European Biotechnology Regulation: Contested Boundaries of Risk. Les Levidow, Susan Carr, and David Wield, Centre for Technology Strategy, Open University, Milton Keynes MK7 6AA, UK, tel. 44-1908-653759, fax 652175, email "L.Levidow@open.ac.uk"
Biotechnological 'risk' as a legitimacy problem
Since the 1980s there has been a wide-ranging risk debate over biotechnology -- in particular, over genetically modified organisms (GMOs) which are designed for commercial use in the environment. Public concerns have included: ethical qualms about 'interfering with nature', environmental harm from the early trial releases, long-term effects on agriculture, and the commercial motives of the agrochemical companies involved. An underlying issue has been the avowed aim of industrializing agriculture -- i.e., treating nature as a 'bioreactor' whose industrial efficiency must be optimized.
For all those reasons, biotechnology has faced a legitimacy problem. According to biotechnologists, they merely correct genetic deficiencies, thus supplementing or improving natural processes; they promise environment-friendly products for benignly protecting crops, even for 'feeding the world'. According to some critics, however, the most likely products will aggravate the familiar problems of intensive monoculture and technological dependency.
The wider public debate has featured incompatible concepts of 'risk'. For proponents, society is at risk of failing to reap the cornucopian benefits of biotechnology. For critics, society is at risk from biotechnology, whose development precludes or marginalizes beneficial alternatives. These divergent concepts of 'risk' each includes a divergent concept of benefit. In this way, the risk debate serves as an implicit technology assessment.
When scientists entered the public debate, they disagreed about what knowledge-base is necessary for assessing the environmental risks. Molecular biologists have portrayed GMOs as familiar products and thus as predictable, even as inherently safe. Ecologists have tended to regard the genetic novelty of GMOs as a source of environmental unpredictability; they have emphasized the need for careful safety testing and for better ecological knowledge. Some have warned that inserted genes might inadvertently confer a selective advantage, by analogy to the behaviour of some non-indigenous organisms entering a new environment Along with environmentalists, some scientists have warned that GMO products could impose a 'genetic treadmill', by analogy to the 'chemical treadmill' of pesticides which have generated resistant pests.
Given the substantial overlap between public and scientific concerns about risk, biotechnology proponents foresaw political dangers from open debate. For example, a 1989 OECD report expressed concern that scientific disagreements 'could be amplified and misunderstood in public opinion'. Yet environmental NGOs have understood such disagreements -- perhaps all too well. In this context, safety regulation has had the implicit role of managing public debate. Regulators seek means for credibly translating the risk debate into scientific terms, while separating risk assessment from technology assessment (TA).
Uncertainty-based regulation
In response to the risk debate, the European Community enacted
the Deliberate Release Directive 90/220. This precautionary legislation
aimed to prevent hazards not yet documented for GMOs. It provided
a framework for clarifying uncertainty about risk -- i.e., about
potential effects which may constitute or lead to environmental
harm. As an EC policy document later explained, 'there are concerns
that this new technology might entail potential risks...', which
were not officially defined.
As stated in the Directive, it aimed to 'establish harmonized
procedures and criteria', especially for market approval of products.
It aimed to 'complete the internal market' by avoiding trade barriers;
it modestly sought 'mutual recognition of data' among European
countries, rather than uniform 'standards'. For risk assessment,
the Directive left open the meaning of key terms, which have since
been interpreted differently by various national authorities.
In particular:
EU-level disharmonies and ambiguities
Member states of the European Union have given different answers to those questions, especially at the commercial stage. Every marketing application has met objections from at least one country, often on grounds that the product could affect agrochemical usage, as a relevant 'secondary effect'. It is mainly the Scandinavian countries which have interpreted the Directive more broadly, thus better accommodating environmental concerns. The EU-level conflicts can be illustrated by two GMO products in particular.
In one case, Plant Genetic Systems (PGS) requested market approval for a herbicide-tolerant oilseed rape. When the UK recommended approval, some member states objected that the herbicide-tolerance gene could spread and generate herbicide-tolerant weeds, thus potentially restricting options for weed-control methods and for government policy. In another difficult case, Ciba-Geigy requested market approval for a maize plant which contained genes for both an insecticide (a Bt toxin) and herbicide-tolerance. When France recommended market approval, some member states objected that product usage could intensify selection pressure for insects resistant to the naturally occurring toxin, thus eliminating a safe alternative to chemical pesticides.
In both cases, member states disagreed on the statutory relevance of so-called 'secondary effects'. Proponents of market approval argued that the Directive encompasses only 'product safety', i.e. the direct ecological effects of the GMO or its inserted genes. Others emphasized the Directive's requirement that all appropriate measures be taken 'to avoid adverse effects on human health and the environment, which they interpreted to encompass agricultural practices in using the product.
Beneath the official government stances, there lay some ambiguity in defining the relevant effects. In the PGS case, UK regulators judged that spread of the herbicide-tolerance gene could not cause harm to agriculture, because other effective herbicides were available. 'A weed is not a problem if you can control it', according to a member of the UK advisory committee. However, he also acknowledged, 'There is a grey area between safety and agricultural strategy... We think about this grey area all the time, so as to get a clearer definition of environmental harm.' 'Thus perceived advantages or disadvantages of a product may influence how regulators define harm; an implicit TA enters the safety judgement.
A similar ambiguity operated in France. For herbicide-tolerant oilseed rape, the French advisory committee perceived 'no uncertainties about an identified or potential risk'. Rather, it emphasized 'unknowns about socioeconomic consequences', i.e. the prospect of generating herbicide-tolerant weeds. Given those unknowns, the committee recommended that market approval be granted only on a five-year basis, with mandatory monitoring of commercial usage. How could this legal restriction be justified, if there is no environmental risk? According to the chairman of the French advisory committee, 'The term "environment" should be understood in a broad sense: herbicide-resistant weeds can be an environmental problem for the relevant agrosystem...' Thus environmental harm has been flexibly defined, depending upon the 'step' to be justified.
Indeed, the 'step-by-step principle' has had inconsistent interpretations. According to a UK regulator, 'If the hazard assessment says that the product is safe.... then there is little point in monitoring'. Nevertheless the UK funded safety trials of the PGS crop -- after recommending market approval. For the Bt maize, France proposed to require no monitoring but privately told other member states that large-scale releases would be necessary to provide extra data for an adequate risk assessment. In effect, the original 'step-by-step principle' was redefined, or even reversed. In effect, market approval would now authorize a large-scale experiment.
Market approvals also involve disputes over what labelling (if any) to require. Some member states argue that labelling may be required only 'in relation to the risk assessment' --which supposedly show that the products would be safe, regardless of how they are used. Other member states ask that products be labelled as 'genetically modified', regardless of their safety. Similarly, NGOs had demanded that any novel food be labelled -- in order to ensure an informed choice for consumers, e.g. the right not to buy food which had been produced via herbicide treatments. Thus the labelling issue compounds the dispute over the statutory relevance of herbicide implications.
A regulatory gap -- or implicit technology assessment?
Some regulators have acknowledged a 'gap' between GMO and agrochemical regulation, given that potential effects on chemical usage lie beyond those administrative boundaries. To call this a 'regulatory gap', however, wrongly implies that the problem lies beyond risk assessment On the contrary, the problem also lies inside risk assessment.
When regulators claim that a GMO product is safe, regardless of how it is used, they try to have it both ways. On the one hand, they assume that particular modes of crop protection are acceptable, or even desirable. On the other hand, they attribute any undesirable effects to those agricultural practices, rather than to the product as such. This distinction ignores (or denies) the in-built link between the use and design of a product. On this basis, their 'safety' claim serves to limits official responsibility for potential effects in commercial agriculture.
In such ways, 'risk assessment' conceptually narrows the range of potential effects which must be prevented, and whose implausibility must be demonstrated. When regulators define the relevant 'adverse effects' and appropriate 'steps', their judgements involve an implicit technology assessment (TA). They link 'risk' with presumed advantages or disadvantages of the product, according to their assumptions about agriculture and society. Such normative judgements were not specified by the Directive but remain essential for implementing it.
Consequently, the regulatory procedure has had contested boundaries -- of risk, adverse effects, steps, and product labelling. These also pose dilemmas. The more narrowly the procedure defines the statutory remit, the more readily it can accept the applicant's proposal, yet the less readily it can accommodate public concerns. On the other hand, the broader the remit, the less readily the procedure can justify market approval, especially as a final 'step'. Ultimate environmental effects of product usage are difficult to test, yet post-market monitoring implies that commercial use is a larger-scale 'experiment'.
In sum, as biotechnology further industrializes agriculture, its R&D devises chemical-like, genetic-level methods of environmental control. Safety regulation faces the awkward task of 'completing the internal market' for products which pre-empt agricultural options, e.g. by accepting an in-built 'genetic treadmill'; indeed, such products impose a model of 'the market' itself. Thus the contested boundaries of 'risk' regulation arise from its origins in the legitimacy problem of biotechnology as an implicit model of socio-natural order.
Acknowledgements: This paper arises from two studies: 'GMO Releases: Managing Uncertainties about Biosafety', funded by the European Commission's DGXII/E-1 during 1994-95; and from 'Precautionary to Risk-Based Regulation: the Case of GMO Releases', funded by Britain's Economic and Social Research Council during 1995-96.
Key references:
Les Levidow, Susan Carr, Rene von Schomberg, David Wield, 'Bounding the risk assessment of a herbicidetolerant crop', in Ad van Dommelen, ed., Coping with Deliberate Release: The Limits of Risk Assessment, pp.81-102, Tilburg: International Centre for Human and Public Affairs, 1996.
Susan Carr and Les Levidow, eds, 'Biotechnology Risk Regulation
in Europe', special issue of Science & Public Policy, June
1996, 23(3), forthcoming.