Risks of Genetic Engineering in Agriculture

E. Ann Clark, Plant Agriculture, University of Guelph, Guelph, ON N1G 2W1 eaclark@uoguelph.ca

Adapted from a talk to the annual meeting of the National Farmers Union, November 97, in Saskatoon

The vision of genetic engineering in agriculture raises many issues, of which the following is a partial list.

For example, in August 97, DuPont reportedly bought 20% of Pioneer Hi-Bred International Inc. for a cool $1.7 billion. By 1997, Monsanto owned over half of Calgene, 40% of DeKalb, and 100% of Agrocetus, and purchased Holdenís Foundation Seeds (source of 35% of the parental lines used by independent corn breeders) for $1 billion.
a) the downstream implications of yet further concentration of power within the already well consolidated agribusiness industry. The proceeds of the agrochemical industry have been and are being used to accumulate control over the seeds industry in North America. Where is a producer to go next year, or 5 years from now, if she or he wants to plant non-GE seed?

b) responsibility/accountability when GE entities either fail to perform or cause harm to society and the environment. For example, society as a whole has had to bear the burden of dealing with the adverse agricultural effects of unintentional as well as intentional introductions of exotic invaders. Simberloff (1996) found that crop losses and control costs for dealing with exotic pests account for about one-quarter of the US agricultural gross national product annually. Examples of exotic species which have become pestiferous in the US include the cotton boll weevil, leafy spurge, the European gypsy moth, purple loosestrife, multiflora rose, and kudzu1. The question is, given that the potential for ecological damage is known from the outset, will the proprietors of GE entities be held accountable for the harm, if any, caused by their products?

The precedent for culpability for industrial-origin environmental and health damage is already well established, if incompletely enforced. For example, mining companies are subject to financial penalties if discharged water quality and gaseous emissions exceed set standards. Tobacco companies are finally, after great societal pressure, being obliged to accept responsibility for the decades of adverse implications of tobacco smoking. But the manufacturers of the roughly 75,000 diverse chemicals that are now used for everything from automobile interiors to pest control are still largely unaccountable for possible harm from their products. According to Steingraber (1997), only 1.5-3% (1200-1500 chemicals) have even been tested for carcinogenicity, let alone other possible adverse effects, as on the endocrine system (Colborn et al., 1996).

Which model will pertain to the marketers of GE products - that of the mining, and lately, tobacco companies or that of the purveyors of chemicals? Will society, once again, be obliged to assume responsibility for the ecological and health repercussions, if any, from GE organisms, leaving the proprietors to capitalize on the benefits without absorbing the risks?

At the producer level, millions of dollars were reportedly lost by individuals in Mississippi and other states who had the misfortune of planting some of Monsantoís new Roundup-Ready cotton2 cultivars in 1997 (Myerson, 1997). The specific problem is in GE-versions of Paymaster varieties #1244, #1215, #1330, and #1220, all of which had performed fine in previous years - before insertion of the GE genes (Lappe and Bailey, 1997). In Mississippi alone, some 30,000 acres which had been sown to these GE cotton cultivars yielded cotton bolls which were deformed or absent entirely, reducing yield by as much as 40%. Some 46 of the 200 farmers who planted the GE cotton in Mississippi have asked the Mississippi Seed Arbitration Council to cover their losses - as compared to a total of 2 farmers who had sought coverage for other reasons since 1989. Lawsuits have reportedly been launched against the two companies across the cotton belt.

Concern about the performance of these GE products is not limited to farmers. The risks and benefits of the Roundup resistance technology are sufficiently obscure for Pioneer Hi-Bred to announce in November 97 its refusal to add Monsantoís Roundup Ready gene to its corn.

And the problem is also not limited to Roundup resistance. Another genetically engineered Monsanto cotton, the Bt cultivar ĎBollgardí, failed to reliably repel its target pest - cotton boll weevil on some farms, causing 25 farmers in Texas to band together and sue the two companies for losses over more than 18,000 ac (Lappe and Bailey, 1997; Myerson, 1997).

In a nutshell, the GE crops have not worked consistently- perhaps because they were rushed to release without the normal 3 year testing that is customary (but not required) in Mississippi3. In fact, state and federal cotton specialists in Mississippi were denied access to the material before it went on the market. The geneticist and research manager for the USDA in Stoneville, Mississippi reportedly asked for just one pound of seed last year for a test, and was told by the company that seed couldnít be spared (Myerson, 1997). As late as September 97, the Director of Biotechnology and Scientific Services of the USDA reportedly stated that he was "totally unaware" of the Bt cotton crop failure underway in the cotton belt.

c) diversion of both scarce research resources and producer ingenuity into symptoms rather than causes of cropping system malfunction. To a hammer, everything looks like a tack, and to a community that sees genetics as the root cause of agricultural problems, genetic engineering may seem like the best - or perhaps - the only solution. Lipson (1997) analyzed USDA funding to determine how much was being spent on organic farming - a more holistic way of addressing the problems of agriculture today. He found that just 34 of 30,000 funded projects - accounting for a maximum of $1.5 million out of the $1.8 billion dollar budget - could be identified as pertaining specifically to organic farming. To a large degree, public research funding is being allocated to problem-solving, largely through purchased inputs, rather than to problem avoidance through environmentally sound management.

For example, pests are created - not born, a fact which does not appear to have been assimilated by those advocating GE approaches to pest and weed management (e.g. introducing genes for BT or Roundup resistance into cultivars). Looking for genetic solutions to management problems has the same likelihood of success as applying N to amend a soil deficient in P.
Most production agriculture departments in North American universities today have an increasingly prominent sector devoted to genetic engineering. With few exceptions, these positions were not added to the existing complement of breeders, agronomists, and management specialists. Rather, they displaced the more traditional specializations upon retirement, leaving less and less in the way of people and resources to address the real issues facing farmers today (see sidebar). Diluting the capacity of the agricultural research community to educate the farmers, extensionists, and researchers of the future and to study the root causes of todayís farming problems may, in fact, be the real, long-term legacy of the current fixation with biotechnology.


But perhaps the greatest concern with GE crops is in the realm of agroecology. From the outset, geneticists, ecologists, entomologists, microbiologists, and others have repeatedly raised concerns about the ecological implications of GE (see Krimsky, 1982). However, early alarms were based more on possibilities than on actual evidence - making it easier to discount and bypass the expressed concerns. More recently, however, targeted research is beginning to identify real world examples that challenge the validity of some of the premises of genetic engineering in agriculture.

One specific issue is the projected, commercially brief lifespan of cultivars genetically engineered for specific pests, because of the inevitable evolution of resistance in the target organisms. The ecological implausibility of the resistance management plans upon which GE products have been approved is discussed more fully in Benbrook and Hansen (1996) and Clark (1997a). In the case of BT cotton, for example, recent research by Gould et al. (1997) has confirmed much higher frequencies of resistance alleles in tobacco bollworm (Heliothis virescens) populations in the field - and hence, much faster potential rates of resistance development in the target insect - than was anticipated in the proponentís resistance management plans. One wonders why the proprietors of this technology would presume to release a targeted technology without knowing the resistance frequencies of the pest they were targeting. As stated by Tabashnik (1997), "Nothing will be gained and much can be lost if we pretend to know more about resistance management than we really do."

But the larger issue is the marvellously persuasive ways that ecological arguments against the rapid commercialization of GE technology have been deflected, bypassed, and ultimately ignored, by reasoning which is itself ecologically unsound. Three lines of reasoning are presented below in an effort to disabuse GE proponents of the comforting notion that GE organisms are generically "ecologically safe" (adapted from Kareiva and Parker, 1994).

1. Ecological risk cannot be framed in general terms - the risks associated with a particular GE trait or cultivar must be assessed on an individual case basis. It is not the process of genetic engineering that is the risk, but rather, the products. Some products may be ecologically benign, such as oil or protein quality traits affecting product value. Other products of genetic engineering, such as crop cultivars endowed with fitness-enhancing traits such as greater stress tolerance, insect resistance, or herbicide resistance, are more likely to become invasive - either in and of themselves or via transfer to other species. Hence, it is not possible to generalize or state categorically that GE crops do or do not pose an ecological risk. Ecological risks must be evaluated systematically and comprehensively for each proposed product, a process which has been attempted only sporadically for GE products currently or soon to be on the market.

2. Blanket assurances of ecological "safety" are without scientific merit, specifically because the criteria used to claim safety - such as Bakerís List of attributes to predict the potential of a cultivar to become "weedy" - or protocols used to test for biosafety (as reviewed by Ingham and Holmes, 1995) are ecologically unsound. For example,

a. Lack of weedy traits? Not reported as a weed? Proponents of GE crops have argued that GE crops are unlikely to become serious weeds, because they lack the attributes associated with weediness by Baker (1965; Table 1) and others. For example, Rissler and Mellon (1996) referenced Asgrow Seed Company as having claimed that a proposed transgenic virus-tolerant squash cultivar possessed only three of Bakerís traits, and hence, posed little risk of weediness. Yet, in a study of 49 British annuals, Williamson (1993) found only a weak correlation - of no predictive value - between Bakerís list of traits and weediness. Some of the worst weeds had few of Bakerís traits, while some with many of "weedy" traits were not, in fact, weedy (Table 1).

A similar table could be made for insect invaders, as for example, by Ehrlich (1986, 1989) as reported in van Lenteren (1992) (Table 2). However, so many exceptions exist to these general observations as to make them without value in anticipating or predicting the potential of a new GE introduction to become pestiferous. Yet, they are nonetheless used in just this way by proponents seeking to rationalize release of GE products.

Table 1. Bakerís List (adapted from Rissler and Mellon (1996)

Characters Associated with Weediness


Seeds germinate in many environments


Seeds remain viable a long time


Plants grow rapidly through the vegetative phase to flowering


Plants produce seeds continuously as long as growing conditions permit


Flowers are self-compatible, but not obligatorily self-pollinated


Pollen from flowers that are cross pollinated is carried by nonspecialized flower visitors (usually insects or by wind)


Plants produce large numbers of seed in favorable environmental conditions


Plants produce seed in a wide range of environmental circunstances


Plants are adapted for both long- and short-distance dispersal


If perennials, the plants have vigorous vegetative reproduction or regeneration from fragments


If perennials, the plants are brittle near the soil line to prevent easy withdrawal from the soil


Plants compete by special means, such as forming rosettes, choking growth, producing toxic chemicals

The potential for an GE entity - whether a cultivar, a microbe, or an insect - to become invasive cannot be predicted without targeted study, and even then, it is wise to recall that "Nature bats last". For example, the infamous Colorado potato beetle (CPB) originated in the eastern Rocky Mountains, from southern Mexico to Colorado, with Solonaceous species such as Solanum rostratum and S. angustifolium as its original, preferred hosts (van Lenteren, 1992). But potato (S. tuberosa) reportedly evolved further south - outside of this zone. Domesticated potatoes eventually reached Colorado - the original host range of CPB - in 1859. Inexplicably, the CPB then switched preferred hosts, becoming a significant, worldwide pest of potatoes from the early 1900's onward, despite having evolved in the absence of potato in the first place (van Lenteren, 1992).

This reality is as true of GE organisms as it is of naturally evolving organisms. Thus, one cannot claim to predict the ecological "safety" of anything without targeted testing throughout the zone of inference. It could be argued, with some merit, that the same reasoning would apply equally to both conventional plant breeding and to genetic engineering. But GE cultivars pose the greater risk, due in part to the emphasis on simply inherited, fitness enhancing traits which could realistically be expected to be selected for and retained by weedy relatives (see below).

Table 2. Characteristics predictive of potential for insect invasion (adapted from van Lenteren, 1992)

Successful Invaders

Unsuccessful Invaders
Large native range Small native range
Abundant in original range Rare in original range
Vagile Sedentary
Polyphagous Mono- or oligophagous
Short generation times Long generation times
Much genetic variability Little genetic variability
Gregarious Solitary
Fertilized female able to colonize alone Fertilized female not able to colonize alone
Larger than most relatives Smaller than most relatives
Associated with Homo sapiens Not associated with Homo sapiens
Able to function in a wide range of environs Able to function only in a narrow range

And finally, the test systems used by proponents to infer ecological safety have too often been chosen without real world considerations. As reviewed by Ingham and Holmes (1995), test systems typically consist of "sterile soil, without plants, and without other organisms present that can be affected or impacted". It is hardly surprising that under such conditions, no ecological impact can be detected.

No Reported Hybrids with Wild Species? Low frequency of outcrossing? It is argued, for example, that the risk of outcrossing to form fertile, weedy hybrids is negligible, either a) because there are no known weedy or naturalized relatives or b) because centuries of conventional plant breeding have failed to create a monster. Both of these premises are readily challenged.

Weedy Relatives: at home and abroad. Harlan (1965) has demonstrated that as many as one in 10 crops have wild races or close relatives that are already significant weeds. Crops such as barley, oats, and potatoes are known to have close weedy relatives, while a crop like alfalfa is already genetically similar to weedy relatives. Both shattercane and Johnson grass are weedy relatives of sorghum, just as wild mustard is a close relative of canola and other domesticated brassicas. The risk of outcrossing to wild relatives is amplified, with potentially devastating repercussions for germplasm conservation, when transgenic crops are grown in the Third World, where most food crops evolved.

And the time of that risk is fast approaching. The 31 January 98 issue of Correjo Braziliense, a major daily newspaper in Brasilia, the capital of Brazil, included an article noting the illegal import and planting of herbicide-resistant transgenic soybeans. The volume imported was sufficient for use by a number of growers. One of the suppliers of the illegal transgenic soybeans was an Argentinean subsidiary of Monsanto.

Evidence of Outcrossing Furthermore, traditional breeding has focused primarily on specific disease and pest resistance rather than on introduction of fitness-enhancing traits such as BT- or herbicide-resistance which could materially strengthen competitiveness should they reappear in weedy hybrids. Recent evidence summarized by Gledhill and McGrath (1997) clearly shows that these traits do move readily into adjoining populations. For example, Mikkelsen et al. (1996) demonstrated the ease of transfer of BASTA (glufosinate)-tolerance from oilseed rape (Brassica napus) (2n = 38 chromosomes) into a wild weedy relative (B. campestris) (2n = 20) under field conditions. In as little as two generations of crossing and backcrossing, they found the transgene for BASTA tolerance in fertile, B. campestris- like plants containing 20 chromosomes.

In another study, Chevre et al. (1997) examined the stability of hybrids (n=28) between transgenic oilseed rape (n=19) and wild radish (Raphanus raphanistrum) (n=9) through four generations of backcrossing between the hybrid and surrounding wild radish plants. Persistence of the herbicide-resistance transgene in 20% of the fourth generation hybrids was considered unusual, because hybrids between different species typically die out after the first generation. The potential for rapid spread of herbicide resistance into wild weedy populations was clearly demonstrated in these and other studies.

Evidence cited in Kling (1996) showed extensive movement of marker genes from cultivated to wild sunflowers and from sown to wild strawberries. He quotes ecological geneticist Norm Ellstrand of the University of California at Riverside as saying "...It will probably happen in far less than 1% of the products....but within 10 years we will have a moderate-to-large-scale ecological or economic catastrophe, because there will be so many products being released."

Mono- vs. Polygenic Traits As discussed by Kareiva and Parker (1994), a further difference is that yield increases through traditional plant breeding have generally involved polygenic traits, which would be difficult to transfer en masse. Conversely, GE traits are typically controlled by a single gene, enhancing the feasibility of outcrossing and gene transfer. Thus, the bold assertion that transgenic cultivars pose no greater risk (than traditionally bred cultivars) of outcrossing to create weedy hybrids ignores the reality that a) wild, weedy, cross-able ancestors do exist for transgenic crops, either here or abroad, b) single gene traits will be easier to transfer than polygenic traits, and c) unlike traditional plant breeding, GE cultivars are being engineering with fitness-enhancing traits that will increase the risk of weed creation.

3. Containment of transgenes is impossible - are the inevitable consequences acceptable?. Approval for commercialization of GE crops has been predicated largely on the notion that transgenes will be contained and controlled within a defined, sown field. Mounting evidence proves this to be a frail foundation for such a risky technology - so much so that Kling (1996) reported that it is already assumed by agricultural officials that gene flow to wild relatives will, in fact, occur. Thus, it is incumbent upon society to shift consideration of costs and benefits from the presumption that containment (avoiding outcrossing or escapes) is possible, to an explicit acknowledgement of the implications of escape. The issue is not if transgenes get into the environment but when (how fast) and "then what"?

The potential for enhanced fitness of weedy relatives is but one risk. Risks to potential ecosystem dysfunction, such as the following, are seldom considered.

a. Doyle et al. (1995) reported that Pseudomonas engineered to degrade 2,4-D also adversely affected soil fungi which can be critical for nutrient cycling and access, particularly in forest soils.

b. In Scotland, Birch et al. (1997) found that ladybugs (Adalia bipunctata) which fed on peach potato aphids ((Myzus persicae) which had in turn fed on GE potatoes4 produced up to 30% fewer progeny and lived only half as long as ladybugs feeding on aphids which had fed on conventional potatoes. A similar finding was reported from Switzerland, where the BT endotoxin from transgenic corn killed most cornborers, but the green lacewings which fed on the cornborers were also killed. Green lacewings, a beneficial insect, also died when fed on African cotton worm, which had survived feeding on BT corn.

In an Internet communication, Altieri (1998) commented that the aphids in the Birch et al. (1997) study had apparently sequestered the BT endotoxin and succeeded in transferring it to predators, in this case, the ladybug. He cited further evidence that the phenomenon of toxin sequestration followed by transfer from herbivore to parasite was not uncommon in nature. Movement of the debilitating impact of BT endotoxins through soil invertebrate communities, with consequent impacts on soil organic matter and nutrient cycling, could have far reaching effects on ecosystem function.

These are but a few documented examples of the types of unintended and potentially devastating ecological implications of genetic engineering, quite apart from generating superweeds.


The risks of genetic engineering of agricultural commodities are many and diverse. The ongoing concentration of power, with the proceeds of the chemical industry being used to absorb the seed trade, is a truly frightening prospect. The question of liability and culpability for adverse downstream implications, not just to the end users - the farmers - but to society and the environment, has barely been considered. Yet, these products are already on the market. The most profoundly harmful impact of genetic engineering in agriculture may well be the diversion of scarce research resources from addressing the real issues facing agriculture today (Clark, 1997b). Without rigorous and ongoing study, including the training of the next generation of researchers, extensionists, and policymakers, the prospects for achieving truly sustainable agriculture will be yet further circumscribed.

Recent research is directly confronting some of the fundamental tenets of the arguments used by proponents to convince regulators of the ecological "safety" of genetically engineered crops. Evidence was presented to support the conclusion that ecological risk cannot be assessed generically, on the process of genetic engineering, but rather, must be applied to the individual products of genetic engineering, on a case by case basis. To date, the criteria and protocols used to assess the potential for ecological risk are hopelessly naive and critically unsound. Genetically engineered cultivars are inherently more likely than conventionally bred cultivars to outcross and disperse transgenes into the wider environment, with potentially adverse consequences. Such consequences could include not simply producing superweeds, but widening impacts through additional trophic levels, including but limited to decimation of beneficial organisms.

It is unconscionable to allow the proprietors of genetically engineered crops to continue to develop, release, and profit from a technology that has been applied hastily and prematurely, without due regard for the consequences to society and environment, both today and tomorrow.


  1. Altieri, M.A. 1998. The Environmental Risks of Transgenic Crops: an Agroecological Assessment. Posted on 11 Jan 98 by info@natural-law.ca
  2. Baker, H. 1965. Characteristics and modes of origin of weeds. In: H.G. Baker and G.L. Stebbins (ed) The Genetics of Colonizing Species. pp.147-168. Academic Press, N.Y.
  3. Benbrook, C.M. and M. Hansen. 1997. Return to the "Stone-Age of Pest Management". Address before the EPA Public Meeting "Plant Pesticides Resistance Management", 21 March 97, Washington, D.C.
  4. Birch, A.N.E., I.E. Geoghegan, M.E.N. Majerus, C. Hackette, and J. Allen. 1997. Interactions between plant resistance genes, pest aphid populations and beneficial aphid predators. Scottish Crop Research Institute Annual Report. pp. 68-72.
  5. Chevre et al. 1997. Nature 389:924
  6. Clark, E. Ann. 1997a. A Ludditeís view of biotechnology. (http://plant.uoguelph.ca/faculty/eclark/biotech.htm)
  7. Clark, E. Ann. 1997b. Reorienting agricultural research back to the farm. Presented to Practical Partnerships: A New England Sustainable Agriculture Conference. Portland, Maine (http://plant.uoguelph.ca/faculty/eclark/maine.htm)
  8. Colborn, Theo, Diane Dumanoski, and John Peterson Myers. 1996. Our Stolen Future. Plume/Penguin Books, New York.
  9. Doyle, J.D., G. Stotzky, G. McClung, and C.W. Hendricks. 1995. Effects of genetically engineered microorganisms on microbial populations and processes in natural habitats. Adv. Appl. Microbiol. 40:237-287.
  10. Gledhill, M. And P. McGrath. 1997. Call for a spin doctor. New Scientist (1 Nov 97)
  11. Gould, F., A. Anderson, A. Jones, D. Sumerford, D.G. Heckel, J. Lopez, S. Micinski, R. Leonard, and M. Laster. 1997. Initial frequency of alleles for resistance to Bacillus thuringiensis toxins in field populations of Heliothis virescens. PNAS 94: 3519-3523.
  12. Harlan, J. 1965. The possible role of weed races in the evolution of cultivated plants. Eupytica 14:173-176.
  13. Ingham, E. And M. Holmes. 1995. Biosafety Regulations: A Critique of Existing Documents. An Occasional Paper of the Edmonds Institute, Edmonds, WA. 27 pp.
  14. Kareiva, P. And I. Parker. 1995. Environmental risks of genetically engineered organisms and key regulatory issues. An Independent Report prepared for Greenpeace International.
  15. Kling,J. 1996. Could transgenic supercrops one day breed superweeds? Science 274:180-181.
  16. Krimsky, S. 1982. Genetic Alchemy. The Social History of the Recombinant DNA Controversy. The MIT Press, Cambridge, MA.
  17. Lappe, M. and B. Bailey. 1997. Genetically engineered cotton in jeopardy. Unpublished mimeo.
  18. Lipson, M. 1997. Searching for the "O-Word". Organic Farming Research Foundation, Santa Cruz, CA 83 pp.
  19. Mikkelson, T.R., B. Anderson, and R.B. Jorgensen. 1996. The risk of crop transgene spread. Nature 380:31.
  20. Myerson, Allen R. 1997. Seeds of discontent: cotton growers say strain cuts yields. New York Times 19 Nov 97.
  21. OTA (Office of Technology Assessment). 1993. Harmful Non-Indigenous Species in the United States. U.S. Congress, Office of Technology Assessment, US Govít Printing Office, Washington D.C.
  22. Rissler, J. and M. Mellon. 1996. The Ecological Risks of Engineered Crops. The MIT Press, Cambridge, MA 168 pp.
  23. Simberloff, D. 1996. Impacts of introduced species in the United States. Consequences 2(2).
  24. Steingraber, S. 1997. Living Downstream: An Ecologist Looks at Cancer and the Environment. Addison-Wesley, N.Y.
  25. Tabashnik, B.E., Y-B Liu, N. Finson, L. Masoson, and D.G. Heckel. 1997. One gene in diamondback moth confers resistance to four Bacillus thuringiensis toxins. Proc. National Acad. Sciences 94:1640-1644.
  26. van Lenteren, J.C. 1992. Insect invasions: origins and effects. Pp.59-80. In: J. Weverling and P. Schenkelaars (eds) Ecological Effects of Genetically Modified Organisms. Netherlands Ecological Society, Arnhem, The Netherlands
  27. Williamson, M. 1993. Invaders, weeds and the risk from genetically manipulated organisms. Experientia 49:219-224.


  1. According to Simberloff, about 15% of the thousands of exotic plant, animal, and microbial species that have invaded the US have become "invaders", causing all manner of difficulty and expense. Estimated direct costs to taxpayers of dealing with exotic invaders in the US varies from the hundreds of millions to the billions of dollars (OTA, 1993).

  2. Developed in conjunction with Delta and Pine Land Company, the worldís largest producer of cotton seed

  3. More fundamental questions about the assumptions underlying the resistance management plan of the new GE crops were reviewed in Clark (1997)

  4. Engineered to contain lectin from snowdrops, which is known to interfere with insect digestion