In GM crop approval practice, as in society in general, the conclusions on adverse effects and on their significance vary. This also pertains to subsequent requirements such as further experimental research, CSM or GS. Ultimately, however, decisions should be science-based , implying the use of data with sound statistical design and power (European Commission, 2001). Lövei and Arpaia (2005) and Perry et al. (2009) state that field studies for the ERA often lack such statistical design and power. Studies with NTOs to check for unexpected direct and/or indirect effects as previewed also by GMO Panel (2010) are largely missing.
Test species for NTOs risk assessment should be selected using a systematic, transparent, scientifically based and stepwise methodology as developed by the GMO ERA Guideline Project (Hilbeck and Andow, 2004; Hilbeck et al., 2006; Andow et al., 2008) and as refined by Hilbeck et al. (2008a). Furthermore, field sites for NTO testing should be selected to be representative for the receiving environments relevant to the market release as developed by Jänsch et al. (2010). Field studies should use plant material from the GMP and be performed in the laboratory, greenhouse or at the semi-field level, and in the field.
To date, the ERA in the application dossiers usually relies on tests originally developed and standardised for chemicals. These tests frequently do not examine the whole GM plants but only specific transgene products. Although this ecotoxicological testing concept is widely used, it does not fulfil the requirements of the Directive 2001/18/EC. A harmonised concept for ecotoxicological testing considering the whole GM plant characteristics is recommended, for example as outlined by Römbke et al. (2010). Furthermore, we recommend a more in-depth examination as to whether experimental data on specific potential adverse effects of GMHR sugar beet cultivation are scientifically-based and statistically sufficient to be upscaled (Breckling et al., 2009; Squire et al., 2009).
A general barrier to premarket tests and field studies is the research and publication control by GM seed companies. Under the threat of litigation, user agreements explicitly exclude the use of the seeds for any independent research (Waltz, 2009; Scientific American eds., 2009), and experimental results must be approved by the companies before being published. Thus, experimental data exhibiting potential adverse effects may not be made public and cannot enter the ERA process.
If potential adverse effects identified in the ERA are selected for further PMEM (Figure 3*), this also requires statistical multi-scale designs (Firbank et al., 2003; Stein and Ettema, 2003); this includes determining the environmental baseline status (European Commission, 2002) to identify adverse effects in a CSM and in GS. Science-based PMEM approaches are available (Graef et al., 2005a,b; Züghart et al., 2008) and should be carried out in a coordinated and harmonised way (Finck et al., 2006; Graef et al., 2008). The PMEM requires both static and flexible elements because agricultural systems are dynamic. They need to be adaptive as new information emerges (Lindemayer and Likens, 2009) and should feed back on the ERA as monitoring data becomes available.