Nature Biotechnology 25, 35 - 36 (2007)

Early-tier tests insufficient for GMO risk assessment

Andreas Lang1, Éva Lauber2 & Béla Darvas2

  1. University of Basel, Institute of Environmental Geosciences, Bernoullistrasse 30, CH-4056 Basel, Switzerland. e-mail:
  2. Plant Protection Institute, Hungarian Academy of Sciences, H-1525, Budapest, Herman Ottó u. 15, PO Box 102, Hungary.

To the editor:

As field ecologists involved in risk assessment research on Bacillus thuringiensis toxin (Bt)-maize in Europe for several years now, we are very pleased that the publication last January of an article by Romeis et al. (Nat. Biotechnol. 24, 63–71, 2006) and an earlier paper by Firbank et al. Nat. Biotechnol. ( 23, 1475–1476, 2005) is encouraging discussion of the development of environmental risk assessment (ERA) of genetically modified organisms (GMOs). We hope that dialog stimulated by these articles will aid the design and further development of ERA. With regard to this, Romeis et al. claim that the first-tier laboratory experiments “are relatively simple in design [...] and the results are easy to interpret.” This may be true, but we question whether this automatically makes them ecologically meaningful, as seems to be the implication. With regard to any ERA, we would also point out that (i) acute exposure testing is not sufficient, (ii) region-specific tests need to be performed and (iii) that field tests need to be carried out.

Of immediate relevance to an ERA of predators and parasitoids is the case of Bt maize and its nontarget effects on (in particular, protected) butterflies, which has attracted considerable attention and provided several useful lessons. After a laboratory study1 describing a potentially adverse effect of Bt-maize pollen consumption on the Monarch butterfly (Danaus plexippus), extensive laboratory and field studies have been funded to investigate the potential risk of Bt-maize cultivation for the Monarch and other butterfly species2. In some of these studies3, 4, first-instar butterfly larvae were fed with Bt-pollen in excess for two to three days and the subsequent effects recorded, claiming “worst case conditions” based on the fact that the larvae were force fed with a huge amount of pollen. The butterfly larvae were unaffected by the pollen of MON 810 and other events (in contrast to pollen of the event Bt176). In the decision-tree proposed by Romeis et al., the ERA for MON 810 would have then stopped. However, Monarch butterflies are negatively affected when first-instar larvae are exposed to naturally occurring concentrations of MON 810 pollen over a continuous and realistic time period5. This may or may not have consequences at the population level; indeed, it was concluded it is more likely not to, because of a low exposure of the entire Monarch population to pollen-shedding Bt-maize fields. Even so, early-tiered tests, as suggested by Romeis et al. would have failed to identify the potential risk. An acute dose, even if several times higher than would be expected in the field, is not equivalent to a low natural dose experienced over a longer period, and there can be a great difference in the effects between acute and chronic exposure.

Results of experiments conducted by Darvas et al.6 have also shown that MON 810 pollen may negatively affect first-instar larvae of a Hungarian protected species, the Peacock butterfly (Inachis io). ERA must be region specific, as required by the Cartagena Protocol on Biosafety. For example, there are 187 protected Lepidoptera species in the Pannon Biogeographical Region, and especially species living on nettles (Urtica spp.) have larvae that are potentially affected by corn pollen shedding7. The corn-nettle association is the third most frequent plant relationship at cornfield perimeters in Hungary7. Tests have also revealed that the first-instar larvae of the Peacock butterfly are extremely sensitive to the Bt spray DIPEL, which contains Cry1 and Cry2 toxins (LC50 = 1 p.p.m.); the same is also true for larvae of the Comma butterfly (Polygonia c-album), another rare species living on nettles (LC50 = 19 p.p.m.). These values appear to be much lower than for lepidopteran pests in corn (suggested concentration of 500–1,000 p.p.m. in plant protection practice)8.

Many laboratory tests for GMO effects consider acute toxicity9, although chronic exposure is typical for Bt-corn and other transgenic insect-resistant crops. Moreover, the “worst case conditions” mentioned and performed by Romeis et al. are not in fact worst-case scenarios for lepidopteran larvae and for other invertebrate test organisms. Laboratory settings with ample food supply and favorable climatic conditions ensure that animals are in good condition, giving them an advantage when exposed to Cry toxins. In a 'real' worst case (under natural conditions), additional stressors, such as low temperatures, rain, food shortage or especially parasites and diseases10, 11 are likely to exacerbate a Bt effect on butterfly larvae. For example, Peacock butterfly larval populations are regularly reduced by a viral pathogen (cypovirus 2) and some parasitoids (Sturmia bella, Tachinidae and some Chalcidoidea) in Hungary. These controlling agents may divide a single population into different susceptible and tolerant subpopulations, modulating the effect of an additional pathogenic factor, such as Cry1Ab toxin-contaminated food.


In conclusion, toxicological early-tier laboratory tests may be simple to run and to interpret, but they will often be overly simplistic with respect to realistic exposure to hazardous agents (as exemplified by temporally delayed effects of Cry toxins, cypovirus 2, Sturmia bella parasitosis) and modulating environmental factors (the latter either increasing or decreasing a potentially adverse GMO effect). In consequence, there is a strong trade-off between simplicity of the ERA and its uncertainty. ERA should be handled flexibly and on a case-by-case basis, but field experiments should not be excluded in principle simply because early-tier tests indicate no harmful effects, at least not without sound and case-specific reasoning. Laboratory tests, as suggested by Romeis et al. are not worst case for Lepidoptera larvae owing to other factors influencing the final impact; therefore, one cannot discontinue investigation when early-tiered tests do not indicate an adverse effect.

Finally, the same Bt-plant (or other transgenic event) can generate different ecological consequences in different biogeographical regions. For these reasons, we strongly advocate additional field or semi-field tests in ERA of genetically modified crops, regardless of early-tier results. In particular, semi-field and/or field tests should be conducted in cases where indirect effects are to be expected, for example, when the genetically modified plant itself is a stressor in addition to the introduced transgenic trait, or where effects are triggered by land-use changes after large-scale cultivation of GMOs (as mentioned by Firbank et al.). This is not motivated by a search for what Romeis et al. term 'subtle' effects; in some cases, it may be the only way to detect any effect at all, including one of considerable significance. To simplify ERA for reasons of convenient administration or management procedure is not acceptable if it means simplifying and even ignoring existing, relevant biological knowledge.



  1. Losey, J.E. et al. Nature 399, 214 (1999). | Article | PubMed | ISI | ChemPort |
  2. Sears, M.K. et al. Proc. Natl. Acad. Sci. USA 98, 11937–11942 (2001). | Article | PubMed | ChemPort |
  3. Hellmich, R.L. et al. Proc. Natl. Acad. Sci. USA 98, 11925–11930 (2001). | Article | PubMed | ChemPort |
  4. Wraight, C.L. et al. Proc. Natl. Acad. Sci. USA 97, 7700–7703 (2001). | Article |
  5. Dively, G.P. et al. Environ. Entomol. 33, 1116–1125 (2004). | ChemPort |
  6. Darvas, B. et al. 1st Hungarian-Taiwanese Entomological Symposium, October 11–12, 2004, Budapest (2004).
  7. Darvas, B. et al. Növényvédelem 40, 441–449 (2004).
  8. Lauber, E. et al. 52nd Hungarian Plant Protection Days, February 23–24, 2006.
  9. Lövei, G. & Arpaia, S. Entomol. Exp. Appl. 114, 1–14 (2005). | Article | ISI |
  10. Pierce, C.M.F. et al. J. Econ. Entomol. 94, 1361–1368 (2001). | PubMed | ChemPort |
  11. Reardon, B.J. et al. J. Econ. Entomol. 97, 1198–1201 (2004). | PubMed | ChemPort |


These links to content published by Nature Research are automatically generated.