Biotech traits have contributed to a significant reduction in the environmental impact associated with insecticide and herbicide use on the areas devoted to biotech crops (Table 1). Since 1996, the use of pesticides on the biotech crop area was reduced by 443 million kg of active ingredient (9.1% reduction), and the environmental impact associated with herbicide and insecticide use on these crops, as measured by the EIQ indicator, fell by 17.9%.

In absolute terms, the largest environmental gain has been associated with the adoption of GM insect resistant (IR) cotton (-23.9% reduction in the volume of active ingredient used and a 26% reduction in the EIQ indicator 1996–2010) and reflects the significant reduction in insecticide use that the technology has allowed, in what has traditionally been an intensive user of insecticides.

The volume of herbicides used in biotech soybean crops also decreased by 34 million kg (1996–2010), a 1.7% reduction, while the overall environmental impact associated with herbicide use on these crops decreased by a significantly larger 16.4%. This highlights the switch in herbicides used with most GM herbicide tolerant (HT) crops to active ingredients with a more environmentally benign profile than the ones generally used on conventional crops.

Important environmental gains have also arisen in the maize and canola sectors. In the maize sector, herbicide and insecticide use decreased by 212.8 million kg (1996–2010), and the associated environmental impact of pesticide use on this crop area decreased, due to a combination of reduced insecticide use (37.7%) and a switch to more environmentally benign herbicides (11.5%). In the canola sector, farmers reduced herbicide use by 14.4 million kg (a 18.2% reduction), and the associated environmental impact of herbicide use on this crop area fell by 27.6% (due to a switch to more environmentally benign herbicides).

In terms of the division of the environmental benefits associated with less insecticide and herbicide use for farmers in developing countries relative to farmers in developed countries, Table 2 shows a 55%:45% split of the environmental benefits (1996–2010) respectively in developed (55%) and developing countries (45%). Over three-quarters (76%) of the environmental gains in developing countries have been from the use of GM IR cotton.

It should, however, be noted that in some regions where GM HT crops have been widely grown, some farmers have relied too much on the use of single herbicides like glyphosate to manage weeds in GM HT crops and this has contributed to the development of weed resistance. Worldwide, there are 21 weed species that are currently resistant to glyphosate⁶ (compared with, for example, 69 weed species resistant to triazine herbicides such as atrazine). A few of the glyphosate resistant species, such as marestail (Conyza canadensis) and palmer pigweed (Amaranthus palmeri) are now reasonably widespread in the US, especially marestail, where there are several million acres infested and palmer pigweed, in Southern states, where over a million acres are estimated to exhibit such resistance. In Argentina, development of resistance to glyphosate in weeds such as Johnson Grass (Sorghum halepense) is also reported.

Where this has occurred, farmers have had to adopt reactive weed management strategies incorporating the use of a mix of herbicides. In recent years, there has also been a growing consensus among weed scientists of a need for changes in the weed management programs in GM HT crops because of the evolution of these weed populations that are resistant to glyphosate. While the overall level of weed resistance in areas planted to GM HT crops is still relatively low (equal to between 5% and 10% of the total US cropping area annually planted to GM HT crops), growers of GM HT crops are increasingly being advised to be more proactive and include other herbicides in combination with glyphosate in their weed management systems even where instances of weed resistance to glyphosate have not been found. This is because proactive weed management programs generally require fewer herbicides and are more economical than reactive weed management programs. At the macro level, the adoption of both reactive and proactive weed management programs in GM HT crops has already begun to influence the mix, total amount and overall environmental profile of herbicides applied to GM HT soybeans, cotton, maize and canola and this is reflected in the data presented in this paper. For example, in the US GM HT soybean crop in 2010, just over a third of the crop received an additional herbicide treatment of one of the following active ingredients^a 2 4 D, chlorimuron, clethodim and flumioxazin. This compares with 13% of the GM HT soybean crop receiving a treatment of one of these four herbicide active ingredients in 2006. As a result, the average amount of herbicide active ingredient applied to GM HT soybeans in the US (per hectare) has increased by about a third over the last five years (the associated EIQ value has increased by about 27%). This compares with the average amount of herbicide active ingredient applied to the conventional (non-GM) soybean alternative which increased by 15% over the same period (the associated EIQ value for conventional soybeans increased by 27%). The increase in the use of herbicides on conventional soybeans in the US can also be partly attributed to the ongoing development of weed resistance to herbicides commonly used and highlights that the development of weed resistance to herbicides is a problem faced by all farmers, regardless of production method.

Assessment of the impact of biotech crops on insecticide and herbicide use requires comparisons of the respective weed and pest control measures used on biotech vs. the “conventional alternative” form of production. This presents a number of challenges relating to availability and representativeness. Comparison data ideally derives from farm level surveys which collect usage data on the different forms of production. A search of literature on biotech crop impact on insecticide or herbicide use at the trait, local, regional or national level shows that the number of studies exploring these issues is limited^7-9 with even fewer,^10,11 providing data to the pesticide (active ingredient) level. Second, national level pesticide usage survey data are also extremely limited; in fact, there are no published annual pesticide usage surveys conducted by national authorities in any of the countries currently growing biotech traits, and the only country in which pesticide usage data are collected (by private market research companies) on an annual basis and which allows a comparison between biotech and conventional crops to be made, is the US.^d

Unfortunately, even where national survey data are available on usage, the data on conventional crop usage may fail to be reasonably representative of what herbicides and insecticides might be expected to be used in the absence of biotechnology. When biotech traits dominate total production (e.g., for soybeans, corn, cotton and canola in the US since the early 2000s), the conventional cropping data set used to identify pesticide use relates to a relatively small share of total crop area and therefore is likely to under estimate what usage would probably be in the absence of biotechnology. The reasons why this conventional cropping data set is unrepresentative of the levels of pesticide use that might reasonably be expected to be used in the absence of biotechnology include:

To overcome these problems in the analysis of pesticide use changes arising from the adoption of biotech crops (i.e., where biotech traits account for the majority of total plantings), presented in this paper, recorded usage levels for the biotech crops are used (based on survey data), with the conventional alternative (counterfactual situation) identified based on opinion from extension advisors and industry specialists as to what farmers might reasonably be expected to use in terms of crop protection practices and usage levels of pesticide.^e This methodology has been used by others.¹⁵ As an approach, this has an additional advantage of providing comparisons of current crop protection practices on both biotech crops and the conventional alternatives and so takes into account dynamic changes in crop protection management practices and technologies rather than making comparisons on past practices alone. Details of how this methodology has been applied to the 2010 calculations, sources used for each trait/country combination examined and examples of typical conventional vs. biotech pesticide applications are provided in Supplemental Appendices 1 and 2.

The most common way in which changes in pesticide use with biotech crops has been presented in the literature has been in terms of the volume (quantity) of pesticide applied. While comparisons of total pesticide volume used in biotech and conventional crop production systems are a useful indicator of associated environmental impacts, amount of active ingredient used is an imperfect measure because it does not account for differences in the specific pest control programs used in biotech and conventional cropping systems. For example, different specific products used in biotech vs conventional crop systems, differences in the rate of pesticides used for efficacy and differences in the environmental characteristics (mobility, persistence, etc.,) are masked in general comparisons of total pesticide volumes used.

In this paper, the pesticide related environmental impact changes associated with biotech crop adoption are examined in terms of changes in the volume (amount) of active ingredient applied but supplemented by the use of an alternative indicator, developed at Cornell University in the 1990s, the environmental impact quotient (EIQ). The EIQ indicator, developed by Kovach et al. and updated annually, effectively integrates the various environmental impacts of individual pesticides into a single “field value per hectare.” The EIQ value is multiplied by the amount of pesticide active ingredient (ai) used per hectare to produce a field EIQ value. For example, the EIQ rating for glyphosate is 15.33. By using this rating multiplied by the amount of glyphosate used per hectare (e.g., a hypothetical example of 1.1 kg applied per ha), the field EIQ value for glyphosate would be equivalent to 16.86/ha.

The EIQ indicator used is therefore a comparison of the field EIQ/ha for conventional vs. biotech crop production systems, with the total environmental impact or load of each system, a direct function of respective field EIQ/ha values and the area planted to each type of production (biotech vs. conventional). The use of environmental indicators is commonly used by researchers and the EIQ indicator has been, for example, cited by Brimner et al. in a study comparing the environmental impacts of biotech and conventional canola and by Kleiter.

The EIQ indicator provides an improved assessment of the impact of biotech crops on the environment when compared with examining only changes in volume of active ingredient applied, because it draws on some of the key toxicity and environmental exposure data related to individual products, as applicable to impacts on farm workers, consumers and ecology. Readers should, however, note that the EIQ is an indicator only and does not take into account all environmental issues and impacts. It is therefore not a comprehensive indicator. Detailed examples of the relevant amounts of active ingredient used and their associated field EIQ values for biotech vs conventional crops for the year 2010 are presented in Supplemental Appendix 2.

The methodology used to assess impact on greenhouse gas emissions combines reviews of literature relating to changes in fuel and tillage systems and carbon emissions coupled with evidence from the development of relevant biotech crops and their impact on both fuel use and tillage systems. Reductions in the level of GHG emissions associated with the adoption of biotech crops are acknowledged in a wide body of literature.^20-27 First, biotech crops contribute to a reduction in fuel use due to less frequent herbicide or insecticide applications and a reduction in the energy use in soil cultivation. Lazarus²⁸ estimated that one pesticide spray application uses 1.31 L of fuel which is equivalent to 3.5 kg/ha of carbon dioxide emissions.^f In this analysis, we used the conservative assumption that only GM IR crops reduced spray applications with the number of spray applications of herbicides remaining the same as for conventional production systems.^g

In addition, there has been a shift from conventional tillage to reduced/no till. This has had a marked impact on tractor fuel consumption due to energy intensive cultivation methods being replaced with no/reduced tillage and herbicide-based weed control systems. The GM HT crop where this is most evident is GM HT soybeans. Here adoption of the technology has made an important contribution to facilitating the adoption of reduced or no tillage farming.^h Before the introduction of GM HT soybean cultivars, no tillage (NT) systems were practised by some farmers with varying degrees of success using a number of herbicides. The opportunity for growers to control weeds with a non-residual foliar herbicide as a “burndown” pre-seeding treatment followed by a post-emergent treatment when the soybean crop became established has made the NT systems more reliable, technically viable and commercially attractive. These technical and cost advantages have contributed to the rapid adoption of GM HT cultivars and the near doubling of the NT soybean area in the US (also more than a 5-fold increase in Argentina). In both countries, GM HT soybeans are estimated to account for over 95% of the NT soybean crop area since 2007/8.

Substantial growth in NT production systems have also occurred in Canada, where the NT canola area increased from 0.8 million ha to 2.6 million ha (equal to about half of the total canola area) between 1996 and 2005 (95% of the NT canola area is planted with GM HT cultivars). Similarly the area planted to NT in the US cotton crop increased from 0.2 million ha to 1 million ha over the same period (of which 86% is planted to GM HT cultivars) and has remained at this share of the total crop since 2007.

The fuel savings resulting from changes in tillage systems used in this paper are drawn from a review of literature including Jasa,²² CTIC,¹⁹ University of Illinois,³⁰ USDA Energy Estimator³¹ and Reeder.³² The adoption of no tillage (NT) farming systems is estimated to reduce cultivation fuel usage by 27.22 L/ha compared with traditional conventional tillage (CT: average usage 49.01 L/ha) and by 9.56 L/ha compared with (the average of) reduced tillage (RT) cultivation methods (average usage 39.45 L/ha). In turn, this results in reductions of carbon dioxide emissions of 72.68 kg/ha for NT relative to CT and 25.53 kg/ha for RT relative to CT.ⁱ

Second, the use of “no-till” and “reduced-till” farming systems that utilize less ploughing increase the amount of organic carbon in the form of crop residue that is stored or sequestered in the soil. This carbon sequestration reduces carbon dioxide emissions to the environment. A number of researchers have examined the relationship between carbon sequestration and different tillage systems.^{24,25,27,33-41} This literature shows that the amount of carbon sequestered varies by soil type, cropping system, eco-region and tillage depth. It also shows that tillage systems can impact on levels of other GHG emissions such as methane and nitrous oxide and on crop yield. Overall, the literature highlights the difficulty in estimating the contribution NT/RT systems can make to soil carbon sequestration, especially because of the dynamic nature of soils, climate, cropping types and patterns. If a specific crop area is in continuous NT crop rotation, the full soil carbon sequestration benefits described in the literature can be realized. However, if the NT crop area is returned to a conventional tillage system, a proportion of the soil organic carbon gain will be lost. The temporary nature of this form of carbon storage will only become permanent when farmers adopt a continuous NT system which itself tends to be highly dependent upon effective herbicide-based weed control systems.

In sum, drawing on the various discussed literature, the analysis presented below uses the following conservative assumptions^j:

These assumptions were applied to the reduced insecticide spray applications data on GM IR crops, derived from separate analysis and reviews of impact literature by the authors,^1,47 and the GM HT crop areas using no/reduced tillage (limited to the GM HT soybean crops in North and South America and GM HT canola crop in Canada^l). Additional detail relating to the estimates for carbon dioxide savings at the country and trait levels are presented in Supplemental Appendix 3.