Case Studies of APHIS Assessments
Chapter 3 provided an overview of the regulations and general procedures used by the Animal and Plant Health Inspection Service (APHIS) to review the large number of transgenic plants coming through the research and development pipeline each year. This chapter examines in detail the specific procedures and judgments made by APHIS in its assessment of a set of cases that have moved through the notification, permitting, and deregulation pathways. The specific set of cases examined was chosen in order to cover a broad array of products, procedures, and potential risks. For each case the types of risks considered by APHIS are noted, and the information and processes used by APHIS in making judgments about these risks are assessed. Risks not considered by APHIS also are pointed out. For all of the case studies, the degree of public and external scientist involvement in the decision-making process also is assessed.
First presented is a case study for a transgenic plant that was field tested through the notification process and one that went through the permitting process. Then four types of transgenic plants that have been deregulated by APHIS are examined. Chapter 5 develops a more general assessment of APHIS oversight and makes recommendations for specific changes.
NOTIFICATION PROCESS CASE STUDY
Notification for Salt- and Drought-Tolerant Bermudagrass
Bermudagrass (Cynodon dactylon and a few related species) is an important grass of lawns and pastures in the United States and elsewhere
(Simpson and Ogorzaly 1995, Taliaferro 1995). Consequently, strategies have been sought to improve the performance of this plant under a variety of environmental stresses (Cisar et al. 2000).
Cynodon dactylon is also considered one of the world’s worst weeds (Holm et al. 1977), especially in the tropics and subtropics but also in warmer parts of temperate zones. It is an especially important weed of sugarcane, cotton, and corn. It is a troublesome weed in some parts of the United States; for example, in the West it has been described as “posing a serious threat to crop production and turf management” (Ball et al. 2000). The species is wind pollinated and reproduces by seed but more frequently by vegetative spread of plant parts (stolons and rhizomes).
Since 1999 APHIS has received and acknowledged the eligibility of five notifications from Rutgers University for New Jersey field tests of salt- and drought-tolerant bermudagrass. One of the notifications (99-308-10n) is discussed here.
The notification application names the transformed organism as a bermudagrass hybrid, Cynodon dactylon × C. transvaalensis. The application also describes the mode of transformation (in this case, particle bombardment). The added genes also are detailed. The gene inserted to confer possible drought and salt tolerance was betaine aldehyde dehydrogenase from Atriplex hortensis, a plant species that shows considerable drought and salt tolerance. The promoter for that gene was maize ubiquitin. The terminator was nopaline synthase polyadenylation sequence from Agrobacterium tumefaciens. The plants also are transgenic for a selectable marker, hygromycin B phosphotransferase from the bacterium Streptomyces hygroscopicus with a rice actin promoter and a 35s polyadenylation sequence from cauliflower mosaic virus (CaMV) for a terminator. The specific planned introduction was a field test at Rutgers University Horticulture Farm II. The applicant certified that the regulated article “will be introduced in accordance with the eligibility criteria and performance standards set forth in 7 CFR 340.3.”
Because the primary concern for the notification process is containment, adherence to performance standards is important. And because the transformed organism is closely related to an important weed and contains transgenes that might confer an advantage to that weed, that containment is not just an academic exercise. Indeed, the applicant took containment very seriously when reporting his containment procedures to APHIS in a letter dated March 26, 2001. The transformed organism is the variety “TifEagle” (Hanna and Elsner 1999), used primarily as a turfgrass for putting greens in golf courses. The variety is a triploid bermudagrass that is both male and female sterile. Thus, dispersal by pollen and seed does not occur. The field test involved a comparison of the performance
of the transgenic grass to nontransgenic strains in an area considerably less than an acre. Twice weekly mowing maintained the grass at a height of 5/32 inches. The borders of the field test were maintained by application of an herbicide treatment every two weeks. The applicant also states that TifEagle cannot withstand central New Jersey’s cold winters but that the plots would be monitored to see if the transgenic plants had developed winterhardiness. The applicant does not describe other methods for preventing accidental spread by fragments of the grass that may attach to equipment or shoes. But the ultradwarf, dense-growing nature of this particular variety probably makes such fragmentation extremely unlikely—especially compared to the easily broken, rambling runners of the wild type.
Environmental Risks Considered by APHIS
APHIS does not conduct environmental assessments on notifications, which are assumed to be safe based on meeting the notification criteria and based on using plant-specific performance standards that minimize any chance of plant or gene escape beyond the confines of the field plot.
Involvement of Potential Participant Groups
There is no public or external scientific involvement for this or any other plant that goes through the notification process.
THE PERMITTING PROCESS
Permitting of Maize-Expressing Proteins with Pharmaceutical Applications
This case is a permit application (00-073-01r, dated March 8, 2000) in which the applicant (ProdiGene) requested permission to grow maize transformed with one or more transgene-expressing proteins with pharmaceutical properties (with the date of intended release 60 days later). The specific phenotype is listed as “antibody production in seed.” A description of the transgenic plant was not available because it is confidential business information (CBI). The purpose of the permit was to grow the transformed maize for seed increase and genetic improvement. The test plots, totaling no more than 2 acres, were to be grown in Nebraska.
Environmental Risks Considered by APHIS
Although APHIS decided that this application did not require an environmental assessment, the permit application itself provides information that addresses a number of environmental issues (see Chapter 3). The following information was gleaned from the permit application and from APHIS’s letter giving notice of its review to the state of Nebraska.
Environmental Impact Related to Donor. Maize was transformed with four genes, one or more of which encode a protein or proteins (one or more antibodies) with pharmaceutical properties. The identities of the genes, their enhancers, their products, and their sources were marked CBI and therefore are not available to the public. The plants were also transformed with the selectable marker, the gene coding for maize-optimized phosphinothricin acetyl transferase (moPAT) derived from the bacterium Streptomyces viridochromogenes, a nonpathogenic soil bacterium. The selectable marker results in tolerance to the herbicide glufosinate. That promotor and terminator for that gene were from CaMV. APHIS concluded that none of these genes contained any inherent plant pest characteristics. APHIS also concluded that the promoter and terminator from CaMV cannot cause plant disease by themselves or in conjunction with any of the genes introduced into the maize plants.
Environmental Impact Related to the Vector and Vector Agent. Transformation was facilitated with Agrobacterium tumefaciens. This is a well-characterized transformation system resulting in stably integrated and inherited transgenes. APHIS found no inherent environmental impact.
Quarantine of Organism and Final Disposition. The focus of the permitting process is to ensure that the transgene escape does not occur by the dispersal of seed, pollen, or plants and that transgenic plants do not persist after the experiment ends. The applicants stated that an isolation distance of 1,320 feet would be used to minimize transgene flow by pollen (this is double the 660-foot isolation distance recommended in the APHIS 1997 user’s guide). The applicants stated that all seed would be harvested and that the plants would be plowed into the soil. The field would then be monitored for volunteers, which would be destroyed by hand or with an appropriate herbicide. APHIS concluded that these measures were adequate to confine the transgene and the transgenic organism. Supplemental permit conditions included monitoring by the appropriate officials, the reporting of field data to APHIS within six months of the termination of the field test, monitoring the field for one year after the test for volunteers, and notification of any changes in protocols.
An issue taken at face value is the isolation distance requirement for corn, a highly outcrossing, wind-pollinated species. Although corn has no weedy or wild relatives in North America with which it can freely cross, isolation between the transgenic corn and other corn crops remains an issue. Isolation distances are set to prevent some minimum level of contamination but were not set up to provide for zero levels of contamination. And zero levels are what would be needed to absolutely prevent escape of the transgene into the environment. For example, the APHIS-recommended isolation distance of 660 feet for corn is presumably derived from that required by the U.S. Department of Agriculture (USDA 1994a) for producing foundation seed (used for seed increase); the maximum proportion of contamination is 0.1%. There is no reason to assume that absolute isolation should be attained at twice that distance. It is likely there would be some very low level of contamination of any corn grown at or near the 1,320-foot isolation distance from the test plots. If adjacent corn were grown for a purpose such that its seeds were not replanted, there would be no permanent escape into the environment. However, as outlined in Chapter 2, some consider that the risk of these genes entering the food supply should be considered an environmental risk.
If contaminated corn were grown such that its seeds were to be replanted, it is possible that the transgenes (for antibody production and glufosinate tolerance) could end up in the genetic stocks. One possible example is adjacent plots of other experimental corn varieties grown for seed increase prior to commercial use. Another example could be nearby plantations of open-pollinated corn grown by a farmer who keeps the seed from year to year. Either of these scenarios could result in perpetuation of the transgenes indefinitely unless they were lost from the breeding stock by random drift. Whether the transgenes for antibody production in seed have an impact in a different corn crop depends on a variety of factors, including the specific pharmaceutical compound created through expression of the transgene; the levels at which that compound is created and stored in tissues that might be consumed by humans, farm animals, or non-target organisms; and the level at which contamination has occurred, and the threshold effects of that compound if used for animal or human food. In this case, because identity of the pharmaceutical compound is not given in the application because it is CBI, it is not possible to judge that impact.
Whether the transgene for glufosinate production has an impact in corn-breeding stocks contaminated by pollen would depend on whether the contaminated stocks would find themselves under selection by that herbicide. The environmental impact of glufosinate tolerance in corn is discussed at length in the Bt corn case study below.
Involvement of Potential Participant Groups in Decision Making
Because APHIS did not see a need for conducting an environmental assessment before approving this introduction, there was no opportunity for public or external scientific involvement in this permit decision.
PETITIONS FOR DEREGULATED STATUS: FOUR CASE STUDIES INVOLVING SIX PETITIONS
Two Virus-Resistant Squash Petitions
Virus-based diseases can sometimes pose important problems for crop production (Hadidi et al. 1998). Virus resistance may be transferred into a crop via conventional breeding methods but only if that resistance already exists in the crop or in a sexually compatible relative. Transgenic virus resistance provides an opportunity for disease resistance in crops whose close relatives are not resistant to the virus in question. Transgenic virus resistance can be obtained by introduction of part of the disease viral genome into the susceptible plant genome; in particular, expression of the viral coat protein (CP) often confers resistance (Powell-Abel et al. 1986, Grumet 1995).
Field trials of dozens of crop species with transgenic-based virus resistance have been conducted (see “Field Test Releases in the U.S.,” Information Systems for Biotechnology online database: www.nbiap.vt.edu). As of April 2001, APHIS had approved six petitions for the deregulation of transgenic crops with virus resistance (see “Current Status of Petitions,” APHIS website: www.aphis.usda.gov/biotech/petday.html). The deregulated crops transformed are papaya, potato, and squash. The case of deregulation of Upjohn/Asgrow’s virus-resistant crookneck squash varieties (application numbers 92-204-01p and 95-352-01p) exemplifies how APHIS evaluated a number of different issues associated with the biosafety of a transgenic product.
Viral diseases are periodically an important problem for growers of squashes and other cucurbit crops (Desbiez and Lecoq 1997). These diseases include those caused by zucchini yellow mosaic virus (ZYMV), watermelon mosaic virus 2 (WMV2), and cucumber mosaic virus (CMV). Aphids act as the vectors of all three viruses.
Interestingly, squash varieties with genetically based virus resistance to WMV2 and ZYMV were developed almost simultaneously by both transgenic and conventional methods. In the same year, 1994, Harris Moran released a conventionally bred virus-resistant zucchini (Tigress)
and Asgrow’s transgenic virus-resistant yellow crookneck squash (ZW-20)—whose resistance was based on expression of viral CP genes—was deregulated (USDA 1994b, Schultheis and Walters 1998). Subsequently, Asgrow’s second transgenic yellow crookneck squash (CZW-3)—containing viral CP genes to confer resistance to WMV2, ZYMV, and CMV—was deregulated in 1996 (USDA 1996). Conventionally created CMV-resistant marrow squash is commercially available from Thompson and Morgan (USDA 1996). As of January 2001, APHIS had received 66 notifications and permit applications for squash varieties with transgenic resistance for as many as five viruses (see “Field Test Releases in the U.S.,” Information Systems for Biotechnology online database: www.nbiap.vt.edu).
APHIS’s action on the Upjohn/Asgrow petition for ZW-20 squash received considerable comment from the public, both pro and con. By the time APHIS made its final decision, the agency had published three Federal Register announcements and conducted a number of public meetings. Feedback on the petition is detailed below under “Involvement of Potential Participant Groups.” APHIS provided a detailed response to commenters who disagreed with its ruling in Response to the Upjohn Company/ Asgrow Seed Company Petititon 92-204-01 for Determination of Nonregulated Status for ZW-20 Squash (USDA 1994b).
The petition for CZW-3 did not generate controversy. APHIS received only a few comments, all favorable to the petition, in response to a single Federal Register announcement. Nonetheless, APHIS detailed its findings by covering much of the same ground as that for ZW-20 in Response to the Asgrow Seed Company Petition 95-352-01 for Determination of Nonregulated Status for CZW-3 Squash (USDA 1996).
Both APHIS response documents on the transgenic virus-resistant squashes considered a number of potential risks in some detail. Below is a highly abstracted overview of APHIS’s arguments for finding “no significant impact.”
Environmental Risks Considered by APHIS
Disease in the Transgenic Crop and its Progeny Directly Resulting from the Transgenes, Their Products, or Added Regulatory Sequences. Some of the DNA sequences used in transforming these squashes were derived from Agrobacterium tumefaciens (the agent of crown gall disease), but the disease-causing genes were removed. Likewise, the viral CP transgenes and additional viral regulatory sequences to control their expression were all derived from disease-causing organisms, but they and their products do not cause disease. CZW-3 also has a selectable marker for kanamycin resistance; APHIS did not consider whether that transgene and its prod-
ucts will cause disease CZW-3, presumably because the source of this gene is not a pathogenic organism. Thus, APHIS concluded that the transgenes, their products, or added regulatory sequences do not result in plant pathogenic properties (USDA 1994b, 1996).
Evolution of New Plant Viruses. For both ZW-20 and CZW-3, APHIS addressed the risks that other viruses would appear with altered host specificities (via transcapsidation, when one virus bears the coat protein of another; also known as “heteroencapsidation,” “genomic masking,” or “masked viruses”) or evolve increased virulence (from recombination with virus-derived transgenes; Matthews 1991).
With regard to the first issue, APHIS pointed out that mixed infections by plant viruses are not uncommon, that these viruses are common viruses of squash, and that transcapsidation is already occurring in infected plants. Given that the amount of coat protein in the transgenic squashes is considerably less than that in naturally infected plants, the chances of transcapsidation are lower in transgenic plants than infected ones. Furthermore, APHIS pointed out that even if masked viruses (i.e., viral nucleic acids enrobed with a coat protein of a different virus produced by the transgenes of the plant) were produced, they would have biological properties identical to those produced in naturally infected plants (USDA 1994b, 1996).
With regard to evolution of virulence via recombination, APHIS concluded that:
because the viral transgene is derived from virus that naturally infects the squash host, is synthesized in the same tissues as in the naturally-infected plants, is produced in less concentration than during natural infections, and if a recombinant was formed would have to be competitive with other squash-infecting viruses. APHIS believes that even if a recombinant virus did occur that [sic] this virus could be managed just like the numerous new viruses that are detected every year in the United States. (USDA 1996; cf. 1994b)
APHIS addressed two additional issues for its assessment of CZW-3: the release of subliminally infecting viruses (those unable to move from the initial site of infection) and synergy (the increased severity of symptoms from multiple infections; Matthews 1991). The agency addressed the concern that infection from a different virus may release subliminally infecting viruses. In the case of plants expressing CP genes, the worry is that those genes might facilitate movement out of the transgenic plants. But “since the CP transgenes in CZW-3 are all from viral strains that routinely infect the curcubit family, it is not expected [that] subliminally infecting viruses will present a problem any more serious than can occur
in naturally infected squash plants” (USDA 1996). Synergy was not considered an environmental risk but rather an agronomic problem. “Asgrow inoculated CZW-3 with several common squash-infecting viruses. No synergistic symptoms were seen in infected plants” (USDA 1996).
In both response documents, APHIS concluded that the transgenic squashes should pose no greater risk of evolution of new viruses than naturally infected plants. In the second document, that conclusion was extended to cover the case of wild relatives that pick up the transgenic traits through introgression.
Increased Weediness in the Transgenic Squash Relative to Convention ally Bred Squash. For both ZW-20 and CZW-3, APHIS addressed the risk that the virus resistance genes would increase the weediness of yellow crookneck squash. Yellow crookneck squash is not listed as a common or troublesome weed anywhere in the United States; for example, it is not on the Weed Science Society of America’s “Composite List of Weeds” (available online at http://ext.agn.uiuc.edu/wssa/). Squash volunteers occur adjacent to squash production fields and, if necessary, are controlled mechanically or with herbicides. They do not readily establish as feral or free-living populations (USDA 1994b, 1996).
For ZW-20, Upjohn/Asgrow supplied APHIS with data comparing the transgenic squashes with their nontransgenic counterparts, showing “no major changes in seed germination, cucurbitin levels, seed set viability, susceptibility or resistance to pathogens or insects (except ZYMV and WMV2), and there are no differences in overwintering survivability” (USDA 1994b). For CZW-3 the APHIS response document stated that “Asgrow has reported that there are no major changes in CZW-3 performance characteristics (except for resistance to CMV, ZYMV, and WMV2)” (USDA 1996). Given that the transgenic squashes would be expected to be grown in the same regions as squash is typically grown, APHIS concluded that “there is no evidence to support the conclusion that introduction of virus resistance genes into squash could increase its weediness potential. Many pathogen and insect resistance genes have been introduced into commercial varieties of squash by conventional means in the past without any reports of increased weediness” and noted that conventionally improved cultivars having resistance genes to viruses had already been developed. In both response documents, APHIS concluded that the virus resistance transgenes are unlikely to increase the weediness of yellow crookneck squash (USDA 1994b, 1996).
Impact on Non-target Organisms Other Than Wild Relatives. APHIS pointed out that both ZW-20 and CZW-3 transgenic squash plants have
no direct pathogenic properties. The protein products of the transgenes are already present in high concentration in naturally infected plants, and the levels of cucurbitin—a naturally occurring plant defensive compound—are likely to be unchanged. Therefore, APHIS concluded that “there is no reason to believe deleterious effects on beneficial organisms could result specifically from the cultivation of” the new transgenic squashes. APHIS noted Upjohn/Asgrow taste tests for cucurbitin levels, but otherwise its conclusions were based on the fact that the coat proteins present in the transgenic squashes are already present in the environment in virus-infected plants. In the second determination, APHIS briefly examined the issue of whether insecticide usage might be reduced by the introduction of the CZW-3 but did not reach a conclusion.
Impacts on Free-Living Relatives of Squash Arising from Interbreeding. Most of the APHIS discussion in the response documents, particularly the 1994 one focuses on whether wild relatives could benefit from virus-resistance alleles, leading to the evolution of increased weediness. The effort was, in part, in response to several negative comments received after the three APHIS Federal Register announcements associated with ZW-20. Many of the comments questioning the decision did so because it marked an important APHIS precedent. As noted in Chapter 2, the sexual transfer of beneficial alleles from a transgenic crop to a wild relative might result in the evolution of a more difficult weed. This issue is perhaps the most widely discussed risk associated with transgenic crops (e.g., Colwell et al. 1985, Goodman and Newell 1985, Snow and Moran-Palma 1997, Hails 2000).
This case study has all three elements that could create such a risk— transgenes of a type that could confer a fitness boost in the wild, a sexually compatible wild relative, and the fact that the wild relative has been classified as a weed. If the crop mates with the wild relatives introducing virus resistance into wild populations and if the primary factor limiting the aggressiveness of wild populations is disease caused by the same viruses, introgression of the transgenes could result in increased weediness of the wild relatives. To obtain more information on the relevant biology of the wild relative, APHIS commissioned a report on the risks that might be posed by crop to wild gene flow by Hugh Wilson, an expert on cucurbit taxonomy and ecology. Wilson (1993) concluded that free-living Cucurbita pepo (FLCP) is a significant weed that might benefit from protection from ZYMV and WMV2. Key information on squash and its weedy North American relatives is summarized below with APHIS’s conclusions and the committee’s evaluation of those conclusions.
Yellow crookneck squash belongs to the species Cucurbita pepo. As discussed above, the squash itself is not a weed. However, squash freely crosses with wild weedy plants known as Texas gourd (originally classified as C. texana but now considered a subspecies of C. pepo) and as FLCP. An experiment by Kirkpatrick and Wilson (1988) demonstrated that squash and FLCP naturally hybridize freely; the crop sired 5% of the seed set by FLCP, growing 1,300m from cultivated squash. Hybrids between the crop and FLCP are fully fertile (Whitaker and Bemis 1964). Clearly, if the crop and FLCP grow in the same region, natural hybridization will occur, and crop alleles will readily enter the natural populations. Indeed, cultivated squash and FLCP co-occur in many regions of Texas, Louisiana, Alabama, Mississippi, Missouri, and Arkansas (Wilson 1993). APHIS concludes that natural hybridization will move the virus resistance genes from the transgenic crop to the wild populations (USDA 1994b, 1996).
FLCP is an agricultural weed in cotton and soybean fields. At one time it was one of the 10 most important weeds in Arkansas (McCormick 1977). APHIS contacted three weed experts for their opinions on the current status of FLCP as a weed. The three experts noted that FLCP plants appeared to be “less a problem” in 1994 than during the 1980s because of new herbicides not available in the 1980s and suggested that new herbicide-tolerant crops would “further expand the tools for effective control of FLCP plants” (USDA 1994b). APHIS concluded that “FLCP plants are not a serious weed in unmanaged or agricultural ecosystems [because] “the registration of new herbicides now allows effective management of these plants” (USDA 1994b). However, two of the three weed experts reported that FLCP is less of a problem; it is not clear how serious a weed they still consider it to be.
The key issue raised by the foregoing data is whether virus resistance genes will provide enough of a benefit that FLCP becomes a more difficult weed. Wild and cultivated C. pepo are susceptible to the same viruses (e.g., Provvidenti et al. 1978). To determine whether viruses limit the population size and number of FLCP, Asgrow conducted a survey in 1993. Fourteen FLCP populations (two in Arkansas, four in Louisiana, and eight in Mississippi—a severe drought precluded sampling in Texas) in nine locations were visited once (when plants were at maturity); no visual symptoms of viral infection were noted. Some of these sites were within a mile of cultivated squash. But it was not reported whether the nearby cultivated plants were infected with virus; that information would have shown whether viruses were present that year. A single plant was sampled from each population. Each plant was subjected to multiple analyses to check for asymptomatic viral infection, and all were found to be virus free. On the basis of these data and qualitative anecdotal reports
from weed experts, APHIS concluded that “there is no scientific or anecdotal evidence to suggest that these viruses routinely infect FLCP plants” (USDA 1996) and that because “FLCP plants are not under significant environmental stress from viral infection, the selective pressure to maintain the virus resistance genes in natural populations of FLCP plants should be minimal” (USDA 1994b).
These conclusions warrant some discussion. One issue is the adequacy of the survey. The committee questions whether a single visit and the small sample of plants are adequate to determine whether a disease is among the important factors that limit population size, number, and niche. The APHIS documents reveal that these viral diseases vary tremendously in the crop from place to place and year to year. Should they be expected to do any differently in closely related plants? Is a single-year survey sufficient? Likewise, is a single visit per site sufficient? APHIS defends the survey as “an appropriate, adequate, and proven means for determining whether a plant is a significant natural host for a particular virus.” The committee questions the phrase “appropriate, adequate, and proven.” Standard sampling procedures of plant epidemiology include sampling several plants per site, repeated visits over time, and a statistically derived basis for determining sample size (e.g., Campbell and Madden 1990). The committee does not see those standards in the survey defended by APHIS. In fact, the timing of the survey might have been important. The committee notes that Provvidenti et al. (1978) found that virus-infected wild plants die prematurely. If wild plants are especially vulnerable to the disease at an early age, by the time the survey was conducted it is possible that the viruses could have wiped out whole populations or that the only plants remaining in populations were those that, for whatever reason, escaped viral infection. The committee also notes that the drop in FLCP frequency reported by the weed experts interviewed by APHIS (see above) is coincident with the arrival and spread of ZYMV in the United States reported by APHIS (USDA 1994b).
Another issue is that of alternative methods for identifying whether virus resistance might provide a fitness boost for FLCP. Some comments suggested experimental approaches for measuring whether virus resistance would confer a fitness benefit to the wild plants. Two types of experiments were proposed:
Inoculating greenhouse-grown plants with ZYMV and comparing the response to infection of the following plants: FLCP, ZW-20, nontransgenic yellow crookneck squash, and F1 and F2 hybrids of FLCP with ZW-20 and nontransgenic yellow crookneck squash. The rationale is that a
change in the weediness potential should be apparent by comparison of plant fitness.
Experiments to exclude the virus from an array of genotypes to determine whether the virus impacts the fitness of FLCP. The rationale is that if the virus is a significant natural enemy, excluding it should result in a fitness boost in FLCP. The comments did not give details of the experiment in the reports to APHIS.
APHIS rejected the “experimental approaches suggested by commenters to determine the impact of movement of virus resistance gene to FLCP plants” as “flawed” (USDA 1994b). APHIS described the limitations of the proposed experiments at length. The first approach was criticized because the greenhouse is an artificial environment. “It is not uncommon for crop plants to be susceptible under controlled conditions to a widely prevalent virus yet [be] rarely infected by that virus under natural conditions.” APHIS concluded that the survey “will provide more reliable information than a greenhouse experimentation based one.” APHIS interpreted the second proposed experiment as one that would involve placing insect-proof cages over plants growing in natural conditions. APHIS criticized that presumed approach as not feasible as well as artificial because cages would exclude all insect pests of FLCP and all insect-borne viruses, not just the two in question.
It is notable that Asgrow itself conducted experiments similar to the type APHIS rejected. Regrettably, the sample sizes in those experiments were too small to draw any conclusions. APHIS also referred to another Asgrow experiment which showed that “FLCP × ZW20 hybrids do not appear to be strong competitors when growing in fields that have not been tilled to remove competing wild plants based on survival and seed set (data report 93-041-01)” (USDA 1994b). APHIS did not state to what the hybrids were compared. The committee does not necessarily support the proposed experimental methods but does question why APHIS is so critical of those methods to measure potential weediness while accepting, apparently without criticism, experimental data of the same general type from Asgrow.
The third issue is whether transgenes will be maintained in natural populations. The APHIS statement above that “the selective pressure to maintain the virus resistance genes” and others in the documents belies an assumption that crop alleles are maintained in natural populations only when they are beneficial. Population genetics theory has demonstrated that even very low levels of gene flow (two successful pollinations per generation) are sufficient to maintain a neutral crop allele at substantial frequencies in a natural population (Wright 1969). Given the hybrid-
ization rate of 5% that Kirkpatrick and Wilson (1988) observed, the committee predicts that neutral and beneficial crop alleles would persist in the FLCP populations. Indeed, at this rate of hybridization even somewhat deleterious alleles are expected to persist in wild populations (Slatkin 1987); that is, a 5% migration rate would maintain alleles that cause a 5% fitness drop. The maintenance of crop alleles in natural populations does not necessarily pose a risk; indeed, the flow of neutral, beneficial, and deleterious squash alleles into natural FLCP populations must have occurred in the past and may still be occurring. However, the APHIS expectation that only beneficial alleles will be maintained in natural populations under gene flow from crops reflects inadequate expertise in population genetics at the time the squash documents were prepared.
APHIS may well be correct that virus resistance would not result in increased weediness of FLCP. But its defense of the Asgrow survey, the “double-standard” for experimental data, and the flawed population genetics arguments are scientifically inadequate to support that conclusion.
Finally, APHIS makes a “substantial equivalence” argument in comparing the product produced by a transgenic method to one produced by conventional, sexual crosses. APHIS notes that because some nontransgenic virus-resistant squashes (see “Background” above) were available at about the same time, it argues that the risks of gene flow of virus resistance alleles from those conventional varieties should be equivalent to the transgenic products in their potential impact on FLCP (USDA 1994b, 1996).
APHIS does not have a mandate to consider the environmental impacts of transgenic plants outside the United States. Because of a negative comment suggesting that ZW-20 would find its way into Mexico, the APHIS response document associated with that regulated article states that “many crucial scientific facts explained in this determination for the United States also apply in Mexico” (USDA 1994b).
Environmental Risks Not Considered by APHIS
As noted in Chapter 2, crop to wild gene flow, transgenic or otherwise, may present an extinction risk to wild crop relatives. Preexisting nontransgenic squashes may already pose that risk to their wild relatives in the United States and Latin America. APHIS did not directly consider whether transgenic squash would have a negative impact on its wild relatives in the United States or Latin America, either in terms of posing an impact that might lead to their extinction (for example, Curcubita okeechobeensis is a federally listed endangered species) or in terms of changes in their genetic diversity. Relative to nontransgenic crops, a transgenic crop could pose an greater extinction risk to a wild relative or,
more likely, alter its genetic diversity if it permits the crop to be grown in closer proximity to the wild plants, thereby increasing interpopulation hybridization rates and detrimental gene flow.
Involvement of Potential Participant Groups
On September 4, 1992, APHIS announced its intent to issue an interpretive ruling on the Upjohn/Asgrow petition that ZW-20 squash did not present a plant pest risk and would no longer be considered a regulated article in the Federal Register (57:40632). During the 45-day comment period, APHIS received 17 comments regarding its proposed ruling; seven were generally supportive of APHIS’s proposed action, and 10 were not.
On March 22, 1993, APHIS published a second Federal Register notice (58:15323) requesting additional information on eight issues raised by commenters to the first notice. At the same time, APHIS commissioned Hugh Wilson of Texas A&M University, a cucurbit taxonomist and ecologist, to prepare a report (see above) related to issues raised in comments to the first Federal Register notice. There were 12 comments to the second notice; 10 were generally supportive of APHIS’s action and two expressed serious reservations. After the close of the official comment period, APHIS received two additional letters urging the agency not to approve the petition.
On May 23, 1994, APHIS published a third notice in the Federal Register (59:266l9–26620) announcing an environmental assessment (EA) and preliminary finding of no significant impact (FONSI) for comment at a public meeting and for written comment during a 45-day comment period. Two individuals, one in favor of the EA and FONSI and one against, spoke and provided written comments at the meeting. During the rest of the comment period, APHIS received 52 additional written comments. Twenty-nine comment letters disagreed with APHIS’s proposal to approve the subject petition; 23 comments supported APHIS’s findings in the EA and FONSI. The affiliations of the final set of commenters were as follows: private individuals (18), universities (12), agricultural experiment stations (11), public policy and public-interest groups (6), industry (2), associations (1), cooperative extension service (1), and federal research laboratory (1). About a third of the final set of comments were detailed and substantial. The affiliations of prior commenters were not published by APHIS.
The second petition for CZW-3 was not nearly as controversial. On February 2, 1996, APHIS published a Federal Register notice (61:3899–3900) announcing that the Asgrow petition was available for public review. APHIS received a total of four comments, all of which were favorable to the petition.
Soybean with Altered Oil Profile
Introduced to North America from China, soybean (Glycine max) produces almost half the world’s vegetable oil. The United States grows almost half the world’s soybean crop (Singh and Hymowitz 1999). Soy is a highly versatile crop, grown for human food, animal feed, and industrial components. It is considered the world’s foremost provider of vegetable oil and protein. Its major product, soybean oil, is used in the production of a wide range of food products, from margarine to salad oil and as an additive in hundreds of processed foods. The seed meal remaining after oil extraction is a nutritious (but incomplete) protein source. In addition, soy is made into tofu, soy sauce, meat substitutes (e.g., “veggie burgers”) and industrial products such as soaps, diesel fuel, and oil-based carriers for other chemicals (Smith and Huyser 1987, Fehr 1989).
Like all vegetable oils, soybean oil is composed of a mixture of several fatty acids. The composition of fatty acids gives the various oils their properties. In developing seeds, fatty acids are synthesized in a known biochemical cascade. Starting with common nutrient building blocks, specific enzymes act to create simple fatty acids, such as the saturated lauric (C12:0) and palmitic (C16:0) acids. Other enzymes can then act on these fatty acids to result in more complex fatty acids, such as stearic acid (C18:0) and the monounsaturated oleic (C18:1) acid, the major constituent of olive oil. Each plant species produces enzymes to act on the fatty acids to synthesize a mixture of fatty acids characteristic of the seed oil for that species. Over half the soybean oil (~54%) consists of the polyunsaturated fats linoleic acid (C18:2) and alpha-linolenic acid (C18:3; ~8%). About a quarter (~23%) of the oil is the monounsaturated fatty acid oleic acid (C18:1). The remainder, about 15%, is saturated fatty acid.
The fatty acid composition of various oil crops has been subject to conventional breeding to create desirable or enhanced properties. Because the incidence of coronary heart disease correlates more strongly with the amount of saturated fat in the diet than with total fat intake (Zyriax and Windler 2000), oils that are low in saturated fatty acids are considered healthier. One example is the oil from linseed flax. Linseed oil is used primarily as an industrial drying oil in paints and varnishes but is also considered a “health food” because of the high proportion (over 60%) of omega-3 polyunsaturated fatty acid (linolenic acid, C18:3). Linolenic acid oxidizes rapidly, which is a good characteristic for paint but not for food because oxidation makes it rancid and unpalatable. Thus, too much linolenic acid causes oil to spoil easily. Using ethylmethane sulfonate-induced mutagenesis, plant breeders in Australia produced linseed flax strains
with substantially reduced linolenic acid (Green 1986). Oil from this mutant is virtually identical to high-quality oils from safflower or sunflower. The mutant linseed varieties are now in commercial production in several countries.
Plant breeders were not as successful developing altered oil-quality soybean genotypes using mutagenesis, although mutant soybean oil varieties do exist (Brossman and Wilcox 1984, Kinney 1994). The exact mechanisms or genotypes of most mutations are unknown. In mutation breeding, plant material is exposed to a mutagenic agent, such as ionizing radiation, in the hope of obtaining a mutation with a desirable phenotype. Mutagenic agents alter the DNA of the subject organism, causing destruction or duplication of a gene or parts of genes. Plant breeders select genetically based phenotypes and rarely characterize mutants at a molecular genetic level. It is almost certain that any mutation resulting in a desired phenotype will also contain mutations at other less obvious genes. These conceivably could affect ecological fitness, nutritional or antinutritional composition, or other important characteristics. In practice, however, such accessory or pleiotropic effects are identified during the breeding evaluation process prior to commercial release.
To avoid some of the problems and uncertainties with mutation breeding, DuPont used transgenic technology to improve the soy oil quality by increasing the proportion of oleic acid from approximately one-quarter to over one-half of the fatty acids. This change was accomplished by inserting another copy of a soybean gene to interfere with the enzyme responsible for converting oleic acid to polyunsaturated fatty acids. Sometimes, introducing an extra copy of a gene results in a phenomenon called cosuppression, the simultaneous silencing of the activity of both the endogenous and the inserted copies of the gene. This enzyme is encoded by a gene called Gm fad 2-1, which catalyses the biochemical reaction converting oleic acid to linoleic acid. DuPont’s strategy was to develop transgenic soybean lines in which the inserted soybean gene interferes with the normal activity of that enzyme through cosuppression, resulting in a buildup of oleic acid and a reduction in linoleic and other polyunsaturated fatty acids. DuPont inserted DNA, including the Gm fad 2-1 gene, into soybean (cultivar Asgrow A2396) using the particle gun delivery method, regenerated transgenic soybean plants, and analyzed them for activity of the Gm fad 2-1 gene.
In the DNA plasmid containing Gm fad 2-1, the gene was linked to the soybean seed-specific promoter of the beta-conglycinin gene and the transcription terminator region from the phaseolin gene of Phaseolus vulgaris (common bean). This plasmid included two additional genes from E. coli: the ampicillin resistance beta-lactamase gene (bla) with a bacterial promoter, which was used in selecting the initial DNA constructs in bacteria
and the beta-glucuronidase gene (uid-A) with the CaMV 35S promoter, which served as a reporter gene in the initial selection of transformed soybean cells. Neither of these E. coli marker genes is active (i.e., they do not generate functional proteins) in the transgenic soybean plants under review. However, the nonfunctional DNA is present in every soybean cell.
A second piece of DNA, a plasmid carrying the dihydrodipicolinic acid synthase (dapA) gene from Cornybacterium glutanicum, also was included in the initial soybean transformation. When active, the dapA gene can increase free lysine content, another potential quality trait in soybean. However, in the soybean transformation event under review, two insertions occurred: one, locus A, consists of the Gm fad 2-1 plasmid and the second, locus B, on a different chromosome, carries both the Gm fad 2-1 plasmid and the dapA plasmid. Subsequent analysis showed the activity of the Gm fad 2-1 transgene at locus A suppressed the endogenous Gm fad 2-1 gene, resulting in accumulation of oleic acid to over 80% of the seed oil. But the activity of the Gm fad 2-1 at locus B caused accumulation of the active enzyme, resulting in the reduction of oleic acid to less than 4% of seed oil. Because reduced oleic acid content was undesirable, conventional plant-breeding methods were used to eliminate the chromosome-carrying locus B (consisting of both the active Gm fad 2-1 transgene and the dapA-containing plasmid). Then, three soybean lines carrying only locus A were selected and designated G94-1, G94-19, and G168. The soybean lines under review therefore lack the dapA gene (as well as the second Gm fad 2-1-containing plasmid). They accumulate oleic acid and do not accumulate additional lysine.
Besides the effects on oleic acid accumulation, DuPont noted changes in the phenotypes of these soybean lines. In addition to a substantial elevation of oleic acid (C18:1) and a reduction of linoleic (C18:2) acid, linolenic acid (C18:3) and palmitic (C16:0) acid, a saturated fatty acid, are significantly reduced. Trace amounts of a linoleic acid isomer also were detected in the oil. In the meal, beta-conglycinin a and a' subunits of the storage protein were replaced by glycinin subunits. DuPont suggested that this change occurred because the recombinant gene in locus A, consisting of the beta-conglycinin promoter fused to the Gm fad 2-1 gene, interfered with both the Gm fad 2-1 gene and the endogenous beta-conglycinin genes. In its petition, DuPont argued, citing Kitamura (1995), that this protein subunit substitution is nutritionally advantageous.
DuPont conducted about 25 confined field trials with the transgenic soybean lines under APHIS notifications in the mid-1990s and an additional trial in Chile over the winter of 1995–1996. DuPont submitted a petition for an APHIS determination of nonregulated status for the soybean lines G94-1, G94-19, and G168, and APHIS approved it. In May 1997, APHIS issued an EA and FONSI on the environment from agricultural
cultivation and use of the DuPont soybean lines (USDA 1997a). Much of the salient discussion and justification for this determination appear in APHIS’s appendix to the USDA (1997a) document.
Environmental Risks Considered by APHIS
APHIS considered the genotypic and phenotypic characteristics of the transgenic soybean lines and how they might pose an environmental hazard in unconfined release. The primary issues include the following:
Potential for the Transgenes, Their Products, or the Added Regulatory Sequences to Present a Plant Pest Risk. APHIS concluded that “neither the introduced genes, their products, nor [the] regulatory sequences controlling their expression presents a plant pest risk in the soybean sublines” (USDA 1997a). Subsequent discussion outlines the basis for this assertion, detailing the composition of the plasmids, the origin and function of each gene segment, the method of insertion (particle bombardment), and the mechanism leading to observed elevated oleic acid content in the seeds. This part also describes the genetic status of the initial transformant, carrying two inserts (loci A and B) as determined by Southern blot analysis, and how DuPont used ordinary plant-breeding methods to select progeny homozygous for locus A but lacking locus B.
DuPont’s analysis of Southern blots over four generations (R1 to R4) led APHIS to conclude there are two copies of the Gm fad 2-1 gene (one endogenous, one inserted) and that they are transmitted to progeny in a normal Mendelian manner. Unspecified evidence was provided that the dapA gene (from the second plasmid and inserted into locus B) is absent.
APHIS concluded that introduction of the vector DNA does not present a plant pest risk in the subject soybean lines because there are no pathogenic DNA sequences present, plus the likelihood of nonsexual transmission of DNA to other organisms is exceedingly small and, even in that eventuality, would not constitute a plant pest hazard. Finally, even though the DNA constructs did use portions derived from known plant pathogens (Agrobacterium tumefaciens and CaMV), those segments were not derived from pathogenic sequences in the respective pathogens. DuPont provided unspecified evidence that no symptoms of either Agrobacterium or CaMV infections were noted during greenhouse and field trial monitoring of the soybeans. Furthermore, DuPont reported no difference in disease and insect susceptibility between the transgenic lines and the parental soybean.
Potential for Increased Weediness of the Subject Soybeans over Conven tional Soybeans. APHIS concluded that the high oleic soybean lines had no significant potential to become weeds. APHIS compared weed charac-
teristics (Baker 1965) to the crop itself, noting that “soybean does not possess characteristics of plants that are notably successful weeds” (USDA 1997a). (See discussion of Baker’s list of weed characters in the next case study.) Citing evidence provided by DuPont, APHIS said the applicant had not observed any significant changes in such characteristics as seed production, germination, standability, overwintering capacity, or pathogen susceptibility that might affect their potential to become successful weeds.
Potential for Outcrossing to Wild Relatives. APHIS noted that there are no free-living close relatives of cultivated soybean in the continental United States but that there are some wild perennial Glycine species in the Pacific territories. However, soybean cannot naturally hybridize with those species. APHIS acknowledged that soybean naturally hybridizes with G. soja, a weed of northeast Asia (including Japan) but not found in the United States (Kwon et al. 1972, Holm et al. 1979). APHIS also noted that soybean is almost exclusively self-pollinating, so hybrids due to natural outcrossing between the subject lines and wild plants is very rare. Even if outcrossing were to occur, they argue, there would be no significant impact because the high oleic trait confers no selective advantage as it does not appear to contribute to enhanced ecological fitness or weediness characteristic to wild populations.
Potential Impact on Non-target Organisms. APHIS considered the possibility that the modified soybeans might have some deleterious effect on other organisms, including beneficial insects and rare or endangered species. APHIS determined that there would be no significant deleterious impact, citing the nontoxic nature of oleic acid and also that the genes and enzymes responsible for the modified phenotype are naturally present in soybeans. The higher concentration of oleic acid is not known to have toxic properties. Oleic acid is a very common food component (and the primary ingredient of olive oil), so there is substantial history of consumption and literature on nutritional and toxicological aspects.
Observations from field trials revealed no negative effects on nontarget organisms, although specific parameters measured were not provided. APHIS also stated that, since there are no novel proteins in the transgenic soybeans, there is no potential for exposing other organisms to new, potentially harmful proteins. The expressed enzyme is common and well characterized in nature, suggesting no potential for harm to beneficial organisms, and the product, oleic acid, should not result in any harm beyond that caused by a high monounsaturated fat diet. No other potential mechanisms for harm to beneficial organisms were identified.
APHIS discussed the indirect metabolic alterations in the transgenic soybeans, including the presence of a linolenic acid isomer, reduction in
beta-conglycinin subunits, and corresponding increase in glycinin subunits. The agency determined that the linoleic acid isomer, present in trace quantities (<1%) in the oil, and the alteration in the seed storage protein subunits should have no adverse effects. The glycinin subunits are common components of soybean and so are unlikely to cause harm. The linoleic acid isomer is also found in other common foods, including partially hydrogenated vegetable oils and human breast milk, so it is unlikely to be harmful in the transgenic soybeans.
Soybean naturally produces antinutritional components, notably trypsin inhibitors, phytic acid, and oligosaccharides. APHIS noted that the presence of antinutritional factors in the transgenic lines were similar to those in conventional soybeans. They also said the presence of DNA from the uidA and bla genes should not have any adverse effects, as they are not expressed in the plant and do not encode infectious agents. APHIS concluded that none of these indirect effects should have any negative impact on non-target organisms.
Potential Impact on Agricultural Commodities. Citing its authority to investigate the potential for plant pest effects, APHIS determined that the subject soybean lines, their components, and their processing characteristics have no indirect plant pest effect on any processed plant commodity.
Based on unspecified evidence provided by DuPont, APHIS concluded that the high oleic soy lines present no threat to raw or processed agricultural commodities. APHIS noted that DuPont consulted with the Food and Drug Administration concerning the safety of consumption of both the oil and meal of these lines by humans and other animals.
Potential Environmental Impacts from Growing the Modified Soybeans outside the United States. Executive Order 12114 (January 4, 1979) allows APHIS to consider the possible environmental impacts of cultivation of regulated articles in other countries. APHIS concluded that the subject soybeans will not have an adverse environmental impact when grown anywhere in the world because there are no significant differences from the parent line for any investigated parameter, except for the higher levels of oleic acid, the glycinin protein substitution, and the linoleic acid isomer. APHIS also noted that “all national and international regulatory authorities and phytosanitary regimes that apply to introductions of new soybean varieties internationally apply equally to those covered by this determination” (USDA 1997a).
Environmental Risks Not Considered by APHIS
APHIS considered the information provided by DuPont and others in reaching its determination, but there is little experimental data in the
decision document or even the appendix to allow an independent critique. One recommendation of another National Research Council study was that “the quality, quantity, and public accessibility of information on the regulation of transgenic pest-protected plant products should be expanded” (NRC 2000c). This recommendation should be expanded to include all new plant products and should be implemented as soon as possible.
Some environmental questions remain. For example, temperate plant species (like soybean) tend to have a higher proportion of polyunsaturated fatty acids in the seed oil than do more tropical or subtropical species (like olive, where monounsaturated fatty acids predominate). Recent studies investigated the effects of fatty acid profile changes on the degree of cold tolerance in plants. Kodama et al. (1994, 1995) transformed tobacco to produce higher amounts of polyunsaturated fatty acids and noted an alleviation of cold-induced growth suppression. If polyunsaturated fatty acids help the seed survive cold winters, perhaps the transgenic soybeans, with such a deficit of polyunsaturates, would have reduced winter survival capacity. The data required to address this question might have been considered and investigated but were not included in the report. APHIS reported that DuPont provided evidence on overwintering but did not elaborate. Possibly, the transgenic soybeans suffered more winterkill than the nontransformed soybeans, but since this segment of the report dealt with the increased weediness potential, that fact might not have been considered important. Also, this question might have been considered irrelevant; that is, reduced overwintering potential of the transgenic soybeans would represent a reduced ecological fitness and hence no increased threat. It is unlikely that the modified-oil soybean changed overwintering capacity because the seed oil was modified, not the plasma membrane lipids, which are more relevant for cold tolerance. APHIS might have been aware of this fact and therefore did not think it necessary to mention it. But by not mentioning the possibility, APHIS leaves itself open to the charge that the agency overlooked it.
An interesting aspect of this case study is the prior presence in the market of nontransgenic soybean varieties with similar (high oleic acid content) attributes. The plant pest risk potential and differences between conventional soybeans and the transgenic varieties are uncertain, not because of incomplete information and regulatory scrutiny of the transgenic lines but because of almost complete lack of information on the conventional ones. The functional assumption is that conventional breeding methods, including irradiation, chemical mutagenesis, somaclonal variation, and other mechanisms of dramatic genetic disruption are safe and environmentally benign.
DuPont provided information on the molecular structure, though not
the nucleotide sequence of the genetic modifications, of the transgenic soybeans. The difference between these soybean lines and regular soybeans is clear; the transgenic soybeans carry a tiny additional piece of DNA, an addition of less than one one-hundredth of 1% of the total DNA in the soybean.
Conventional mutant soybeans (Kinney 1994) manifest an unstable increase in oleic acid in the seed oil. The genetic mechanism by which this trait is achieved is unknown, but it must be the result of genetic modifications in the soy genome. Mutagenesis may have altered the amount of DNA, either by destroying portions of the genome or by causing a duplication (perfect or imperfect) of portions of the genome. In any case, it is likely that several genes were altered, not just those regulating oleic acid content. Any number of genes relating to environmental fitness, production of antinutritional compounds, or other undesirable consequences also may have been altered. But APHIS does not assess these new crop cultivars because, as noted above, the current trigger for regulatory review limits oversight to plants altered using rDNA breeding methods. This is not to suggest that new crop varieties developed solely by conventional breeding should be regulated as stringently as transgenic varieties currently are. Despite the genetic uncertainties and unknown consequences of conventional breeding, as noted in Chapter 1, real damage to the environment from new crop varieties is rare. Nevertheless, the additional knowledge of the genetic changes in transgenic varieties allows, in this particular case study, more confident and reliable predictions of environmental (as well as health and nutritional) effects than conventional varieties. There is no indication that the risks associated with these transgenic soybean lines differ in any material way from those of the same species with similar but non-transgenic based attributes. High oleic acid oil soybean varieties, regardless of how the varieties were derived, appear to present similar degrees of hazard.
Also, APHIS did not consider the effect of large-scale commingling of the high-oleic soybeans with regular soybeans, as that might adversely affect the quality of the processed commodity (soybean oil) but not the plant pest characteristics. The transgenic lines will be grown in a segregated identity-preserved manner to avoid commingling with regular soybeans. Any significant inadvertent mixing of the transgenic soybeans and conventional ones could be problematic because the resulting oil would be a blend of high and low oleic acids, the exact proportions depending on the ratio of the blend. The uncertainty of the final oil composition could adversely affect the value of the commodity oil product. While not an environmental risk, it is a potential impact on other agricultural commodities that are part of APHIS’s charge to review. This is true of all identity-preserved crops being grown.
Involvement of Potential Participant Groups
The designated public comment period on the petition ended, with no submissions, on April 29, 1997. There was no indication from the documents of any other public involvement.
Two Bt Corn Petitions
Insect pests of corn (Zea mays L.) comprise members of many different orders, including seed maggots, rootworms, wireworms, grasshoppers, flea beetles, aphids, and leafhoppers, which feed on kernels, roots, leaves, silks and vascular tissue (Hill 1983, 1987; Davidson and Lyon 1987; Straub and Emmett 1992; Steffey et al. 1999). Some leafhoppers and aphids are vectors for plant pathogens. The main Lepidopteran pests of field corn in the United States are borers (European corn borer, ECB, Ostrinia nubilalis (Hübner); southwestern corn borer, SWCB, Diatraea grandiosella, and corn earworm, CEW, Helicoverpa zea (Boddie) and defoliators (e.g., armyworms). Some of these pests are a chronic problem; others periodically account for severe yield losses (e.g., Ostlie and Hutchison 1997, Mississippi State University Extension Service 1999).
According to various extension service fact sheets for corn growers (Mississippi State University Extension Service 1999, Swanson 2000, University of Illinois 2001), management options for pest control include (1) early planting; (2) crop rotation; (3) resistant varieties; (4) varieties that attract natural enemies of target pests; (5) soil quality management; (6) noncrop management to reduce alternate hosts for pests, e.g., barnyard grass for armyworms, and to enhance naturally occurring predators and parasitoids; (7) conservation of natural enemies through intercropping or insectary plantings of noncrop vegetation; (8) augmentation (field releases) of natural enemies (e.g., Trichogramma spp. egg parasitoids); (9) destruction of corn stalks after harvest; and (10) insecticide treatments. Organically managed corn in the Midwest relies primarily on crop rotation for pest control (Swanson 2000). Price premiums compensate for lower yields under organic management. A relatively small proportion of the total U.S. field corn acreage is treated with insecticides, in part because sprays are often not a reliable or cost-effective means of controlling some of the major pests. For example, less than 5% of U.S. field corn acreage was sprayed for ECB before adoption of Bt corn (Gianessi and Carpenter 1999).
Genetically based pest resistance in corn was developed through the use of conventional breeding methods utilizing mechanical, chemical-repellent, or antibiotic properties of the corn plant. Commercial varieties
with Lepidopteran resistance traits have been available since at least 1902 when Burpee Seeds released “Golden Bantam.” Varieties with antibiotic properties have been developed by selecting for high levels of the toxin 2,4-dihydroxy-7-methoxy-1,4-benzoxazin-3-one (DIMBOA) and related hydroxamic acids, which vary widely in concentration in different corn plant tissues (Houseman et al. 1992). Structural traits such as highly lignified stalks have also been selected to serve for increased pest resistance. However, many of these varieties are variably or moderately resistant and may have relatively lower yields than susceptible corn lines. For example, DIMBOA levels in initially resistant varieties tend to decrease with plant age (Barry et al. 1994, Frey et al. 1997). Thus, some conventionally bred varieties resistant to ECB have been replaced by higher-yielding varieties with increased total yields despite heavier losses from ECB (Gianessi and Carpenter 1999).
Transgenic insect pest resistance can be obtained by transferring part of a bacterial genome (from various strains of Bacillus thuringiensis) into a plant genome. The genetic sequence encodes production of a toxic protein so that plant tissues become lethal to target caterpillar pests feeding on those tissues. At least three major opportunities are afforded by transgenic insect resistance traits in crops: (1) pest resistance in crops whose close relatives are not resistant to the target pests, (2) more effective pest control than pesticides and/or conventionally available resistance, and (3) the trait can be transferred into varieties showing excellent agronomic performance (e.g., Wiebold et al. 2000), thus avoiding the yield drag that sometimes accompanies conventional breeding efforts. For caterpillar pests of corn, all of these factors can be important. Therefore, starting in the 1980s, transgenic Bt corn lines were developed and field-tested for efficacy against target Lepidopteran pests, including ECB, SWCB, and, when toxins are expressed sufficiently in the corn ears, CEW.
Other crops have shown promise for Bt-based insect resistance. By October 2000, 25 petitions for nonregulated status (from Monsanto, Northrup-King, Ciba-Geigy, DeKalb, AgroEvo, and Calgene) had been submitted to APHIS for Bt crops (corn, potato, tomato, and cotton). Sixteen were approved for deregulation of transgenic crops with Bt-based insect pest resistance, seven petitions were withdrawn, and two were pending a decision. Field trials of dozens of other transgenic Bt-based insect-resistant plant varieties (including cotton, potato, rapeseed, poplar, broccoli) have been conducted (see “Field Test Releases in the U.S.,” Information Systems for Biotechnology online database: www.nbiap.vt.edu).
For comparison, one of the earliest Bt corn petitions for deregulation by APHIS and one of the most recent are included in this case study (see “Current Status of Petitions,” www.aphis.usda.gov/biotech/petday.html). Transformation Event 176 corn (Maximizer), with marker genes for resis-
tance to the antibiotic kanamycin and tolerance to glufosinate, was the first Bt corn to be deregulated (USDA 1999). The petition for Bt corn Transformation Event CBH-351 (StarLink) was approved in 1998 (USDA 1998). These two events expressed different kinds of plant-produced toxic proteins. The kind of toxin expressed by Event 176 has been commercially available in formulations of bacterial insecticides, but Event CBH-351 expresses a Cry toxin that has not been available commercially in insecticides. Below are highly abstracted overviews and comparisons of the assessments by USDA-APHIS for its findings of “no significant impact.” To the extent possible without access to the original research designs and test data submitted by the petitioning companies, this summary illustrates how APHIS evaluated a number of different issues associated with the biosafety of these transgenic products.
Environmental Risks Considered by APHIS
Disease Resulting Directly from the Transgenes, Their Products or Added Regulatory Sequences. For Event 176, two promoter sequences were used to allow high levels of Cry1Ab protein expression in both green tissue and pollen, which in combination were expected to be most effective in controlling ECB. Some of the DNA sequences used in transforming these plants were derived from cauliflower mosaic virus (CaMV), but the disease-causing genes of this virus were not involved. The marker gene bar from the soil bacterium Streptomyces hygroscopicus, which encodes for an enzyme resulting in glufosinate tolerance, was used as a selective marker. Plasmid pUC19 also harbors ampicillin resistance but is expressed only in a bacterium, not in plant cells.
Event CBH-351 was developed with two pUC19-based plasmids, which contained the modified cry gene and the bar gene, respectively. As in Event 176, the bar gene was cloned from S. hygroscopicus, and its expression results in whole-plant glufosinate tolerance. Event CBH-351 expresses modified Cry9c proteins, using sequences from Bacillus thuringiensis subsp. tolworthi (isolated from grain dust in the Philippines). Event CBH-351 uses regulatory regions from petunia and two plant pathogens, CaMV and Agrobacterium tumefaciens. In both cases, APHIS concluded that although pathogenic organisms were used in their development, the transgenes, their products, or added regulatory sequences do not result in plant pathogenic properties (USDA 1995, 1998). In the CBH-351 determination document, a discussion follows this conclusion, providing some information on the genetic constructs, including the number of insertion sites, the copy number and expression level of the transgenes, and evidence of stable inheritance of the transgenes in crosses of CBH-351 into several different genetic backgrounds. Unpredicted, unexplained effects
are also mentioned, such as leaf striping on transgenic plants as compared to nontransgenic plants in Puerto Rican field trials and more predictable but secondary effects, such as reduction of stalk rot in transgenic plants.
In contrast to the document for CBH-351, the APHIS determination for Event 176 did not mention or evaluate descriptions of the event regarding pleiotropic effects or other unpredictable consequences of the random insertion of these genetic sequences, the number of copies, and stability of the transgenes. The determination for CBH-351 was more detailed than was the earlier determination, suggesting either that APHIS received more information, adjusted its evaluation procedures, reported more details in its summary, or some combination of the three.
Weediness of the Crop Plant Resulting from the Transgene and Associ ated Gene Sequences. For both Event 176 and Event CBH-351, APHIS addressed the risk that expression of the insect control protein might provide a selective advantage to the plant, sufficient to make it a plant pest. For Event 176, APHIS first compared the characteristics of nontransgenic cultivated corn with a list of ideal characteristics of weeds published by Baker (1965) and found little overlap. Though noting that some ecologists have criticized this list, APHIS relied on it anyway because no more broadly accepted suite of characteristics was available. Second, APHIS assessed the weed status of corn, consulting several weed compendia and found that corn was not listed in most of them. Third, APHIS considered the trait itself—insect resistance. APHIS concluded that since insect resistance was not among the weedy characteristics listed by Baker, that trait would not likely contribute to weediness in transgenic Bt corn. The herbicide tolerance trait of Event 176 was not perceived as one that could reverse the plant’s nonweed status, primarily because glufosinate had not yet been registered for use in corn. Finally, “APHIS evaluated field data submitted by Ciba Seeds which specifically demonstrates that Event 176 corn is no more weedy than the non-modified recipient.” These field data were not available to this committee, so it cannot comment on the adequacy of the study in supporting that conclusion.
APHIS used the same arguments for the later determination (USDA 1998) for Event CBH-351, citing Baker (1965), and noting that insect resistance was not listed as a characteristic of weeds. APHIS also found no differences between CBH-351 and its nontransgenic counterpart or a nontransgenic standard line in field tests conducted from 1995 to 1997 in 17 states and territories of the United States in Bt corn’s ability to compete or persist as a weed. The CBH-351 determination pointed out that other corn genotypes with resistance to certain Lepidopteran pests, including ECB, have been in use for decades and have not been reported to cause increased weediness. Given these data and observations, APHIS concluded
that the introduced genetic constructs and new traits, Lepidopteran insect resistance and tolerance to glufosinate, are not expected to release Bt corn volunteers from any constraint that would result in increased weediness. APHIS added that CBH-351 corn is still susceptible to other non-Lepidopteran pests and diseases of corn.
Although the committee believes that APHIS has come to a reasonable conclusion for this case, APHIS’s analysis lacked scientific rigor, balance, and transparency. First, APHIS’s statement that most weed lists did not include corn surprised the committee because corn was on the first list consulted. Corn is listed as a common or troublesome weed in the United States according to the Weed Science Society of America’s “Composite List of Weeds” (available online at http://ext.agn.uiuc.edu/wssa/). Although the accompanying APHIS environmental assessment listed some of the authorities used to determine that corn is not a weed, the committee suggests that APHIS cite all sources it consults. The environmental assessment should then explain the reasoning for adhering to some but not other authorities.
Second, the herbicide tolerance trait in Event 176 corn was considered not to pose a hazard in terms of “weedier” free-living relatives because glufosinate had not yet been registered for use in corn and exerts no selection pressure for this trait in nature (USDA 1995). There was no discussion about herbicide-tolerant Event 176 corn volunteers persisting where glufosinate is applied, such as adjacent habitats or rotational crops. By the time CBH-351 was evaluated, however, glufosinate was registered for use on corn. The subsequent determination provided a more thorough discussion of the movement of this trait via gene flow and listed possible changes in agronomic practices needed to control resulting corn volunteers, including alternative herbicides or mechanical control measures (USDA 1998). Such herbicide-tolerant volunteers, however, were not considered weedier, even though they would require additional control measures. Later in the determination, however, APHIS implies that if a herbicide tolerance trait were to introgress into teosinte, the descendents would become weedier in the presence of glufosinate herbicide selection. Different conclusions for corn and teosinte descendants could be considered contradictory.
Third, APHIS relies extensively on Baker’s list of weed characteristics in both determinations. Although an excellent heuristic tool for understanding weed ecology, Baker’s list, in itself, is not an effective predictive tool. APHIS mentioned that some ecologists question the effectiveness of using the list for that purpose. If APHIS uses that list, the committee suggests that the agency include more discussion of such studies as Williamson (1994) and Perrins et al. (1992), which demonstrate that known weeds cannot be separated from nonweeds based on that list.
Another argument (used by APHIS in these determinations) is that the “lack of reported incidences” is evidence that something has not happened. For example, the determination for CBH-351 mentioned that increased weediness of corn due to an insect resistance trait has not been reported, as far as APHIS is aware. The committee agrees that if a phenomenon is dramatic and occurs over a short time period, it is likely to be noticed and even reported. That is, a lack of reporting of obvious phenomena may be a strong indicator that such phenomena have not occurred. However, a slight increase in fitness due to insect resistance is an invisible phenomenon, and any increase in weediness of corn due to that trait would likely be slow and subtle, especially since the conventionally bred traits were only moderately effective in conferring resistance.
To demonstrate that corn volunteers are more common as a result of conventionally bred ECB resistance, one would need a detailed comparison between, for example, the incidence of volunteer corn in the 1940s and the 1990s, including a way to isolate the resistance mechanism from other changes in corn production and management. Such an approach could be used to test for environmental effects of pest resistance factors. However, because a “lack of evidence” is used to conclude that there are no effects, there is no encouragement for such testing in the APHIS determinations. In general, use of the term “lack of evidence” can mean anything from “detailed, replicated, long-term experimental studies found no evidence” to “there are reasons to expect a problem but no one has tested this so there is a lack of evidence.”
Finally, the committee found the argument that insect resistance cannot release a plant from ecological constraints to be both weak and inconsistent. Indeed, APHIS is the federal agency that evaluates and approves biological control applications for the importation of insects (including Lepidoptera) specifically to increase herbivore pressure and thereby reduce the incidence of certain weeds. The embedded assumption in the biological control of weeds is that the presence or absence of herbivory, by even one species, can determine whether a plant species behaves as a weed. Biological control was not mentioned in either environmental assessment. A stronger argument for why Lepidopteran resistance would not release corn volunteers would have included an explanation of why corn and its Lepidopteran herbivores are different from those plant-herbivore pairs that result in biological control of weeds. (APHIS cited Gould  as saying that corn is incapable of sustained reproduction in feral populations.)
Impacts on Free-Living Relatives of Corn Arising from Interbreeding. APHIS considered the potential for gene flow from Event 176 transgenic corn and, should it occur, two of its possible consequences. Those conse-
quences were increased weediness in wild relatives and population changes that would lead to reduced genetic diversity. Glufosinate resistance was not considered to pose a hazard in either regard, first because selective pressure for this trait would not occur either in natural habitats or cornfields, where glufosinate was not registered for use (USDA 1994a), and later because alternative management techniques for glufosinate-tolerant corn exist (USDA 1998).
First, however, APHIS discussed the possibility for introgression from transgenic corn into other corn cultivars via wind pollination. Gene flow would be reduced, according to USDA (1995), by low survival rates of corn pollen after 30 minutes. Some controversy exists now, however, about corn pollen lifetimes; the EPA (2000b) claims corn pollen is viable for several hours in good conditions. APHIS pointed out that maintaining an isolation distance between cultivars also reduces gene flow, but no particular separation distance is suggested. Additionally, APHIS noted that the primary use of corn kernels is food or feed rather than seed; thus, any hybrid seed resulting from cross-pollination with Bt corn would more likely be eaten than sown. APHIS did not discuss any indirect effects of cross-pollination on the food quality of recipient non-Bt corn. This eventually became a concern for the CBH-351 (StarLink) Event. As discussed in Chapter 2, there is disagreement about when gene flow to other non-propagating plants should be considered an environmental risk.
The analysis of Event 176 for impacts on corn relatives was not conceptually different from the analysis of corn’s weediness. No other wild relatives, in the United States or abroad, were mentioned in this section of the determination despite the fact that sexual transfer of beneficial alleles from a transgenic crop to a wild relative that might result in a more difficult weed had been a widely discussed risk associated with transgenic crops prior to that time (e.g., Colwell et al. 1985, Goodman and Newell 1985, Darmency 1994, Kareiva et al. 1994). The environmental assessment for deregulation of Event 176 noted that hybrids between corn and Tripsacum weeds are sterile. For wild Zea (teosintes), which are native to Mesoamerica, the environmental assessment states that “in the wild, introgressive hybridization does not occur because of differences in flowering time, geographic separation, block inheritance, developmental morphology and timing of the reproductive structures, dissemination, and dormancy (Galinat 1988).”
The environmental assessment for CBH-351 discussed the possible avenues for transgene introgression into populations of corn’s wild relatives to a greater extent. It mentioned that corn’s wild relative, Eastern gamagrass, Tripsacum dactyloides, which is native to the United States, is grown as a new crop in the Midwest. The strong statement against the occurrence of introgressive intrageneric hybridization for Event 176 was
altered for CBH-351 with less absolute language (e.g., “does not occur” was replaced with “is limited, in part, by several factors” citing Doebley  as well as Galinat ). The statement in the environmental assessment for Event 176 regarding the sterility of Zea × Tripsacum hybrids was revised for Event CBH-351 to the effect that resulting hybrids are often sterile or have greatly reduced fecundity. The latter information suggests that hybrids with some fertility might occur at a very low frequency if Tripsacum and field corn are grown in close proximity, such as in the Midwest. For both Event 176 and Event CBH-351, APHIS concluded that environmental impacts anywhere in the world should not be significantly different from those arising from the cultivation of any other variety of insect- and herbicide-tolerant corn.
The key questions posed by APHIS in the foregoing two analyses are: (1) Is it likely that the Bt endotoxin trait would introgress into at least one population of a wild relative of corn? (2) Can pest resistance genes against certain Lepidopterans provide sufficient benefit to result in making corn or its hybrids a more difficult weed? (3) Are Bt varieties similar enough in conventionally bred varieties to use the latter as a model for the former? Whereas the absence of selection pressure by herbicides was mentioned for the herbicide tolerance trait and weediness, no clear analysis of release mechanisms for wild relatives was provided by APHIS with respect to herbivore selection pressure.
APHIS’s view of whether introgression of the transgene into wild relatives could occur seems to have changed from impossible for the first Bt corn determination to being unlikely in the United States and likely in Mesoamerica for the most recent determination. This shift in interpretation shows that APHIS adapts its arguments to take into account additional or overlooked scientific information as it becomes available. Certainly, compared to a crop like Sorghum bicolor, which has an abundant cross-compatible wild relative in the United States, corn has fewer avenues for gene exchange with wild relatives in this country. But it apparently has a slight chance given that populations of Zea and Tripsacum species are reported by APHIS (USDA 1998) to occur in the United States.
Morphologically intermediate plants between teosinte and corn appear to be spontaneous hybrids in Mesoamerica (Wilkes 1997), but they could also be crop mimics; actual hybridization and introgression rates are unknown (Ellstrand et al. 1999). Should there be crop-to-wild relative gene flow, in either the United States or Mesoamerica, introgression and maintenance of the transgene can occur even if hybridization rates are low. Despite APHIS arguments that a positive selection pressure must exist to maintain a transgene in a wild population, under modest gene flow pressure, Bt-based resistance could, in fact, be maintained at reasonably high frequencies in populations of wild corn relatives (see ar-
guments above for virus-resistant squash) even if it were a neutral or deleterious allele. That is, relatively rare gene flow events provide an opportunity for persistence, perhaps leading to changes in the genetic diversity of natural populations (Ellstrand et al. 1999). Thus, the consequences of persistence of the transgene in natural populations are worthy of a serious assessment.
When considering whether transfer of a Lepidopteran resistance trait could allow an ecological release (i.e., an increase in population size, density, or range) of any wild relatives of corn, two further questions arise. Which Lepidopteran species frequently use wild and cultivated Zea and Tripsacum as host plants? Which of those species are susceptible to the Bt toxins? APHIS does not provide specific information on these questions. With respect to the range of Lepidopterans that feed on corn and/or its wild relatives, two contradictory statements and an extrapolation from a conference discussion comment are provided. In the determination document for CBH-351, APHIS stated that Cry9C protein has insecticidal activity against ECB and members of the families Pyralidae, Plutellidae, Sphingidae, and Noctuidae. Another statement in the same document states: “CBH-351 corn plants were generally indistinguishable from control corn plants for disease susceptibility and insect susceptibility except for tolerance to European corn borer, where a clear advantage was noted for CBH-351 corn” (USDA 1998). Perhaps these statements seem contradictory because in one case the plants were challenged with various Lepidopteran pests and in the other case plants were exposed to natural levels of pests; however, no such explanation is given in the APHIS documents.
APHIS cited Sánchez González and Ruiz Corral (1997) to suggest that teosinte is susceptible to most of the same pests and diseases (including Lepidopterans) that feed on Zea mays. The statement is actually less clear; in the conference discussion following the paper by Ruiz and Sánchez, Bruce Benz is quoted as saying, “In the case of insects, certain Coleoptera (Macrodactylus murinus) infest both Zea mays ssp. diploperennis and maize, although the impact seems to be different, while in the case of Homoptera and Lepidoptera, even though both species of plants may be infested, it seems that the damage tolerance of teosinte is lower than that of maize.” Lepidopterans feeding on corn’s wild relatives may indeed be primarily those that feed on corn; if so, they may be the very species that limit the ranges and aggressiveness of such wild grasses. However, few records are available in the most comprehensive Lepidopteran host plant database (Robinson 1999, Robinson et al. 2000). Thus, it is premature to make a conclusion or to suggest which species feed on which wild relatives.
Letourneau et al. (2001) list 376 Lepidopteran species recorded as feeding on Zea mays, two feeding on Z. mexicana (Schrad.;=Z. mays ssp.
mexicana), and no records for Lepidopterans feeding on other teosintes (e.g., Z. diploperennis). Of the 376 Lepidopteran species feeding on Z. mays, fewer than 15 are listed as having been tested for commercial Bt toxin susceptibility in the most comprehensive database (van Frankenhuyzen and Nystrom 1999). Most testing has been carried out on target pests; thus, little is known about which nonpest Lepidopterans are susceptible to any or all Bt-based proteins. Even less is known about whether these herbivores limit the population size of corn’s wild relatives. Nevertheless, APHIS found it unlikely that potential introgression of Lepidopteran resistance will cause teosinte to become weedier (USDA 1998).
Whether or not conventionally bred insect resistance traits in corn and the Bt-based resistance trait inserted in CBH-351 corn are similar enough to make conclusions about the selective pressure of the Bt traits on feral progeny or hybrids with wild relatives is unknown. The genetic novelty of Bt toxins and herbicide resistance is likely to be outside the range of natural variation found in populations of wild relatives, whereas some of the traits responsible for causing Lepidopteran resistance in conventional corn hybrids may correspond to traits already present in populations of corn’s wild relatives (e.g., tough stalks, low protein content, high DIMBOA content). Appropriate measures of environmental effects of conventionally bred resistant varieties are not provided by APHIS, though some may be available; at least some discussion of the kinds of resistance mechanisms present in conventional varieties and their relationship to the range of naturally occurring levels of resistance in wild relatives should be included in this discussion. This would aid in determining if the Bt proteins constitute fundamentally different mechanisms of resistance.
While APHIS might be correct that introgression of the Bt-resistance and glufosinate tolerance traits into feral corn and wild corn relatives is unlikely and that Lepidopteran resistance and herbicide tolerance would not result in increased weediness of these plants, APHIS’s data on Lepidopteran pressure in natural populations, their population genetic arguments, definitions of weediness, and model of conventionally bred resistant corn varieties are inadequate to support such a strong conclusion.
Impacts on the Components, Quality, and Processing Characteristics of Corn Raising from the Event of Inserting Btk Insecticidal Protein Gene. APHIS concluded from extensive field tests that the Event 176 and Event CBH-351 corn plants had agronomic characteristics typical of nontransgenic corn such that, except for the desired effect of insect resistance, the introduced sequences did not confer any disease-specific property of the donor organism or any other plant pest characteristics.
Potential to Harm Organisms Beneficial to Agroecosystems. The Btk protein (Cry1Ab) expressed in Event 176 was tested by Ciba Seeds for toxicity to non-target organisms, as either an extracted leaf protein powder, intact pollen grains, or a bacterial cell paste, as well as in a comparative field trial. No information is given on the experimental design of any of these tests, except that the appropriate form of Btk protein described above was used for different non-target organisms, including earthworms, aquatic invertebrate Daphnia magna, and bobwhite quail. The field test focused on predators and parasites in three insect orders and included at least ladybird beetle larvae, a fly, and honeybee larvae. In the section on threatened or endangered species, APHIS summarized results of tests and observations on quail, the aquatic and soil-inhabiting invertebrates, a moth, and a number of butterflies. Bobwhite quail showed no adverse effects on feed consumption, body weight, or mortality when fed corn protein powder from Bt or isogenic non-Bt hybrids. Indirect effects of Bt corn on quail through its changes in its food supply were considered by APHIS to be negligible if the effect of Bt corn is restricted to the European corn borer (however, it is known that Btk affects many Lepidoptera). No data were mentioned on the toxicity of Btk proteins on endangered Lepidopterans, presumably because exposure to the toxins in nature was deemed unlikely. From these data, and data in the literature APHIS concluded that only a specific group of Lepidopterans should be affected and that the majority of quail’s prey species should remain unaffected by Event 176 corn.
APHIS concluded that, because the specific receptors for Bt toxin in the midgut of target insects were not expected to occur in other invertebrates or any vertebrates, Event CBH-351, although not derived from a type of Bt available as a commercial formulation, should not adversely affect non-target taxa, including humans. Tests for effects on non-target organisms for Event CBH-351 (Cry9c) by AgroEvo (formerly Plant Genetic Systems, now Aventis) differed somewhat from tests on non-target toxicity for Event 176. Specifically, test organisms included adult honeybees, ladybeetles, earthworms, Folsomia candida springtails, bobwhite quail, and mice. Details on experimental design are not given, so a scientifically based review of APHIS evaluation of the data is not possible.
The fact that the committee cannot review the actual methodology and data summaries of any of the Bt corn testing is unfortunate for two reasons: (1) the committee’s evaluation will be superficial, as if it were asked to review a scientific report but were given only the introduction and discussion sections without the methods or results, and (2) the committee’s questions may be misguided because of incomplete information provided in the APHIS descriptions of tests and results. Nevertheless, the committee has several questions about what seem to be the methods used
by the petitioners to test for environmental effects of the proteins expressed in Event 176 and Event CBH-351. It is curious that adult honeybees were tested, since adults both ingest pollen and feed it to their brood. APHIS justified the use of adult honeybees by noting that corn is not a nectar-producing plant, and so honeybees would visit it infrequently compared to other plants; thus relatively low concentrations of Bt pollen would be expected in adult pollen loads; the committee remains unconvinced that larval tests are unnecessary. Specific criticisms about the adequacy of Bt tests for lethal and nonlethal effects on natural enemies were discussed by Hilbeck et al. (1998a, 1998b). They noted, for example, that aphids were not an appropriate test prey for determining indirect effects of Bt proteins on natural enemies since the aphids may not ingest Bt proteins. Indeed, thorough testing for non-target impacts of Bt pollen might also include larvae of Lepidopteran species that might encounter corn pollen and feed on it more frequently in nature (Losey et al. 1999, 2001, Wraight et al. 2000).
The committee also questions the ability to detect community-level differences in the comparisons described. Field tests seem to have been conducted using a factorial design, with three replicated split plots of Bt corn and non-Bt isolines with or without pyrethroid insecticide application. Arthropod predator comparisons were based on 5 five-minute visual observations (biweekly) and seven weekly sticky trap samples per split plot. On three of seven dates, trapped predators were significantly more abundant in unsprayed non-Bt corn than on unsprayed Bt corn; no effect of Bt corn was found on the other four dates. Nor were differences found between plants that were and were not sprayed with pesticide. Information to decipher whether the trap samples showing higher levels of predators in non-Bt corn plots reflected predator colonization rates or resident predators emerging as adults from the corn (thus reflecting performance). Visual observations showed no difference in predator abundance between plots with and without Bt corn. Similar to the findings for Ciba Seeds field comparisons, the diversity of predators in experimental plots did not differ between Bt and nontransgenic lines. It is not clear what measure of diversity was used or at which taxonomic level. APHIS mentioned possible problems with the reliability of the field test due to small plot sizes (10 rows by 20 feet). Another problem might be the statistical power of the test to detect differences in predator diversity, especially if many of the arthropods were identified taxonomically at levels higher than species (genus or family).
Hoy et al. (1998) have pointed out that there could be complex indirect effects in crop fields that could cause shifts in the relative abundance of natural enemies. Also, indirect impacts of Bt corn on pests might occur. For example, a reduction in pyrethroid use for corn borers on Bt corn
might minimize outbreaks of spider mites by allowing their natural enemies to survive. In contrast, minor pests may become more predominant as ECB is controlled and if foliar insecticides are reduced (Ostlie and Hutchison 1997). Community-level tests were restricted to above-ground species; given the persistence of Bt proteins in the soil from root exudates or after incorporation of plant material (Stotzky 2000, 2001), any evidence for no significant effect on communities of soil organisms should also be discussed.
Potential for an Adverse Impact on Threatened or Endangered Species. APHIS consulted the list of threatened and endangered species (50 CFR 17.11). None of these species feed on corn, so APHIS concluded that Bt corn would not affect these species. No mention of toxic effects of corn pollen on any sensitive Lepidopteran species was made for either Event 176 or CBH-351. Nor was there mention of any possible effects to nontarget susceptible Lepidopterans that may be dependent on Zea host plants. The latter, though beyond the scope required, may be a proactive consideration for assessing possible environmental consequences should transgenes escape and affect population levels of non-target moths and butterflies, which are also plant pollinators (Letourneau et al. 2001). The possibility that threatened or endangered plants could rely on Bt-susceptible Lepidopteran pollinators was not mentioned. While this possibility may not be high, it is best to be thorough in these assessments.
Potential for an Adverse Impact on the Ability to Control Non-target Insect Pests. In the CBH-351 determination, APHIS briefly examined the issue of whether insecticide usage might be reduced by the introduction of Bt corn Event CBH-351 but did not reach a conclusion (USDA 1995, 1998). Perhaps because the APHIS assessments expected no adverse effects on natural enemies and possibly positive effects due to the curtailed usage of broad-spectrum pesticides, the notion of secondary pests is not discussed directly. If, in response to low levels of target pests, non-target insect pests increased and became secondary pests in Bt corn, those species would need to be controlled by alternative measures.
Effects of the Cultivation of Bt Corn on the Ability to Control Insects and Weeds in Corn and Other Crops. APHIS considered evolution of pest resistance to Cry proteins for Event 176. These proteins are similar to those used for ECB control in commercially available crystalline powder formations. Based partly on experimental demonstrations of resistance evolution in Lepidoptera to Cry toxins, APHIS predicted that resistant insects would probably evolve in response to Bt corn. However, a resistance management strategy was outside the scope of the determination.
APHIS summarized statements from CIBA Seeds—that Bt corn be used within the scope of an integrated pest management (IPM) strategy, that populations of ECB be monitored for resistance, that high-dose expression coupled with non-Bt corn refugia and development and use of new insect control proteins would delay the evolution of pest resistance, and that farmers be educated about resistance management strategies. APHIS suggested that, should these measures fail, resistant ECB populations might be controlled by agronomic practices such as rotation and alternate insecticides. The determination document concluded that evolution of resistance to Bt-based insecticides is a potential risk associated with Event 176 but that this risk was no greater than that posed by applying insecticides themselves. For Event CBH-351, which expresses a protein different from those in commercial formulations of bacterial sprays, similar arguments were tendered. Cross-resistance was considered unlikely due to separate receptors in some species, so the Cry9C protein was suggested as a useful alternative when resistance evolves to other Bt toxins (such as that in Event 176). APHIS concluded that CBH-351 should pose no greater effects in resistance evolution than the use of ECB-tolerant corn cultivars, chemical insecticides, or biological insecticides (USDA 1998). The possible consequences of herbicide tolerance in affecting weed control, addressed only for Event CBH-351, were predicted to be positive, allowing more choice among postemergent herbicides and no-till options.
Although APHIS makes some comments about resistance, the agency has apparently relied on the Environmental Protection Agency (EPA) to formalize and/or enforce resistance management plans, encouraging its consideration. The committee suggests that APHIS either indicate that resistance management is beyond its scope and not discuss it or provide detailed analysis of the practical issues of resistance management efforts in field corn. Otherwise, the outcome can be perceived as different levels of scrutiny between the EPA and APHIS.
Environmental Risks Not Considered by APHIS
APHIS did not directly consider whether transgenic corn would have a negative impact on corn’s wild relatives in the United States or elsewhere, either in terms of changes in their genetic diversity or in terms of posing an impact that might lead to their extinction. This is likely due to the conclusions that hybrids with Zea or Tripsacum would rarely occur, especially in the United States. Mortality of non-target Lepidoptera should susceptible species ingest toxin-containing pollen on their host plants is not discussed. Although threatened or endangered Lepidoptera were considered, the link between threatened or endangered plants and Bt-susceptible Lepidopteran pollinators was not explored.
The determination documents describe comments received on each of the two petitions (Event 176 and Event CBH-351) during the designated 60-day period after posting in the Federal Register. Most (2,271 of 2,309) were form letters (source not specified), and 2,307 either favored deregulation or endorsed the concept of an ECB-resistant corn variety. One letter pointed out that CBH-351 controls third- and fourth-stage ECB larvae. The two commenters expressing reservations were concerned about resistance management and the establishment of refugia of nontransgenic corn where the 176 Event would be grown. There was no indication from the documents of any other public involvement in APHIS’s decision-making process.
Herbicide-Tolerant and Insect-Resistant Cotton
Cotton production has historically relied on heavy use of both insecticides and herbicides. On a per-acre basis during the 1990s, the number of pounds of insecticide used in cotton was three to eight times more than in corn and about 100 times higher than in soybean (NRC 2000b). A number of key insect pests of cotton such as the tobacco budworm have evolved resistance to many insecticides. In the mid-1990s insecticide resistance threatened the economic viability of cotton farming in a number of areas of the United States (e.g., Luttrell et al. 1994).
Herbicide use in conventional cotton has been high and on par with that for other row crops such as soybean and corn. However, the available herbicides have been difficult for farmers to work with because of the limited time period for high efficacy and the limited spectrum of weeds killed by each herbicide. As a result, cotton farmers have often had to use multiple herbicides to control weeds. Any technology that increases the efficiency of weed control is of interest to farmers.
As discussed in the prior case study, transgenic cultivars expressing insecticidal proteins derived from the soil bacterium Bacillus thuringiensis (Bt) have been successful in limiting damage by a number of Lepidopteran insect pests (e.g., European corn borer, pink bollworm, tobacco budworm, cotton bollworm; Gould 1998). It was therefore not surprising that many cotton farmers whose livelihood was threatened by insecticide resistance embraced transgenic Bt-cotton, which caused nearly 100% mortality of the tobacco budworm.
Transgenic cultivars with herbicide tolerance and/or insect resistance were planted on over 40 million hectares in 2000 (James 2000), making
these two crop traits the most widely used products of agricultural biotechnology in the world. Commercial sale of insect-resistant transgenic cotton began in 1996, and the sale of herbicide-tolerant cotton began in 1997 (USDA 1999; see also Biotech Basics 2001). Increasingly, cotton cultivars are being produced that have both herbicide tolerance and insect resistance. In 2000, one-third of all transgenic cotton in the United States had both traits (USDA-NASS 2001).
There are two approaches for gaining regulatory approval of a cotton cultivar that is both herbicide tolerant and insecticidal. The most common approach since 1996 has been to obtain regulatory approval for each trait individually. For example, a herbicide-tolerant cotton genotype is developed and a petition is sent to APHIS asking for deregulated status. A Bt-producing cotton genotype is developed separately and goes through the EPA regulatory process as well as a petition for deregulated status with APHIS. Once the herbicide-tolerant cotton and Bt-producing cotton are granted nonregulated status, APHIS has no authority over those plants. Because the herbicide-tolerant cotton is not in itself a pesticide, EPA has no authority to govern its sale (EPA does regulate the sale of all herbicides). Therefore, anyone with legal access to the deregulated insecticidal cotton germplasm and to the deregulated herbicide-tolerant cotton germplasm can cross the two types of cotton and produce a new cultivar with both traits by this conventional breeding technique. This multitrait (generally referred to as “stacked trait”) cotton can then be commercialized without further regulatory oversight.
In the case study examined here (petition 97-013-01p for determination of nonregulated status for Event 31807 and Event 31808), Calgene took a different approach to developing a cotton plant with herbicide tolerance and insect resistance. Although not clearly stated in the APHIS environmental assessment (USDA 1997b), it appears that Calgene developed a single construct for insertion into the cotton genome that contained both a gene for bromoxynil tolerance and the Bt gene, Cry1Ac, for insecticidal activity. The breeding advantage for using a single construct with both genes tightly linked is that the probability of segregation of the two genes during backcrossing to other cotton cultivars is extremely low. Because the two genes are essentially inherited as a unit, Calgene had both traits reviewed simultaneously by APHIS. The committee selected this environmental assessment as a case study of multiple genes because it is a case in which APHIS examined a petition for a plant with two genes, each governing a different agriculturally important trait. Many petitions for deregulation involve plants with multiple transgenes. In most cases one gene produces the phenotype of commercial interest and a second gene acts as a selectable marker. In the case of virus-resistant
squash, multiple genes are present for resistance to a number of viruses. This cotton case stands out because both genes have distinct commercial uses.
The environmental assessment and determination documents for this petition were relatively short. The assessment formally considered only two alternative actions: “no action,” which would mean refusal to grant nonregulated status, or a determination of a “finding of no significant impact,” which would result in complete deregulation. These alternatives contrast with other recent environmental assessments. For example, in the environmental assessment of a Bt corn petition that was also reviewed in 1997 (96-317-01p), three alternative actions are stated. The additional action listed is to “approve the petition with geographical limitation.” No explanation was given in this case study’s assessment about why only two options were considered.
Environmental Risks Considered by APHIS in Its Environmental Assessments and Determination Documents
Disease in the Transgenic Crop and Its Progeny Resulting from the Trans genes. Because the herbicide tolerance and Bt genes were inserted using Agrobacterium tumefaciens, and because a cauliflower mosaic virus 35S promotor and a chimeric 35S promotor were part of the inserted DNA, APHIS examined the potential for risk from these sequences that came from plant pest species. The potential for these sequences to result in risks was dismissed because the disease-causing genes were not present.
Potential Environmental Impacts. APHIS recognized the potential for transgenic cotton to cross with wild cottons in some parts of the continental United States but concluded that “none of the relatives of cotton in the United States show any definite weedy tendencies” (USDA 1997b). (APHIS acknowledged that judgment of weediness based on the 12 traits listed by Baker (1965) or subsequent modifications are “imperfect guides to weediness.” (The utility of Baker’s list as a regulatory guide is discussed at length in the previous case study and is not repeated here.) Furthermore, APHIS stated that gene flow to wild relatives would not be a problem because (1) “any potential effects of the trait would not significantly alter the weediness of the wild cotton; and (2) wild cotton populations have not been actively protected, but have in fact been, in some locations such as Florida, subject in the past to Federal eradication campaigns because they serve as potential hosts for the boll weevil” (USDA 1997b). The EPA, which has also reviewed transgenic Bt cotton, came to a different conclusion. EPA allowed the planting of cotton in all areas of the continental United States except southern Florida because of the presence
of feral populations of Gossypium hirsutum that can cross with commercial cultivars. However, the EPA stated that commercial cotton is not presently grown in that area of Florida. The agency also found a risk of transgenic cotton crossing with wild cotton in Hawaii. In this case the agency was specifically concerned with the risk that hybridization of the transgenic cotton with wild G. tomentosum could threaten that species’ biodiversity and put restrictions on all but isolated breeding nurseries. The EPA noted that nontransgenic cotton would pose a similar threat but is not regulated (EPA 2000a).
Potential Impacts on Non-target Organisms. APHIS concluded there is no reason to believe that transgenic cotton lines would have deleterious effects on non-target species, based in part on EPA’s finding that “foliar microbial pesticides indicated no unreasonable adverse effects on nontarget insects, birds, and mammals” (EPA 1995). Also, APHIS argued that “invertebrates such as earthworms, and all vertebrate organisms, including non-target birds, mammals and humans, are not expected to be affected by the Btk insect control protein because they would not be expected to contain the receptor protein found in the midgut of target insects” (USDA 1997b; see also BOX 4.1).
The comparison of Bt cotton with Btk-sprayed pesticides is not appropriate in this case because the Btk pesticides degrade very rapidly in the field, due in large part to ultraviolet light exposure, while the Bt toxin in cotton is expressed constitutively and is tilled into the soil. Furthermore, no information is given to indicate whether the Cry1Ac toxin produced by the plant is a protoxin, as in the pesticide, or if it is an activated toxin that could have different ecological impacts (see discussion in Chapter 2).
Potential Impacts on the Development of Insect Resistance to the Btk Insect Control Protein. APHIS considered the issue of insect pests evolving resistance to the Bt toxin. The environmental assessment indicated that the EPA’s active resistance management program should delay the onset of resistance. APHIS also concluded that if resistance to Bt does evolve in insect pests, the ability to control the insects will not be reduced because conventional insecticides will still be available.
At the time this assessment was written, the tobacco budworm had become highly resistant to pyrethroid insecticides in major cotton-growing areas, and the cost of chemically novel replacement pesticides was about triple the cost of pyrethroids. APHIS did not discuss the fact that one of the major pests affected by Bt cotton is the corn earworm (H. zea). That species feeds on many vegetable crops and is treated with Bt sprays by organic farmers. One of the factors that led to developing resistance management programs for Bt crops was concern that in the absence of Bt
BOX 4.1 The Mysterious Ecological Role of Bt Toxins
Over 500 scientific journal articles have been published on Bacillus thuringiensis since 1999 (www.WebofScience.com). Of these, only a handful discuss topics related to the natural ecology of this bacterium. An examination of the older literature reveals a similar trend. Most ecological studies tend to examine persistence of the bacterium in the soil (e.g., Addison 1993) or competition between B. thuringiensis and related species (e,g,, Yara et al. 1997). Most of the toxicological testing of B. thuringiensis isolates have pragmatically focused on pest insects (see Schnepf et al. 1998). Some papers examined toxicity to non-target organisms, including collembola, honeybees, and daphnia, as part of the process for regulatory approval (e.g., Sims). Most of these tests indicate that the common B. thuringiensis toxins are specific to small taxonomic groups of insects, and there is therefore a tendency to conclude that insects and B. thuringiensis have coevolved with each other (see Yara et al. 1997).
Indeed, there is no basis for such a conclusion. As a case in point, many of the commercialized B. thuringiensis toxins are considered specific to Lepidopteran larvae (Schnepf et al. 1998). However, general knowledge of the ecology of these larvae and this bacterium indicates that they rarely come in contact. B. thuringiensis is considered a soil bacterium and is rapidly killed when exposed to ultraviolet from direct sunlight. In order to be toxic, B. thuringiensis must be ingested. Most Lepidopteran larvae feed on leaves, fruits, flowers and plant stems. The few that feed on plant roots only ingest soil, and the bacteria in it, as a contaminant of their diet. Many Lepidopteran larvae pupate in the soil, but the prepupal stages in the soil do not feed. Furthermore, epizootic of B. thuringiensis in Lepidopteran populations are rare. The only habitat where these larvae and bacteria could commonly come in contact is in grain bins where a small number of Lepidopteran species live.
If Lepidopterans are an unlikely natural host for B. thuringiensis, is there some other more likely host? It certainly seems unlikely that this bacterium would use over 10% of its protein to make a toxic crystal of protein unless it had some function. One candidate for a host that has received minimal attention is the bacteriophagous nematode. As the name implies, these nematodes eat bacteria, including B. thuringiensis. It has been known for over 10 years that some B. thuringiensis isolates are toxic to C. elegans (Feitelson et al. 1992) but only recently have studies begun to look at other nematodes (Marroquin et al. 2000, Griffitts et al. 2001). Given that bacteriophagous nematodes are one of the most diverse groups of soil invertebrates, there is at least a reasonable expectation that B. thuringiensis has evolved in interaction with these organisms. Of course, until more studies are done on the ecological interactions of B. thuringiensis and soil-dwelling organisms, it will not be known what is the most common host or food of B. thuringiensis.
Without this knowledge, our ability to develop tests to examine non-target effects of B. thuringiensis toxins will at least be inefficient and at most totally misguided. In order for APHIS to develop more rigorous environmental assessments, it would be helpful to accumulate knowledge about the natural ecology of B. thuringiensis.
organic farmers would have no means to control this insect. The environmental assessment does not comment on whether the return to the use of conventional insecticides would cause environmental problems. As indicated in the previous case study on Bt corn, it would seem best for APHIS to consider this issue in more depth or to completely defer to EPA authority.
Environmental Risks Not Considered by APHIS in its Environmental Assessments and Determination Documents
Potential Impacts on Non-target Organisms. Bacillus thuringiensis is a soil-dwelling organism that would rarely seem to come in contact with foliage- and fruit-feeding insects. Bt protoxin created by this bacteria must be ingested before its insecticidal properties can be activated. Many Lepidoptera pupate in the soil, but Lepidoptera with soil-dwelling feeding phases are very rare. Based on the lack of interaction between the bacteria and Lepidopteran-feeding stages, there is no obvious ecological or evolutionary explanation for B. thuringiensis producing a Lepidopteran-specific toxin. Presumably, the bacteria produce endotoxins for another purpose, but this purpose has not been determined (see BOX 4.1). APHIS presented no data on tests that the applicants might have conducted on impacts of the truncated Bt toxin on organisms in the soil, including microbes and nematodes that could interact ecologically with the Bt bacterium and its toxin.
Potential to Cross with Wild Species in Some Geographic Areas in the United States. As stated in the “Background” to this case study section, the environmental assessment presents only two alternative responses to the petition for a finding of nonregulated status. APHIS did not mention the option of approving the petition with geographical limitations, even though this option was presented in other APHIS environmental assessments. In the current assessment, APHIS makes the decision to grant complete approval of the petition. The transgenic cotton lines under consideration were deregulated throughout the United States.
Impacts of Commercialization of Transgenic Cotton on Environments Outside the United States. Other APHIS environmental assessments discuss concerns about the impact of a transgenic crop approved for use in the United States being planted in other countries. One example is the squash case study. The environmental assessment of Calgene’s transgenic cotton mentioned the existence of wild cotton in Mexico but does not really assess potential impacts on those species. Furthermore, this environmental assessment did not consider the fact that the specific Bt toxin
gene under review might be useful for resistance management in the United States but might also facilitate resistance in pests that occur beyond U.S. borders. At least one cotton pest, H. zea, is known to move between Mexico and the United States each year (Raulston et al. 1986, Pair et al. 1987). Therefore, inappropriate planting of Bt cotton in Mexico could select for resistant pest individuals that would then migrate to the United States.
Interactions among Multiple Transgenic Traits. APHIS treated herbicide tolerance and production of the Bt insecticide as two separate traits. It did not consider that there might be interactions between the two traits that could have a detrimental effect.
Integrated pest management (IPM) emerged in the late 1950s as an effort to put pesticide use on a more ecological footing. One of the tenets of IPM is that natural processes can be manipulated to increase their effectiveness, and chemical controls should be used only when and where natural processes of control fail to keep pests below economic-injury levels (NRC 1996). Even with crops that have only Bt toxin genes, it is difficult to follow IPM guidelines because seed must be purchased in the spring before pest abundance can be predicted (Gould 1988). When two traits are combined in a single cultivar and it is impossible to purchase cultivars with only one of the two traits, farmers are forced to buy a cultivar with both traits even if they need only one for their farming operation. In the case of the stacking of herbicide tolerance and Bt toxin production, a farmer who needs herbicide tolerance may end up planting cotton with the Bt trait, even if the densities of the Bt target pests on the farm do not warrant control with the Bt trait. While it is difficult to determine how many farmers have specifically begun to use cotton with the Bt trait based on their desire to use herbicide-tolerant cotton, interviews with North Carolina farmers indicate that it may be over 20% (Bacheler 2000, North Carolina State University Extension, personal communication). This approach to the use of a pest control tool is clearly not an appropriate way to achieve the goals of IPM.
In addition to negating progress in adopting IPM farming methods (NRC 2000b), overuse of pesticidal crops due to a lack of seed choice could lead to more intense selection for Bt-resistant pest strains. The EPA has developed regulations to delay the evolution of Bt-resistant pest populations. In 1998 the EPA’s Scientific Advisory Panel recommended that resistance management for Bt crops must include the following two components: (1) the transgenic plants must produce a high enough dose of toxin to kill partially resistant individuals (this dose was set at 25 times the dose needed to kill susceptible individuals) and (2) enough non-
transgenic hosts must be planted on each farm to produce 500 susceptible pest individuals for each resistant individual produced in the Bt crop.
In the case of cotton, EPA-registered transgenic plants do not produce a high dose of Bt toxin for the cotton bollworm, so a large proportion of partially resistant individuals could survive on the Bt cotton. This registration was not appropriate according to the EPA Scientific Advisory Panel, and current requirements for on-farm, non-Bt acreage are not expected to produce the desired 500:1 ratio. Only a substantial increase in the refuge acreage could ameliorate this problem. While the EPA may not increase the on-farm refuge requirements, concern over resistance evolution in H. zea due to an inadequate resistance management program was reemphasized by the most recent EPA Scientific Advisory Panel (EPA 2000b). If the stacked trait cottons such as the one approved by the environmental assessment discussed here are commercially successful, they could increase regionwide adoption of Bt cotton, further accentuating the risk of rapid evolution of Bt resistance in H. zea.
This case study identifies only two negative environmental effects that could be caused by the interaction of two transgenic traits. If outcrossing to weedy relatives was more of a problem with cotton, the interaction between herbicide tolerance genes and Bt genes could exacerbate an additional risk—the transfer of the Bt trait to noncrop plants. A Bt gene inserted into a crop along with a herbicide tolerance gene could be transferred to wild relative populations much faster than in cases where the Bt gene was inherited separately from herbicide tolerance. A potential scenario is as follows: Pollen for the stacked trait cultivar crosses with a weedy relative in a cotton field. The next year progeny from the cross as well as other individuals of the weed species germinate in the cotton field. The farmer sprays bromoxynil. This kills most of the weeds without the herbicide tolerance gene, but those with the gene increase in frequency. Although the Bt gene confers no direct advantage with regard to survival against herbicide spray, there is a major increase in the frequency of weeds with the Bt gene because it is linked to the herbicide tolerance gene. This results in a large fraction of weeds that are now protected from insect feeding. If APHIS reviews transgenic plants with weedy cross-compatible relatives in the United States, such as canola, with stacked herbicide tolerance and insecticidal genes, it would definitely need to consider this interaction.
The APHIS environmental assessment indicated that the agency received no responses to its Federal Register announcement of this petition
for deregulation. The committee is unaware of any other attempt to involve the public in this specific assessment.
This chapter has reviewed case studies of the three primary APHIS regulatory pathways for field release of transgenic organisms as well as a representative sampling of the vast array of transgenic species, phenotypes, and molecular mechanisms designed to obtain those phenotypes. In many cases the committee simply reports, without much comment, how and with what information APHIS made a specific decision. The committee has little to add in those cases. In certain cases, it has pointed out situations in which APHIS might have improved its assessments. The committee has supplied substantial supporting text to explain how those improvements might have been made. While it is recognized that a few of those suggestions benefit from hindsight, most of the suggestions are based on scientific information available, but not utilized, at the time of assessment. The opportunities for improvement of assessment provide a context for the committee’s recommendations in the next chapter.