Select Committee on European Communities Minutes of Evidence

Can the risks from transgenic crop plants be estimated?—Tibtech, December 1996 (vol. 14)

  Companies have been developing transgenic plants for a decade or so. Some lines and some products are getting into commercial production. On supermarket shelves there is a tomato paste that declares itself to be genetically modified (i.e., transgenic). Companies have also been complaining about the tiresomeness of European regulation. Although pressures of trade, and from the OECD, will make rules increasingly similar worldwide, the European political structure has its own distinctive effect. Proposals for commercial release are first assessed by the "competent authority" in one member state. If a favourable opinion is reached, the proposal is sent to the competent authorities in the other member states. At this stage, objections can be raised, as illustrated by two transgenic plant lines that have run into trouble when voted on at the European level.

  The first transgenic plant line was PGS's herbicide tolerant oil-seed rape (Brassica napus). Although it was approved for production1, but not for human or animal consumption, several Scandinavian states had objected because of the implications for the use of herbicides2. The European Commission decided this objection was "not within the scope of" the Directive1. There appears to be a gap in the regulatory system, which may cause problems when proposals bridge the scopes of two different Directives, 91/414 (pesticides) and 90/220 [release of genetically modified organisms (GMOs)].

  The second transgenic plant line was Ciba's corn-borer tolerant maize (Zea mays), which also contained a gene for herbicide tolerance and a gene for an antibiotic marker under a bacterial promoter. This line failed to gain a qualified majority in the European Commission's Article 21 Committee. Competent authorities voted against or abstained for various reasons, including the risk of anti-biotic tolerance spreading in bacteria in the rumen of cattle, a possibility that Ciba had not considered adequately.

  These cases show that the competent authorities take hazards seriously but vary in their assessment of the risk. Is it possible to quantify the probability of a hazard happening; to quantify the risk to agricultural systems, the general environment and society as a whole? With the controversy surrounding bovine spongiform encephalopathy (BSE) buzzing in our heads, it would be a brave regulator who did not examine any hazard, particularly those declared remote (which usually means unknown).

What are the hazards?

  Of the ten or more potential hazards relating to the introduction of GMOs, the obvious agricultural and environmental ones are the spread of the transgenic plant, and the spread of the target transgene through hybridization. In this article, I will focus on recently published results that relate to these two hazards. Two other biological ones are the spread of associated transgenes (as in Ciba's maize mentioned above) and toxicity. For the latter there is already a cautionary tale, involving a lack of sufficient knowledge of food allergies. The development of soy bean containing a gene from Brazil nuts had to be abandoned at a late stage3.

  Transgenic plants may also accelerate the evolution of various resistant pests4. In addition, there are social, economic and legal hazards5 such as the use of a multiplicity of constructs for the same product6, conflicts of interest7, legal liability8, effects on the diversity of crops, effects on farming practices and problems from international trade9. The present regulatory systems only address the safety of biological effects, and there is no official standing forum for discussing the other problems5. Some of these problems are common to agriculture in general but are more acute for GMOs.

How do we quantify the risks?

  The natural way to quantify the agricultural and ecological risks is to do field trials and to consult the literature. But small-scale field trials are normally so constrained as to give little information on hazards, and frequently are not designed to allow sound quantitative comparisons9. A recent claim10 that 91 per cent of trials have had minimal risk, and the remaining ones a low one, depends on classifying the risks (a prejudgment) rather than on the experimental results, and on combining probabilities in a dubious way11.

  As an example of the problems involved in quantifying risks, I will describe the situation for oil-seed rape (canola), one of the commonest plants in European trials of transgenic plants. Surveys of the literature accompanying proposals have claimed that there is effectively no risk of hybridization with other species, and that pollen spread to other cultivars will be minimal. Experiments have now shown that there can be appreciable and effective hybridization with Brassica campestris12, the wild and weedy form of B. rapa (turnip, turnip-rape and some American canola), and with hoary mustard Hirschfeldia incana13, with effective gene flow from crop to weed. The hybrids are fertile despite large differences in chromosome number. In addition, studies of pollen flow show that experimental design can lead to order of magnitude discrepancies14 in the prediction of large-scale (kilometres) dispersal; well designed trials show that significant quantities of pollen travel long distances15. The mathematics of wind dispersal may be consistent with scale-free dispersal, i.e., dispersal to any distance16. Altogether, the literature and some trials have been misleading.

  Does oil-seed rape become naturalized, does it produce established, self-perpetuating, populations? It is now one of the most conspicuous road-side plants in Britain. One study17 around the M25 (the London orbital motorway) found that the populations were transient, as stated by the literature on these species. But surrounding fields in Scotland, at least, there are persistent populations18. Despite these demonstrations that much of the biology of B. napus had been misdescribed, the earlier risk assessments have not been revisited and revised.

Can we predict? What are the concerns?

  Well-designed experiments could give reliable information on a range of characteristics such as seed production, growth rate and survivorship at all stages of the life cycle9? These could be related to the fitness of feral crops and hybrids9. The major environmental and agricultural concern is of producing an intractable problem of invasiveness or weediness. In Europe there have been few such cases (an example is the spread of the ornamental shrub Rhododendron in Britain7) but there are many major problems in other parts of the world7. It is commonly thought, and supported by the practice of biological control, that many plants that are a pest as invaders, but not as natives, have lost their herbivores and pathogens in transit. Transgenics that are tolerant or resistant to pests are being produced in numbers and so, by analogy, a few may well become feral pests. The situation might be similar for other novel constructs.

  Two points are well established. The first is that although relatively few new feral plants become problems, perhaps about 1 per cent (Ref. 7), almost all crops occur as transient feral individuals. The other is that most measured characteristics fail to predict weediness or invasiveness7. We have to use unreliable tools, weak statistical relationships, to predict rare major events with long-term consequences.

  The analogy of transgenic plants with introduced invaders is often challenged because the introduction of a gene is said to make the plant less fit. Here I will just add that the intention of crop breeders is to produce novel plants, and that plants with novel genes will sometimes be fitter than their parents19, or may evolve to normal or superior fitness20. They might also give rise to new agricultural practices. Plants differing by rather few genes can differ markedly in invasiveness7. Familiarity is not a defence against environmental problems, still less against problems of toxicity, or social or economic woes.

  So there is a need for better experiments19, for regulators to be more critical of experimental designs and results, and for the different risks to be brought together for consideration. The costs, which are not negligible, should be regarded as an insurance against the types of disaster that have often followed invasions7. These costs have been, and will continue to be, borne by regulatory agencies and industry rather than research councils. They will do so in the interests of safety and consumer concern. Without these costs, many of the benefits of biotechnology could be lost.


  I am most grateful to F Amijee, J Bergelson, A Gray and L Levidow for providing information and comments.


  1. Commission for the European Communities (1996) Off J Eur Commun L 37, 30-31.

  2. Levidow, L, Carr, S, von Schoberg, R and Wield, D (1996) in Coping with Deliberate Release (van Dommelen A D, ed), pages 81-102, International Centre for Human and Public Affairs.

  3. Nestle, M (1996) New Engl J Med 334, 726-728.

  4. Mortimer, A M and Putwain, P D in Proceedings of a Workshop on the Environmental Impact of Genetically Modified Crops (Gray, A J, Gliddon, C and Amijee, F, eds), Department of the Environment GMO Research Reports, London (in press).

  5. Askew, M F in Proceedings of a Workshop on the Environmental Impact of Genetically Modified Crops (Gray, A J, Gliddon, C and Amijee F, eds), Department of the Environment GMO Research Reports, London (in press).

  6. DoE and MAFF (1995) Report of a Joint Workshop on Herbicide Tolerant Crops, Biotechnology Unit, Department of the Environment, London.

  7. Williamson, M (1996) Biological Invasions, Chapman and Hall.

  8. Green Alliance (1994) Why are Environmental Groups Concerned About Release of Genetically Modified Organisms into the Environment?, Green Alliance.

  9. Purrington, C B and Bergelson J (1995) Trends Ecol Evol 10, 340-342.

  10. Ahl Goy, P and Duesing, J H (1996) Biotechnology 14, 39-40.

  11. Landbo, L and Mikkelsen, T R (1996) Nat Biotechnol 14, 406.

  12. Mikkelsen, T R Andersen B and Jorgensen, R B (1996) Nature 380, 31.

  13. Lefol, E et al (1995) J Appl Ecol 32, 803-808.

  14. Crawford J W, Squire G and Burn D in Proceedings of a Workshop on the Environmental Impact of Genetically Modified Crops (Gray, A J, Gliddon C, and Amijee F, eds), Department of the Environment GMO Research Reports, London (in press).

  15. Timmons, A M et al (1996) Nature 380, 487.

  16. Shaw, M W (1995) Proc R Soc London Ser B 259, 243-248.

  17. Crawley, M J and Brown, S L (1995) Proc R Soc London Ser B 259, 49-254.

  18. Charters, Y, Robertson, A and Wilkinson, M in Proceedings of a Workshop on the Environmental Impact of Genetically Modified Crops (Gray, A J Gliddon, C and Amijee, F, eds), Department of the Environment GMO Research Reports, London (in press).

  19. Bergelson, J and Purrington, C B (1996) Am Nat 148, 536-558.

  20. Schrag, S J and Perrot, V (1996) Nature 381, 120-121.

previous page contents next page

House of Lords home page Parliament home page House of Commons home page search page enquiries

© Parliamentary copyright 1999