The study focused on the quantities of pesticides used on GM varieties as compared with conventional crops. There are other important indicators of environmental impact - such as the ecotoxicity of different pesticides and the timing of spray applications in relation to critical periods for wildlife - but pesticide volumes remain a major bone of contention in the GM debate.
The first generation of GM crops have affected pesticide use in two ways. Some were engineered to produce the bacterial toxin Bacillus thuringiensis (Bt) which kills some insect pests, enabling farmers to dispense with certain insecticides. Others are herbicide-tolerant crops which can be sprayed as they grow with broad spectrum herbicides such as glyphosate, simplifying weed control regimes.
Most previous studies of the impact of these crops on pesticide use in the US covered their initial commercialisation in the late 1990s. They generally found reductions in pesticide consumption - and the crop biotech industry was not slow to claim this as a major environmental asset for GM technology.
However, the picture appears to have changed significantly since then. The new study, by the Northwest Science and Environmental Policy Centre, is claimed to be the first to review pesticide usage trends over the eight years since commercialisation began in 1996.
The findings are based on official data on acreages and pesticide consumption for maize, cotton and soya, the three most important GM crops. In 2003, GM varieties accounted for between 40% and 81% of the total US plantings of these crops, and covered almost 110 million acres in all.
The key findings of the study (see figure) were that pesticide usage on the three crops was reduced by a cumulative 25 million pounds in the first three years of GM commercialisation. However, in the last three years, GM crops increased pesticide consumption by some 73 million pounds.
To put these figures in context, the biggest reduction in pesticide usage made possible by GM technology occurred in 1997, when total applications on the three crops decreased by about 3%. In contrast, estimated pesticide usage this year was some 12% higher as a result of GM crops.
The trend is not universally negative. Modest reductions in insecticide usage with Bt maize and cotton have been sustained over the eight years, with no sign yet that the targeted pests are developing resistance to the toxin.
The story is very different with herbicide-tolerant crops, where a combination of factors have pushed up pesticide usage on soya in particular.
One important factor was the ending of patent protection for Monsanto's glyphosate herbicide, the main broad spectrum product which crops have been engineered to tolerate. According to the report, competing production has halved prices of the herbicide since the mid-1990s, creating "an economic windfall for soyabean farmers who are paying less for herbicides despite spraying substantially more."
At the same time, the dominance of glyphosate in weed control strategies has put farmers on a familiar treadmill, as weed communities have shifted towards species less sensitive to the herbicide and the first glyphosate-resistant weeds have emerged. Farmers have responded by spraying more glyphosate - with the average application rate increasing by 22% on GM soya in 2002 alone. Similar increases have occurred in herbicide-tolerant GM maize.
The differences in herbicide usage between GM and conventional crops have been widened further by declining herbicide applications on conventional varieties, due to regulatory restrictions on high-dose products such as atrazine and the introduction of lower-dose herbicides such as S-metolachlor.
The report's authors commented: "For years, weed scientists have warned that heavy reliance on herbicide-tolerant crops would trigger ecological changes in farm fields that would incrementally erode the technology's effectiveness. It now appears that this process began in 2001 in the US in the case of herbicide-tolerant crops."
The US experience will not necessarily be mirrored in other countries due to differences in agronomic, economic, regulatory and environmental conditions. Nevertheless, it suggests that the recently completed farm-scale evaluations in the UK, which revealed a significant potential for herbicide-tolerant oilseed rape and beet to cause ecological damage (ENDS Report 345, pp 27-31 ), were too short to disclose their full environmental effects.