Comments Header
Experimental Design and Statistical Analysis:
Bt Corn, Lignin, and ANOVAs

by Eric Ribbens

I am not a teacher, but rather a research analyst at Friends of the Earth.  I recently presented Saxena and Stotzky's paper at a National Academy of Sciences forum on potential unintended health effects of genetically engineered foods.  I have a few comments on points you raised.

First, I find it interesting that hybrids derived from two distinct transformation events (MON810 and Bt11) exhibit the same unintended effect—increased lignin levels.  These plants were both transformed by gene gun, maximizing variability in chromosomal insert location.  One would think different insertion sites would yield different unintended effects (if the inserted promoter is somehow driving the increase in lignin, that is).  I suppose Cry1Ab protein could disrupt cellular metabolism somehow, producing the same effect in MON810 and Bt11.  This would require looking closely at the lignin biosynthetic pathway, which involves shikimic acid, an intermediate in the synthesis of aromatic amino acids.  This seems unlikely, though.

An interesting observation, perhaps unrelated, is that I have seen reports that Roundup Ready soybeans have increased lignin levels in some circumstances, and reduced levels of aromatic amino acids in others.  Could a non-specific effect of the engineering process be at play?  The problem here is that RR soybeans, I believe, are transformed via Agrobacterium.

I find your mechanical explanation unconvincing.  I don't think the paper says anything about corn borer infestation or damage; therefore why would the Bt varieties grow larger, especially when grown indoors where presumably there are no pests?

I would be interested to hear your comments.  I spoke with Stotzky, and he didn't have answers to the questions raised above.

Comments submitted 02/13/2003 by:

Bill Freese
Senior Policy Analyst
Friends of the Earth

Case Author's Reply:

Thank you for your interesting comments on the Saxena and Stotzky paper.  I agree that it is interesting that two of the three hybrid transformations do indeed seem to increase lignin concentration.  The evidence for the third transformation is incomplete, and it is interesting that the "Maximizer" hybrid with Bt actually has lignin levels comparable to the non-Bt hybrids.  I also think you are right that a strictly mechanical explanation is unlikely, particularly since the growth room levels also were increased.  Clearly there is a genetic component to lignin production, which was increased by the Bt transformation.  Equally clear is that there is an environmental component, as indicated by the lower lignin levels in the growth room (where plants probably received less mechanical stress / damage).

My main point in writing the Saxena and Stotzky case was not to examine Bt or lignin per se, but rather to examine an interesting recent paper with a flawed experimental design and a weak statistical analysis.  The authors probably overstate their results, given the very small sample sizes and limited samples (only from one node, only from one date).  Their use of t-tests is probably inappropriate, and in particular enables them to completely overlook potential interactions between site and transformation.  They also have no statistical analysis of different transformations overall, which could have been done.  Finally, their use of SEM as a measure of variability, particularly given that they did not report sample sizes, is in my opinion a mistake, because SEM is so highly dependent on sample size and carries little or no intrinsic information about variability around a mean.

By the way, my stats class just did this case last week.  My students were particularly interested in how this paper really does have some interesting stuff, but the potential punch is reduced by the mistakes.  We ended up concluding that part of the blame should be placed on the reviewers' shoulders.

Reply submitted 03/08/2003 by:

Eric Ribbens
Biological Sciences
Western Illinois University