Impacts of Genetically Engineered Crops on Pesticide Use

Genetically engineered, herbicide-resistant and insect-resistant crops have been remarkable commercial successes in the United States. Few independent studies have calculated their impacts on pesticide use per hectare or overall pesticide use, or taken into account the impact of rapidly spreading glyphosate-resistant weeds. A model was developed to quantify by crop and year the impacts of six major transgenic pest-management traits on pesticide use in the U.S. over the 16-year period, 1996–2011: herbicide-resistant corn, soybeans, and cotton; Bacillus thuringiensis (Bt) corn targeting the European corn borer; Bt corn for corn rootworms; and Bt cotton for Lepidopteron insects. Herbicide-resistant crop technology has led to a 239 million kilogram (527 million pound) increase in herbicide use in the United States between 1996 and 2011, while Bt crops have reduced insecticide applications by 56 million kilograms (123 million pounds). Overall, pesticide use increased by an estimated 183 million kgs (404 million pounds), or about 7%. Contrary to often-repeated claims that today’s genetically-engineered crops have, and are reducing pesticide use, the spread of glyphosate-resistant weeds in herbicide-resistant weed management systems has brought about substantial increases in the number and volume of herbicides applied. If new genetically engineered forms of corn and soybeans tolerant of 2,4-D are approved, the volume of 2,4-D sprayed could drive herbicide usage upward by another approximate 50%. The magnitude of increases in herbicide use on herbicide-resistant hectares has dwarfed the reduction in insecticide use on Bt crops over the past 16 years, and will continue to do so for the foreseeable future.


Executive Summary
In a recent story tracking the emergence of weeds resistant to glyphosate (Roundup) herbicides, a North Carolina farmer said that "Roundup is the greatest thing in agriculture in my lifetime." A retired weed scientist admits in the same story "In hindsight, we screwed up. We can't rely on the same thing over and over." But farmers did, turning glyphosate herbicide and genetically engineered (GE) corn, soybeans, and cotton into the most stunning and profi table market success story in the history of the pesticide and seed industry.
Th is report documents some of the key impacts of GE crops on their way to market dominance and explains why the total pounds of herbicides applied on GE crops has spiked so sharply in recent years, with more increases to come.
But fi rst, some key terms are defi ned.
A "pesticide" is a chemical that controls pests. Th e term encompasses herbicides applied to control weeds, insecticides used to manage insects, and fungicides sprayed to manage plant diseases.
A pesticide "active ingredient" (AI) is the chemical (or chemicals) in a pesticide that is responsible for killing or otherwise controlling target pests.
"Pesticide use" is usually measured as pounds of pesticide "active ingredient" applied per acre, or on a given crop over some period of time.
A "trait" in a genetically engineered crop is the unique characteristic or attribute added to the genetic makeup of the crop using recombinant DNA (gene-splicing) technology. Th e capacity of a plant to withstand applications of a particular herbicide is an example of a GE crop trait.
"Stacked" GE seeds are those expressing two or more distinct traits.
"Trait acres" are the number of GE crop acres that contain a particular trait. One acre planted to a single-trait GE crop represents one trait acre, an acre planted to a "stacked" crop with two traits is equivalent to two trait acres, and so on. (Th is is why GE "trait acres" planted exceeds total GE crop acres planted).
GE seeds were introduced commercially in 1996 and now dominate the production of corn, soybeans, and cotton in the United States. GE crops contain one or both of two major categories of traits: • Herbicide-tolerant (HT) crops are genetically engineered to survive direct application of one or more herbicides during the growing season, chemicals that would otherwise kill or severely stunt the crop. Th e major HT crops are soybeans, corn, and cotton. Nearly all HT trait acres are planted to "Roundup Ready" (RR) seeds that tolerate applications of Monsanto' s glyphosate (Roundup) herbicide, the active ingredient in Roundup herbicide.
• Bt crops are engineered to produce toxins derived from the natural bacterium Bacillus thuringiensis (Bt) in plant cells. Th ese toxins are lethal to certain agricultural insect pests.

A. Th is Report
Th is report focuses on the impacts of GE crops on pesticide use, as measured by the total pounds applied on HT and Bt corn in contrast to conventional corn, HT soybeans in contrast to conventional soybeans, and HT and Bt cotton compared to conventional cotton.
Offi cial U. S. Department of Agriculture (USDA) surveys are the source of most of the data used in this report on the acres planted to each GE trait in corn, soybeans, and cotton. Annual "trait acreage" reports from Monsanto provide more nuanced data on the acres planted to crops with specifi c traits and trait combinations.
Th e data in this report on the acres planted to crops with each major GE trait are of high quality and are not controversial.
Pesticide use data come from annual surveys done by the USDA' s National Agricultural Statistics Service (NASS). Th ese surveys encompass the percentage of crop acres treated with each pesticide active ingredient, average rates of application, the number of applications, and pounds of active ingredient applied.
NASS pesticide use data are also of high quality and have stood the test of time, but NASS surveys do not report pesticide use separately on crop acres planted to GE seeds, in contrast to acres planted to conventional seeds. Hence, a method was developed for each GE crop and trait to estimate from NASS data how much more or less pesticide was used on a GE acre versus an acre planted with conventional seeds (for more methodological details, see Chapters 2, 4, and 5).
Th ese diff erences in pesticide use per acre are calculated by crop, trait, and year. Th e result is then multiplied by the acres planted to each GE crop trait in a given year. Last, the model adds together the diff erences in the total pounds of pesticides applied across all crops, traits and years, producing this report' s bottom line. It' s a big number --an additional 318 million pounds of pesticides were applied due to the planting of GE crops from 1996 through crop year 2008.

B. Key Findings
Farmers planted 941 million acres of GE HT corn, soybeans, and cotton from 1996 through 2008. HT soybeans accounted for two-thirds of these acres.
Bt corn and cotton were grown on 357 million acres, with corn accounting for 79% of these acres.
Th us, about 1.3 billion trait acres of HT and Bt crops have been grown between 1996 and 2008. HT crops account for 72% of total GE crop trait acreage. Th e actual number of acres planted to GE soybeans, corn, and cotton over this period is considerably less than 1.3 billion due to the prevalence of "stacked" versions of GE corn and cotton.

Impacts on Pesticide Use
GE crops have increased overall pesticide use by 318.4 million pounds over the fi rst 13 years of commercial use, compared to the amount of pesticide likely to have been applied in the absence of HT and Bt seeds.
Th e 318.4 million pound increase represents, on average, an additional 0.25 pound of pesticide active ingredient for every GE trait acre planted over the fi rst 13 years of commercial use.
Bt corn and cotton have delivered consistent reductions in insecticide use totaling 64.2 million pounds over the 13 years. Bt corn reduced insecticide use by 32.6 million pounds, or by about 0.1 pound per acre. Bt cotton reduced insecticide use by 31.6 million pounds, or about 0.4 pounds per acre planted.
HT crops have increased herbicide use by a total of 382.6 million pounds over 13 years. HT soybeans increased herbicide use by 351 pounds (about 0.55 pound per acre), accounting for 92% of the total increase in herbicide use across the three HT crops.
Recently herbicide use on GE acres has veered sharply upward. Crop years 2007 and 2008 accounted for 46% of the increase in herbicide use over 13 years across the three HT crops. Herbicide use on HT crops rose a remarkable 31.4% from 2007 to 2008. GE crops reduced overall pesticide use in the fi rst three years of commercial introduction (1996-1998) by 1.2%, 2.3%, and 2.3% per year, but increased pesticide use by 20% in 2007 and by 27% in 2008.
Two major factors are driving the trend toward larger margins of diff erence in the pounds of herbicides used to control weeds on an acre planted to HT seeds, in comparison to conventional seeds: • Th e emergence and rapid spread of weeds resistant to glyphosate, and • Incremental reductions in the average application rate of herbicides applied on non-GE crop acres.

Resistant Weeds
Th e widespread adoption of glyphosate-resistant (GR), RR soybeans, corn, and cotton has vastly increased the use of glyphosate herbicide. Excessive reliance on glyphosate has spawned a growing epidemic of glyphosate-resistant weeds, just as overuse of antibiotics can trigger the proliferation of antibiotic-resistant bacteria.
GR weeds were practically unknown before the introduction of RR crops in 1996. Today, nine or more GR weeds collectively infest millions of acres of U.S. cropland. Th ousands of fi elds harbor two or more resistant weeds. Th e South is most heavily impacted, though resistant weeds are rapidly emerging in the Midwest, and as far north as Minnesota, Wisconsin, and Michigan. In general, farmers can respond to resistant weeds on acres planted to HT crops in fi ve ways: • Applying additional herbicide active ingredients, • Increasing herbicide application rates, • Making multiple applications of herbicides previously sprayed only once, • Th rough greater reliance on tillage for weed control, and • By manual weeding.
In the period covered by this report, the fi rst three of the above fi ve responses have been by far the most common, and each increases the pounds of herbicides applied on HT crop acres.
GR pigweed (Palmer amaranth) has spread dramatically across the South since the fi rst resistant populations were confi rmed in 2005, and already poses a major threat to U.S. cotton production. Some infestations are so severe that cotton farmers have been forced to abandon cropland, or resort to the preindustrial practice of "chopping cotton" (hoeing weeds by hand).
Resistant horseweed (marestail) is the most widely spread and extensive glyphosate-resistant weed. It emerged fi rst in Delaware in the year 2000, and now infests several million acres in at least 16 states of the South and Midwest, notably Illinois. GR horseweed, giant ragweed, common waterhemp, and six other weeds are not only driving substantial increases in the use of glyphosate, but also the increased use of more toxic herbicides, including paraquat and 2,4-D, one component of the Vietnam War defoliant, Agent Orange.
Growing reliance on older, higher-risk herbicides for management of resistant weeds on HT crop acres is now inevitable in the foreseeable future and will markedly deepen the environmental and public health footprint of weed management on over 100 million acres of U.S. cropland. Th is footprint will both deepen and grow more diverse, encompassing heightened risk of birth defects and other reproductive problems, more severe impacts on aquatic ecosystems, and much more frequent instances of herbicide-driven damage to nearby crops and plants, as a result of the off -target movement of herbicides. crop year has tripled on cotton farms, doubled in the case of soybeans, and risen 39% on corn. Th e average annual increase in the pounds of glyphosate applied to cotton, soybeans, and corn has been 18.2%, 9.8%, and 4.3%, respectively, since HT crops were introduced.

Lower-Dose Herbicides Used with Conventional Crops
Th e second key factor responsible for the increasing margin of diff erence in herbicide use on HT versus conventional crops is progress made by the pesticide industry in discovering more potent active ingredients that are eff ective at progressively lower average rates of application. As a result of these discoveries, the average per acre amount of herbicides applied to conventional crops has steadily fallen since 1996. In contrast, glyphosate/ Roundup is a relatively high-dose herbicide and glyphosate use rates have been rising rapidly on HT crop acres, as clearly evident in the NASS data presented above.
Th e average rate of herbicides applied to conventional soybean acres dropped from 1.19 pounds of active ingredient per acre in 1996 to 0.49 pounds in 2008. Th e steady reduction in the rate of application of conventional soybean herbicides accounts for roughly one-half of the diff erence in herbicide use on GE versus conventional soybean acres. Th e increase in the total pounds of herbicides applied to HT soybean acres, from 0.89 pounds in 1996 to 1.65 pounds in 2008, accounts for the other one-half of the diff erence.
A similar trend is evident with insecticides. Corn insecticides targeting the corn rootworm (CRW) were applied at around 0.7 pound per acre in the mid-1990s and about 0.2 pound a decade later. Th e exception to this rule of dramatically falling pesticide use rates has been cotton insecticides targeting the budworm/bollworm complex. Th e rate of these products has fallen marginally from 0.56 to 0.47 pounds per acre.

C. Th e Road Ahead for GE Corn, Soybeans, and Cotton
Th e vast majority of corn, soybean, and cotton fi elds in the U.S. in 2010 will be sown with GE seeds. Th is is not a bold prediction because the non-GE seed supply is so thin now that most farmers will be purchasing GE seeds for the next several years, whether they want to or not.
Th e GE corn, soybean, and cotton seeds planted over the next fi ve to 10 years will, if current trends hold, contain increasing numbers of stacked traits (usually three or more), cost considerably more per acre, and pose unique resistance management, crop health, food safety, and environmental risks. HT crops will continue to drive herbicide use up sharply, and those increases in the years ahead will continue to dwarf the reductions in insecticide use on Bt crop acres.

Tipping Point for RR Crops
Crop year 2009 will probably mark several tipping points for RR crops. Th e acres planted to HT soybeans fell 1% from the year before, and will likely fall by a few additional percentage points in 2010. Farmer demand for conventional soybeans is outstripping supply in several states, and universities and regional seed companies are working together to close the gap.
Reasons given by farmers for turning away from the RR system include the cost and challenges inherent in dealing with GR weeds, the sharply increasing price of RR seeds, premium prices off ered for non-GE soybeans, the poorer than expected and promised yield performance of RR 2 soybeans in 2009, and the ability of farmers to save and replant conventional seeds (a traditional practice made illegal with the purchase of HT/RR seeds).
In regions where farmers are combating resistant weeds, especially Palmer amaranth and horseweed in the South, university experts are projecting increases of up to $80 per acre in costs associated with HT crops in 2010. Th is increase represents a remarkable 28% of soybean income per acre over operating costs, based on USDA' s bullish forecast for 2010 soybean income (average yield 42 bushels; average price, about $9.90).
Th e economic picture dramatically darkens for farmers combating resistant weeds under average soybean yields (36 bushels) and market prices ($6.50 per bushel). Such average conditions would generate about $234 in gross income per acre. Th e estimated $80 increase in 2010 costs per acre of HT soybeans would then account for one-third of gross income per acre, and total cash operating costs would exceed $200 per acre, leaving just $34 to cover land, labor, management, debt, and all other fi xed costs. Such a scenario leaves little or no room for profi t at the farm level.

Crop Effi cacy
Th e future of Bt transgenic crops is brighter, but if and only if resistance is prevented. Th e seed industry, the Environmental Protection Agency (EPA), and university scientists have collaborated eff ectively in the last 13 years in an eff ort to closely monitor and prevent resistance to Bt crops.
But now, some experts argue that the emphasis on resistance management in Bt crops can be relaxed. Th ey point out that the trend in the seed industry toward stacking multiple Bt toxins in corn and cotton varieties should reduce the risk of resistance. Th e EPA has apparently been persuaded by this argument, since it has approved several recent Bt crops with substantially relaxed resistance management provisions.
History suggests that lessened diligence in preventing Bt resistance is premature. It took 10-15 years for corn and cotton insects to develop resistance to each new type of insecticide applied to control them since the 1950s.
Bt cotton has now been grown for 14 years, but the acreage planted to it did not reach one-third of national cotton acres until 2000. Plus, the fi rst populations of Bt resistant bollworms were discovered in Mississippi and Arkansas cotton fi elds in 2003, about when experts predicted fi eld resistance would emerge.
Bt corn for CRW control has been planted on signifi cant acreage for only three years (2007)(2008)(2009). Bt corn hybrids for Eastern corn borer (ECB) control are still planted on just a little over one-half national corn acres. For both types of Bt corn, and especially in the case of Bt corn for CRW control, it is far too early to declare with confi dence that resistance is no longer a signifi cant threat.

Future Trends
Agricultural biotechnology fi rms have thus far devoted the lion' s share of their R&D resources to the development of only two biotech traits: herbicide tolerance and insect resistance. Pest control systems largely based on these traits are in jeopardy, biologically and economically, for the simple reason that they foster near-exclusive reliance on single pest control agents -season-long, year after year, and over vast areas of cropland. Th ese are "perfect storm" conditions for the evolution and spread of resistance.
Th ere is no serious dispute that RR crops have been popular, for the most part eff ective, and about budget-neutral for farmers. But they have fostered unprecedented reliance on glyphosate for weed control, and overreliance has spawned a growing epidemic of glyphosate-tolerant and resistant weeds.
Two major players in the industry -Monsanto and Syngenta -are now off ering to pay farmers rebates on the order of $12 per acre to spray herbicides that work through a mode of action diff erent from glyphosate. Monsanto' s program will even pay farmers to purchase herbicides sold by competitors, a sign of how seriously Monsanto now views the threat posed by resistance to its bread and butter product lines.
While corn, soybean, and cotton farmers view the spread of resistant weeds as a slow moving train wreck eroding their bottom line, the seed and pesticide industry sees new market opportunities and profi t potential arising in the wake of resistant weeds. A large portion of industry R&D investments are going into the development of crops that will either withstand higher rates of glyphosate applications, or tolerate applications of additional herbicides, or both. In short, the industry' s response is more of the same.
One major biotech company has applied for and received a patent covering HT crops that can be directly sprayed with herbicide products falling within seven or more diff erent herbicide families of chemistry. 2 Th ese next-generation HT crops will likely be sprayed with two or three times the number of herbicides typically applied today on fi elds planted with HT seeds, and the total pounds of herbicides applied on HT crops, and the cost of herbicides, will keep rising as a result.
Addressing the rapidly emerging problem of resistant weeds in this way makes as much sense as pouring gasoline on a fi re in the hope of snuffi ng out the fl ames. Despite these ill-conceived eff orts, unmanageable weeds with their roots in the Southeast will almost certainly continue to spread north and west, fi rst into the fringes but eventually throughout the Corn belt.
2 Herbicides within a "family of chemistry" work through the same mode of action.
Farmers will have to diversify weed management tactics and systems to deal with HR weeds. Deep tillage (left photo) buries weed seeds; cover crops (center) can repress weed germination and growth; and mechanized cultivation between plant rows (right) is a proven alternative to herbicides.

November 2009
The First Thirteen Years 8 Major weed management problems in the cotton industry in the Southeastern U.S. will not have a dramatic impact on U.S. agriculture or national well being, but what if the same fate lies ahead for corn and soybean producers? It well might in the absence of major changes in weed management systems and regulatory policies.
Instead of just spraying more, farmers must diversify the tactics embedded in their weed management systems, alter crop rotations, scrupulously follow recommended herbicide resistance management plans, and utilize tillage more aggressively to bury herbicide-tolerant weed seeds deep enough to keep them from germinating. . Sustaining the effi cacy of Bt crops is both important and possible. Th e emergence in 2003 of the fi rst, isolated fi eld populations of a major cotton insect resistant to Bt is troubling, but also reinforces the importance of today' s resistance management plans, which have kept the resistant populations found in Mississippi and Arkansas from spreading. Th e industry has recently proposed, and EPA has approved, backing away from Bt resistance management practices, steps that recklessly place the future effi cacy of Bt crops and Bt insecticide sprays at risk.
Overall pesticide use is bound to continue rising on GE corn, soybeans, and cotton. Even if the new, multipletoxin versions of Bt corn and cotton prove more eff ective in reducing insect pressure and feeding damage, the reduction in pounds of insecticides achieved as a result will be dwarfed by the continuing surge in herbicide use on HT crops.
Th e immediate and pressing goals for farmers, scientists and the seed industry include developing weed management systems capable of getting ahead of resistant weeds, assuring no lapse in the commitment to preserving the effi cacy of Bt toxins, and expanding the supply and quality of conventional corn, soybean, and cotton seeds. Th e last goal will likely emerge as the most vital, since the productivity of our agricultural system and the quality of much of our food supply begins with and depends on seeds.
Weeds, insects, and plant diseases can signifi cantly reduce the yield and quality of crops. Since the dawn of agriculture and around the world, managing pests has been a constant, annual, and unavoidable challenge for farmers. Th e eff ectiveness of steps taken to keep pest losses to a minimum has often meant the diff erence between life and death for families, tribes, communities, and even some civilizations.
Since World War II, pesticides have become the major tool employed by U.S. farmers to combat weed competition and insect damage. Th e term "pesticide" encompasses any chemical designed to control, manage, or kill a pest. Th ere are three major types of pesticides: herbicides to control weeds, insecticides to manage insects, and fungicides to control plant disease. Th ere are several other types of pesticides including rodenticides, nematicides (nematodes), antibiotics (bacteria), plant growth regulators, and miticides (mites).
All pesticides contain one or more "active ingredients" (AI). Th ese are the chemicals within pesticide products that are responsible for either killing a target pest outright, or undermining the ability of a target pest to thrive or do damage to a growing crop. "Inert ingredients" are added to pesticide products to improve the effi cacy and stability of a pesticide.
Pesticides work through many diff erent modes of action. Some modes of action disrupt one or more essential physiological processes within the target pest suffi ciently to kill the pest in a short period of time. Other modes of action involve blocking how a pest is able to digest food, impeding growth, or impairing reproduction.
Natural biochemicals like insect pheromones (scents that attract insects), botanicals, bacteria like Bacillus thuriengensis (Bt), and horticultural oils are also classifi ed by the Environmental Protection Agency (EPA) as "pesticides" because of their ability to help manage pests. Most of these work through a non-toxic mode of action and many are approved by the United States Department of Agriculture' s (USDA' s) National Organic Program (NOP) for use on certifi ed organic farms.

A. Tracking Pesticide Use and Risk
Th ere are two basic ways to track changes in reliance on pesticides: fi rst, the number of diff erent pesticides applied on a given acre, and second, the total pounds of pesticide active ingredient applied per acre in a given year. applied in 1964 for each pound of herbicide on major U.S. fi eld, fruit and vegetable crops. 1 Just seven years later in 1971, 176 million pounds of herbicides were applied, in contrast to 128 million pounds of insecticides.
Since 1971, the shift to much lower-dose insecticides has reduced the total pounds of insecticides applied to under 40 million in 2004. Herbicide use, on the other hand, rose from 176 million pounds in 1971 to 363 million pounds in 1997, despite the registration of several lower-dose herbicides starting in the early 1980s.
In 2004 across major fi eld crops, the ERS reports that 7.6 pounds of herbicides were applied for each pound of insecticide. Th e unmistakable dominance of herbicides in measures of the total pounds of pesticides applied is why the performance of herbicide-tolerant GE crops determines, for the most part, the impact of GE technology on overall pesticide use. Table 2.1 provides an overview of the acres planted and pesticide use from 1996 through 2008 for the three major GE crops: corn, soybeans, and cotton. Across these three crops and the 13 years covered in this analysis, 3.8 billion pounds of herbicides were applied, compared to 409 million pounds of insecticides -9.3 pounds of herbicides for each pound of insecticide. Cotton is clearly an exception in that insecticide use accounts for 43% of the total pounds of pesticides applied to that crop.
Environmental and public health problems with pesticides began to attract the attention of both scientists and citizens in the 1960s. Rachel Carson' s famous 1962 book Silent Spring deepened public Two decades later in 1991, corn farmers applied on average about two diff erent herbicides per acre. Since 1991 reliance has gradually increased and reached a peak of 2.78 herbicides applied to the average acre in 2001.
Corn growers have been less reliant on insecticides than on herbicides, as clear in Figures 2.1 and 2.2. Between 29% and 39% of national corn acres have been treated with an insecticide since 1971. Th is lessened reliance compared to herbicides refl ects two facts on the ground: • Weeds are a problem every year in every fi eld, while corn insects are episodic pests that cause problems serious enough to warrant treatments in only some regions and in some years; and • Planting corn and soybeans in a crop rotation is typically very eff ective in suppressing most important corn insect pests.
Increasing reliance by soybean farmers over time on a greater number of herbicides is evident in Figure 2.1, until the introduction of Roundup Ready (RR) glyphosate-resistant soybeans in 1996. Th e number of herbicides applied per acre fell from 2.7 in 1996 to 1.38 in 2005, although the number of herbicides applied on soybean acres is now rising as a result of the emergence of weeds resistant to glyphosate. Very few soybean acres are treated with insecticides.
In terms of the volume, or pounds of pesticide active ingredient applied per acre, there were about three pounds of insecticides awareness and concern over the impact of persistent, c h l o r i n a t e d hydrocarbon insecticides. Government scientists and r e g u l a t o r y agencies focused more attention on pesticide use and risks, both confi rming the existence of signifi cant environmental impacts from pesticide use, especially insecticides, and gaining insight into how pesticides were harming birds and other wildlife, as well as people.
As pesticide use grew in the 1970s and 1980s, so did evidence of adverse impacts on exposed wildlife populations and people. Th e regulation of pesticide use and risks became one of the dominant areas of focus for the EPA and the environmental community in the 1980s and through much of the 1990s. An overview of pest management, pesticide use and risks, and eff orts to move toward more preventionoriented pest management systems is provided in the 1996 Consumers Union book Pest Management at the Crossroads (PMAC). 2 A key theme of PMAC is that changes in crop rotations and other farming practices can sharply reduce pest pressure and reliance on pesticides.

B. Milestones and Major Impacts of GE Crops
Th e application of recombinant DNA technology in crop breeding, popularly known as genetic engineering, has been promoted by the biotechnology industry as another means to reduce pesticide use. Genetically engingeered (GE) crops were introduced commercially in the U.S. in 1996 and were rapidly adopted by corn, soybean, and cotton farmers.
By 1998, concern and controversy over the health and environmental impacts of GE plants had, for the most part, overshadowed long- standing worries over pesticide use and risk, both in the U.S. and Europe.
In part for this reason, there has been surprisingly little rigourous independent analysis of the pesticide use implications of GE crop technology. Th is lack of solid data is all the more surprising given that: 1) nearly all commercially grown GE crops have pest management traits that directly impact pesticide use practices; and 2) the technology is being implemented and promoted by agrichemical fi rms that have acquired a signifi cant share of the world' s seed supply.
Th is report attempts to fi ll an important gap in understanding of the impacts of GE crop technology by answering the following question: How have GE crops impacted pesticide use in the United States? We begin by providing brief overviews of the two major traits introduced into the three primary GE crops: herbicide tolerance and insect resistance in corn, soybeans, and cotton. GE crops with these traits comprise roughly 99% of all biotech crops grown (by acreage) in the U.S. from 1996 to 2008. 3

Herbicide Tolerance
Herbicide-tolerant (HT) crops are engineered to survive direct "post-emergence" application of one or more herbicides. Th e herbicide kills or severely stunts all or most growing weeds, while leaving the crop undamaged, or just modestly impacted for a short period of time. 3 GE canola has been planted on no more than 1 million acres annually; GE papaya is grown only in Hawaii on roughly 1,000 acres (and no where else in the world); the acreage of GE squash is unknown but almost certainly miniscule. GE sugar beets were not planted on a commercial scale until 2009.
A handful of HT crops was introduced prior to the advent of genetic engineering. Th e fi rst such crop, canola resistant to atrazine and related triazine herbicides, was commercialized in 1984. Interestingly, it was developed through recurrent backcrossing of canola with a related weed (Brassica campestris) from a population that had previously evolved resistance in the fi eld through repeated application of triazine herbicides. 4 Most other non-GE HT crops were developed through use of mutagenesis to be resistant to sulfonylurea and/or imidazolinone herbicides that inhibit the acetolactate synthase enzyme (ALS inhibitors). ALS inhibitor-resistant corn, soybeans, and canola were commercialized in 1992, 1994, and 1997, respectively, followed in the early years of this decade by resistant varieties of wheat, rice and sunfl ower. 5 It is worth noting that these crops were endowed with resistance to the two classes of herbicides to which weeds, at the time, had developed the most widespread resistance, in terms of both number of resistant biotypes and acreage infested. Th e fi rst major wave of herbicide resistance that began in the 1970s involved 23 species of weeds resistant to atrazine and related herbicides of the photosystem II inhibitor class, which have been reported to infest up to 1.9 million acres of cropland in the U.S. Th e second major wave began in the 1980s, and involves 37 species of weeds resistant to ALS inhibitors. Scientists have confi rmed that these resistant weeds now infest up to 152,000 sites covering 9.9 million acres (see Figure 2.4). 4 Tranel, P. J., and Horvath, D. P., (2009     Th ough acreage fi gures are diffi cult to come by, a market research fi rm recently estimated that non-GE herbicide-resistant crops were planted on roughly 6 million acres in 2007. 6 It was not until the advent of genetic engineering that HT crops became prevalent. Th is report deals only with GE HT crops. GE HT soybeans, cotton, and corn were introduced beginning in 1996 on just over 7 million acres, and their use expanded by nearly 20-fold to cover more than 132 million acres by 2008. In 2008, HT soybeans, cotton, and corn represented 92%, 93%, and 63% of total acres planted to each crop, respectively (see Figure 2.5, A major factor driving adoption of glyphosate-resistant (GR) crops has been the declining effi cacy of popular ALS inhibitors. Control problems emerged with ALS inhibitors as a result of the development of resistant weeds beginning in 1987, just fi ve years after the fi rst ALS inhibitor herbicide was brought to market in 1982. 8 As noted above, weeds resistant to ALS inhibitors were more prevalent than any other class of herbicide-resistant weeds in the U.S.
Another reason for the dominance of RR crop systems is ease of use and the effi cacy of glyphosate, an herbicide that kills a broad spectrum of weeds including annual and perennial broadleaf and grass species. RR-based cropping systems have been well received by farmers because they are simple, fl exible, and forgiving.
Prior to the commercial introduction of RR HT crops, glyphosate use was restricted to either before a crop was planted or new seedlings have emerged, or after a crop was harvested. Any direct applications on a growing crop were certain to cause signifi cant damage. RR technology widened the application window to allow post-emergence applications over the top of growing plants throughout the season, thus leading to dramatically increased use of and reliance on glyphosatebased herbicides. As discussed further below, RR crop systems have fostered a third wave of resistant weeds that poses a serious threat to agriculture, and are also profoundly shaping the biotech industry' s product pipeline. As yet, there has been no regulatory response to the growing epidemic of GR weeds.

Insect Resistance
In contrast to herbicides, insecticide use in American agriculture has declined sharply since the mid-1960s as a result of the shift away from chlorinated hydrocarbon and carbamate insecticides applied at about one pound per acre, to synthetic pyrethroid and other insecticides applied at one-half to one-tenth pound per acre, or less.
Insect-resistant cotton and corn varieties are genetically modifi ed to produce one or more truncated and activated forms of the toxins (e.g., Cry1Ab) derived from the soil bacterium Bacillus thuringiensis (Bt). Th ese so-called Bt crops were introduced in 1996, and the percentage of national crop acres planted has grown rapidly, as shown in Figure  2.6.
Acreage planted to Bt crops grew from 1.8 million acres of cotton in 1996 to 55.8 million acres of corn and cotton in 2008, as shown in Supplemental Table 6. Th e fi rst Bt corn varieties, and all Bt cotton varieties, repel above-ground Lepidopteron pests such as the European corn borer (ECB), Southwestern corn borer (SWCB), and cotton bollworm. Bt corn to control corn rootworm (CRW) and other soil-borne insects was introduced in 2003.
Bt toxins are biosynthesized continuously throughout the tissues of Bt plants, although genetic engineers have some ability to preferentially target (i.e., increase) expression levels in those plant tissues where the toxin is most needed to fend off insect feeding. Bt plant-incorporated toxins exert profound selection pressure for development of resistant insects by virtue of the plant' s continual production of toxin, in contrast to the intense but short-lived exposure characteristic of Bt insecticidal sprays.
Th e mode of action of Bt sprays and toxins is not completely known. Foliar Bt sprays contain inactive Cry protoxins (about 130-140 kDa in size) which exist in a crystalline form, when ingested. Th e alkaline nature of the fore-and mid-gut dissolves the crystal and cleaves it one or two times in the fore and midgut to create a truncated, activated toxin (about 60-65 kDa in size). Th e activated Cry toxins poke a hole in the gut epithelium, but it is unclear what causes insect death. Th e two proposed mechanisms are: 1) disruption of the mid-gut epithelium causes insects to stop feeding and starve to death, or 2) extensive cell lysis provides the Bt access to the hemocoel, where they germinate and reproduce, leading to septicemia and death. 9 Th e toxicity of Bt sprays is limited to those insects with the alkaline gut pH required to cleave and activate the protoxin. In Bt plants, the Cry toxins are already activated, increasing the potential for adverse impacts on populations of benefi cial insects. 10 Even before their commercial introduction, many scientists were concerned that Bt crops would accelerate the evolution of pest resistance to Bt toxins. 11 In response to such clear warnings from scientists and in the hope of delaying the emergence of resistance, the EPA mandated that Bt cotton and corn growers plant blocks of conventional (non-Bt) crop "refuges" amidst Bt fi elds to help slow development of resistance. Refuges work by maintaining populations of susceptible insects, some of which will mate with resistant insects, thereby diluting the presence of Bt-resistant genes in insect populations. EPA encourages "high-dose" Bt crops as another resistance management strategy; high levels of expression of Bt toxins lead to a more complete kill of target insects, and hence fewer surviving insects with the potential to pass along resistant genes. SmartStax corn varieties will be sold for the fi rst time in 2010 expressing six diff erent Bt toxins, three for the ECB and SWCB, and three more for the CRW.
New issues arise in assessing risks associated with the stacked versions of crops that have more than one Cry protein. Th ere may be a synergistic eff ect between the various Cry proteins which could aff ect the effi cacy of the various Cry proteins against their target and non-target organisms. Cross-resistance could emerge as a new challenge in managing resistance. Additional data will also be needed for human toxicity and environmental eff ects. 13 For instance, the EPA recently funded research to develop an animal model of allergenicity to better assess the potential for Bt insecticidal proteins to trigger food allergies. 14

C. Data Sources and Complications
Th is report is based on surveys of agricultural chemical use conducted by USDA' s National Agricultural Statistics Service (NASS). We chose to base this analysis on USDA data for several reasons. First, NASS supplies highly reliable data through use of transparent, rigorous methods and statistically representative sampling procedures. 15 Second, because the NASS program has collected annual pesticide usage data on soybeans, corn, and cotton for most of the years covered by this report, it off ers a consistent dataset that facilitates accurate, year-to-year comparisons. Finally, the public availability of NASS data (free of charge) facilitates open review and criticism of any analysis utilizing them.
NASS data are considered the gold standard of pesticide use information in the U.S. NASS reports provide a solid basis to study trends in the intensity of pesticide use across crops and  From 1991 through 2001, NASS surveyed pesticide use on major fi eld crops including corn, soybeans, and cotton on an annual basis. Annual summary reports have been issued with a set of tables covering pesticide use in all "Program States," 19 as well as at the national level.
Each standard table for a given crop reports the percentage of acres treated with a specifi c pesticide active ingredient, the average rate of application in pounds of active ingredient per acre; the average number of applications; the average rate per crop year, which is simply the one-time application rate multiplied by the number of applications; and the total pounds applied.
Benbrook Consulting Services (BCS) and Ecologic, Inc. have moved NASS survey data into a database program to carry out additional computations. For instance, average fi gures for individual and aggregate pesticide use in the Program States are applied to the small proportion of acres that NASS does not survey to arrive at estimates of total pesticide use for all crop acres in any given year. 20 19 "Program States" are those surveyed that year by NASS, and typically represent 85% or more of the national acreage planted to a given crop. 20 Th is is accepted practice, e.g. see "Agricultural Resources and Environmental Indicators: Pest Management Practices," USDA Economic Research Service, Report No. AH722, September 2000, Table 4.3.1, footnote 1, accessible at http://www.ers.usda. gov/publications/arei/ah722/arei4_3/DBGen.htm. "Th e estimates assume that pesticide use on acreage in non-surveyed States occurred at the same average rate as in the surveyed States."

The Organic Center
In years when a given crop was not surveyed by NASS (e.g., cotton in 2006), average values are interpolated between the previous and following year to fi ll in such data gaps. For corn since 2005, soybeans since 2006, and cotton for 2008, herbicide and insecticide use rates were projected from recent trends and in light of published reports on university websites regarding levels of pest pressure and the emergence of resistant weeds or insects.
Spikes upward in pesticide use are readily apparent in NASS data and have alerted farmers, scientists, and USDA to pestinduced problems in specifi c crops and regions. Such problems might be triggered by the emergence of resistance to a onceeff ective pesticide or the introduction of a new invasive species. Likewise, reductions in the frequency and intensity of pesticide use are regarded as evidence that farmers have made progress in adopting prevention-based Integrated Pest Management (IPM), perhaps through the planting of a new crop variety or adoption of a more complex crop rotation.
By combining NASS pesticide use data with EPA data on the toxicological potency of pesticide active ingredients, pesticide risk indices specifi c to diff erent classes of organisms, like birds or bees, have been calculated by the Economic Research Service (ERS) and other analysts. Such indices provide a useful early-warning system to detect changes in pest pressure, or pesticide effi cacy over time and in diff erent regions that may lead to "unreasonable adverse eff ects on man or the environment," the basic standard embedded in U.S. pesticide regulatory policy.
Scientists studying the emergence of resistance to a specifi c pesticide, or family of chemicals, rely heavily on pesticide-use data to determine the degree of selection pressure required to trigger resistance. 21 Epidemiologists exploring associations between pesticide use, exposure and patterns in birth defects or cancer often use NASS data in constructing retrospective estimates of exposure levels.

Impacts of USDA Decision to Stop Collecting Pesticide Use Data
NASS has dramatically scaled back its program in recent years. First, NASS replaced its annual surveys of major fi eld crops with less frequent ones beginning in 2002. Th en, in the 2007 growing season, data collection was limited to just two crops-cotton and apples. NASS did not collect pesticide use data on any crops during the 2008 growing season, citing a shortage of funds and the availability of private sector survey data as reasons for cutting the program. 22 Of the three major crops covered in this report, NASS data are available in most years for cotton through 2007, through 2006 for soybeans, and through 2005 for corn.
Th e absence of a continuous series of NASS data since 2005 for the three major GE crops hampers the ability of independent analysts and government scientists to track the performance and impacts of GE crops. Th e lack of NASS pesticide-use data covering recent crop years is a special concern, given the dramatic impact of resistant weeds on the number and volume of herbicides applied to HT crops.
USDA' s decision to drop the pesticide-use surveys led to strong protests from a wide range of groups, including Th e Organic Center, Center for Food Safety, Union of Concerned Scientists, Natural Resources Defense Council, and many other organizations, including several with close ties to the pesticide industry. 23 In 2008, the administrator of the EPA voiced concern to the Secretary of Agriculture about the loss of NASS data, joining several government offi cials at the state and federal levels.

D. Methodology
In this report a four-step methodology is used to calculate the diff erences in the amount of pesticides applied to GE crops versus conventional crops in a given year.
First, the total number of acres of each crop planted to conventional, HT and/or Bt varieties is derived from standard USDA sources: NASS for soybeans and corn, the Agricultural Marketing Serviee (AMS) for cotton.
Monsanto' s "Biotechnology Trait Acreage" reports are used to disaggregate total Bt corn trait acres to those planted to varieties engineered to control the ECB, the CRW, or both.
Second, the average amount of pesticides applied per acre per crop year is estimated for conventional GE crop acreage (detailed results in Supplemental Table 7).
Th ird and by year, the average amount of herbicides or insecticides applied to an acre planted to a conventional seed variety is subtracted from the corresponding amount for the GE crop.
Finally, in the fourth step, the diff erence in pesticide pounds applied per acre for each GE trait is multiplied by the acres planted to the GE crop in that year (full results appear in Supplemental Table 8). Th e impacts of herbicide tolerant and Bt crops on pesticide use per acre are then added together across the three crops over the 13 years of commercial use, producing the overall impact of today' s major GE crops on herbicide, insecticide, and all pesticide use.

Estimating Herbicide Application Rates on Conventional and HT Soybeans, Corn, and Cotton
Because the USDA does not report herbicide-use data separately on acres planted to conventional varieties, in contrast to GE varieties, an indirect method was developed that draws on NASS data. Th e method involves the use of a standard formula to estimate what is not known, from variables that are known from NASS and other data sources.
Th e average pounds of herbicides applied on all corn, soybean, or cotton acres in a given year are easily calculated from NASS data. Data are readily accessible on the share of total crop acres in a given year that were planted to conventional crop varieties, as well as the percentage planted to GE varieties. Th ese two percentages add up to 100% and can be used in a weighted-average formula, along with average herbicide use on GE crop acres, to calculate the pounds of herbicides applied on non-HT acres.
Th e basic weighted average formula, as applied to the pounds of herbicides used in producing HT and conventional acres of crop x , contains the following fi ve data elements, the fi rst four of which are known or can be projected from USDA. For a given crop and year, we calculated the impact of HT technology on herbicide use by subtracting the average rate applied to conventional acres (number 5 in above list) from the average rate applied to HT acres (number 4 in above list). When this number is negative, HT technology reduced herbicide use in that year for that crop; when it is positive, average herbicide use was higher on HT acres.

Insecticide Application Rates on Conventional and Bt Corn and Cotton Acres
In the case of Bt corn, two steps are required to estimate the impact of an acre planted to Bt corn for ECB/SWCB and/or CRW control on corn insecticide use. First, the average rate of application per crop year must be calculated for insecticides targeting the ECB and the CRW. Th is process is complicated by the fact that several insecticides are applied for control of both the ECB and CRW. For these insecticides, the portion of acres treated for control of ECB versus the CRW must be estimated. We reviewed pesticide labels, treatments recommended in university spray guides, and consulted with experts in corn IPM in carrying out this step (see Supplemental Table 9 for the share of insecticide acres treated targeting the ECB and Supplemental Table 10 for the share targeting the CRW).
Th e percentage of national corn acres treated with each insecticide for ECB/SWCB and CRW control was used to calculate a weighted average rate of insecticide application across all corn acres treated per crop year. Based on these calculations, the weighted average rate of insecticides applied on conventional acres for ECB control drops from 0.2 pounds of active ingredient per acre in 1996 Th e second step in calculating the pounds of insecticides displaced by the planting of Bt corn is to estimate the portion of acreage planted to Bt corn for ECB and/or CRW control that would have been treated with an insecticide if the corresponding Bt crop had not been planted. Th is step is required since Bt corn is now planted on far more acres than were ever treated with insecticides. Historically, USDA data show that before the advent of Bt corn, just 6% -9% of national corn acres were typically treated for ECB/SWCB control, while 27% +/-4% were treated for CRW control.
Supplemental Table 11 provides the details of this step and the resulting estimates of insecticide use averted through the planting of Bt corn for ECB and/or CRW control.
In the case of Bt acres targeting the ECB/SWCB, the likely share of acres planted to Bt corn that would have been sprayed for ECB control begins at 90% in 1997, the fi rst year of commercial planting, and drops incrementally to 45% in 2008, a year when over half of corn acres were planted to a Bt corn variety engineered for ECB control.
Th e high initial percentage is based on the assumption that early adopters of Bt ECB corn were more likely to have been farmers contending with serious ECB and/or SWCB infestations, triggering the need for insecticide applications. Th e falling percentage refl ects the progressively wide adoption of Bt corn by farmers with lesser ECB/SWCB problems, many of whom likely did not spray prior to the commercial launch of Bt corn.
In the case of Bt corn for CRW control, the percentage of acres planted that would likely have been treated with an insecticide targeting the CRW begins at 95% in 2003, the fi rst year of commercial sales, and declines to 60% in 2008, a year when 35% of corn acres were planted to a Bt corn for CRW control and another 9% of corn acres were sprayed for CRW control with an insecticide (i.e., about 44% of corn acres were either sprayed or planted to a Bt variety for CRW control, well above the 27% +/-4% level treated with insecticide for CRW from 1964 through 2008).

The Organic Center
Th is higher projected level of CRW treatment of corn acres is justifi ed in part by the emergence in the late 1990s of a variant of the CRW that learned to overwinter in soybean fi elds, thus undermining the effi cacy of corn-soybean rotations in reducing CRW populations.
Bt cotton targets the budworm/bollworm complex, but does not appear to have signifi cant eff ects on other insect pests, including the boll weevil, plant bugs, white fl ies, and stink bugs. Growers typically apply broad-spectrum insecticides to control both the budworm/bollworm complex and other insects. Bt cotton will reduce the use of insecticides for budworm/bollworm complex, but not applications of insecticides targeting other insects.
Supplemental Table 12 reports the basis for estimating the pounds of insecticides averted by each acre planted to Bt cotton. First, university insect management guides and experts were consulted to estimate the portion of total acres treated with each cotton insecticide for control of the budworm/bollworm complex versus other insects. Th en the number of acres treated with each insecticide is calculated from NASS data, as well as the share of total acres treated that was accounted for by a given insecticide.
Finally, weighted average use rates were calculated using the shares of total acre treatments with each individual insecticide. In the case of cotton, this weighted average insecticide application rate falls modestly from 0.56 pounds per acre in 1996 to 0.47 in 2007-2008.

E. Assumptions and Caveats
Th e methodologies used to project pesticide use on conventional and GE-crop acres require a number of assumptions and projections. Here, a brief description is provided of the major assumptions embedded in the Supplemental Tables that form the operating core of the model used to estimate the impact of GE crops on pesticide use. Each assumption or projection is also assessed in terms of its impact on our analysis of pesticide-use levels.

Farmers planting GE-crop varieties take advantage of the novel traits they are paying for.
For example, in the case of herbicide-tolerant plants, it is assumed that farmers build their weed management program around glyphosate herbicide. Likewise, a farmer purchasing a stackedtrait corn or cotton variety will alter both weed and insect pest management systems in accord with the purchased traits.
Th ese assumptions closely refl ect reality up to the 2009 crop season, but may not in the future as the seed industry moves toward more multiple-trait stacked varieties.

A small acreage of corn and cotton planted to GE herbicidetolerant varieties other than those resistant to Roundup are included in the herbicide-tolerant acreage estimates from the NASS and AMS. Herbicide use on these non-RR acres, however, is analyzed as if the acres were planted to a RR variety.
Perhaps 15 million acres have been planted to non-RR HT varieties over the last 13 years, a period during which approximately 941 million acres of RR crops have been planted. Accordingly, these non-RR HT acres account for just one out of every 63 acres of HT crops. In addition, the diff erences in herbicide use on non-RR HT crops, compared to RR crops, are modest. As a result, this assumption has virtually no impact on the outcome of the analysis.

The Organic Center
Th is assumption assumes close to 100% control of target pests, and overstates effi cacy in regions with high pest pressure, especially where multiple generations of target pests are common. As a result, the displacement of insecticide use is likely overstated in the case of some acres planted to Bt crops. For example, University of Illinois entomologists have documented spotty performance of Bt corn for CRW control, especially under high population pressure, and reported that some growers have applied soil insecticides on Bt-corn acres. 26 In fact, there was so much farm press media attention on the benefi ts of applying a soil insecticide on corn acres planted to a Bt corn for CRW control that the top entomologists in the University of Illinois felt compelled to ask -and answer "No" to -the following question in a widely read bulletin for growers: "Does it always make sense to use a soil insecticide in conjunction with a Bt [CRW] Hybrid?" 27 Accordingly, this assumption overstates the reduction in insecticide use on some Bt corn acres. But because corn insecticides are applied at relatively low rates, the impact of this assumption is modest. Th is could change dramatically, of course, if resistance emerges to the Bt toxins engineered into corn for CRW control, and farmers are forced to apply higher-rate insecticides to prevent serious CRW feeding damage.

It is possible to estimate the shares of the pounds applied of a given, broad-spectrum insecticide across multiple target insects, so that these shares can be used in estimating the rate of insecticide applications displaced by a given Bt trait.
Bt varieties have many complex impacts on insect communities and populations. In some fi elds, lessened insecticide use allows secondary pests to reach damage thresholds, triggering the need for additional insecticide sprays. 28 In other fi elds or perhaps in certain years, the reduction in insecticides targeting key Lepidopteron insects creates an opening for populations 26  Th us, crediting Bt corn for ECB/SWCB control with displacement of all the pounds of organophosphate or synthetic pyrethroid insecticides applied would overstate the impacts of the technology, since a portion of most of these insecticides are applied by farmers for the control of other insects, including the CRW.
Th rough consultation with insect pest management guides and entomologists, these shares were approximated for the key target pests of Bt-crop varieties. In some cases the shares used in the model likely overestimate displacement, while in others, displacement is likely underestimated. Given that most insecticides now applied to corn and cotton acres are low-dose products, discrepancies in these shares will have a modest impact on the pounds of insecticides displaced by Bt crops, especially relative to changes in the pounds of herbicides applied on HT acres.

Some portion of the acres planted to Bt corn do not displace insecticides because before the commercial availability of Bt-corn seed, farmers were not treating their fi elds with insecticides.
Historically, around 35% +/-4% of corn acres have been treated each year with an insecticide for control of the ECB, SWCB, CRW, and other insect pests. In 2008, 57% of corn acres were planted to a Bt variety, including many acres planted to a dual-Bt variety. For this reason, crediting each acre of corn planted to a single Bt trait with displacement of an insecticide acre treatment would substantially overestimate the reduction in insecticide use attributed to the technology.
As previously noted, corn insect pressure, however, has also changed in recent years as a result of the emergence of a new subspecies of the CRW that overwinters in soybean fi elds and disrupts the effi cacy of the corn-soybean rotation in reducing CRW populations.
Th is variant of the CRW was taken into account by increasing the share of Bt-corn acres assumed to displace insecticide applications to well above historic levels of insecticide use. Th e projections of Bt corn impacts on insecticide use refl ect a near doubling of the percentage of acres that farmers would likely spray with an insecticide, in the absence of Bt corn.
Th is assumption likely leads to a modest overestimate of the displacement of insecticide use caused by Bt corn, since corn farmers have other proven alternatives to reduce CRW populations through IPM systems. Regrettably, some corn farmers have lost interest in the multi-tactic approaches used in successful IPM systems as one consequence of the planting of Bt corn.

Th e Bt toxins manufactured within the cells of Bt crops are not counted as insecticides "applied" on Bt-crop acres.
Clearly, this assumption underestimates the pounds of insecticidal compounds required to manage insects on Bt crop acres. Opinions diff er among entomologists, the industry, and other experts on whether it is appropriate to count Bt toxins manufactured inside GE plants as equivalent to a liquid Bt insecticide sprayed on the outside of the plant. Uncertainty over the exact mode of action of Bt insecticides and GE toxins is part of the reason for diff ering opinions.
Th ose who argue that plant-manufactured Bt toxins should not count as equivalent to an applied insecticide assert that a Bt variety is just like any other new plant variety that has been bred to express some plant protein or phytochemical useful in combating insect-feeding damage.
Th ose skeptical of this position point to major diff erences in the two Bt delivery systems and in the source of the Bt toxin. Bt liquid sprays are applied only when and as needed, consistent with the core principles of IPM. Liquid sprays expose pest populations to short-lived selection pressure, thereby reducing the risk of resistance.
Bt plants, however, produce the toxin continuously during the growing season, not just when needed, and in nearly all plant tissues, not just where the toxins are needed to control attacking insects. In a year with low pest pressure, farmers can decide not to spray insecticides on a corn fi eld, but they cannot stop Bt hybrids from manufacturing Bt toxins in nearly all plant cells. 29 Th ere is another key diff erence that rarely is acknowledged. When plant breeders develop a new variety with a higher level of resistance to a given insect through traditional breeding techniques, they do so by selecting a top-yielding variety to crossbreed with another variety that expresses relatively higher levels of natural phytochemicals that discourage pest feeding, disrupt pest development or reproduction, or in some way reduce the viability of pest populations.
It is extremely rare for a new crop variety developed through conventional breeding to reduce insect feeding damage by killing the target insects. Instead, the elevated levels of phytochemicals in the new variety work through one or more non-toxic modes of action.
29 Moreover, from a food safety perspective, Bt toxins in liquid sprays break down relatively quickly in the fi eld when exposed to sunlight and hence do not end up in the harvested portion of crops.
Bt toxins in GE plants are inside plant cells, including the cells of the harvested portion of the crop fed to animals or consumed by people. Th is is a second reason why some entomologists reject the notion that there is nothing diff erent between a crop variety genetically engineered to synthesize Bt toxins within plant cells, and a new variety from conventional breeders that has improved resistance to an insect pest because of altered levels of natural phytochemicals that work through a non-toxic mode of action.
No resolution is in sight for this complex debate within the entomological community. In addition, no method exists to estimate the pounds of Bt toxins produced by a corn or cotton plant during a growing season. Hence, there is no way to project the pounds of Bt produced by an acre of Bt corn or cotton. Work is needed to develop such a methodology. It will likely show that there is a surprisingly large amount of toxin synthesized by plants during a typical growing season, especially in the new corn varieties engineered to produce six Bt toxins. In the case of GE cotton, USDA' s Agricultural Marketing Service (AMS) has a more accurate breakdown of trait categories by acreage than NASS/ERS. AMS' s annual "Cotton Varieties Planted" reports 3 are favored for these data by cotton experts, 4 and also provide fi gures that are in closer agreement with the information on GE cotton trait acres released periodically by Monsanto.

Th e last NASS survey of soybean herbicide use was in
Supplemental Table 2 reports the ERS data on the percent of corn acres by state and nationally planted to HT varieties, Bt varieties, and stacked varieties (one or more Bt genes, plus herbicide tolerance). Supplemental Table 3 covers herbicide tolerant soybeans, and Supplemental Table 4 presents both percent of national acres and absolute acreage planted to various GE cotton trait categories.

A. Acres Planted
Th e percent of national corn, soybean, and cotton acres planted to GE crop traits is presented in Figure 3.1. Soybean and cotton HT seeds were adopted rapidly by farmers.  Table 3.1 come  from Supplemental Tables 5 and 6, where HT and Bt crop acreage, respectively, is reported for all years.
HT crops clearly account for the lion' s share of total GE trait acreage -72% over the fi rst 13 years of commercial use and around three-quarters in most years. HT soybeans account for almost one-half of all GE trait acres. Th is is why HT soybeans are so important in terms of the overall impact of GE crops on the pounds of pesticides applied.
As discussed in Chapter 2, we assume in this report that when a farmer purchases a variety with a given trait, the farmer relies on that trait in carrying out his/her pest management program. Yet this is not always the case, either because the trait does not perform well enough, or because it is not utilized by the farmer.
Some traits do not perform well enough to allow the farmer to completely forego pest management measures more typical of the conventional grower. For example, several Midwestern universities have documented the need for insecticide applications to avoid serious root damage in fi elds planted to Bt corn for CRW control, and many farmers are making such applications. 5 In other cases, superfl uous traits go unutilized. For example, corn hybrids engineered to tolerate two diff erent herbicides are on the market, yet only one HT trait will likely be utilized by most farmers. 6 Many corn hybrids express the Bt gene for both ECB and CRW control, yet many farmers buying these hybrids face economically damaging levels of only one, or neither, of these insects, in most years. Why would farmers buy corn seed with unnecessary traits? Because such varieties are the only ones available with other valuable genetic traits matched to a particular farm' s soils, maturity zone, and production system.
Th is tendency to under-utilize GE traits is likely to increase markedly in frequency (i.e., the number of fi elds impacted) Th e strategy of off ering farmers more multiple-trait stacked varieties and fewer single-trait varieties is referred to in the industry as "biotech trait penetration." 9 Th is strategy is, in turn, driven by the fee-per-trait pricing structure used across the industry. For instance, Monsanto and Dow AgroSciences recently announced a collaboration to develop so-called "SmartStax" corn hybrids that contain eight GE traits stacked 7 Monsanto (2009 Th e commercial introduction of these varieties raises several new issues and questions, some of which are addressed in Chapter 7.

New Challenges in Tracking GE Traits and Acres
Th e trend toward stacked traits also raises analytical challenges. In corn and cotton, the total number of GE trait acres now far exceeds the total number of acres planted. Th e tracking of GE seed traits will be complicated by other factors. As the trend toward more multiple-trait varieties continues, seed companies may begin to neither announce, nor charge, for the presence of certain traits, including those that become obsolete (e.g., the RR trait will become obsolete if and when, and wherever the spread of resistant weeds renders the herbicide ineff ective).
In other cases, farmers will be forced by lack of choice to buy a variety that contains traits of little or no use. For this reason, future surveys of GE crop traits will need to explore ways to distinguish between total pest management related trait acres and "functional" trait acres, where a given trait actually changes how the farmer manages pests and the crop.
Glyphosate herbicide, marketed as Roundup by Monsanto, has been and remains the backbone of HT cropping systems. Th e effi cacy of RR technology was excellent in the fi rst few years of commercial use. A single application was often all that farmers needed for season long control in corn and soybeans. Typically, an additional application of Roundup or another herbicide was necessary in cotton growing areas, because of the longer growing season and many aggressive weed species in cotton country.
Shifts in weed communities favoring those species not as fully controlled by Roundup started occurring after just a few years of use on the same acre of cropland. After four to six years of applications, such weed shifts to more glyphosatetolerant species had led to higher rates of Roundup and/or additional applications. In areas where farmers grew RR crops in rotation, like RR soybeans followed by RR cotton, weed populations resistant to Roundup began to emerge and spread.

Impacts of Herbicide-Tolerant Crops on Herbicide Use
Th ese changes in weed communities -shifts to more GT species and evolution of glyphosate-resistant biotypes -have driven the incremental increases in both the rates and number of applications of glyphosate and other herbicides required on HT acres.
Th e title of a recent university extension report to Illinois farmers about the utility of glyphosate-based weed management systems states: "Turn Out the Lights -Th e Party' s Over." 1 In the article, Aaron Hager asserts that: "Th e rapid adoption of glyphosate-resistant corn hybrids and weed spectrum changes in response to near-ubiquitous use of glyphosate in soybean suggests the following theses: the ability of glyphosate to be a stand-alone herbicide for weed management in soybeans  will (continue to) decline. In other words, the ' simplicity' of glyphosate as a stand-alone weed management tool soon will be relegated to the annals of history."

[Emphasis in original]
Th is ecological adaptation to the RR system was predictable and openly discussed well before the fi rst RR crop was planted. A publication issued in 1990 by the Biotechnology Working Group focused on the impacts of HT crops on sustainable agriculture. It stated nearly 20 years ago that: "If a shift to herbicide-tolerant crops led to greater use of certain herbicides,... problems associated with resistant weeds would likely increase." 2 In the 1996 Consumers Union book Pest Management at the Crossroads (PMAC), 3 the "special caution" needed in managing GE crops was highlighted. After discussing the possibility that gene fl ow could create "super" HT weeds, 4 the report warns that: "A more widespread concern with herbicide tolerant plants is the likelihood they will accelerate the emergence of resistant weed species… "Gaining the ability to apply the herbicides more frequently or possibly at higher rates is the major reason farmers are willing to pay the higher cost for transgenic seed. Such changes in the pattern of herbicide use, though, are almost custom-made for accelerating resistance." (page 220, emphasis added) Th e impact of shifts to weed species more tolerant of glyphosate and the evolution and spread of GR populations is unmistakable in USDA pesticide use data over the last 13 years. In corn, the pounds of glyphosate applied rose "only" 4.3% per year. Th e reason is clear --RR corn was adopted much more slowly than HT cotton and soybeans. Market penetration did not reach a third of national corn acres until 2006. Accordingly, corn farmers are just now entering the time period when substantial increases are likely in glyphosate application rates, unless farmers switch to other herbicides and weed management technology.

A. Herbicide-Tolerant Soybeans
Th e general procedure for estimating herbicide use on conventional and GE acres was described in the methodology section in Chapter 2. Here, the methodology is briefl y summarized and issues specifi c to each crop are discussed.
Th e average number of pounds of herbicides applied to HT acres is composed of the volume of Roundup applied plus an estimate of the pounds of other herbicides needed to achieve eff ective control.
Total herbicide applications on acres planted to conventional seeds is calculated by use of a weighted-average formula computing the average pounds of herbicides applied on all acres from the pounds applied on conventional and GE acres, coupled with the shares of acres planted to HT and conventional varieties.
Th e average pounds applied on acres planted to conventional seeds is then subtracted from the average pounds applied to HT acres, producing the diff erence in herbicide use on an acre of HT crop, in contrast to acres planted to conventional varieties.
Herbicide use rates on all soybean acres, HT acres, and conventional acres are computed in Supplemental Table 15 and are displayed graphically in Figure 4.1.
Th e values in the line "Glyphosate on RR Acres" in Supplemental

Special ERS Tabulation in 1998
Th e ERS carried out a series of special tabulations of herbicide use data on HT and conventional soybean acres drawing on crop sample points in the 1998 Agricultural Resource Management Survey (ARMS). Th is tabulation was requested and paid for by Benbrook Consulting Services. In this tabulation, ERS analysts divided all soybean acres into four categories: * Conventional varieties, no glyphosate applied; * Conventional varieties, glyphosate applied (mostly on no-till acreage); * RR varieties; and * Other HT varieties.
From the ARMS soybean dataset, ERS calculated both the percent of total soybean acreage by category, as well as the average number of herbicides and pounds of herbicides applied in each category. Th is information was used to calculate total herbicide use per acre on conventional and HT soybeans in 1998, using the weighted average formula described previously, as shown in Table 4.2.
Th e rates and percents of acres planted to conventional varieties treated and not treated with glyphosate were used to calculate the overall conventional soybean rate of 1.13 pounds per acre. Conventional acres treated with glyphosate were planted using either no-till or conservation tillage systems in which the glyphosate is applied before soybean seeds germinate.
Th e average rate of all herbicides applied on HT acres was calculated at 1.2 pounds per acre. Accordingly, the average acre of HT soybeans in 1998 required 0.07 pounds more herbicide than the average acre of conventional soybeans.

B. Herbicide-Tolerant Corn
Adoption of HT corn increased more slowly than HT soybeans and cotton, in large part because of several cost-eff ective, herbicide-based weed management alternatives. By 2001, 68% of soybeans and 74% of cotton acres were planted to HT varieties, whereas just 8% of corn acres were planted to HT seeds.
Farmers were slower to adopt the higher cost HT corn varieties because, in general, corn weed management is simpler than soybean or cotton weed management. Corn germinates and grows quickly, producing a "closed canopy" earlier in the crop season than in soybean and cotton fi elds. A crop has a "closed canopy" when the foliage of the crop fully shades the ground from direct sunlight. Weed germination and growth slow dramatically once a crop canopy is closed.
As in the case of soybeans, projections of herbicide use on HT corn acres are based on the performance of the RR system. NASS data on corn herbicide use suggest that between 2% and 5% of corn acres in some years were treated with glufosinate, the active ingredient associated with HT LibertyLink corn varieties. An unknown portion of these corn acres was planted to HT varieties. On these glufosinate HT acres, the average rate of herbicide use was likely somewhat lower than on the average RR acre, because glufosinate is applied at about one-half the glyphosate rate. Still, LibertyLink acres have had a very modest impact on overall HT corn herbicide use. Th e volume of herbicides other than glyphosate applied to HT corn acres was estimated from university weed management recommendations. Th e volume of "Other Herbicides on HT Acres" decreased modestly from 1.2 pounds per acre in 1996-1997 to 1.1 in 2005. Th e volume applied then increases about 7% over three years to 1.18 pounds in 2008 as a result of changes in weed communities and the growing presence of resistant weeds.
From 1996 through 2008, total herbicide use on HT corn acres rose from 1.88 pounds of active ingredient to 2.27, a 21% increase. During this period, glyphosate use is projected to increase from 0.68 pounds per acre to 1.09 pounds, a 60% increase (fi ve percent per year).
Total herbicide applications on conventional and other non-HT 7 corn acres trended downward from 1996 through 2008, falling from 2.67 to 2.02 pounds per acre, refl ecting the gradual shift to lower-dose herbicides, as well as regulatory limits on the rate of atrazine that can be 7 Note that single trait Bt corn without herbicide tolerance is treated as "conventional" for the purposes of this HT corn discussion. applied. Th e registration of s-metolachlor also contributed to a reduction in average corn herbicide application rates. Th is product is a more active stereoisomer of metolachlor, and is eff ective at an application rate about 35% below metolachlor' s typical rate of application.
Overall, herbicide use per acre on all corn acres also trended downward during this period from 2.65 pounds in 1996 to 1.9 pounds in 2002. Herbicide use per acre then began rising, from 1.9 pounds in 2002 to 2.05 pounds in 2005, the last year NASS surveyed corn pesticide use. During this three-year period, average use per acre rose 2.7% annually. From 2005 through 2008, total herbicide use was projected to increase 2% per year. Herbicide use per crop year for all corn, HT, and conventional corn varieties is shown in Figure 4.2., covering the full thirteen year period. Th e diff erence in total herbicide use on HT corn acres, compared to conventional corn acres, gradually changes from a reduction of 0.79 pounds per acre in 1996 to an increase of 0.25 pounds per acre in 2008. Th is shift from a signifi cant reduction per acre of HT corn to a moderate increase in herbicide use is driven by a combination of factors: * Increased average annual glyphosate use rates on HT acres; * An approximate 30% increase in the average number of applications; and * Steady reductions in the average pounds of herbicides applied on conventional corn acres.

November 2009
The First Thirteen Years 34

C. Herbicide-Tolerant Cotton
Of the three crops covered in this report, cotton farmers face the most diffi cult challenge in managing weeds. Th e space between cotton rows is greater than in corn and soybeans fi elds. Th e canopy closes more slowly in cotton fi elds, and sometimes never fully closes. Th e cotton growing season is longer than corn and soybeans, giving weeds an extended window of opportunity to germinate and grow. Th is requires conventional farmers to make more applications of generally longer-acting herbicides. In

D. Impacts of Resistant Weeds on Herbicide Use and Risks
Th e Weed Science Society of America (WSSA) and the industry-sponsored Herbicide Resistance Action Committee maintain a registry of resistant weed species around the world (accessible at www.weedscience.org). Th e WSSA defi nes weed resistance as "the inherited ability of a plant to survive and reproduce following exposure to a dose of herbicide normally lethal to the wild type." Scientists use a simple test to screen for levels of resistance. Th e amount of herbicide required to reduce plant growth by 50% is measured, producing a value called the GR50, for "Growth Reduction by 50%." A case of resistance is regarded as clear cut when the GR50 herbicide dose in a weed population is at least 10-fold higher than the GR50 in a susceptible weed population.
Widespread use of HT technology has turned the U.S. into the resistant weed epicenter of the world. Th e WSSA records 125 resistant biotypes of 68 weeds, infesting up to 18 million acres in the U.S., while Australia is a distant second with 53 resistant biotypes.
Th e actual number of resistant weed populations and the acreage infested with them are likely higher, since the WSSA system is a passive reporting system that depends on academic weed scientists to upload their data on resistant populations. WSSA also has strict standards that must be met for verifying resistance before a resistant weed report is listed, which in some cases may delay or prevent likely cases from being reported.
In addition, WSSA does not report cases of ecological weed shifts -the selection and increasing predominance of weed species that are naturally more tolerant of an intensively used herbicide. For instance, a number of GT weed species are becoming more prominent in GR cropping systems, including common lambsquarters, velvetleaf, Asiatic dayfl ower and tropical spiderwort, among others. 8 Some weed scientists have called for more active and intensive surveillance of resistant weeds in HT cropping systems. 9

Dramatic Increases Reported in Glyphosate Resistance
Glyphosate was fi rst introduced in 1974, and for the next 22 8   Th e emergence and rapid spread of GR weeds has driven rising herbicide use in all three HT crops, especially in recent years. Increasing glyphosate application rates and/or the number of applications will usually buy a little time, but invariably accelerates the emergence of full-blown resistance. Th is is the classic defi nition, and regrettable outcome, of what scientists call the "pesticide treadmill." Below, we present case studies of three particularly troubling GR weeds: Palmer amaranth (pigweed), horseweed, and giant ragweed.

Th e "Perfect Weed"
GR Palmer amaranth has been called "the perfect weed." It has spread rapidly across the southern U.S. in the wake of RR cotton, soybeans, and more recently, corn. Preserving the effi cacy of this last line of defense is now a priority for weed scientists in the region. One scientist asserts that an eff ective resistance management plan for the PPOs is all that stands between GR Palmer amaranth and "…the ability to do economic weed control in cotton and soybeans.'" 26

Glyposate-Resistant Horseweed
Horseweed, or marestail, is a second "high impact" GR weed that has spread rapidly over the past two years. First 25  forcing farmers to rely more heavily on mechanical tillage for weed control, in the process reducing substantially the cotton acreage planted using conservation tillage. 35 As farmers increase their use of tillage, average soil erosion rates increase. For this reason the emergence of GR weeds both increases pesticide use and erosion losses, negating two of the often-claimed benefi ts of HT technology. resistant weeds spread, the presence in any given fi eld of weeds resistant to herbicides in multiple families of chemistry will become commonplace. Th is will compel farmers to rely more heavily on tillage and herbicides, including many older ones such as 2,4-D, that work through still eff ective modes of action.

HT Crops Accelerate the Pesticide Treadmill
Farmers have been creating, and then dealing with HR weeds since the use of herbicides became prevalent in the 1970s. As discussed in Chapter 2 , weeds resistant to triazine and later ALS-inhibitor herbicides (among others) emerged well before the introduction of GE HT crops in the mid 1990s (see also Figure 2.4). Th is fact has led some, notably the biotechnology and seed industries, to assert that there is nothing new or diff erent with GR weeds. In fact, the causes and consequences of the emergence of GR weeds are diff erent in many ways.
HT crop technology allows herbicides (in this case, glyphosate) to be applied in ways and at times not previously possible. Crops can be sprayed over an extended period of time, instead of during one optimal application window. Th is leads to multiple applications of the same herbicide in the same season. Th e rotation of one RR crop following another creates near-continuous selection pressure on weed populations over two or more years. Higher rates of application can be made, increasing the volume sprayed.
Th e sheer scope of introduction of GR crops has fostered such unprecedented reliance on a single chemical for weed control that one leading one expert has remarked that "Glyphosate is as important to world agriculture as penicillin is to human health." 39 Th is extreme reliance makes the threat of GR weeds far more menacing than herbicideresistant weeds of the past. As discussed in Chapter 7, the responses to this threat proposed thus far will likely make matters worse.
Already in some regions, only one herbicide mode of action remains eff ective and available to manage resistant weeds. Ramping up use of herbicides in still-eff ective families of chemistry will buy farmers and industry some time, but it will also bring on more resistant weeds. Unless steps are taken to break the underlying ecological conditions favoring the selection and spread of resistant weeds, this vicious circle will grind through the list of registered herbicide products until there are no longer any economically viable herbicidebased options.
No one can predict with confi dence when such a breaking point for herbicide-based weed management systems will occur for a given crop and region. Attempts to deal with resistant weeds through development of GE crops tolerant to a longer list of herbicides and more overall use of herbicides will almost certainly shorten the path to such breaking points.
Failure to act on the lessons learned in regions heavily reliant on HT crop technology that are now infested with two or more diffi cult to control weeds resistant to multiple herbicides will virtually guarantee that the tipping point will come sooner rather than later, and when it arrives, farmers will be forced to make systemic changes in farming systems that will be costly in multiple dimensions.

Impacts of Bt Crops on Insecticide Use
• Th e amount of Bt toxins manufactured within plant cells during a growing season; and • Th e volume of insecticidal seed treatments used to help plants thrive through the early stages of growth.
As discussed in Chapter 2, there is no way to accurately project the volume of Bt toxins produced by a GE plant. Moreover, there is unresolved debate over whether these toxins should be counted as an "insecticide applied" for purposes of estimating the impact of GE crops on insecticide use.
In order to estimate the total pounds of Bt toxins manufactured by a Bt plant, as well as by all plants on an acre of corn or cotton, scientists need to gain better understanding of Bt gene expression levels in diff erent plant tissues, how long Bt toxins persist in plant cells, and how the toxins break down. Such information will also prove useful in conducting more refi ned dietary risk assessments and to assess impacts of Bt toxins on soil microbial communities.
Seed treatment technology has dramatically changed in recent years. Th e number of pesticide active ingredients utilized in seed treatment mixtures has gone up. Most seed treatment pesticides are now encapsulated around the seed in slow release formulations that markedly extend and improve their eff ectiveness. Th e increasing use of more potent pesticides in seed treatments tends to lower the total volume of active ingredients applied as seed treatments.

November 2009
The First Thirteen Years 42

A. Stacked Traits and Multiple Insecticide Formulations Muddy the Water
Projecting the impact of Bt traits on insecticide use has grown more complicated as a result of the trend toward stacked traits. Since 2005, a growing portion of Bt corn has contained both the Bt gene for ECB control (Monsanto' s YieldGard corn) and the Bt gene for CRW control (Monsanto' s YieldGard for CRW). Varieties expressing both Bt traits are referred to as "YieldGard Plus." It is diffi cult to project with certainty how the three forms of Bt corn -YieldGard, YieldGard for CRW, and YieldGard Plus -aff ect insecticide use. Many insecticides applied by corn farmers are sold in more than one formulation. One formulation, a liquid spray for example, might be labeled for control of the ECB/SWCB, while a granular formulation of the same insecticide(s) is labeled for control of the CRW and other soil-borne insects.
In its annual pesticide use reports, NASS provides data by active ingredient (not formulation) on the percent of acres treated, the rate, number of applications, and pounds applied. For active ingredients in formulations eff ective against both the ECB/SWCB and CRW, there is no accurate way of apportioning use (i.e., share of acres treated, amount) between them, and hence a degree of uncertainty is unavoidable in identifying the insecticide acre treatments displaced by the planting of a particular kind of Bt corn.
Another source of uncertainty can skew estimates of the number of insecticide applications displaced by Bt corn. Many acres of Bt corn are planted on farms where conventional varieties of corn were previously planted and not routinely sprayed with insecticides for either the ECB/SWCB or CRW.
As evident in Supplemental Table 9, generally 6% to 9% of national corn acres have been sprayed for ECB/SWCB control in any given year. 1 Yet by its third year of commercial use in 1998, Bt corn for ECB control was planted on 19.1% of national corn acres -more than twice the average acreage typically sprayed to control the ECB/SWCB.
In 2009, over one-half of national corn acres were planted to Bt corn for ECB/SWCB control. Clearly, many of these acres were not previously sprayed for ECB/SWCB control; hence, the planting of Bt corn on these acres did not reduce insecticide use. For this reason, annual estimates are made of the percent of Bt corn acres that would likely have been treated with an insecticide if conventional hybrids had been planted instead, and this estimate was used in calculating the pounds of insecticides actually displaced by Bt corn.
In the case of Bt corn for CRW management, historically 27% +/-4% of national corn acres have been treated with soil insecticides for CRW control, a share close to the 35% market penetration in 2008 of Monsanto' s Bt corn for CRW control. Clearly, however, the availability of CRW Bt corn has not eliminated the use of corn soil insecticides.

B. Insecticide Use Displaced by Two Types of Bt Corn
Bt corn for ECB/SWCB control has had a modest, but positive impact in reducing insecticide applications to corn, while Bt corn for the CRW is having a more signifi cant impact. Th ere is a signifi cant degree of uncertainty in the estimates of the impacts of Bt corn for CRW control on insecticide use. Only 5% of national corn acres were planted to CRW hybrids in the last year NASS collected corn insecticide use data (2005). Th e big jump upward in Bt corn acres for CRW control came in 2007 and 2008.
Th ere is little publicly accessible information on corn insecticide use in recent years as a result of the decision by NASS in 2007 to suspend the annual pesticide use surveys in major fi eld crops like corn.

Bt Corn for ECB Control
Th e introduction of Bt corn in 1997 increased research focus and funding for work on ECB/SWCB management and heightened grower awareness of the damage caused by these insects in some seasons. As a result, many farmers became more aggressive and pro-active in managing ECB/SWCB. While sound advice, more and more corn farmers will be unable to act on it since the majority of corn hybrids off ered for sale now include the Bt gene for ECB/SWCB control. Supplemental Tables 9 and 11 set forth the basis for estimating the impact of Bt corn for ECB/SWCB control on corn insecticide use. Supplemental Table 9 projects the average rate of insecticides applied on conventional corn to control the ECB/ SWCB, relying on NASS data on corn insecticide use. Since no NASS data have been collected since 2005, insecticide use rates for 2006-2008 were assumed to remain unchanged. No important new active ingredients have come on the market and attained signifi cant corn use in this period, so it is very likely that average use rates have changed little since 2005.
University experts and insect-control guides were consulted to determine which corn insecticides target the ECB largely or exclusively, and which insecticides are partially applied for ECB control. Th e same was done for the CRW insecticides.
Th ese percentages are incorporated in Supplemental Tables 9 (ECB/SWCB rates) and 10 (CRW rates). Average insecticide use rates for products targeting the ECB/SWCB were then calculated based on the weighted shares of total national corn acres treated for ECB/SWCB control.
Th e average rate of application of corn insecticides targeting the ECB fell gradually from 0.21 pounds in 1996 to 0.13 pounds in 2008, consistent with the long-term downward trend in the application rates of registered pesticides. Farmers relied less heavily on organophosphate insecticides applied at rates of 0.5 to 1.2 pound per acre, and more heavily on synthetic pyrethroid insecticides applied at rates between 0.01 and 0.1 pounds per acre. Figure 5.1 shows the generally downward trend in the rate of insecticide applications displaced by the planting of Bt corn for ECB/SWCB control, as well as Bt corn for CRW management.
Supplemental Table 11 calculates the percent of corn acres planted to ECB Bt corn, the number of acres planted each year, and the likely number of acres planted that would previously have been treated with an insecticide. As a result of this adjustment, Bt-ECB acres that would have been sprayed absent Bt technology changes from 90% in 1997 to 45% in 2007-2008 (see Chapter 2D for the rationale behind these adjustments).
Th e line in Supplemental Table 11 labeled "Adjusted Volume of Insecticide Displaced by a Bt-ECB Acre" is the estimated rate of insecticide applications for ECB/SWCB control from Supplemental Table 9 multiplied by the percent of Bt corn for ECB control that would have previously been treated with an ECB insecticide. Th is step addresses the previously described source of upward bias in estimates of insecticide applications displaced by Bt corn (i.e., the fact that not all acres planted to a Bt hybrid would have been sprayed with an insecticide if conventional corn had been planted).

Bt Corn for CRW Control
Th e impact of Bt corn for CRW control is projected in the same way as the impact of ECB Bt corn, as shown in Supplemental  Tables 10 and 11. Bt corn for CRW control was introduced as a single-trait variety in 2003 and was planted on less than one percent of national corn acres in that year. By 2008, over onethird of national corn acres were planted to a variety expressing the CRW Bt gene.
Th e average pounds of insecticides applied per acre of corn to treat the CRW and related soil-borne insects are calculated in Supplemental Table 10. Th e volume of insecticides applied for CRW control fell from 0.29 pounds per acre in 2003 to 0.19 in 2005-2008, as shown in Figure 5.1. Th is reduction was driven by the shift away from relatively high dose insecticides to lower-dose active ingredients applied at rates between 0.01 and 0.1 pound per acre.
As with ECB Bt corn, the percent of corn acres under active management for the CRW -seed treatments, Bt genes, and conventional insecticides --has far outpaced the historic percent of corn acres sprayed with an insecticide for CRW control. Accordingly, the percent of acres planted to CRW Bt corn that would have previously been treated with an insecticide is adjusted from an estimated 95% in the fi rst year of adoption in 2003, to 60% in 2008, for reasons discussed further in Chapter 2. Accordingly, the model projects in 2008 that 18 million acres of corn were not sprayed for CRW as a result of the planting of Bt corn for CRW control (0.6 x 30.1 million acres of Bt CRW corn). 3 In addition to these Bt acres, an estimated 8 million more acres were sprayed with a CRW insecticide, for a total of 38 million acres that were directly treated during the growing season. In addition, essentially all national corn acres were treated with a seed treatment targeting the CRW.

C. Bt Cotton Continues to Perform Well
Essentially 100% of the acres planted to Bt cotton were previously sprayed for control of the budworm/bollworm complex of insects -the prime target of Bt cotton. Moreover, Bt cotton is highly eff ective, so each acre planted is assumed to displace the average pounds of insecticides previously sprayed on an acre of conventional cotton for budworm/bollworm control.
A c c o r d i n g l y, estimating the diff erence in insecticide use on acres planted to Bt and conventional cotton varieties is simpler than in the case of Bt corn. Plus, NASS surveyed cotton pesticide use in 2007, reducing the need for assumptions in extrapolating current use rates.
Estimates of the average pounds of insecticides displaced by each acre of Bt cotton are shown in Supplemental Table 12.
Th e percent of cotton acres planted to Bt varieties rose from 12% in 1996 to 52.5% in 2004 and reached 73% in 2008.
NASS pesticide use data includes the percent of crop acres treated with 11 insecticides known to target the budworm/ bollworm complex, including organophosphates, synthetic pyrethroids, carbamates, liquid Bt sprays, and two reducedrisk insecticides, emamectin benzoate and indoxacarb. Th e extremely toxic carbamate insecticide aldicarb was the market leader throughout this period, accounting for one-half to twothirds of the acres treated over the 13-year period.
Many of these insecticides were applied multiple times, and hence it is necessary to calculate the number of cotton acretreatments with each insecticide, in order to calculate the weighted average rate of application per crop year (taking into account multiple applications). In 1996, the year Bt cotton was introduced, aldicarb accounted for 28% of the acre-treatments, followed by methyl parathion at 25%. Th e share of total acretreatments accounted for by each of the 11 insecticides was used in calculating the weighted average rates in the last line in Supplemental Table 12.
Th e average budworm/bollworm insecticide application rate in 1996 was 0.56 pound per acre. Th e rate has dropped gradually to 0.47 pounds in 2008. Th e limited decline in cotton insecticide rates refl ects the growing percentage of acre treatments accounted for by aldicarb, an insecticide applied at the rate around 0.6 pounds per acre. By 2008, the percent of cotton acres treated with insecticides for the budworm/ bollworm complex had fallen from 48% to 25%, but aldicarb' s share of the total number of acre-treatments had risen from 28% to 67%.

Aggregate Impacts of GE Crops on Pesticide Use: Th e First Th irteen Years
Corn, cotton, and soybeans account for nearly all GE crops grown in the U.S. since 1996. About 941 million acres have been planted to corn, soybeans and cotton with herbicide tolerance, while 357 million acres of corn and cotton have carried the Bt trait, for a total of 1.3 billion GE trait acres over the 13 years covered by this study (see Figures 6.1 and 6.2). As explained in Chapter 3, the actual area planted to GE crops over this period is substantially less than 1.3 billion acres due to the growing prevalence of stacked crops that contain both HT and Bt traits.

A. Major Findings and Conclusions
Diff erences in the pounds of pesticides applied on acres planted to GE varieties, compared to acres planted to conventional seeds, are reported in Supplemental Table 7.
HT corn reduced herbicide use in its fi rst year of introduction by almost 0.8 pounds per acre. Over time, increases in the average rate of application of glyphosate drove herbicide use upward on HT acres.
By 2005, herbicide use on conventional and HT corn acres was essentially identical and by 2006, the average pounds applied on an HT corn acre had risen to 0.08 pounds above the average pounds of herbicides applied to an acre of corn planted to a conventional variety.
Th e same pattern is evident with HT cotton. Each acre of HT cotton in 1996 reduced herbicide use by three-quarters of a pound, but by 2001, rising glyphosate use on HT acres had overtaken the average pounds applied on conventional acres.
Today, each acre of HT cotton increases the average pounds of herbicides applied by about two-thirds relative to conventional cotton. RR soybeans reduced average herbicide use by 0.3 pounds per acre planted in 1996. Just two years later, USDA data show that average herbicide use on HT soybean acres had already risen above the average rate on acres planted to conventional soybeans. By 2008, the diff erence had increased to 1.16 pounds per acre.
Th is dramatic change in herbicide application rates is unmistakable in USDA surveys of pesticide use on soybean farms. Th ere is also general agreement on why the performance of RR soybeans has changed so dramatically over the yearsintense selection pressure from excessive reliance on glyphosate has triggered weed shifts to species more tolerant of glyphosate, as well as evolution of glyphosate-resistant biotypes.
As is the case with corn and cotton, steady reductions over the 13 year period in average soybean herbicide application rates per acre also contributed to the growing margin of diff erence in overall herbicides applications on RR versus conventional crop acres. Th ese reductions were brought about by the registration and growing market penetration of several low-dose herbicide products. Figure 6.3 portrays these trends in the diff erences in pesticide use on an acre planted to a GE crop, compared to an acre planted to a conventional variety.
Estimates of the impacts of GE crops on pesticide use have been calculated by crop, trait, and year. Th e annual change in the volume of pesticide use triggered by the planting of an acre of GE crop (Supplemental Table 7) is multiplied by the acres planted to each GE trait, producing the values in Supplemental Table 8. A graphic depiction of the overall impact of GE crops on pesticide use from 1996 through 2008 appears in Figure  6.4.

Key Conclusions
Over the fi rst 13 years of commercial planting of major GE crops in the United States, this analysis shows that: • GE crops increased overall pesticide use by 318.4 million pounds, or by 7.5% of combined use on the three crops; • Herbicide tolerant crops increased herbicide use by 382.6 million pounds, while Bt crops reduced insecticide use by 64.2 million pounds; • Herbicide tolerant soybeans accounted for 92% of the increased herbicide use across the three HT crops; • GE crops reduced pesticide use in the fi rst three years of commercial introduction by 1.1%, 2.3%, and 2.3% per year, but rising rates per crop year of glyphosate on RR varieties increased aggregate pesticide use across all GE traits and acres beginning in 1999; • Rates of corn and soybean herbicide and corn insecticide applications on cropland planted to conventional varieties trended downward during the study period by 24% to over 90% as a result of the shift toward lower-dose pesticides; • Th e 26% increase in the pounds of pesticides applied on GE crops in 2008, compared to acres planted to conventional varieties, was almost fi ve-fold greater than the 5.8% increase just fi ve years earlier, in 2003; and • Th e upward trend in pesticide use on GE crops has been driven almost solely by the rapid emergence and spread of weeds tolerant of or resistant to glyphosate.
Moreover, further increases in overall pesticide use on GE crops is inevitable in 2010 and for the foreseeable future in the U.S. because of the further emergence and steady spread of weeds resistant to glyphosate.

U.S. Department of Agriculture
Th e USDA has done very little research on the impacts of GE crops on pesticide use, and has been essentially silent on the topic for about a decade. A report by the ERS was issued in May 2002 entitled Adoption of Bioengineered Crops. 1 A short section addresses the impacts of GE crops on pesticide use between 1997 and 1998 for HT soybeans and cotton and Bt cotton, and between 1996-1997 for HT corn. Across the three major crops, the ERS analysts estimated a reduction of 2.5 million pounds of pesticides applied, very close to the 2.2 million pounds reduction estimated in this report for the corresponding years.
Th is 2002 ERS report concluded that herbicide use on HT soybeans went up in 1998 because 13.4 million pounds of glyphosate were substituted for 11.1 million pounds of other herbicides. Th e ERS projection of a 2.3 million pound increase in herbicide pounds applied on HT acres is also very close to the 2.2 million pound increase based on the methodology used in this report. Th ere is no discussion of the impact of GE crops on pesticide use in the current version of the "Agricultural Biotechnology" Briefi ng Room on the ERS website. 3 No other offi cial reports have been issued by USDA addressing the overall impact of GE crops on pesticide use.

National Center for Food and Agriculture (NCFAP) Policy Studies
Several simulation studies by the National Center for Food and Agriculture Policy (NCFAP), an organization funded in part by the biotechnology industry, have addressed the impact of GE crops on pesticide use. Th e most recent report was released in November 2006 and projects impacts in crop year 2005. 4 NCFAP' s general method is to simulate pesticide use on GE and non-GE crops by simply extrapolating from particular pest management systems recommended by university extension agents for adoption on all GE and non-GE crop acres. Such simplistic models are highly vulnerable to error, since actual pest management systems often deviate considerably from those recommended by university specialists. Th e results from such models need to be checked against real-world pesticide use data whenever possible.

Herbicide-Tolerant Corn
NCFAP estimates that genetically engineered HT corn was planted on 35% of corn acres in 2005, a considerably higher share compared to NASS' s corresponding fi gure of 26%, a discrepancy that is not noted or explained by NCFAP. Based on this 35% fi gure, NCFAP estimates that GE HT corn reduced herbicide use by 21.8 million pounds in 2005, or about 0.8 pounds per acre.
Th is fi nding rests largely on two faulty assumptions that exaggerate the amount of herbicide applied to conventional/ non-HT corn acres, which in turn infl ates the "reduction" from a switch to HT corn. Th ese faulty assumptions relate to the extent and rate of use of two high-dose herbicides, atrazine and s-metolachlor/metolachlor, that are used on both HT and conventional/non-HT 5 corn.
With regard to extent of use, NCFAP assumes that all non-HT corn farmers apply two premixed products: fi rst, a mixture of the high-dose herbicides s-metolachlor and atrazine (preemergence), followed post-emergence by a product consisting of mesotrione, nicosulfuron and rimsulfuron.
NASS data demonstrate clearly that the atrazine-metolachlor premix could not have been used by a majority of, much less all, farmers planting non-HT corn. According to NCFAP, non-HT corn comprised 65% of national corn acres, while NASS reports that just 25% of all corn was treated with either s-metolachlor or metolachlor, so that at most 25% of corn acres were treated with this premix (atrazine was applied to 66% of corn acres). At most, 38% of non-HT corn acres could have been treated with this high-rate premix (25% maximum treated, divided by 65% planted). NCFAP also overestimates the rate of herbicide applied to non-HT acres. NCFAP assumes that non-HT corn farmers apply the s-metolachlor/atrazine premix at 3.16 pounds of active ingredients per acre, and the low-dose post-emergence mix at 0.07 pounds per acre, for a total of 3.23 pounds per acre. However, NASS reports that the average amounts of atrazine and s-metolachlor applied to all corn in the 2005 season were 1.13 and 1.35 pounds per acre, respectively. Accordingly, the combined average rate of atrazine and s-metolachlor applied via the premix was at most 2.48 pounds of active ingredient per acre, much less than the 3.16 pounds assumed by NCFAP.
NCFAP projects that an average of 2.5 pounds of herbicides were applied on RR corn acres in 2005, resulting in a 0.73 pound per acre reduction (3.23 pounds on conventional acres, minus 2.5 pounds on RR acres). NCFAP would have projected a 0.02 pound increase on HT acres had it used the more realistic NASS application rates for atrazine and s-metolachlor on conventional corn. Th e methodology in this report projected a 0.01 pound reduction in per acre herbicide use on HT acres in 2005.

Herbicide-Tolerant Soybeans
In the case of soybeans, NCFAP both underestimates herbicide use on HT acres and overstates the amount applied to conventional acres. Th ese faulty assumptions result in a simulated and illusory "reduction" of 20.5 million pounds nationally from the planting of HT soybeans in 2005. HT soybeans -all Roundup Ready -were planted on nearly 90% of national soybean acres in 2005.
NCFAP wrongly assumed that one application of glyphosate suffi ced for over 80% of Roundup Ready soybean acres, resulting in a simulated 1.18 glyphosate applications to the average RR soybean acre for the year. In contrast, NASS reported an average of 1.5 applications of glyphosate (28% higher), a fi gure that refl ects the need for two or more glyphosate applications to control resistant weeds in many states (see Chapter 4). Similarly, NCFAP' s estimate of total herbicide applied to RR soybeans -1.03 pounds per acres per year -does not even match the annual NASS fi gure for glyphosate alone, which is 1.1 pounds per acre, much less account for non-glyphosate herbicides applied to RR soybeans.
The Organic Center NCFAP assumes, for reasons not explained, that herbicides in addition to glyphosate were applied to RR soybeans in just one state (Iowa). In Iowa, NCFAP assumes that soybean farmers apply 0.19 pounds per acre of Canopy (a premix of chlorimuron and metribuzin), in addition to one application of glyphosate. In contrast, this report more realistically estimates that non-glyphosate herbicides were applied to RR soybean acres at an average rate of 0.12 pounds per acre in 2005.
NCFAP also vastly overstates the amount of herbicides applied to conventional soybean acres in 2005, assuming average total applications of 1.35 pounds per acre (all presumed to be nonglyphosate herbicides). Th is presumed rate for herbicides applied to conventional soybean acres is more than twice the rate of 0.59 pound per acre on conventional soybeans estimated in this study, based on NASS data. NCFAP' s estimate of average herbicide use on conventional soybeans is clearly out of step with the trend toward lower-dose herbicides, some of which are applied at rates well below 0.1 pound per acre.
If NCFAP had used NASS data to calibrate its estimates of herbicide use on RR and conventional soybean, it would have arrived at a result much closer to the one in this report: an estimated increase in herbicide use of 41.5 million pounds in 2005 due to the planting of RR soybeans (see Supplemental  Table 8).

PG Economics Ltd
A UK based consulting fi rm, PG Economics Ltd., has carried out several studies of GE crops funded by the pesticide and biotechnology industries. Th eir latest was released in May, 2009. 6 Th e PG Economics report uses methods and sources similar to NCFAP, and claims its estimates are based on "the average performance and impact recorded in diff erent crops." Th e PG Economics report estimates a 4.6% reduction worldwide in herbicide use attributable to GE crops from 1996 through 2007 (the fi rst 12 years of commercial use). Th is report estimates that GE HT corn, soybeans, and cotton have increased herbicide use in the U.S . by 382 million pounds over 13 years, or by about 10% (NASS reports that 3.82 billion pounds of herbicides applied to these three crops from 1996-2008). It is worth noting that the increase in 2008 -the extra year covered by this analysis -was 100 million pounds, or about 26% of the total increase over the 13 years.
Th e methodology in the PG Economics report is worth a closer look. HT soybeans are by far the most important GE crop in the U.S. in terms of impacts on pesticide use, and so the focus herein is on the PG Economics analysis of herbicide use on conventional and HT soybeans, as set forth in Chapter 4 of their above-cited report.
Th e authors begin by noting that there are two primary sources of data on pesticide use in the U.S. -NASS surveys and private farm-level surveys (survey data from DMR Kynetec was used in the PG Economics report).
Th eir Table 33 reports herbicide use on HT and conventional soybeans for 1998 through 2007 in the U.S., based on Kynetec survey data. In every year, herbicide use was higher on HT soybeans than conventional soybeans. Th e margin was typically less than 0.2 pounds until 2002, when the margin increased to around 0.3 from 2003-2007, as shown in Table 6.1.
Estimates of herbicide use on HT soybean acres as reported in the PG Economics report and this analysis diff er modestly, and are accounted for largely by the rate per crop year of glyphosate herbicides. Likewise, the PG Economics and this report' s estimates of total herbicide use on conventional soybean acres, and the diff erences between HT and conventional acres, are relatively close for 1998 through 2004. Th e Kynetec dataset then projects increases in the total rate of herbicide application on conventional acres from 2004 through 2007, despite the continued trend toward greater reliance on relatively low-dose herbicides, as evident in the projections based on NASS data.
Th is deviation in estimates of herbicide use on conventional soybeans accounts for this report' s progressively larger margin of diff erence in herbicide use rates on HT in contrast to conventional soybean acres.
Despite some diff erences, it is signifi cant that the industrysponsored Kynetec survey, as reported by PG Economics, supports the same basic conclusion as this report -HT soybeans have increased herbicide use by a substantial and growing amount.

The Organic Center
But curiously, right after reporting the Kynetec results in Table  33, the authors of the PG Economics report state: "Th e comparison data between the GM HT crop and the conventional alternative presented above is, however, not a reasonable representation of average herbicide usage on the average GM HT crop compared with the average conventional alternative for recent years." (page 66) Th e PG Economics analysts disavow their own data-driven estimates, asserting that herbicide use is lower on conventional soybean acres in the Kynetec dataset because the majority of farmers planting conventional soybeans must be among those facing the lightest weed pressure. Th is creative argument, however, is incompatible with the pattern of adoption of HT soybeans across the states. Since 2006, the rate of adoption of HT soybeans varies modestly between states from 81% to 97%, with no clear pattern between states with relatively low weed pressure (Minnesota, South Dakota) and states with much higher levels of weed pressure (Mississippi, Arkansas). 7 After rejecting the Kynetec survey fi ndings that were based on real data, the PG Economics team then turns to another source for supposedly more reliable estimates -the National Center for Food and Agricultural Policy (see previous section for a critique of NCFAP' s estimates). Th e PG Economics team revises its soybean herbicide use projections drawing on NCFAP' s faulty simulations, and reaches the basic fi nding of a 6.8% reduction in herbicide use as a result of HT soybeans.
Similarly creative -and highly questionable -methodological strategies are employed by the PG Economics team in projecting the impacts of other GE crops on pesticide use. Like the NCFAP, the PG Economics team never explains the discrepancies between their estimates and those based on NASS data. 7 Supplemental Table 3 presents HT soybean adoption rate data by state, and shows that some relatively low weed pressure states have high adoption, while others have lower adoption. Several relatively low pressure states have higher adoption rates than states with high levels of weed pressure. Further evidence was provided at BIO 2009 on the many benefits of agricultural biotechnology. Graham Brookes, Director of PG Economics (UK) released key findings from its Global Impact Study that showed that farmers around the world are growing more biotech crops with significant global economic and environmental benefits. Key highlights of the report include: Biotech crops contribute significantly to reducing the release of greenhouse gas emissions from agricultural practices -mainly from less fuel use and additional soil carbon storage from reduced tillage. In 2007, the reduction of carbon dioxide from the atmosphere by biotech crops was equivalent to removing nearly 6.3 million cars from the road for one year; Biotech crops reduced pesticide use (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007) by 359 million kg (-8.8 percent), and as a result, decreased the environmental impact associated with herbicide and insecticide use on the area planted to biotech crops by 17.2 percent; Herbicide tolerant biotech crops have facilitated the adoption of no/reduced tillage production regions -especially South America; There have been substantial net economic benefits to farmers amounting to $10.1 billion in 2007, and $44.1 billion since 1996. Of the $44.1 billion, 46.5 percent ($20.5 billion) was due to increased yields and the rest to reductions in the cost of production.
The report countered a recent Union of Concerned Scientists (UCS) report that attempted to make the case that biotech crops have not significantly increased yields since their introduction 1996. However, the UCS report suffers from a very flawed, superficial and inconsistent analysis.
The UCS report is very selective in the data it chose to use and does not account for variation in yield, country and region. The UCS report does -in factstate that Bt corn has increased yields in the United States, but states just the opposite in its executive summary. In addition, the report did not take into consideration the significant decrease in costs of production from biotech crops that are just as important to farmers as yield. And, the report did not include canola and cotton that have had significant yield increases over the past decade. Genetically engineered corn, soybeans, and cotton now dominate the market. Across these three crops, the supply of conventional, non-GE seed is so thin now that GE seeds will continue to account for the majority of crop acres planted for at least several years to come.
Th e quantum leap in seed industry profi ts associated with the marketing of GE seeds, coupled with control of the seed supply by companies holding the patents on GE technology, virtually guarantee this outcome. But there are clouds on the horizon for both the biotech industry and corn, soybean, and cotton farmers. Resistant weeds will continue to emerge and spread, and the current pressure to relax resistance management plans applicable to Bt corn and cotton could undermine long-term effi cacy.
Over the next decade, GE seeds will increasingly contain multiple traits, cost considerably more per acre, and pose unique and not well understood resistance management, food safety and environmental risks. Th ese factors will assume ever greater importance in assessments of the costs, benefi ts, and risks of GE crop technologies.

A. Th e Tipping Point for RR Crops
In the 2009 crop year, the percentage of national soybean acres planted to Roundup Ready varieties decreased for the fi rst time since their introduction in 1996. Th ough the decline in adoption was slight (92% to 91%), there are reasons to believe 2009 may mark the tipping point for RR soybean market penetration. Th ese include the slipping effi cacy of the RR system as glyphosate-resistant weeds spread, steeply rising production costs (RR seed, herbicides), early evidence that the 7% to 11% yield increase promised by Monsanto on farms planting Roundup Read 2 soybeans is not occurring in the fi eld 1 ; and the increasingly attractive economics of growing conventional soybeans.
Th e spread of glyphosate-resistant weeds is largely responsible for the sharply increased use of glyphosate on soybeans documented in this report. While incrementally higher glyphosate application rates, and more applications, on RR crop acres will further increase overall glyphosate use, resistant weeds will force a growing number of farmers to resort to additional herbicides as well. As an Iowa State University weed scientist argues in a prescient article entitled "Turn Out the Lights --Th e Party' s Over," the days have passed when a single, properly timed application of glyphosate controlled all weeds, all season long. 2 In the future, most RR acres will be treated with two herbicide active ingredients including glyphosate, and many will be sprayed with three or more, often in multiple-product premixes. As a result, growers planting RR crops will fi nd themselves facing weed control regimes that are more complex, timeconsuming, and expensive than those utilized by conventional corn, soybean, and cotton farmers.
Some farmers have already decided to explore life after RR soybeans. "Interest in Non-Genetically Modifi ed Soybeans Growing" is the title of an April, 2009 story posted by the Ohio State University extension service. Growing interest stems from "cheaper seed and lucrative premiums [for non-GE soybeans]." 3 In anticipation of this growth in demand, the Ohio State extension service reports that seed companies are doubling or tripling their conventional soybean seed supply for 2010.
Similar reports are coming in from Missouri and Arkansas, 4 where demand for cheaper conventional soybeans that yield as well as or better than RR soybeans is outstripping supply.
Agronomists in these states point to three factors driving this renewed interest in conventional soybean seed: • Th e high and rising price of RR seed; • Resistant and tougher-to-control weeds; and • Regaining the option and freedom to save and replant seeds, a traditional practice prohibited with Monsanto' s patented RR soybeans.
Th e cost of soybean seed has risen from around $10 per bushel in the early 1980s to around $50 for RR seed in 2008. Monsanto recently announced that the newly introduced RR 2 soybean seed will cost $74 an acre in 2010, a remarkable "Roundup is the greatest thing in agriculture in my lifetime." "In hindsight, we screwed up. We can't rely on the same thing over and over." Quotes from a North Carolina farmer and a retired scientist 7

B. Industry's Response to Resistant Weeds
While biotechnology companies generally downplay the severity and adverse impacts of glyphosate-resistant weeds, they are nonetheless working aggressively to come up with responses to the problem. Th ree of these responses are discussed below: subsidies for use of herbicides with diff erent modes of action, crops with enhanced resistance to glyphosate, and herbicide-resistant stacks that include resistance to toxic but inexpensive herbicides like 2,4-D.

Subsidies for Use of Non-glyphosate Herbicides
Since 1996 Monsanto has encouraged farmers to rely exclusively on glyphosate for control of weeds in Roundup Ready crops, 8 and discounted the possibility of signifi cant problems triggered by glyphosate-resistant weeds. 9 Now that resistant weeds are threatening the viability of the RR crop system, however, Monsanto and other companies are responding with unprecedented initiatives that subsidize the purchase of competitors' products in a belated eff ort to deal with already-resistant weeds and/or slow the emergence of newly resistant weeds.
Monsanto' s "Start Clean, Stay Clean Assurance Plan" is part of the Roundup Rewards program, 10 which off ers farmers rebates and incentives for those farmers who agree to exclusively purchase specifi c, bundled Monsanto seed and herbicide products. 11 Under this program a farmer can receive a rebate up to $13 per acre for the purchase of a competitor' s herbicide that works through a mode of action diff erent from Roundup' s.
Th e "Roundup Ready Cotton Performance Plus" program also off ers rebates from Monsanto to growers to cover the cost of competitors' herbicides. Th is program pays up to $12 per acre and is designed to encourage the rotation of herbicide modes of action, a core resistance management practice. 12 Syngenta, too, has recently announced a plan, the "2009 AgriEdge Corn and Soybean Program" 13 that off ers rebates for the purchase of herbicides that work through a mode of action other than glyphosate' s.
Although the rotation of herbicide modes of action is an important strategy for sustaining herbicide effi cacy, the rotations must be done carefully. As the pesticide industry The Organic Center moves to more multiple-herbicide premix products, farmers will have a more diffi cult time following recommended herbicide-resistant management plans. In addition, several GR weed biotypes are also already resistant to herbicides in one, two, or more herbicide families of chemistry, as documented in Chapter 4.

Enhanced Glyphosate Resistance
A second strategy to respond to the rapid spread of glyphosate-resistant weeds is engineering crops with enhanced resistance to glyphosate. Such crops will tolerate the use of higher rates of application, in the hope that more glyphosate will control increasingly resistant weeds. While of limited eff ectiveness in the short term, this strategy will accelerate the emergence of weeds with higher levels of glyphosate-resistance, and is, for farmers, like pouring gasoline on a fi re in the hope of putting it out. Th e higher glyphosate application rates made possible by and expected with these new, enhanced glyphosateresistant crops will almost certainly accelerate the evolution and spread of resistant weed populations. Th e only viable alternative for conventional farmers to delay the unraveling of RR technology, whether enhanced or not, is to diversify their weed management tactics to include more tillage, altered crop rotations, the planting of cover crops, and more reliance on alternative herbicide modes of action.

Crops Resistant to Multiple Herbicides
Th e third approach being employed by industry is to develop crops that are resistant to more than one herbicide. Since there are relatively few new herbicides in the development pipeline, this strategy requires companies to engineer resistance to older and often higher-risk herbicides like 2,4-D, paraquat, and dicamba. A review of the scientifi c literature, the farm press, and petitions for deregulation of herbicide-tolerant crop varieties pending at the USDA shows that the industry is investing heavily in the development of crops with resistance to multiple herbicides.
DuPont-Pioneer' s Optimum GAT soybeans and corn combine resistance to glyphosate with resistance to herbicides that inhibit the acetolactate synthase (ALS) 22  enzyme (ALS inhibitors). Optimum GAT crop technology does not seem a promising approach in that it combines resistance to the two classes of herbicides (glyphosate and ALS inhibitors) to which weeds have already developed the most extensive resistance (see Figure 2.4). BASF has also developed ALS inhibitor-resistant soybeans, 23 which will likely also be "stacked" with resistance to glyphosate in the context of a Monsanto-BASF joint-licensing agreement (see below).
From an environmental and human health perspective, the most troubling new resistance traits will allow the use of relatively inexpensive, but toxic herbicides that have not been used widely in corn, soybean, and cotton production for many years because of the initial effi cacy of glyphosate in the RR system. In collaboration with the University of Nebraska, Monsanto has developed soybeans that are tolerant to the chlorophenoxy herbicide dicamba. 24 Th ese dicamba-tolerant soybeans are to be stacked with resistance to glyphosate in collaboration with BASF, the largest producer of dicamba. 25 Dicamba-resistant corn and cotton are also under development, with potential triple-stacking of herbicide tolerance to dicamba, glyphosate, and glufosinate. 26 Dow AgroSciences recently petitioned USDA for commercial approval of a GE-corn variety resistant to a second chlorophenoxy herbicide -2,4-D, a component of the Vietnam War defoliant Agent Orange. Th is 2,4-D-resistant corn will be stacked with resistance to aryloxyphenoxypropionate grass herbicides of the ACCase inhibitor class. 27 Dow projects introduction of this dual herbicide-resistant corn in 2012, and a corresponding soybean variety in 2013 or 2014. 28 Finally, Monsanto and Dow are collaborating to produce "SmartStax" corn, which combines resistance to glyphosate and glufosinate, together with six Bt insecticidal toxins. 29 Moreover, the multiple HT crops described above are just the tip of the iceberg. Th e major players in the industry have discovered or developed at least 12 genes conferring resistance to most major classes of herbicides. 30 One scenario for the future of biotech crops is provided by a 2009 patent granted to DuPont-Pioneer, describing a single plant that is tolerant to at least two, three, four, fi ve, six, or seven or more diff erent herbicide families of chemistry, Th e rationale stated in patent applications and other seed industry documents supporting the development of multiple herbicide-resistant crops is that they will provide farmers new options to deal not just with resistant weeds, but also volunteer plants in a subsequent crop season that also happen to be herbicide tolerant. For instance, glyphosate-resistant weeds and RR corn in a soybean fi eld planted to a variety with dual tolerance to glyphosate and ALS inhibitors could be treated with an over-thetop application of an ALS inhibitor. Likewise, Dow' s dual-tolerant corn could be sprayed directly with 2,4-D to control weeds or soybeans resistant to glyphosate, and perhaps other herbicides.
Managing resistant weeds triggered by GE crops by developing new varieties tolerant of multiple herbicides is 31 Use of the word "type" in this context refers to a herbicide mode of action that might encompass a dozen or more registered active ingredients, and hundreds (even thousands) of products. "Novel Glyphosate-N-Acetyltransferase (GAT) Genes," U.S. Patent Application Publication, Pub. No. US 2009/0011938 A1, January 8, 2009, paragraph 33.
Farmers are now dealing with a new "weed" -volunteer RR corn plants in RR soybean fi elds. Th e converse is also a growing problem -volunteer RR soybeans in RR corn fi elds.
The Organic Center appealing to biotech seed companies, because each herbicidetolerant trait qualifi es the patent holder for a technology fee premium. Progress down this road, however, will draw farmers onto an increasingly costly herbicide treadmill that will erode net farm-level returns and pose signifi cant new public health and environmental risks.
Plus, it likely won't work for long, if at all. Weed biotypes that are resistant to two or three diff erent herbicide modes of action, and literally dozens of herbicide products, are already common. Weeds resistant to glyphosate, ALS inhibitors, or both comprise by far the majority of herbicide-resistant weeds, as measured by both acreage infested and number of resistant biotypes. 32 Multiple-herbicide-resistant crops will also facilitate more frequent applications of 2,4-D, paraquat, and dicamba, as well as higher rates of application. Th e two phenoxy herbicides, 2,4-D and dicamba, have been linked to reproductive problems and birth defects in the Midwest, and pose signifi cantly higher risks to a range of organisms than most other contemporary herbicides. 33 Paraquat is a known risk factor for Alzheimer' s disease, Parkinson' s disease, and other neurological diseases of aging. 34 Already, and before the introduction of any 2,4-D resistant crops, the spread of glyphosate resistant weeds has markedly increased 2,4-D use. NASS data show 2,4-D applications on soybeans rising from 1.73 million pounds in 2005 to 3.67 million pounds in 2006, a 112% increase. In Louisiana in 2006, soybean farmers sprayed 36% of their acres with paraquat, 19% with 2,4-D, and applied 2.3 applications of glyphosate to 87% of planted acres. In summary, glyphosate-resistant crops were rapidly adopted by farmers, who were encouraged to rely exclusively on glyphosate for weed control. Farmers were assured by experts that resistant weeds would never be extensive or diffi cult to control. Voluntary resistance management guidelines weakly advanced by Syngenta, Monsanto, and others have largely failed, while federal regulators have done essentially nothing to stem the rapid emergence of resistant weeds. Now that glyphosate-resistant weeds infest millions of acres of cropland and are threatening the viability of the RR system, the industry is proposing "solutions" that are, in truth, technical fi xes that are almost certain to make matters worse by creating a greater number of weeds resistant to multiple herbicides. It is also inevitable that there will be further, signifi cant increases in herbicide use, including relatively more toxic herbicides like 2,4-D, dicamba, and paraquat.
Increased use of chlorophenoxy herbicides will also lead to much more serious and frequent problems with off -target movement of herbicides and damage to crops, shrubs, and other valuable vegetation. Not only are these herbicides prone to drift during application, they also re-volatilize after application under certain weather conditions. Th e heat of the sun can transform these herbicides back into vapor phase, allowing them to fl oat on the wind and come into contact with non-target plants, such as the wheat or alfalfa in a neighbor' s fi eld, or roses in a garden. At low doses, susceptible plants usually do not die, but often suff er harm to their reproductive functions. Pollen and nectar sources for bees and habitat for benefi cial insects can collapse due to movement of dicamba into hedgerows and uncultivated land.

The Organic Center
Some high-value crops like grapes and tomatoes can be damaged by chlorophenoxy herbicide drift at levels that are essentially undetectable. Factoring this often hidden and always diffi cult to diagnose damage into the GE crop cost-benefi t equation is going to be a major challenge.
Avoiding damage in crop fi elds from off -target movement and carryover of herbicides is one reason the biotechnology industry is moving toward coupling resistance to glyphosate with resistance to chlorophenoxy and other herbicide modes of action. In fact, some have already advanced the troubling proposition that farmers should purchase chlorophenoxyresistance traits precisely in order to defend their crops against drift and revolatilization, problems that will be greatly exacerbated if the industry aggressively markets corn, soybean, and cotton varieties engineered for resistance to these herbicides. 35

C. Resistance Management Still Key in Sustaining Bt Crop Effi cacy
Th e future of Bt crops is brighter than the future of RR crops. Unlike glyphosate, Bt was recognized from the beginning as a valuable, relatively benign insecticide whose continued effi cacy required government action to protect against the evolution of resistant insects. As a result, the EPA established programs to preserve the effi cacy of Bt toxins through the use of refuges for susceptible insect populations and close monitoring of pest populations.
Th e program has been successful, especially in the case of Bt cotton. Th e attention focused by university entomologists on resistance management, the mandatory resistance management plans imposed by the EPA, and the introduction of Bollgard II cotton that expresses two Bt toxins have proven eff ective, thus far, in delaying the emergence of resistance in cotton pests in most regions.
However, the discovery of several Bt-resistant populations of bollworms in Mississippi and Arkansas between 2003 and 2006 by Dr. Bruce Tabashnik and colleagues stands as a reminder that Bt resistance must be closely monitored and aggressively managed.
History, too, suggests that continued diligence in cotton Bt resistance management is warranted. Since the 1950s, it has taken 10-15 years for key cotton insects to develop resistance to each new type of insecticide applied to control them. Th is cycle began with the organochlorines from the early 1960s to mid-1970s, and then repeated itself with the carbamates in the 1970s and 1980s and the synthetic pyrethroids in the 1980s and 1990s. Th e Bt cotton varieties have been in use for about 10 years. Researchers have recently shown that cross-resistance can develop in some cotton insect pests to the two Bt toxins in Bollgard II varieties. 36 As a result, prudence dictates waiting a few more years before determining whether contemporary resistance management plans are excessive.
Bt corn also remains highly eff ective for control of ECBs and SWCBs, but is being used in ways that impose signifi cant selection pressure on insect populations. Unfortunately, the industry has convinced the EPA to relax resistance management requirements applicable to recently approved, stacked Bt corn varieties expressing two or more modes of action for ECB/ SWCB control.
Th e industry has also asked for reduced resistance management requirements for corn hybrids expressing Bt for control of the CRW, an insect notorious for its ability to develop resistance. 37 Scientists convened by the EPA to assess future CRW resistance management plans questioned the science supporting such requests by industry to relax 36 Tabashnik, B. et al., 2009. "Asymmerical cross-resistance between Bt toxins Cry 1Ac and Cry2Ab in pink bollworm," Proceedings of the National Academy of Sciences, www.pnas.org/cgi/doi/10.1073/ pnas.0901351106. 37 Th e CRW is resistant to insecticide active ingredients in nearly all major insecticide families of chemistry. In addition, the corn rootworm is the fi rst and only insect known to have developed resistance to crop rotations. Th e western CRW is listed as resistant to 11 insecticides in four families of chemistry in the Arthropod Pesticide Resistance Database at Michigan State University. Details on western CRW resistance are at http://www.pesticideresistance. org/search/12/0/558/0/ The Organic Center resistance management provisions, 38 but the requests were nevertheless approved.

D. Why the Dramatic Increase in the Number of Toxins Needed to Grow Corn?
Another way of looking at pesticide dependence is to track the number rather than the amount of insecticides used on a crop. Th e combination of nicotinyl and other insecticide seed treatments and the increasing number of toxins in stacked Bt corn varieties represents a stunning increase in the number of diff erent pesticidal toxins now being used to bring the nation' s corn crop to harvest.
Eight-stack corn hybrids will be planted in 2010 expressing three diff erent Bt toxins for control of the ECB/SWCB, and three more to control CRWs -a total of six Bt toxins. Th e seeds will be coated with two insecticides, including one nicotinyl insecticide that will move systemically throughout the tissues of the corn plant. A portion of the acres planted to these varieties will still be treated with one or more conventional corn insecticides.
Accordingly, nine or more chemicals will be used to manage corn insects on many fi elds in 2010. But on other conventional and organic farms, millions of acres of corn will receive no insecticide, and several million more, just a seed treatment. Traditionally, about two-thirds of corn acres have not required an insecticide spray application.

E. Stacking Traits Poses New and Poorly Understood Risks
Th ere has been virtually no independent fi eld research on the ecological and food safety implications when widely planted Bt corn varieties are simultaneously expressing two, three, or six Bt toxins. Current USDA and EPA approvals are based on the assumption that multiple genes producing diff erent Bt toxins in corn plants will operate exactly as they do in varieties engineered to produce just a single Bt toxin.
Current EPA policy also apparently assumes there are no interactions in GE plants between the novel DNA introduced in the plant, the novel proteins produced in the plant as a result, and the systemic insecticides and fungicides now routinely used as seed treatments.
Th ese are critical assumptions grounded upon very little science, that also require suspension of common sense. If interactions do, in fact, occur under some circumstances, or if the stability of gene expression patterns is reduced as the number of traits engineered into a plant increases, unpleasant surprises will lie ahead. For this reason, the government and industry should pursue deeper understanding of the impacts of multiple-stacked GE traits, and hopefully before hundreds of millions of acres are planted to them.
Th ere is urgent need for more rigorous and independent scientifi c examination of the unique risks posed by stacked crop varieties. Multiple-trait varieties are already on the market and will gain a much larger share of the market in 2010. Within a few years, single-trait GE varieties will account for only a fraction of GE-planted acres.
Assessment of the risks of multi-trait crops faces a new and deeply troubling obstacle. Because genetically engineered crops are considered inventions under the patent law, patent holders control their use and sale. Patent rights plus market control give the biotechnology industry extraordinary control over the corn, soybean, and cotton seed supply. Th rough technology agreements that every buyer or user of GE seeds must sign and comply with, the seed industry also controls who can conduct research on GE seeds, what topics receive research attention, and how, The Organic Center and sometimes even whether, the fi ndings of independent scientists can be reported publicly. 39 Under such a system there simply is no way that scientists can objectively assess the risks of new biotechnology crops, including the new stacked varieties.
Compared to 15 years ago when the fi rst GE crop was planted, farmers and the public have, for the most part, lost control over the seed supply. Until public plant breeding programs and seed companies re-emerge that are dedicated to producing conventional seeds, farmers will have to accept and plant what the seed industry chooses to off er, and the public will have to live with considerable uncertainty over the novel food safety and environmental risks posed by these new crops. 39  For the foreseeable future, this study confi rms that one direct and predictable outcome of the planting of GE corn, soybean, and cotton seed will be steady, annual increases in the pounds of herbicides applied per acre across close to one-half the nation' s cultivated cropland base. Farm production costs and environmental and health risks will rise in step with the total pounds of pesticides applied on GE crops.
Vastly expanded use of 2,4-D and other older, relatively more toxic herbicides on fi elds infested with glyphosateresistant weeds will increase human and environmental risks, and greatly increase off -target movement of herbicides, in some instances leading to more damage to plants on nearby farms and in neighboring areas.
As glyphosate-resistant weeds spread, farmers are forced to return to deep tillage in an eff ort to bury resistant weed seeds. Th e tillage renaissance unfolding in the Southeast, in step with the spread of resistant weeds, increases soil erosion, energy use, non-point source water pollution, agriculture' s contribution to global warming, and grower production costs. Th ese consequences must now be incorporated in the GE-crop risk-benefi t equation.
The Organic Center