The Real Scoop by Doug Gurian-Sherman
April 30, 2009
More Response to Criticism of Failure to Yield
I wrote in my last post about why Failure To Yield didn’t include studies on the performance of genetically engineered crops in the developing world. Here are my responses to two other critiques.
Critique: You should have included GE cotton in your analysis
Failure to Yield was motivated in large part by the “global food crisis” of the past few years. So we wanted to examine the ability of GE to address the challenges for food production given a growing global population, changing consumption patterns, and climate change impacts. For this reason, we decided to look at major GE food or feed crops in the United States, and this means soybeans and corn. We didn’t include canola, an oilseed crop, because the acreage devoted to canola, about a million acres, is only 0.6 percent of the acreage devoted to corn and soybeans in 2008.
Cotton was excluded because it is primarily a fiber crop. Cotton seed meal may also be used as animal feed, and the plant itself is used as fodder in some places, but these uses are secondary to fiber production. In other words, we did not look at GE cotton because the report is intended to inform the solution of the global food crisis, not a global clothing crisis.
Critique: The yield of corn [or soybeans] has increased faster since the introduction of GE varieties, and faster than in several European countries that do not grow GE varieties. And therefore GE has been responsible for increasing yields in the United States.
This argument, made by University of Georgia agronomy professor Wayne Parrott in a ten-page critique of Failure to Yield represents an elementary error in scientific reasoning: it attributes causality (as in: GE caused the increased yield) to a mere correlation between the coincidence of the introduction of GE traits and observed yield increases.
As the eminent late evolutionary biologist Stephen J. Gould wrote in a section of The Mismeasure of Man that is devoted to the problem of mistaking correlation for cause: "The invalid assumption that correlation implies cause is probably among the two or three most serious and common errors of human reasoning."
In this case the inference that GE traits caused observed increases in yield is unfounded because there are many other factors that could have caused yield changes.
For example, genomic-based methods (non-GE) of conventional breeding were also coming into widespread use in the 1990s, at the same time as GE. These methods can often cut the time it takes to breed new traits, including yield, in half, which in turn could greatly accelerate the rate of yield increase.
But to attribute the observed corn yield increases to genomic breeding methods without additional experimental data would be a big mistake—the same mistake made by Parrott.
In fact, the major reason we wrote Failure to Yield, was to try to tease apart the contributions of various factors contributing to yield. Failure to Yield sought out the controlled, published, field experiments that could be used to assess the contribution of GE traits to the yield of GE soybeans and corn in the United States apart from other factors. This approach allows us to subtract the GE yield contribution from U.S. Department of Agriculture data on yields for these crops over time, providing an estimate of the contribution of conventional breeding and improvements other than the addition of transgenes to the yield increases of these crops.
Parrott’s critique of Failure to Yield contains other factual errors and errors of reasoning that I will address in my next entry.
Previous: Failure to Yield Turns on Biotech Spin Machines Next: Responses to Failure to Yield Critics, Part II
