Invasive species are any “alien species whose introduction does or is likely to cause eco­nomic or environmental harm or harm to human health” [43]. Valery et al. [44] clarify the concept of “alien” at the ecosystem scale, rather than at the scale of geopolitical bound­aries (e. g., switchgrass is native to grasslands of the central United States, but alien to California grasslands). Biomass production for bioenergy has the central aim of maximiz­ing harvestable dry mass per unit land area, labor and input expense. As bioenergy crop candidate species have been evaluated for their ability to fulfill these production criteria, a suite of traits characterizing a bioenergy crop ideotype has been identified. Ideal bioenergy crops feature a C4 photosynthetic system, long canopy duration, perennial life history, no known pests or diseases, high relative growth rate to suppress competing vegetation, sterile seeds, storage of nutrients in underground organs prior to biomass harvest, and high water use efficiency [42]. As noted in Raghu et al. [45], with the exception of sterile seeds and perennial life history, all of these traits are risk factors associated with increased likelihood that a plant species will become invasive when introduced into favorable habitats beyond its native range.

Many scientists have expressed concern about the invasive potential of bioenergy crops over the past five years [45,46-52]. To date, such reports have largely approached the issue in one of three ways: literature reviews providing background on species being considered as potential bioenergy feedstocks [51]; bioclimatic envelope models to determine potential ranges for introduced crops [46, 53]; and qualitative analyses of risk using expert decision support systems, such as the Australian Weed Risk Assessment (WRA) or adaptations thereof, tailored to specific locales [47, 52]. Such approaches are a necessary beginning for evaluation of invasive potential of bioenergy crops, but the power of inferences made with these methods is limited by the lack of empirical evidence from within proposed areas of introduction, and further limited by variation in expert opinion driving these tools [54]. For those bioenergy crops nearing deployment, quantitative risk analysis based on field studies in the proposed production area will provide site-specific information on invasive potential [50]. A small number of such experiments have begun to appear in the scientific literature [55, 56], hopefully providing a more comprehensive understanding of invasiveness in coming years.

Impacts of invasive plants in their new habitats can range from modest effects on commu­nity composition to wholesale reorganization of ecosystem structure and function [57,58]. For example, invasion of montane forests in Hawaii by the fire-adapted grass Schizachyrium condensatum resulted in a more than fivefold increase in fire frequency and severity, altering species composition and nutrient flows [57]. The estimated combined economic impact of invasive species worldwide is in the order of $190 billion annually in lost revenues, ecosys­tem services and cleanup efforts [59]. Among current perennial bioenergy crop species, several are already known as invaders within the continental United States. These include Arundo donax [60], Miscanthus sinensis [51], Phalaris arundinacea [61], Phragmites australis [62], and Triadica sebifera [63]. The choking rhizomatous mats of vegetation produced by A. donax, P. arundinacea or P. australis in riparian corridors displace native vegetation and make lavish use of water resources [64]. Potential impacts of invasions by bioenergy crop species on wildlife populations are difficult to predict, since such studies are few and most draw conflicting inferences depending upon crop species, wildlife species and habitat of concern [65-67]. In addition to the scenario of bioenergy crops becoming invaders themselves, there is also the possibility that they will facilitate the invasion of other organisms. One such scenario includes augmentation of agricultural pest populations by providing them with over-winter habitat. For example, M. x giganteus has been found to serve as an alternate host for the Western corn rootworm (Diabrotica virgifera virgifera), thereby creating the potential for increased severity of outbreaks of this insect pest and exacerbated crop yield losses [68].

The amount of biomass necessary to meet renewable energy goals is enormous and will, therefore, require huge land areas [69]. Pilot projects are already underway to develop biomass production potential, such as the initiative to grow M. x giganteus on marginal arable land in the Midwest United States, sponsored by the Biomass Crop Assistance

Program of the USDA Farm Services Agency [70]. This project calls for four 20 000 ha areas to be planted to M. x giganteus in Arkansas, Missouri, Ohio and Pennsylvania. Current qualitative evaluations of M. x giganteus traits suggest that it has low invasive potential in California and Florida [52,53]. However, even if the probability of a given bioenergy crop species becoming invasive is low, if it is greater than zero there will likely be escapes when production is fully scaled up by 105-106 ha, and 109 plants are involved.

Reducing the frequency and impact of biological invasions resulting from bioenergy production is essential to the sustainability of the enterprise. Three complementary types of actions are necessary to prevent and ameliorate bioenergy crop invasions: (1) germplasm screening; (2) production best management practices; and (3) containment. Most pre-introduction screening of bioenergy crop cultivars to date has been accomplished using variants of the Australian WRA [47,48, 52]. Following such initial screens with an empirically-based demographic modeling approach in planned areas of introduction will likely provide much more robust inferences on how much of a threat different crop cultivars are likely to be [50]. Such a system would be helpful not only for evaluating existing crop germplasm but would also help to define non-invasive crop ideotypes to guide breeding efforts [71].

When scaling up biomass production from test plots to production fields, best man­agement practices for plantation design, production and harvesting should all contribute to lower risks of invasion. A basic ground rule for siting plantations is that rhizomatous perennial grasses, which are easily dispersed by water, should not be planted adjacent to riparian areas [60, 62]. Quantitative knowledge of dispersal processes of the crop species is critical to designing effective buffer areas for production fields. A buffer strip surrounding the bioenergy crop should be sown to a turf or agronomic crop species for which weed management practices are well-characterized. This will form a containing perimeter for the bioenergy crop that is easily maintained as a pure stand and for easy monitoring of possible escapes. The width of the surrounding buffer area should be estimated as the product of the annual rate of vegetative spread of the crop and the number of years a production field will be maintained, possibly increasing the buffer area by some margin of error. If the bioenergy crop species has viable, wind-dispersed seed, as with Miscanthus sinensis [51, 56], it may be safer to embed a smaller biomass production area within a larger matrix of agronomic crop to form a barrier against seed dispersal. Such a design will help to provide containment of the bioenergy crop species even if monitoring efforts fail in some years.

Monitoring is essential for any containment strategy and should be performed annually along the entire perimeter of the bioenergy crop production field. Escapes should be flagged and terminated, and revisited for several years thereafter to ensure complete eradication

[72] . For wind dispersed species, monitoring efforts will need to extend beyond buffer areas into surrounding habitats that are likely to allow establishment of the bioenergy crop species

[73] . Such efforts will be aided at a local scale by empirical data on potential establishment of bioenergy crops in various types of non-arable lands, and at regional and larger scales through the use of climate-matching models [53].