populations large enough to reveal their invasive qualities (Wolfenbarger and Phifer, 2000). These realities increase un­certainty in long term safety predictions.
      Transgene flow also creates potential liabilities (Smyth et al., 2002). The liability is realized when the flow results in traditional, economic or environmental damage (Kershen, 2004; Heinemann, 2007). Traditional damage is harm to hu­man health or property. Economic damage could occur if a conventional or organic farmer lost certification and there­fore revenue because of adventitious presence. Environmental damage could result from, for example, harm to wildlife.
       There are a limited number of properly designed and independently peer-reviewed studies on human health (Do­mingo, 2000; Pryme and Lembcke, 2003). Among the stud­ies that have been published, some have provided evidence for potential undesirable effects (Pryme and Lembcke, 2003; Pusztai et al., 2003). Taken together, these observations cre­ate concern about the adequacy of testing methodologies for commercial GM plants fueling public skepticism and the possibility of lawsuits. A class-action lawsuit was filed by USA consumers because they may have inadvertently con­sumed food not approved for human consumption (a GM variety of maize called Starlink) because of gene flow or an­other failure of segregation. The lawsuit ended with a settle­ment against the seed producer Aventis. This suggests that consumers may have grounds for compensation, at least in the USA, even if their health is not affected by the transgenic crop (Kershen, 2004).
       Farmers, consumers and competitors may be the source of claims against, or the targets of claims from, seed produc­ers (Kershen, 2004; Center for Food Safety, 2005; Eicher et al., 2006). For example, when non-GM corn varieties from Pioneer Hi-Bred were found in Switzerland to contain novel Bt genes, the crops had to be destroyed, and compensation paid to farmers (Smyth et al., 2002).
       Even if liability issues could be ignored, the industry will remain motivated to track transgenes and their users because the genes are protected as IP. Transgene flow can create crops with mixed traits because of "stacking" (two transgenes from different owners in the same genome) or mixed crops (from seed mediated gene flow or volunteers), resulting in potential IP conflicts. IP protection includes par­ticular genes and plant varieties as well as techniques for creating transgenic plants and product ideas, such as the use of Bt-sourced Cry toxins as plant-expressed insecticides. Broad IP claims are creating what some experts call "patent thickets"; the danger of thickets is that no single owner can possess all the elements in any particular transgenic plant (Thomas, 2005).
        Release of insect resistant GM potatoes in South Af­rica illustrates the complexity that IP and liability create for transgenic crops. The potato has elements that are claimed by two different companies. One of the IP owners has been unwilling to license the IP to South Africa for fear of liability should the potatoes cross into neighboring countries (Eicher et al., 2006).
        The harms associated with transgene flow might be ad­dressed by a combination of physical and biological strate­gies for containment (for a comprehensive list, see NRC, 2004). However, no single method and possibly no combi-

nation of methods would be wholly adequate for preventing all flow even though for some genes and some environments, flow might be restricted to acceptable levels (Heinemann, 2007). Future strategies for containment involving steril­ization (i.e., genetic use restriction technologies, GURTs) remain highly controversial because of their potential to cause both unanticipated environmental harm and threaten economic or food security in some agroeconomic systems (Shand, 2002; Heinemann, 2007).
       For transgenic approaches to continue to make signifi­cant contributions in the long term, a substantial increase in public confidence in safety assessments will be needed (Eicher et al., 2006; Herrero et al., 2007; Marvier et al., 2007); conflicts over the free-use of genetic resources must be resolved; and the complex legal environment in which transgenes are central elements of contention will need fur­ther consideration.

Epigenetic modification of traits. Epigenes are defined as units of inheritance that are not strictly based on the or­der of nucleotides in a molecule of DNA (Strohman, 1997; Heinemann and Roughan, 2000; Gilbert, 2002; Ashe and Whitelaw, 2007; Bird, 2007). A growing number of traits are based on epigenetic inheritance, although at present most of these are associated with disease, such as Mad Cow Disease and certain forms of cancer.          In the future, it may be possible to introduce traits based on epigenes. For example, double-stranded RNA (dsRNA) is the basis of at least two commercial transgenic plants and is proposed for use in more (Ogita et al., 2003; Prins, 2003). Small dsRNA molecules appear to be the basis for the trait in "flavr savr" tomatoes—even though at the time of devel­opment the epigenetic nature of the modification was prob­ably not known or fully understood (Sanders and Hiatt, 2005)—and the basis for viral resistance in papaya (Tennant et al., 2001). In these cases, the epigene is dependent upon a corresponding change at the DNA level, but in time it will be possible to use the epigenetic qualities of dsRNA to infec­tiously alter traits without also altering the DNA content of the recipient genome using rDNA techniques. Such prom­ise has already been demonstrated using nematodes where feeding, or soaking the worm in a liquid bath of dsRNA, was sufficient for systemic genetic modification of the worm and the stable transmission of the epigene for at least two generations (Fire et al., 1998; Cogoni and Macino, 2000). The effects of dsRNA also can be transmitted throughout a conventional plant that has been grafted with a limb modi­fied to produce dsRNA (Palauqui et al., 1997; Vaucheret et al., 2001; Yooetal., 2004).
        RNA-based   techniques  will   accelerate   research   de­signed to identify which genes contribute to complex traits and when and where in the organisms those genes are ex­pressed ("turned on"). Generally, dsRNA causes transient, long-term, sometimes heritable gene silencing (turns genes "off"). While silencing that occurs by the general pathways controlled by dsRNA molecules are targeted to sequence matches between the dsRNA and the silenced genes, there are often effects on nontarget genes as well. The number of genes simultaneously silenced by a single dsRNA (includ­ing the targets) can number in the hundreds (Jackson et al.,