Turning a GM yield loss into a gain

April 2013

Arpad Pusztai has voiced his frustration that the kind of 'science' promoted by the biotech industry and its lobbyists is called 'science' at all:
“Science does not 'advocate' anything but seeks the truth and science cannot be 'junk' because then it is no longer science. Why don't we just use the term such as commerce research or something similar.”
His point is illustrated by an item published recently in a top science journal.

Early in 2013 a correspondence appeared in Nature Biotechnology. It described a statistical analysis of yields recorded for maize during 1990 to 2010, spanning the introduction of GM into agriculture. The data used were collected from annual commercial hybrid maize field trials* carried out by the University of Wisconsin Department of Agronomy.

*Note that such field trials provide valuable independent information to State farmers when choosing seeds. They compare new varieties coming to market with others historically available. This is research, but not science.

GM-freeCymru describe the study as “very strange”.

Its objectives were to “measure” the effect of GM insect- and herbicide-resistant traits on maize yield (specifically on yield variation), and on risk to farmers. Strange, indeed, that a study with these objectives should manage to conclude that “this technology seems important given current concerns about the effects of climate change on production uncertainty in agriculture”.

This leap of logic is based on statistical techniques which demonstrate less variability in the GM crop-types compared with all the conventional 'benchmarks' trialled. Calculation of the simple mean yield of each GM trait shows that two thirds had a higher yield than the conventional crops while one-third had a lower yield, which doesn't actually sound too good. However, the 'risk to farmers' was then measured using “hypothetical and essentially meaningless” figures (GM-free Cymru) to “document” how the lower yield variation in GM crops translates into a lower risk to farmers which translates into a yield gain, even if the actual yield is lower. (If you can't make sense of that, don't worry about it.)



The correspondence radiates positivity about GM crops all the way through.

However, since this isn't science, and there's no comparison of like-for-like going on here, what did it actually show?

The low variability in GM yields doesn't seem particularly surprising. Genetic transformation isn't an efficient technique, with the result that the low numbers of successful outcomes are generated into crops with limited underlying genetics. Since the yield potential depends on the underlying genetics, while the insect- and herbicide-tolerance traits only reduce losses when pest levels are actually troublesome, optimally-tended GM crops should have relatively stable yields. Also, they haven't been around very long (most of them less than 10 years). Conventional crops trialled by the University, especially over a 20-year period, will have very widely varying genetics.

The genetic range mis-match between GM and conventional is further widened by the comparison of nearly 20,000 different 'benchmark' observations with GM observations numbering from only 3,484 to as few as 36. 

Note that it's a well-known biotech industry trick to use the wide variation associated with conventional crops to 'prove' GM crops are 'substantially equivalent'. Here we seem to have the same trick to 'prove' GM yields are more stable. Neither are reasonable comparisons, nor are they 'science'.

Add to this that conventional breeding over the last two decades has increased yields by around 1.8 bushels per acre per year, quite independently of any GM tactics. Only five of the twelve GM trait-types studied have been around for even half that time. This means that the earlier samples, which were all conventional will have had lower yields than the more recent ones, and their overall mean used as a 'benchmark' must be an underestimate of contemporary conventional crop performance against which the GM yields are being judged.

Note that it may also be the case that the quality of conventional breeding programmes has declined since the advent of GM. This would make recent yield increases less than what would be expected or possible; it would also depress the conventional 'benchmark' and so inflate the GM yields.

GM-free Cymru has voiced its suspicion that there is a “deliberate bias in the paper, which should have been picked up in any responsible peer review process”.

Interestingly, the agronomist on the team which produced this paper (the other two authors seem to be economists) comments “It (yield) depends on how the transgene interacts with the underlying germplasm. My message to farmers is that every hybrid has to stand on its own”. This seems an acknowledgement that, statistical 'proof' notwithstanding, the underlying comparisons are pointless.

You can draw your own conclusions as to why this agronomist, who clearly knows his subject exceedingly well, should allow his hard-earned data to be used to present apparently misleading spin about how valuable GM is.

  • Quote of the month, GM Watch Review No.326, February 2013
  • Guamming Shi, et al., 2013, Commercialized transgenic traits, maize productivity and yield risk, Nature Biotechnology Correspondence, 31:2, February
  • Brian John, Do GM crops really have higher yields? The answer is no, GM-Free Cymru, 18.02.13
  • Nicole Miller, Value of modified corn is more in reducing losses than boosting yields, University of Wisconsin-Madison Press Release 14.02.13

No comments:

Post a comment

Thanks for your comment. All comments are moderated before they are published.