ANNEX

The proposal put forward by EFSA in its new guidelines is not sufficiently rigorous. Indeed,methodology for the risk assessment of GMOs is an area of debate and controversyamongst scientists. However, in its guidelines, EFSA mostly adopts the methodologyfavoured by scientists who work closely with the agro-biotech industry. If the current draftguidelines are adopted by Ministers, priority is given to quick and easy market approval byignoring the need for a high level of protection for consumers and the environment.

1. Improvements to the new EFSA guidelines

“Problem formulation" is the essential prerequisite for a systematic risk assessment. In thisinitial phase of risk assessment, critical elements such as the determination of the natureand severity of the damage analysed (e.g. death or reduced mobility of an organism), oracceptable thresholds for damages should be determined in a consultative approach. Since"problem formulation" is also a normative and not exclusively scientific procedure, a broadrange of stakeholders should be involved. In its chapter on "problem formulation" EFSA onlyrecognises approaches that have been established by or with scientists working with theagro-biotechnology industry. Approaches by research groups developed outside of industrynetworks are not considered by the guidance. According to the guidance, the applicantcompanies have full freedom to determine these essential elements of risk assessment in away to receive quick and easy approval for their products. The guidance needs to beamended to ensure more transparent and credible problem formulation.

Choose appropriate starting point:

As laid down in the guidance document,1 EFSA introduces a "comparative safetyassessment" (also called concept of substantial equivalence or concept of familiarity) as adecisive step in the environmental risk assessment. The comparative risk or safetyassessment starts with a comparison between the genetically engineered plant and itsconventional counterpart. It ignores that many risks are specific to the genetically engineeredplant: the methods for introducing DNA are not based on the mechanisms of common generegulation and heredity. The newly introduced gene constructs have a specific potential toescape and /or disturb the process of normal gene regulation that is unique for this specifictechnology. Applying a risk assessment that is based on a concept of “comparison” as astarting point can be described as the same as comparing apples with oranges.The comparison between conventional breed plants and those produced by geneticengineering can only be a tool, but it is not a scientific concept or an adequate starting point.The EU should reject the idea that substantial equivalence can serve as safety assessmentin itself or that the concept of familiarity can be applied to environmental risk assessment.Instead, GM plants should be subjected to a risk assessment 'per se' that starts with a broadrange of data concerning the regulation of the inserted additional genes and its interactivitieswith the plants genome, its metabolism and its reaction to confined stress reactions.

Acknowledge and define scientific uncertainty

It is a first step in the right direction that the guidance requires applicants to expressuncertainty with respect to the risks of the cultivation of GM crops. However, the EFSAdocument does not describe any concept for how to assess and qualify these uncertainties1

and their consequences in the overall risk assessment. Such guidance would be crucial, asthere are many unknowns, i.e. unexpected pathways and components in the plants orunexpected toxicity to organisms. It is important that these uncertainties are addressedproperly to enable the Commission to apply the precautionary principle as required by EUlaw.

Define reasons for rejection

The guidance lacks clarification under what conditions a GM crop will be rejected onenvironmental safety grounds. For instance, if the persistence of a GM crop is not taken intoconsideration, it would make it extremely difficult to recall this GM crop if something goeswrong. The guidance should define as reason for rejection if genetically engineered plantsare showing pattern of persistence and / or invasiveness and cannot be recalled. In addition,unsustainable methods of cultivation intrinsically linked to a particular type of geneticallyengineered plant have to be defined as a reason for rejection of the market application.

Sound assessment of combinatorial and synergistic effects

According to the guidance document, combinatorial and synergistic effects, e.g. of differenttoxins produced by a GM plant or of different herbicides, do not have to be investigated.Empirical testing of these effects before any release is considered should be mandatory.Synergistic effects cannot be predicted from the properties of the single substances, that iswhy combinatorial effects have to be tested in empirical testing. Further, a closer interplaywith the assessment of pesticides, its residues and additives have to be established. If theplant produces pesticides (e.g. insecticides), reliable and comparable methods have to bedefined and made available in order to determine its content in the different parts of theplants and in the environment.

Full assessment of stacked events

Each so-called stacked event - GM crops combining different genetically engineered traits,e.g. tolerance to several herbicides and/or producing several pesticides – should be treatedas a new application and should undergo a comprehensive risk assessment includingempirical tests for combinatorial effects. So far EFSA is basing its risk assessment mainly onconsideration based on the assessment of the single genetically plants that were used tocreate stacked plants. This ignores the fact that synergistic effects (for example betweeninsecticidal toxins produced by the plants and the residues stemming from the application ofherbicides) cannot be predicted. Further, the risk assessment of the single geneticallyengineered plants will always leave uncertainties and cannot be regarded as a sufficientlysafe basis for all subsequent combinations of these plants.

Comprehensive assessment

Concerning risks for non-target organisms at all levels in the food web, such as soilorganisms, insects, aquatic organisms and red list organisms and wildlife species, have tobe included. It is not sufficient to reduce the risk assessment just to some low levels of thefood web (so called tiered approach) and then conclude that other levels do not need furtherinvestigations. Furthermore, long-term effects and accumulative effects have to be fullyintegrated into the risk assessment.