection model and their calculation from lower-level demographic parameters.

Table S2. Results of elasticity analysis of Miscanthus × giganteus sterile matrix projection model.

Table S3. Results of elasticity analysis of Miscanthus × giganteus fertile matrix projection model performed during simulations presented in the four corner panels of Fig. 3.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.