Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
6 Biological and Operational Considerations for Bioconfinement This chapter summarizes and analyzes what has been presented in the foregoing chapters. First, the biological opportunities and constraints for confinement are reviewed, with special emphasis on bioconfinement. Next, the operational implications of confinement are considered. Then, confine- ment failure and its mitigation are discussed. Finally, there is a look to the future and the need to explore unanswered questions and promote research that will build better avenues for the confinement of genetically engineered organisms (GEOs). WHAT BIOLOGY TELLS US ABOUT CONFINEMENT AND BIOCONFINEMENT As explained in Chapters 35, a wide array of bioconfinement measures has been proposed for limiting the movement of transgenes. Some of them are hypothetical, some have been examined in the laboratory, and a few take advantage of well-known biological phenomena. All of them share some features. Each method has strengths and weaknesses, and all vary in efficacy depending on circumstances. No one method will achieve 100% confinement in the real world. Straightforward conclusions follow from the observations presented in the preceding chapters. Case-by-Case Evaluation GEOs represent a heterogeneous class with regard to biosafety. Some 180
BIOLOGICAL AND OPERATIONAL CONSIDERATIONS 181 present minimal risk, others moderate risk, and yet others considerable risk. As noted in Chapter 2, the decision of whether and how to confine a GEO depends on factors that range from the phenotype associated with the transgene to the environment into which the organism would be released. Confinement options will vary with the precise species chosen for transfor- mation because there are so many differences in size, genetics, ecology, and dispersal biology. In some cases where confinement is necessary, physical and physicochemical confinement options will suffice; in others, biological confinement might be necessary. Clearly, there is no universal option, and case-by-case evaluation is a necessity. Finding 1. The efficacy of bioconfinement will depend on the organism, the environment, and the temporal and spatial scales over which the organ- ism is introduced. Finding 2. In many cases GEOs will not require bioconfinement. Recommendation 1. Evaluation of the need for bioconfinement should be considered for each GEO separately. Early Evaluation The evaluation of whether and how to confine a GEO cannot be an afterthought in the process of development of a transgenic organism. Making biosafety a primary goal from the start of any project will be a more effective and efficient way to prevent safety failures and it will increase commercial investment ratings and reduce financial risks posed by possible liability claims and loss of consumer confidence (Kapuscinski et al., 2003). If biosafety considerations are delayed until after a product is developed, the need to receive a return on the investment made to create that product could cloud the judgment of those who determine whether and how it should be used. Similar considerations (including reducing liability and avoiding public relations problems) make it preferable for noncommercial GEO developers, such as universities or international research centers, to make biosafety a primary goal at the outset. Dispersal biology and the opportunities for the unintentional move- ment of transgenes must be considered as part of the process of finding the best organism to modify to create a product. For example, the evaluation should consider whether the organism is to be released near or distant from other organisms of the same species. Early evaluation permits the consider- ation and comparison of simpler, traditional confinement techniques along- side the more complex, and sometimes more expensive, bioconfinement options. The constant and iterative evaluation of confinement options during
182 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS the development of a GEO should optimize both the efficacy and the cost effectiveness of the confinement options once they are deployed. Hurried consideration just before the deployment of a GEO is apt to create a makeshift and expensive plan that might work better in theory than in practice. Recommendation 2. The need for bioconfinement should be considered early in the development of a GEO or its products. Redundancy Because methods can fail, a single confinement method will not neces- sarily prevent transgene escape. Therefore, it is sometimes necessary to employ more than a single method. In many technological applications, the principle of redundancy reduces the occurrence of predictable hazards while achieving the benefits of technological application. Redundancy involves applying two or more safety measures to product design and use, each with fundamentally different strengths and vulnerabilities, so that the failure of one safety measure is counterbalanced by the integrity of another. In other cases, it may be possible to combine two barriers of the same type but whose failures would be independent events, such that a failure of one barrier does not trigger a failure of the other. This does not necessarily require using different bioconfinement methods, as long as the measures are independent. By mixing confinement measures with different vulnerabilities, the chances improve that failure of one safety measure will not breach the target level of confinement. When choosing redundant confinement techniques (including bioconfinement), measures should be chosen to compensate for each other's weaknesses. In many cases, this will involve application of an appropriate mix of biological, physical, and physicochemical confinement measures tailored to the GEO in question (Agricultural Biotechnology Research Advisory Com- mittee, 1995; Kapuscinski, 2001; Scientists' Working Group on Biosafety, 1998). For example, the U.S. Department of Agriculture (USDA) has devel- oped requirements for growing transgenic corn for pharmaceutical and industrial chemical production that mandate spatial and temporal isolation from corn grown for other uses (Federal Register, 2003). One feasible application of the principle of redundancy in aquaculture would be to combine physical barriers, such as floating cages, with bioconfinement con- sisting of the use of all-female lines of sterile, triploid fish. Finding 3. It is unlikely that 100% confinement will be achieved by a single method.
BIOLOGICAL AND OPERATIONAL CONSIDERATIONS 183 Finding 4. Redundancy in confinement methodology decreases the probability of failing to attain the desired confinement level. Experimental Information on Efficacy The discussion of redundancy implies that some information should be available on how well a confinement method works. The effectiveness of many confinement methods, particularly bioconfinement methods, will depend on the genotype and the environment. Thus, the efficacy of the planned combination of confinement methods should be tested in represen- tative genotypes under development to ensure that the plan is effective. Also, the planned combination of confinement methods should be tested in every environment in which it is anticipated that a GEO will be released including any environment that the GEO could be foreseen to occupy. For example, if strict confinement is desired for a corn genotype that is to be grown from seed, it is important to test the efficacy of the confinement technique in all environments to which that seed might accidentally be dispersed. Likewise, before field release, the reproductive biology of the novel genotype should be measured relative to its progenitor to evaluate changes that might affect its rate of gamete and progeny production and their dispersal. Studies have shown that some transgenes could allow wild or weedy relatives of the crop to be more successful and that other transgenes will not (Burke and Rieseberg, 2003; Snow et al., 2003). Although new genotypes generally do not have reproductive phenotypes that are different from those of their parents, any changes that occur can be dramatic and have important consequences. For example, hybridization between non- GEO sugar beets and wild sea beets introduced an allele into a crop that increased its rate of premature flowering, and the crop became a noxious weed (Boudry et al., 1993, 1994; Viard et al., 2002). Changes in reproduc- tive biology might not be an anticipated phenotype associated with a novel genotype. Pleiotropy--the unanticipated phenotypic effects of a single allele--is not rare. BeetSwiss chard hybrids with a transgene construct for virus resistance showed a decreased rate of premature flowering relative to nontransgenic control plants (Bartsch et al., 2001). Recommendation 3. Confinement techniques should be tested experi- mentally, separately and in combination, in a variety of appropriate envi- ronments, and in representative genotypes under development before they are put into application. Recommendation 4. To evaluate changes in reproductive biology, the novel genotype should be compared with that of its progenitor before field
184 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS release. For long-lived species, such as trees, it may be necessary to begin field tests before such comparisons are possible, with a realistic plan to mitigate any unexpected and dramatic increase in reproduction. Changes of Efficacy with Scale Typically, precommercial evaluation of GEOs starts at a small scale and then is often expanded to larger scales before release. Even with the largest precommercial field trials involving up to 100 sites and 1000 acres (or less) per site over a two- to three-year period, the scale of these may be dwarfed by the regional or continental scale at which these GEOs may be produced. It is well known that many environmental concerns cannot be addressed prior to commercialization. An example of a response to these concerns is the monitoring requirements for Bt resistance in target insects (NRC, 2002a). Similarly, the spatial or temporal scale of a field release can influence the potential for confinement failure. The appropriate confine- ment option will depend on scale. Under a very limited field release--a tenth of an acre or over a few hoursone or two methods of confinement might suffice. However, the same genotypes released over 100 acres or for many years could require several methods to obtain the same level of con- finement. Alternatively, field release might not be a safe option on a large scale or for a long period. If possible, empirical data (experimental or otherwise) should be used to determine whether the confinement plan is adequate for the anticipated scale of field release. Recommendation 5. Bioconfinement techniques should be assessed with reference to the temporal and spatial scales of field release. How Much Bioconfinement is Enough? The foregoing sections suggest the need to define "adequate level of bioconfinement" early on. This requires an evaluation of failures and their consequences under worst-case scenarios. It also requires an assumption that escaped genes have the opportunity to multiply. In some cases, the escape of 10 individuals per year into the ambient environment might not be a problem; in other cases, 10 would be too many. Recommendation 6. An adequate level of bioconfinement should be defined early in the development of a GEO, after considering worst-case scenarios and the probability of their occurrence.
BIOLOGICAL AND OPERATIONAL CONSIDERATIONS 185 Unacceptability of Some Methods under Some Circumstances Some bioconfinement methods will be unacceptable under some cir- cumstances. Apomictic seed production by absolutely sterile male plants could be a multiple benefit in ensuring true-to-type seed without an oppor- tunity for transgene escape by pollen. Combined with multiple confinement methods, the use of apomixis could be acceptable. However, apomictic organisms with some male fertility that are released close to wild relatives pose an opportunity for the transgenic genome to sweep through the wild population, replacing it with a clonal transgenic lineage (van Dijk and van Damme, 2000; see Chapter 3). In that case, the use of apomixis should be rejected as a confinement method. Finding 5. Some bioconfinement methods are unacceptable under some circumstances. Options Based on Technology and Gene-Specific Compounds Bioconfinement methods that are based on transgenic technology have received considerable recent attention (e.g., Daniell, 2002), and this com- mittee also has identified the potential of bioconfinement by external admin- istration of gene-specific compounds (Chapter 4). Although those methods hold great promise, none has been tested in an array of organisms and for a variety of environments. Indeed, some methods are still theoretical. Even those transgenic bioconfinement methods that have been created have yet to be tested adequately in a single organism under a variety of field condi- tions. Statistically adequate experiments still are necessary to measure their efficacy. Transgenic and gene-specific bioconfinement technology still is in its infancy and has not yet been proven as effective as have nontransgenic confinement methods that already are in use. Finding 6. Many types of bioconfinement are still in the early stages of development, especially those based on transgenic methods and gene-specific compounds. EXECUTION OF CONFINEMENT The foregoing considerations suggest that the field release of a GEO constrained by confinement should follow a straightforward pathway: Decision Making Once the phenotype of the organism has been identified, its biosafety must be appraised: What risks does it pose? What would be the worst
186 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS possible scenarios created by those risks? How are those risks balanced by anticipated benefits? Is some confinement necessary? If so, how much? Are the risks large enough to warrant the use of a different organism or aban- doning the project altogether? What are the possible opportunities for the escape of the gene or the organism, including human error in handling living propagules or gametes? What confinement methods are available for this organism? What is the potential for spread of the GEO if it escapes? Given what is known about the methods, the organism, the novel pheno- type, and the spatial and temporal scale of anticipated field release, which combination of methods--physical, physicochemical, and biological-- should suffice to obtain sufficient confinement to make the risk acceptable? Research Assuming that the decision is made to proceed with a project and that confinement is warranted, experiments designed to answer the questions above should be conducted. For example, the efficacy of the proposed combination of confinement methods should be tested in the field before their use with genotypes that are as similar as possible to the novel genotype in question. The proposed combination of confinement methods also should be tested in an appropriate range of environments to which the new geno- type will be released or to which it might escape. Similarly, the reproductive biology of the novel genotype should be compared with that of its progenitor before field release to evaluate changes in reproductive biology. Integrated Confinement System If the tests of the proposed confinement technique suggest that it will be successful, it will still be necessary to establish an integrated confinement system (ICS) for the deployment of the organism to ensure confinement efficacy, especially as the new genotype is spatially and temporally deployed. ICS is a systematic approach to the design, development, execution, and monitoring of the confinement of a specific GEO. This recommendation is in keeping with system safety management as it is widely practiced in the management of many modern technologies (Roland and Moriarity, 1990). System safety management is a forward-looking, comprehensive, long-term approach that ensures that systems and techniques have safety designed in from the outset (McIntyre, 2002). Necessary elements of ICS include the following: · Commitment to confinement by top management · Establishment of a written plan for redundant confinement measures
BIOLOGICAL AND OPERATIONAL CONSIDERATIONS 187 to be implemented, including documentation, monitoring, and remediation (in case of failed confinement) · Training of employees · Dedication of permanent staff to maintain continuity · Use of standard operating procedures for implementing redundant confinement measures · Use of good management practices for applying confinement measures to pharmaceutical-producing GEOs or the equivalent · Periodic audits by an independent entity to ensure that all elements are in place and working well · Periodic internal review and adjustment to permit adaptive manage- ment of the system in light of lessons learned · Reporting to an appropriate regulatory body For an ICS to work, it should be supported by a rigorous and comprehen- sive regulatory regime that is empowered with inspection and enforcement. Recommendation 7. An integrated confinement system approach should be used for GEOs that warrant confinement. Monitoring and Detection Technology The efficacy of the confinement system must be monitored constantly. Several detection techniques are available to determine whether transgenes move to organisms or environments as the result of confinement failure. Some are associated with portions of the transgenic construct. The creation of a transgenic organism usually involves a selectable marker, such as resistance to a specific antibiotic or herbicide. Because the chosen trait is unlikely to be present in nonengineered members of the species, it can serve as a reliable marker. Likewise, the creation of a transgenic organism some- times involves inserting a reporter gene to confirm that the promoter is working effectively. For instance, the so-called GUS construct is a reporter gene that creates a blue color when cells are soaked in the appropriate solution (Jefferson et al., 1987). Although still at the research level, product developers may be able to use the Cre/lox-mediated recombination tech- nology (see Chapter 3) in the future to remove the unwanted selectable marker genes. Finally, there are methods for testing directly for the geno- type, product, or phenotype of the transgene. For example, a standard tool for amplifying a specific DNA segment (polymerase chain reaction, or PCR) facilitates testing for the presence of specific transgene constructs in the genotype of an organism. A standard testing method for detection of a specific protein (enzyme-linked immunosorbent assay, or ELISA) is avail-
188 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS able for testing for the presence of Bt protein. Herbicide resistance can be tested by direct application of the appropriate herbicide at the appropriate concentration (e.g., Lefol et al., 1996). The committee notes that our ability to detect transgenes with PCR and other devices may currently exceed our ability to characterize the risk or consequences from such transgene con- tamination. As noted earlier in the report, an adequate and appropriate characterization of such events will almost always be on a case-by-case basis--depending, for example, on the transgene, its function, the environ- ment where the contamination occurs, the species carrying the transgene, and the number of GEOs involved. In the future, organisms might be transformed with additional con- structs for the purposes of monitoring them. The addition of a gene derived from jellyfish that expresses green fluorescent protein has been used to monitor insects released for biological control (Staten et al., 2001), and it has been proposed for tracking transgenic plants (Leffel et al., 1997). Like- wise, insertion of DNA sequences that can be used as "bio-barcodesTM" to identify specific transgene constructs has been proposed (Gressel, 2002). Ideally, the development of monitoring methods that can identify escapes through remote sensing and that use Geographic Information System tech- nology would make monitoring more feasible. Monitoring of bioconfinement will not be a simple matter. It will involve looking for what will often be a rare event over a potentially large area. Under such circumstances, sampling becomes a challenge (Marvier et al., 1999). The seeds, eggs, pollen, sperm, spores, or other dispersal propagules of many organisms often are too small to collect or analyze in any statisti- cally meaningful way. The expense and effort of adequate monitoring could outweigh the perceived benefits of introducing a GEO to the field. Even with the best and most thorough monitoring scheme, some events will be missed, and, given enough genotypes over enough time, some fraction of those events will have negative consequences. However, even failures of monitoring can offer benefits. A monitoring failure can be used as an example for developing better confinement and monitoring techniques. Catching a mistake too late may still allow the identification of the source of the product. In the case of realized harm, it can be used to assign responsibility. Nonetheless, monitoring should be seen as a complement to confine- ment, not as a replacement for it. That is, the act of monitoring should not result in complacency about the possibility of escape. Effective confinement and adequate monitoring are often easier to manage than eradicating a reproducing organism once it has reached critical numbers (Simberloff, 2003).
BIOLOGICAL AND OPERATIONAL CONSIDERATIONS 189 Recommendation 8. Easily identifiable markers, sampling strategies, and methods should be developed to facilitate environmental monitoring of GEOs. Eradication or Control of Escaped Organisms It can be worthwhile to attempt to eradicate or control escaped GEOs or transgenes. If individuals can be identified easily and the escape is local- ized, eradication can be possible, depending on the organism. If the escaped organisms do not appear likely to cause the harm originally anticipated, it can be worth considering whether control is necessary at all, especially if control will be difficult. If detection has come too late, however, and if the organisms are creating problems and are too widespread for eradication, control is the only option. Increasing the Efficacy of Confinement Three issues that can significantly affect the efficacy of bioconfinement measures are not directly related to natural science: transparency and public participation, compliance, and international considerations. Transparency and Public Participation The public's right to informationoften called transparencyand to participate in decision making, are fundamental principles of democracy. Each right complements the other, and each can improve the effectiveness of confinement. For example, public participation can bring otherwise unknown information to the decision-making process. Transparency can increase acceptance of bioconfinement measures (and of the GEOs being confined) by building public trust in the decision-making process. Trans- parency and public participation also can improve the quality of decisions about GEOs and confinement in terms of protecting human health and the environment. This is true at various stages of decision making about GEOs and confinement. Confinement considerations should come into play at a number of stages in the "life cycle" of a GEO, including research to develop and characterize the genetic and phenotypic traits of a GEO, risk analysis and risk reduction, field testing, commercialization, large-scale production, processing, transportation (domestic or international), and disposal. The analysis associated with selection of confinement methods for GEOsincluding the decision to proceed or notwould benefit from having a public component. Public participation can identify hazards, raise important questions, and provide information about specific conditions that can lead to more realistic assumptions (NRC, 1996; Hails and
190 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS Kinderlerer, 2003). Members of the public, who often will not be scientific experts or otherwise involved in the field of genetic engineering, can offer information that is indispensable to the clear understanding of social values and other factors that affect the significance of potential effects of a con- finement failure (Chapter 2; NRC, 2002a). Transparency also is important to the assessment of environmental or health effects. Transparency about novel GEOs and their confinement also could yield significant benefits in the face of failure. In some cases, when human health or the environment could be at risk, transparency would increase the likelihood that failure can be averted or mitigated early enough to prevent harm. The committee emphasizes that, to its knowledge, no significant harm to health or the environment has resulted from GEO confinement failure. Nonetheless, the StarLink and Prodigene incidents (described in Chapters 1 and 2), are examples of failures in a system that is intended to maintain safety. This is the same system that the American public expects to ensure food safety and environmental protection as a growing array of new GEOs comes into production. Greater challenges for risk assessment and manage- ment (Chapters 35) will be faced as the probability of a confinement failure increases with use of GEOs on larger spatial and temporal scales and with their growing application to produce an increasingly wide array of products. A lack of transparency could increase the likelihood of failure of confinement and exacerbate its consequences. The committee believes that, because of the fundamental need for trans- parency and public participation, the close connection between them, the need to safeguard the environment, and the desirability for increased cred- ibility with respect to GEOs and their confinement, close cases should be called in favor of transparency. When the need for intellectual property protection of a bioconfinement method arises it will influence how trans- parency is maintained; however, transparency should remain a priority. The committee also believes that appropriate transparency and public par- ticipation should be promoted in designing and implementing the system- atic approach to confinementthe ICS described earlier. The appropriate degree and nature of transparency and public participation could vary at the different points in the system. Recommendation 9. Transparency and public participation should be important components in developing and implementing the most appro- priate bioconfinement techniques and approaches. Compliance Compliance is critical to the success of confinement. If the method in question is not followed, bioconfinement will failregardless of its theo-
BIOLOGICAL AND OPERATIONAL CONSIDERATIONS 191 retical efficacy. The committee considered a few of the many factors that can influence compliance: the nature of bioconfinement methods and the state of verification and monitoring technology, human error, natural events, the cost, and increases in spatial and temporal scale. Confinement Methods, Verification, and Monitoring Compliance with a chosen or prescribed confinement measure is affected by how difficult that measure is to apply. Compliance can be expected to increase--other things being equal--as the ease of applying the confinement measure increases. The efficacy of confinement also would vary with the human processes involved, because of human error, discussed below, and because a properly designed management process can improve implementation of prerelease verification and postrelease monitoring. A related factor is the difficulty for those who produce and use GEOs, as well as for regulators, of verifying and monitoring confinement efficacy. This could depend on the characteristics of the GEO or of the confinement method and on the technology available to test for the presence, or measure the effectiveness, of confinement methods. Verification and monitoring tech- nology are discussed earlier in this chapter. Some bioconfinement methods are more amenable than others to veri- fication and monitoring. It is easier to verify a bioconfinement technique that has an obvious physical manifestation, such as one that involves a visually identifiable phenotype, than it is to verify a bioconfinement tech- nique that does not. Finding 7. The efficacy of bioconfinement will vary with the human processes involved in applying the methods, with the characteristics of the GEO, and with the confinement method itself. Human Error Humans make mistakes, and experience with GEO confinement bears this out. In the 19992000 StarLink situation (see Box 2-1, Chapter 2), corn for human consumption was contaminated by a genetically engineered variety approved only as an animal feed (Taylor and Tick, 2003). The commingling probably occurred because the U.S. commodities system does not keep bulk grain separated. There was considerable speculation as to how and why the varieties were mixed, but there was no doubt that human error was a major cause. The committee recognizes the difficulty of predicting when and where human error will occur, particularly for bioconfinement, for which there is no history of mishaps from which to try to generalize or predict. The
192 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS committee nevertheless believes that the probability of human error should be considered, for instance, by drawing on methods of system safety (Roland and Moriarity, 1990) or organizational analysis, which could be refined as data about confinement accumulate over time. A peculiar form of human error, which has its roots in kindness, also could affect confinement. For example, many goldfish owners do not wish to kill pets they no longer want, and instead might release the fish into bodies of water or flush them down a drain. This is an error in the sense that the actor presumably is ignorant of the possible consequences of the action, which are not always benign. Red-eared slider turtles (Trachemys scripta elegans) and snakehead fish (Channa marulius, C. argus, C. striata, C. micropeltes) have been introduced throughout their nonindigenous range through pet releases (Mayell, 2002; USFWS, 2002; USGS, 2002). Similar actions could affect confinement of genetically engineered animals that have reached the end of their "production lives," yet remain healthy. The omnipresent risk of human error was an important factor underly- ing the committee's conclusions that the implementation of confinement methods should be systematic and integrated. Redundancy in confinement methodologies is essential. The committee is aware that intentional human actions, such as bio- terrorism or unethical business practices, might result in a failure of bio- confinement and release of GEOs into the environment. These topics are beyond the scope of this study; another NRC study has considered one aspect of this issue (NRC, 2003b). Recommendation 10. The possibility of human error should be taken into account as a factor when determining bioconfinement methods and evaluating their efficacy. Natural Events Compliance also can be affected by natural events. A hurricane, tor- nado, or tsunami can wreak havoc with physical confinement, for example, by destroying fish cages. Similarly, natural vectors such as insects, rodents, and birds that carry seeds can affect dispersal. If a bioconfinement system is dependent on physical confinement, a natural disaster could expose the organism to an environment where the bioconfinement technology would no longer function optimally. Cost of Compliance One would expect that--all other things being equal--compliance will increase as the cost of a prescribed bioconfinement regime decreases. The
BIOLOGICAL AND OPERATIONAL CONSIDERATIONS 193 committee is aware, however, that cost can vary significantly. Switching from a well-known and genetically well-characterized crop, such as corn, to a less well understood plant for chemical production could add greatly to the cost of its confinement. Many bioconfinement techniques are expensive simply because they are untested or are still in the early stages of develop- ment. As new techniques that emerge from laboratories and field trials are put into use, the cost of implementation should change. The committee was thus not able to determine how the cost of compliance might favor specific bioconfinement techniques. Private Litigation Business entities have an incentive to comply with bioconfinement to reduce their risk of liability from private litigation arising from damage to human health or the environment. Private actions alleging liability for damage caused by the escape of a GEO (a confinement failure) can be brought under state tort or nuisance law, although few cases have been filed (Chapter 2). The strength of any disincentive effect will depend not only on the extent of liability provided by relevant tort and nuisance laws but on the interplay with intellectual property law. Under some circumstances, con- finement failure can lead to allegations that intellectual property rights associated with GEOs have been violated by a third party. Those cases will thus help set precedent in cases that pertain to bioconfinement failure. Compliance also can be affected by private suits that are brought to enforce federal laws or to challenge the way federal agencies respond to citizen petitions regarding GEO confinement. Increases in Spatial and Temporal Scale Increases in the spatial and temporal scale of GEO production and the use of bioconfinement techniques could affect the incidence of compliance. This is discussed in detail in previous chapters and earlier in this chapter. INTERNATIONAL ASPECTS GEOs have several significant international dimensions that are relevant to confinement, as described in Chapter 1. The biotechnology industry is international, and development, testing, and use of GEOs actively (and increasingly) occurs throughout the world. GEOs are traded internation- ally. GEOs also can move across national boundaries by a wide range of mechanisms. Therefore, no single country can regulate all of the confine- ment issues that could affect its citizens, its economy, or its environment.
194 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS The international obligations of government must be linked to a case-by- case analysis of the GEO. When confinement fails, GEOs can move from one nation to another. Addressing confinement here in the U.S. thus requires considering efficacy, concerns, and consequences not only in this country but in other countries to or from which GEOs are likely to move. International mechanisms and regimes that apply to such movement also are of concern. The U.S. thus has an interest in international cooperation on appropriate bilateral or multi- lateral regulatory regimes and in appropriate activities for standardizing regulatory approaches in countries throughout the world. Recommendation 11. Regulators should consider the potential effects that a failure of confinement could have on other nations, as well as how foreign confinement failures could affect the United States. Recommendation 12. International cooperation should be pursued to adequately manage confinement of GEOs. BIOCONFINEMENT FAILURE Bioconfinement measures occasionally will fail, for example, because of human error or because of an unpredicted response in the GEO. Poten- tial problems can be addressed at two different times--before and after a failure occurs. The committee focused on preventive actions that might be taken to prevent escape of GEOs and their genes and to questions that would be most important: What, if any, bioconfinement measures (possibly used in concert with other confinement measures) should be used to pro- vide the desired confinement? And, if that amount of prevention is not achievable, should the genetic engineering of the organism proceed at all? The optimal choice of bioconfinement method for any particular situa- tion will be unique to that case. Chapter 2 discusses the points to consider: the desired or mandated level of protection; the organism; the novel trait; the available confinement techniques, biological and otherwise; the relevant genomic, physical, and biotic environment; behavioral factors; social values; the resources potentially available for prevention or remediation of bio- confinement failure; and the competing demands for those resources. In addition, the applicable regulatory regime could impose requirements or constraints regarding what confinement techniques may be used. Decisions about confinement--and remediation if confinement fails-- depend on the judgments, values, instincts, and skills of the people and organizations involved in decision making, as well as on the political situa- tion. Chapter 2 presents an approach that attempts to provide guidance to decision makers based on a rough analysis of the severity of consequence
BIOLOGICAL AND OPERATIONAL CONSIDERATIONS 195 and the probability of occurrence. That and other chapters present several rules of thumb that could assist those who must make judgments. As indicated above, the need for whether confinement is necessary should be considered, and an adequate level of confinement should be defined, early in the development of a GEO (Recommendations 2 and 6). This reflects the fact that it is essential to consider possible preventive action, including the use of biological and other confinement measures. With respect to harm to human health and the environment generally, it is well recognized that prevention typically is less expensive and more effec- tive than post-failure remedial action, and that some consequences (e.g., death of a human, extinction of a species, and destruction of a major ecosystem) cannot be undone at all. Indeed, the choice of what confinement technique or techniques to use should be made very early in the process of developing a GEO as part of a broader analysis of possible preventive actions: confinement may be precluded if not undertaken early, and that analysis may determine that the desired level of protection cannot be attained through the use of biological and other confinement measures. Thus the proposed GEO may not be developed at all. LOOKING TO THE FUTURE: STRATEGIC PUBLIC INVESTMENT IN BIOCONFINEMENT RESEARCH The need for continuedand increasedpublic support of agricultural research has been articulated in previous National Research Council reports (NRC, 1989b; 2000; 2003a). One NRC report (NRC, 2002c) defines publicly funded agricultural research as any agricultural research performed with financial or material support from the public sector. For agricultural research in general, the reasons given for public support include · To improve human health and well-being through advances that lead to higher quality and nutritional value in the food supply and greater food safety · To sustain the quality and productivity of natural resources · To preserve biological resources that are the endowment for future generations Publicly funded research on bioconfinement methods is needed for all of these reasons. The institutions that conduct and fund public agricultural research have been widening their agendas to support broad public policy goals. Environmental issues, sustainable production systems, and resource conser- vation are among the new emphases (NRC, 2002c; 2003a). That shift is occurring simultaneously with an increase in industry-funded agricultural
196 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS research. In the past, publicly funded basic and applied researchmainly at the nation's land grant colleges and universities and in USDA and state laboratoriesfocused on productivity. The result was the underpinnings of the large agricultural output we enjoy today. Although strong public-sector crop and animal-breeding programs continue, over the past 25 years much of the productivity research has moved to the private sector, as is especially apparent in applied research in biotechnology. Of necessity, industrial research is primarily market-driven, whereas publicly funded research need not be. Long-range and non-market-driven publicly funded research can help ensure the continuation of the fundamental biological discoveries that will lead to innovative bioconfinement methods not envisioned today. Publicly funded research also should lead to new ways to assess the risks of various bioconfinementand other confinementmethods, and to new ways to monitor confinement. Finally, publicly-funded research on bioconfinement will help train professionals who will manage this powerful technology. In addition to these broad reasons for publicly funded research on bioconfinement, the committee recommends support for additional scien- tific research that · characterizes ecological risks and consequences and develops methods and protocols for assessing the environmental effects of confinement failure (Recommendation 13). More data are needed on the nature of potential ecological effects: their probability, their severity, and the potential for remedial action should confinement fail. Those research needs also were identified in recent reports (NRC, 2002a; 2002b) that noted the need for developing deeper theoretical and empirical understanding of the kinds of environmental effects that could result from transgene movement and the conditions under which such effects would be likely to occur. Many novel transgenic organisms are likely to be developed, and it will be useful to fund research to identify and investigate environmental hazards associated with a range of transgenic plants, animals, and microbes (NRC, 2002a; 2002b). · develops reliable, safe, and environmentally sound bioconfinement (Recommendation 14). Clearly, this is especially important for GEOs that have a high potential for escape, such as perennial plants (turfgrasses and trees), aquatic organisms, insects, microbes, and viruses. The need for con- tinued research on bioconfinement of genetically engineered crops was noted in an earlier report of the National Research Council (2002a). This committee suggests that a special case can be made for research aimed at identifying and developing new hosts for transgenes involved in the production of chemicals and pharmaceuticals. Those hosts should have features that prevent them from threatening the environment and its ecol-
BIOLOGICAL AND OPERATIONAL CONSIDERATIONS 197 ogy, biodiversity, and the food and feed supply. For example, tropical plant hosts that have no known temperate relatives and an inability to overwinter in temperate regions might be grown in those regions in the summer. Those plants also could be grown in greenhouses year-round, and their ability to escape would be limited by winter temperatures. The research could involve everything from basic investigations of the biology and genetics of the new hosts to their cultivation, harvest, and processing. By its nature this would be long-range and high-risk work, and it is therefore unlikely to be attrac- tive to private industry. The committee notes that such research would be expected to lead to the development of new niche crops and to new indus- tries to process the products. Furthermore, the committee also recommends support for scientific research that · identifies and develops methods and protocols that assess the efficacy of bioconfinement (Recommendation 15). It is important to know how a given method performs in various environments and across different spatial and temporal scales. In this contextand for environmental and safety studies as wellnew methods and approaches are needed for monitoring confinement. Easily identifiable markers for GEOs would be particularly useful. · identifies economic, legal, ethical, and social factors that might influence the application of particular techniques, as well as their regulation (Recommendation 16). Evaluating confinement requires a multidisciplinary approach that includes the natural and social sciences. No collection of such expertise exists in the U.S.--or anywhere else--to the committee's knowledge, and the use of social science information in this area is particu- larly weak. Specific issues on which research could shed useful light include social factors that affect the significance of potential hazards; behavioral patterns of those who grow or use GEOs with respect to the willingness or ability to apply appropriate bioconfinement techniques; ways to reduce human error and instill a strong confinement ethic in those engaged in bioconfinement; and ways the federal regulatory system could be simpli- fied, strengthened, and made more credible. · develops a better understanding of the dispersal biology of organisms targeted for genetic engineering and release, where sufficient information does not exist or where questions have arisen (Recommendation 17). In particular, the following issues have been neglected: seed dispersal patterns, the significance of rare long-distance dispersal, the population genetic impacts of repeated and unilateral migration, the relative fitness effects of transgenes in introgressed organisms, improved verification technology, and the theoretical and empirical bases of monitoring. · develops a better understanding of invasion biology (Recommenda- tion 18). In particular, the ability to predict invasiveness is weak. Research
198 BIOCONFINEMENT OF GENETICALLY ENGINEERED ORGANISMS should address what ecophysiological changes or other phenotypic alter- ations pose significant risks of increased invasiveness, and thus will inform regarding the assignment of traits for which confinement will be necessary. This research should include reviews, analyses, and experimental tests of fundamental assumptions of ecological and evolutionary principles. The committee briefly considered the ramifications of increased cooperation between public- and private-sector researchers. Intellectual property issues permeate current agricultural research and development, especially in biotechnology (NRC, 2003a), and research on bioconfinement methods is no exception. Indeed, development of bioconfinement methods to protect investment by preventing unlicensed use of GEOs has spurred much industry research in recent years (Chapter 3). Continued private support of applied research on some approaches to bioconfinement is to be expected, and the result should be the development of increasingly sophis- ticated and reliable methods. Various changes during the past 20 years in U.S. law concerning intel- lectual property rights--together with political, economic, social, scientific, and technological developments--have led to increased collaboration between private industry and publicly funded research institutions. This has had favorable consequences, including bringing useful products to market more rapidly, raising new funds for public research and education, introducing academic scientists to the challenges of product development and the regu- latory approval system, and providing access for academic researchers to proprietary information held by private industry. However, the increased mixing of public and private support has the potential to compromise fundamental agricultural research at public institutions--research which can be aimed uniquely at the public good (NRC, 1997; 2003a). The long- term effect could be to hamper the growth of the very research and innova- tion base upon which industry will rely (NRC, 1997). The committee emphasizes the need for scientists and administrators of publicly funded research programs to devise ways to work with industry for the public good, while at the same time recognizing their unique roles and importance in biological research.