First, we would like to remind you that the report produced by Monsanto had in no way tackled this question by gender and week after week, expect for the ANOVA.
Our method for tackling this issue has been the following:
- Graphic representation of the experimental curves on the growth charts of rats by gender and for the only groups fed on MON863 at a level of 11% and 33% and the corresponding groups fed on an isogenic maize (control groups).
- This representation highlighted differential discrepancies function of the gender and the dose: males fed on 11% GMOs were growing on average less quickly than those of the control groups, whereas the females fed on 33% GMOs were growing faster than those of the control groups. After a bibliographical research, we found out that in the case of the growth of mammals, Gompertz curves could be used to illustrate their growths.
- We decided then to adjust a Gompertz curve to the empirical curves of interests. The parameters were then assessed by non-linear regression on the basis of 14 average weight values corresponding to 14 weeks, with a view to plot a theoretical average curve corresponding to the empirical curve (in our paper, we talk about “the experimental and corresponding theoretical curves”). We have then represented the average theoretical curves by highlighting the error margins associated to each of the average values. In that case, the residues were reduced to the 14 differences between the average values and the Gompertz curve, which after testing could be considered as being in accordance with the usual conditions.
- Then these curves were actually compared; they were found to be significantly different. Our deduction was that the average growths of the concerned rats were different, and this element was then related to the haematological, chemical and urinary variables as a whole, which were apparently expressed in a differential manner (tables 2 & 3); this constitutes the most important element in our paper.
We would like to remind you that our approach did not consist in conducting a statistical study on the comparison of the models as the external CGB rapporteur did, but in comparing the average theoretical curves on the observed average values.
The three authors of the paper, Dr. D. Cellier, Pr. G.E. Séralini and Dr. Joël Spiroux de Vendômois