Bt is indisputably the most important biopesticide registered for use in the United States, indeed worldwide. Bt foliar insecticide sales in the United States today are pushing $60 million annually and account for the lion's share of biopesticide sales (Gianessi, 1995). The prospect for steady sales growth were bright, at least until the emergence of Bt-transgenic plants that promise to quickly lead to widespread resistance to all Bt products, hence removing this uniquely effective and safe natural product from the "pesticide tool-kit."
Bt spray formulations are available for use in managing many difficult-to-control lepidopteran and coleopteran insects. Lepidopteran insects are major problems for farmers producing field crops like corn (corn ear worm), cotton (bollworm, budworm, pink bollworm), and potatoes (Colorado potato beetle). But of even greater concern from the perspective of human health, lepidopteran species are also typically the dominant insect pests in fruit and vegetable crops. By "dominant," we mean the species that, over the years, requires the most frequent applications of the generally most toxic insecticides.
Most fruit and vegetable growers design IPM systems around tactics and techniques needed to control lepidopteran species. Bt and/or pheromones form the backbone of many such systems, especially on farms well along the IPM continuum toward biointensive IPM -- the kind of IPM both the EPA and USDA are trying to promote. Use of these soft biopesticides makes it possible for farmers to avoid the need for damaging, broad-spectrum sprays that often trigger secondary pest population explosions, along with a host of other problems (farm worker poisonings, wildlife kills, dietary health risks, etc.).
Indeed, because of the effectiveness and safety of Bt compared to the insecticides it displaces, Bt is probably the single most important insecticide ever discovered. Its importance is also about to exponentially rise as the EPA moves forward with the implementation of "The Food Quality Protection Act of 1996."
By crop season 2000, the FQPA will surely lead to substantial restrictions on many of the organophosphate (OP), carbamate, and synthetic pyrethroid insecticides that farmers would have to return to if the effectiveness of foliar Bt insecticides is undermined by resistance. The prospect of the loss of Bt led an experienced Florida crop consultant to remark that the loss of Bt would "... force us back to the stone-age of pest management in a heart-beat."
In 1994, the average cotton acre in the United States was sprayed with 3.5 different insecticide active ingredients, 5.4 times with 2.48 pounds of insecticide a.i. (USDA Cropping Practices Survey, 1994). Based on Monsanto data and statements from consultants who managed Bollgard or Ingard fields in 1997, Bt-transgenic cotton in 1996 made it possible for farmers to reduce the pounds of insecticide active ingredients applied by about 50 percent on average – or by about 1.25 to 1.5 pounds. Given that 1.8 million acres of cotton were planted to Bt varieties in 1996, about 2 million pounds of insecticide active ingredient applications were saved.
People do not eat cotton. Cottonseed meal plays a very minor role in the human diet. When Bt is no longer effective on major fruit and vegetable crops, we project that, at a minimum, between 2 and 3 million acres of fruits and vegetable crops will need to be sprayed an average four more times for lepidopteran insect control, at an average rate of application of 0.5 pounds active ingredient per acre. The increase in insecticide use – 4 million to 6 million pounds on crops that people eat --will significantly expand total dietary pesticide residue intake.
In addition, consultants report that Bollgard varieties are slower to develop and stop growth, "So you have to put defoliants on in cooler temperatures, which forces you to use a higher rate, increasing costs," according to Louisiana cotton consultant Roger Carter (Ag Consultant, November, 1996, page 10).
The EPA's mission is to reduce overall pesticide exposure and risks. Hence, in light of new evidence from crop year 1996, steps are now needed to assure that a temporary reduction in cotton insecticide use is not achieved at the expense of significant longer-term increases in insecticide use on human food crops.
We now offer our views on the four questions EPA raised.
1. What are the implications of the bollworm control failure for resistance management plans?
We and many others felt that EPA placed too much confidence in theoretical models and heroic assumptions in initially accepting the resistance management plans (RMP) for Bt cotton and corn. Now that extensive performance data from plantings in the United States and Australia in crop season 1996 are available, there is a foundation for evaluating the assumptions and predictions that support the current Bt-transgenic plant resistance management plan.
In a nutshell, performance in the field, coupled with good science removes any douBt about whether the current RMP might work. It has not and will not. And it is not a matter of just working out a few "kinks." Accordingly, EPA should act to suspend all existing Bt-transgenic plant registrations until such time as a proven, science-based resistance management plan is developed and submitted for review by independent scientists. In light of recent evidence and key scientific findings reviewed below, we douBt such a plan will ever be forthcoming and so urge EPA to take action now to save Bt, before it is too late.
Monsanto made clear when it got approval for Bollgard that its RMP consisted of the "high dose" strategy plus the use of refugia. They had lab data showing that the level of Bt endotoxin expression was one to two orders of magnitude (i.e. 10- to 100-fold) higher than the level needed to kill early instar larvae of the major lepidopteran pests. The "high dose" strategy was and is still is the primary foundation for the proposed RMP. The role, and potential contribution of refugia is to simply delay resistance. When EPA approved commercial use of Bollgard, there was inherent in EPA's decision a leap of faith that the high dose strategy would work. While Monsanto had laboratory data confirming that plants could express Bt at levels far in excess of the minimum lethal dose for feeding insects, Monsanto lacked extensive field data and experience to verify that their RMP would work under field conditions.
Indeed, Monsanto made a number of assumptions associated with the RMP that were not extensively checked by field testing, including that --
i) the "high dose" expressed in plants in the lab would also be achieved in plants in the field, and evenly across all plant tissues and environmental and soil conditions;
ii) the susceptibility of lepidopteran larvae to "high dose" exposures was the same or similar for the captive populations used in most of the testing and wild population of lepidopteran larvae (e.g. cotton budworm, pink bollworm, and cotton bollworm in the field);
iii) the refugia were of the proper size and would work as intended in the field;
iv) resistance to Bt is a recessive trait;
v) resistance to the different strains of Bt endotoxin would require separate independent mutations (i.e. that cross resistance does not occur, or is very rare, with the Bt endotoxins); and
vi) that resistance to Bt is a rare trait in the wild, i.e. the frequency of a resistance allele is quite low.
Now there are extensive field observations by crop consultants in the United States, significant scientific data from Australia, and an important paper just published by Dr. Bruce Tabashnik and colleagues with which to revisit each of these assumptions. But the bottom-line is indisputable -- the high dose plus refugia theory has been proven incompatible with reality in the field, because of variable field conditions and the variability of Bt-endotoxin expression in the field, and hence there is now NO resistance management plan for Bt-transgenic plants.
Assumption # 1 -- "high doses" expressed in plants in the lab would also be achieved in plants in the field, and evenly across all environmental and soil conditions.
Field data show clearly that Bt expression starts out high in most fields but tapers off as the season progresses. Moreover, expression tends to be much lower -- and sometimes below a lethal dose -- in the bottom 25 percent of the cotton plant. For example, Clayton, Louisiana crop consultant Roger Carter reported 90 percent to 95 percent control in the top portion of plants, but only 20 percent to 25 percent below the terminal (Ag Consultant, November, 1996, page 10). An open letter sent to Monsanto by the Bourke Cotton Growers Association in Australia also highlights the poor performance in the lower parts of plants --
The loss of these nodes is, moreover, an economic issue, since the lower nodes on cotton often produce the highest quality cotton.
There is virtually no Bt expression in the pollen, so some insects that feed on pollen and/or lower parts of plants can escape lethal doses.
The best results in the United States were achieved in irrigated areas with very flat, homogenous soil conditions conducive to very even growth of cotton plants. The more even and steady the growth of a cotton crop, the more likely the Bt toxin will be expressed at lethal levels, evenly throughout the architecture of the plant. But where soil conditions vary because of compaction, uneven irrigation or mineral imbalances, for example, plants may not grow evenly, and parts of some plants will not express lethal levels. Again, some insects will survive.
Physical injury from equipment, or wind and rain, or drought conditions, weed competition and a host of root diseases can also disrupt the physiology of the cotton plant enough so that the level of Bt expression is not high enough in parts of plants, or again, is uneven. Uneven expression of Bt could accelerate the emergence of genetic resistance because of the capacity of some insects to sense, and then avoid, parts of plants with lethal levels of Bt expression -- what Dr. Marvin Harris cites as an example of "behavioral resistance" (Letter in Science, date unknown, circa. November, 1996).
In an early summer (1996) press release, the Chemical and Agricultural products Division of Abbott, which sells a foliar Bt product called DiPel, alerted growers that while they could not spray foliar Bt products on refugia, they could make mid- and late-season applications on Bollgard fields. In explaining why such applications were warranted, both to protect the crop and delay resistance, the Abbott press release states that "It seems that while Bollgard usually provides good control, recent field results are showing variable toxin expression and lack of worm control, especially late in the growing season."
Assumption # 2 -- the susceptibility of lepidopteran larvae to "high dose" exposures was the same or similar for captive populations and wild population of lepidopteran larvae (e.g. cotton budworm, pink bollworm, and cotton bollworm).
A variety of studies have shown that this assumption clearly has not held up. Dr. J.R. Bradley Jr., Department of Entomology, North Carolina State University, wrote in a letter to Science Magazine --
"Our data were summarily ignored in favor of data acquired when there were low numbers of 'wild' bollworms or from test sites artificially infested with laboratory-cultured larvae. The wave of euphoria created by Bt cotton swept across the cotton-belt and carried many entomologists with." (Science, date unknown, about November, 1996).
Assumption # 3 -- refugia were of the proper size and would work as intended in the field.
The purpose of refugia is to delay resistance; no one contends that refugia will prevent resistance. Moreover, to be effective, refugia need to be quite large -- as much as 40 percent of a field (assuming other measures are taken to control insects, as they obviously must be with refugia of this size).
According to Dr. Fred Gould, a leading authority on refugia and Bt resistance, the movement of adult insects like the corn earworm, seriously undermines the high-dose plus refugia theory (seminar presented January 31, 1997, University of Wisconsin-Madison). Gould stated during the seminar that models suggest Bt resistance would emerge in cotton insects in about 12 generations -- or in about three to four years. But with optimal r efugia, some modelers project that resistance could be delayed up to 115 generations -- or about 30 years. Gould suspects that the refugia requirement now in place will fall far short of extending the effective life-span of Bt varieties ten-fold. In a 1991 letter to Science, Dr. Marvin Harris, Texas A+M University, predicted resistance in "25 to 75 generations (3 to 9 years)" and that "...resistance is inevitable." Harris also warns that "...resistance developed in response to transgenic Bt will transfer to fermented Bt products and render them useless against resistant pests." (Science, Vol. 253, September 6, 1991, page 1075).
Two widely respected crop consultants, Harold Lambert and Roger Carter, both believe the refugia requirement is too low, and would instead prefer a 60 percent Bt-cotton, 40 percent refugia requirement (Ag Consultant, November, 1996, page 10).
In a Technical Comment in Science entitled "Evolution of Insect Resistance to Bacillus thuringiensis-Transformed Plants," Dr. Anthony Ives, a University of Wisconsin-Madison zoologist, showed that the population genetics models used to s upport the role of refugia in Bt-transgenic corn varieties contained an inappropriate assumption regarding the preferential migration of corn borer within corn fields. Based on a more accurate set of assumptions, Ives recalculated the impacts of refugia and different spatial patterns for the planting of Bt-transformed and conventional varieties. His major finding --
Assumption # 4 -- resistance to Bt is a recessive trait.
Assumption # 5 -- resistance to the different strains of Bt endotoxin would require separate independent mutations (i.e. that cross resistance to the Bt endotoxins is rare).
Important new evidence has just been published this week in the Proceedings of the National Academy of Sciences (Tabashnik et al., 1997). Applying state-of-the-art methods, the researchers showed that two of the fundamental assumptions behind ALL Bt RMPs -- that resistance to Bt is a rare recessive trait and that resistance to the different endotoxins is independent (i.e. cross resistance is not important and/or common) -- are false.
Dr. Bruce Tabashnik and colleagues decided to test diamondback moth (Plutella xylostella) for resistance to Bt endotoxins. They decided to concentrate on 4 of the most important Bt endotoxins against lepidopteran larvae-- Cry1Aa, Cry1Ab, Cry1Ac, and Cry1F. These are also the most important Bt endotoxins in the approved transgenic crops; Bollgard cotton contains Cry1Ac, while the transgenic Bt corn approved in 1996 contained Cry1Ab. A key finding of the study was that a single "autosomal recessive gene conferred extremely high resistance to four Bt endotoxins (Cry1Aa, Cry1Ab, Cry1Ac, and Cry1F)" (Tabashnik et al., 1997: page 1640). Thus, the notion that resistance can be delayed through the use of two or more endotoxins has been dealt a serious blow, since there is no plausible reason to suspect that the multiple resistance gene will only be found in diamondback moth.
This work has even more disturbing implications for polyphagous (insects that feed on many crops) lepidopteran larvae such as Helicoverpa zea. This means H. zea that develops resistance to the Cry1Ac endotoxin by feeding on Bollgard cotton could become resistant quickly to the Cry1Ab endotoxin used in transgenic corn, or vice versa. Such a population would also, of course, have high level of resistance to the foliar Bts. This would be a particularly damaging and costly loss for organic farmers.
Dr. Fred Gould has reported basically the same thing -- cross-resistance rapidly evolves to the three main strains of Bt, and summed up his discussion of cross-resistance at the January 31, 1997 University of Wisconsin seminar by stating "You have one shot with Bt and then you're done..." In his laboratory, he has observed 20,000-fold resistance ratios, such that insects could literally swim in foliar Bt.
Assumption # 6 -- resistance to Bt is a rare trait in the wild, i.e. the frequency of a resistance allele is quite low.
According to Dr. Fred Gould, most studies to date suggest that there is a 0.0015 frequency of resistance genes in target insect populations. Given the number of insects moving through a field in a season, even this frequency is high enough to require aggressive resistance management. But the recent work of Dr. Bruce Tabashnik and colleagues provides a very different, and far more worrisome picture.
Tabashnik et al. showed that the frequency of a multiple-toxin resistance allele in susceptible populations of the diamondback moth (Plutella xylostella) was an astonishing 0.120, some "10 times higher than the most widely cited estimate for the upper limit for the frequency of resistance alleles in susceptible populations" (Tabashnik et al., 1997: 1640), and almost 100 times higher than the figure that Dr. Gould suggests! The susceptible populations of diamondback moth that Dr. Tabashnik worked with had been reared for over 100 generations without exposure to Bt or any other pesticide. The fact that the resistance allele had such a high frequency strongly suggests that the resistance gene carries very little, if any, genetic load. As Dr. Tabashnik et al. concluded, "Extended maintenance of a resistance allele frequency close to 0.10 without exposure to Bt implies that in the absence of Bt [sic], heterozygotes have little or no fitness disadvantage relative to susceptibles" (Tabashnik et al., 1997: 1643).
The finding that there is little or no fitness cost (i.e. disadvantage) to the multiple-toxin resistance allele has grave (perhaps even fatal) implications for the high dose plus refugia RMP. Since there is little, if any fitness cost, the resistance allele could have far higher frequencies in wild populations of lepidopterans than was previously thought.
A key assumption of high dose plus refugia strategy is that only homozygous susceptible (i.e. RR) moths will emerge from the refugia, so that any resistant (i.e. rr) moths that survive the "high dose" of Bt in the transgenic plant will mate with a homozygous susceptible (i.e. RR) moths and produce only heterozygous susceptible (i.e. Rr) offspring. However, if a significant percentage of the moths that emerge from the refugia are heterozygous susceptibles, then one quarter of the offspring of a mating between such moths (i.e. Rr) and resistant moths (rr) would be resistant to Bt (i.e. rr).
Thus, the combination of much higher than expected survival of cotton bollworm larvae (Helicoverpa zea) on transgenic cotton, coupled with the existence of multiple-toxin resistance could mean that cotton bollworm would develop resistance far faster than previously imagined – perhaps in just two to three seasons.
In light of these real world factors, it is perhaps not surprising that the "high-dose" strategy failed against the cotton bollworm in many areas in crop season 1996. Monsanto has admitted that up to 50 percent of the Bt-cotton acreage had to be sprayed for cotton bollworm (Gary Barton, Monsanto biotechnology spokesman, quoted in Southern Sustainable Farming # 12, September 1996). Harold Lambert, an Innis, Louisiana crop consultant, says the product was only 60 percent effective on bollworms. He calculates that 80 percent of his clients had to spray Bollgard fields 1.5 to 2 times (Ag Consultant, November, 1996, page 9).
It is important to remember that the foundation of current Bt-transgenic plant RMPs is that high doses will kill virtually 100 percent of feeding insects, and hence there will be no survivors to pass along resistance genes. Monsanto, and entomologists working for or on behalf of Bollgard varieties, have argued repeatedly that high populations caused the control failures in 1996, but this makes no real sense, since if the Bt toxin were in fact expressed at 25-times a lethal dose, even the higher populations feeding on the plants would still be controlled.
What went wrong? The relatively widespread failure to control H. zea could be due to (1) widespread Bt resistance in the bollworm; (2) lower levels of expression of the endotoxin in the cotton plant than expected (based on lab studies); (3) variable expression of Bt endotoxin in the plant combined with behavioral tolerance by the bollworm; (4) field populations of H. zea being far more resistant to Bt than lab populations; (5) accidentally planting susceptible cotton instead of Bt cotton, or in all likelihood, various combinations of the above in different settings.
Since Monsanto has refused to release the identity of the farmers using Bt cotton, so that independent researchers can check those fields, we do not know the true extent of the problem in the United States. Inadequate and uneven levels of expression of the endotoxin is suspected in the U.S., but we have no firm evidence. There is evidence that this has been a problem in Australia. Indeed, the view of the Australian researchers is that the Bt cotton has not lived up to expectation.
Australian Experience The first year of commercialization in Australia was more like a widespread field test, and included extensive, open monitoring of field results by crop consultants and other independent experts as the identity and location of all farms planting Bt cotton was known. According to Dr. Neil Forrester, researcher in NSW, Australia --
Dr. Bruce Pyke, of the CRDC (Commonwealth Research and Development Corporation), and member of TIMS (Transgenic and Insect Management Strategies Committee), makes a similar point about the Ingard trials in NSW and central Queensland --
2. Should resistance management plans be mandatory or voluntary?
In the case of any uniquely valuable, public good biopesticide like Bt, resistance management must of course be mandatory. But they also need to work.
For any properly conceived RMP to work, all farmers must follow the plan closely. Cooperation by 95 percent of growers, perhaps even 99 percent, is not good enough, since it takes only one field to produce a small population with resistant genes, which will then spread through the population rapidly in the presence of continued selection pressure.
If a RMP is voluntary, we cannot assume that all farmers will understand the absolute necessity of 100 percent compliance, and lacking such understanding, it is not hard to imagine some cutting corners on at least some fields. For example, imagine the normal reaction of a farmer watching his or her refugia -- 4 percent of cotton acreage planted and not sprayed -- being devastated by lepidopteran larvae. Some may consider it immoral to sit back and watch the crop be destroyed, and will take action consistent with their beliefs about right and wrong.
3. What scientific data are needed to evaluate the resistance management plans?
The one thing clearly learned from the 1996 Bt cotton crop season in the U.S. and Australia is that laboratory data do not adequately predict what will happen in the field. Thus, extensive field data are crucial for testing and validating a RMP.
Furthermore, EPA needs better criteria for deciding what constitutes an adequate RMP. CU feels that the bottom-line criteria should be no increase in resistance among pest populations. Furthermore, EPA needs clear, immediate and enforceable procedures to deal with the failure of a RMP. A finding that a RMP has failed should be triggered by –
Credible evidence of an increase in resistance in a target pest of the engineered plant pesticide, or in a pest species otherwise controlled by the active agent in the plant pesticide, either within natural systems or as a result of the application of another pesticide formulation containing the active agent.
4. Is Bt a "public good"?
YES, by anyone's definition of a public good. Bt is a public good because it is a natural resource, a part of the biotic community, that has evolved over the millennium shaped by the forces of nature. It is a biological resource that serves uniquely valuable functions within biotic communities, both in natural systems and those managed by humans. It is a product of organisms widely dispersed in nature, owned by no one.
In fact, Bt has been playing a role in agricultural pest management systems long before people came to understand its properties some three decades ago. No person or company created Bt. No company or person has a right to undermine its effectiveness in either natural or managed ecosystems. Any individual or company that does should be held liabel for any economic harm suffered by others.
Because Bt is not like other synthetic pesticides discovered by agrichemical companies or other entrepreneurs, the EPA is correct in insisting upon a proven, effective resistant management plan before allowing widespread planting of Bt-transgenic plants. Since such a plan clearly does not exist, EPA must act to restrict plantings to the modest experimental plots required to continue research on resistance management strategies.
As one tangible action in response to the devastating new evidence presented today, EPA should move forward quickly in promulgating a process to rescind approval of a plant pesticide on the grounds of a failed RMP. We suspect they will need to invoke such procedures sooner rather than later, and quite possibly before the end of crop season 1997.
As recommended in Pest Management at the Crossroads (PMAC), CU feels that EPA should delay any further approval of Bt-transgenic plant varieties, and that previous approvals should be reversed when evidence points to the imminent failure of a RMP, a condition which has now clearly been met for Bt cotton, and which may be met for Bt-corn in crop season 1997.
Furthermore, to evaluate RMPs in the case of any transgenic plant varieties, CU feels that EPA should require registrants to submit a detailed annual report on the status of resistance, and on how well the RMP is faring, for at least the first five years of commercial use. The data cited in such reports should be accessible for independent review.
As part of this report, the companies should submit the names and addresses of farmers planting transgenic crops, so that independent researchers can study these farms to develop a complete and accurate assessment of resistance levels and trends. (It is noteworthy that the identity and location of the farmers planting Bt-cotton in Australia is known and public; public release of this information made it possible for regulators, consultants and researchers to appraise performance in Australia much more fully and systematically than possible in the U.S.) In addition, EPA should develop its own, independent capacity to monitor resistant trends in the field, and deploy this capacity in the event additional, uniquely valuable natural biopesticides are introduced into plants via genetic engineering.
Our strong opposition to Bt-transgenic plant varieties is pragmatic. We see many promising applications of biotechnology in advancing progress toward biointensive IPM. Several are described in Chapter 6 of PMAC (Benbrook et al., 1996). But genetic engineering is like "The Force" in Star Wars, it has a dark side with the power to return pest management to the "stone-age." Sometimes the rush to get products onto the market overwhelms the disciplined consideration of need, scientific uncertainties and longer-term consequences. Public policy must fill the void. In the case of public good biopesticides like Bt, EPA can best do so by enforcing the requirement for effective, science-based resistance management plans.
The Australian Cotton Grower, Vol. 18, No. 1, January-February, 1997
Benbrook, C., Groth, E., Hansen, M., Halloran, J., and Sandra Marquardt. 1996. Pest Management at the Crossroads, Consumers Union. Available from PMDS, 301-617-7815; or via the PMAC web page, http://www.pmac.net.
Gianessi, L. 1995. "An Economic Profile of the U.S. Crop Protection Pesticide Industry," National Center for Food and Agricultural Policy, Washington, D.C.
Tabashnik, B., Ltu, Yong-Biao, Finson, N., Masson, L., and David G. Heckel. 1997. "One gene in diamondback moth confers resistance to four Bacillus thuringiensis toxins," Proceedings of the National Academy of Sciences, Vol. 94: 1640-1644.