HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (19) Citing Articles via Google Scholar Google Scholar Articles by Cerdeira, A. L. Articles by Duke, S. O. Search for Related Content PubMed PubMed Citation Articles by Cerdeira, A. L. Articles by Duke, S. O. Agricola Articles by Cerdeira, A. L. Articles by Duke, S. O. Related Collections Weed Management Ecological Risk Assessment Agricultural Pesticides Plant and Environment Interactions Water Pollution

REVIEWS AND ANALYSES

The Current Status and Environmental Impacts of Glyphosate-Resistant Crops

A Review

^a Brazilian Department of Agriculture, Agricultural Research Service, Embrapa/Environment, C.P. 69, Jaguariuna-SP-13820-000, Brazil
^b USDA-ARS, Natural Products Utilization Research Unit, P.O. Box 8048, University, MS 38677, USA

^* Corresponding author (sduke{at}olemiss.edu)

Received for publication October 3, 2005.

	ABSTRACT

TOP ABSTRACT INTRODUCTION GLYPHOSATE AND GRCs EFFECTS ON HERBICIDE AND... EFFECTS ON SOIL EFFECTS ON AIR EFFECTS ON WATER EFFECTS ON OTHER NON-TARGET... FOOD AND FEED SAFETY GLYPHOSATE-RESISTANT WEEDS... GENE FLOW (INTROGRESSION)... CONCLUSIONS REFERENCES

 
Glyphosate [N-(phosphonomethyl) glycine]-resistant crops (GRCs), canola (Brassica napus L.), cotton (Gossypium hirsutum L.), maize (Zea mays L.), and soybean [Glycine max (L.) Merr.] have been commercialized and grown extensively in the Western Hemisphere and, to a lesser extent, elsewhere. Glyphosate-resistant cotton and soybean have become dominant in those countries where their planting is permitted. Effects of glyphosate on contamination of soil, water, and air are minimal, compared to some of the herbicides that they replace. No risks have been found with food or feed safety or nutritional value in products from currently available GRCs. Glyphosate-resistant crops have promoted the adoption of reduced- or no-tillage agriculture in the USA and Argentina, providing a substantial environmental benefit. Weed species in GRC fields have shifted to those that can more successfully withstand glyphosate and to those that avoid the time of its application. Three weed species have evolved resistance to glyphosate in GRCs. Glyphosate-resistant crops have greater potential to become problems as volunteer crops than do conventional crops. Glyphosate resistance transgenes have been found in fields of canola that are supposed to be non-transgenic. Under some circumstances, the largest risk of GRCs may be transgene flow (introgression) from GRCs to related species that might become problems in natural ecosystems. Glyphosate resistance transgenes themselves are highly unlikely to be a risk in wild plant populations, but when linked to transgenes that may impart fitness benefits outside of agriculture (e.g., insect resistance), natural ecosystems could be affected. The development and use of failsafe introgression barriers in crops with such linked genes is needed.

Abbreviations: AMPA, aminomethylphosphonate • EPSPS, 5-enolpyruvyl-shikimate-3-phosphate synthase • GR, glyphosate-resistant • GRC, glyphosate-resistant crop • HRC, herbicide-resistant crop

	INTRODUCTION

TOP ABSTRACT INTRODUCTION GLYPHOSATE AND GRCs EFFECTS ON HERBICIDE AND... EFFECTS ON SOIL EFFECTS ON AIR EFFECTS ON WATER EFFECTS ON OTHER NON-TARGET... FOOD AND FEED SAFETY GLYPHOSATE-RESISTANT WEEDS... GENE FLOW (INTROGRESSION)... CONCLUSIONS REFERENCES

 
HERBICIDE AND INSECT RESISTANCE are the two categories of transgene-conferred traits for crops that have had significant effects on agriculture (Gutterson and Zhang, 2004). Herbicide-resistant crops (HRCs) (Duke, 1996), sometimes called herbicide-tolerant crops, are crops made resistant to herbicides either by transgene technology or by selection in cell or tissue culture for mutations that confer resistance. Most of the success and controversy about safety of HRCs surrounds glyphosate-resistant crops (GRCs), the topic of this review. Glyphosate-resistant crops and their environmental impact have never been the sole subject of an in-depth review, although HRCs have been extensively reviewed (Dekker and Duke, 1995; Duke et al., 1991, 2002; Duke, 1998, 2002, 2005; Duke and Cerdeira, 2005a,b; Dyer et al., 1993; Gressel, 2002b; Hess and Duke, 2000; Silvers et al., 2003; Warwick and Miki, 2004) and have been the topic of two edited books (Duke, 1996; McClean and Evans, 1995) and a special issue of the journal Pest Management Science in 2005. A recent review covered agronomic and environmental aspects of HRCs (Schuette et al., 2004). Other reviewers have discussed the environmental impacts of all transgenic crops, with coverage of GRCs (Carpenter et al., 2002; Uzogara, 2000). Lutman et al. (2000) and Kuiper et al. (2000) published brief reviews of environmental consequences of growing GRCs. Dill (2005) briefly discussed the current status of GRC products. Meyer and Wolters (1998) reviewed the ecological effects of the herbicide use associated with GRCs. None of these publications have focused solely on an in-depth assessment of the potential toxicological and environmental impacts of all aspects of GRCs.

We will discuss what we consider the most important and germane literature, along with other selected examples. The regulatory process for approval of HRCs in the many countries that regulate their approval will not be discussed. The chemical, transgene, and weed management (e.g., changes in tillage) aspects of GRC environmental impacts will be covered. Unfortunately, we cannot access the potential environmental risks of the compounds with which glyphosate is formulated, due the lack of literature on this topic. This problem is exacerbated by the fact that formulations have varied through time and between geographic areas. This herbicide no longer has patent protection, so many formulations are now sold under several trade names. The potential environmental impact of a technology is often geography and/or time dependent. Thus, extrapolation of the results and conclusions of studies to all situations is impossible. The best we can do is generalize from reported studies that may not cover every situation. Analysis of impact cannot be done in a vacuum. Thus, we will at times contrast certain risks of GRCs with the risks that they displace. The viewpoints in this analysis are those of the authors and are not meant to reflect those of our employers.

	GLYPHOSATE AND GRCs

TOP ABSTRACT INTRODUCTION GLYPHOSATE AND GRCs EFFECTS ON HERBICIDE AND... EFFECTS ON SOIL EFFECTS ON AIR EFFECTS ON WATER EFFECTS ON OTHER NON-TARGET... FOOD AND FEED SAFETY GLYPHOSATE-RESISTANT WEEDS... GENE FLOW (INTROGRESSION)... CONCLUSIONS REFERENCES

 
Glyphosate is a very effective non-selective herbicide. Before introduction of GRCs, glyphosate was used in non-crop situations, before planting the crop, or with specialized application equipment to avoid contact with the crop (Duke, 1988; Duke et al., 2003a; Franz et al., 1997). Even before the introduction of GRCs, glyphosate was a very successful herbicide. However, glyphosate use increased more than six-fold between 1992 and 2002, to become the most used herbicide in the United States (Gianessi and Reigner, 2006). Most of this increase in use is due to the adoption of GRCs.

Glyphosate is the only herbicide that acts by blocking the shikimate pathway through inhibition of 5-enolpyruvyl-shikimate-3-phosphate synthase (EPSPS). Inhibition of EPSPS results in reduced aromatic amino acids and deregulation of the shikimate pathway (Duke et al., 2003a). The latter effect causes massive flow of carbon into the pathway, with accumulation of high levels of shikimic acid and its derivatives. Glyphosate is particularly effective because most plants metabolically degrade it very slowly or not at all, and it translocates well to metabolically active tissues such as meristems. Its relatively slow mode of action allows movement of the herbicide throughout the plant before symptoms occur. Glyphosate is only used as a post emergence herbicide, as it has little or no activity in soil. It is not a low dose rate herbicide, with recommended application doses between 0.21 and 4.2 kg ai ha^–1, depending on the use (Vencill, 2002). In most agronomic crops, it is used at rates higher than 1 kg ha^–1. Glyphosate is an anion and is sold as a salt with different cations (e.g., isopropyl amine, trimethylsulfonium, diammonium).

At this time, glyphosate-resistant (GR) soybean, cotton, canola, maize, alfalfa, and sugarbeet are available to farmers of North America (Table 1). Other GR crops and turf grasses are under development. Three transgenes are used with these crops. A gene (CP4) encoding a GR form of EPSPS from Agrobacterium sp. was found to be very effective at producing GRCs (Padgette et al., 1996a). A gene from the microbe Ochrobactrum anthropi that encodes a glyphosate-degrading enzyme (glyphosate oxidase, GOX) is used in GR canola with CP4 EPSPS (Padgette et al., 1996a). GOX degrades glyphosate to glyoxylate, an ubiquitous and safe natural product, and aminomethylphosphonate (AMPA). A form of GR EPSPS from maize was produced by site-directed mutagenesis for use as a transgene in maize (Dill, 2005).

There is still tremendous potential for growth of GRC market share. The worldwide area of major crops is around 700 million hectares, not counting pasture (Fig. 1 ). GRCs represent only a small fraction of this crop area. Worldwide, almost all of the approximately 65 million ha of herbicide-resistant crops (ISAAA, 2005) planted in 2004 were GRCs.

View larger version (12K): [in this window] [in a new window]   Fig. 1. Worldwide area of major crops. (Food and Agriculture Organization, 2003).

View larger version (22K): [in this window] [in a new window]   Fig. 2. Adoption of glyphosate-resistant soybean and cotton in the U.S.A. by year. (Duke, 2005; USDA–ERS, 2005a).

Approximately 75% of canola acreage in the U.S.A. was planted in GR varieties in 2003 (Gianessi, 2005). In Australia, an economic analysis of GR canola showed significant economic advantages (Monjardino et al., 2005).

Despite great success with other GRCs, GR sugarbeet (Beta vulgaris L.) was not grown by North American sugarbeet farmers after its first approval for use, due to concerns about acceptance of sugar from transgenic plants by the confectionary and other prepared food industries. Similar and other concerns resulted in a decision by the company developing GR wheat (Triticum aestivum L.) technology not to ask for deregulation in 2004 (Dill, 2005). GR sugarbeet was deregulated again in 2005, for planting in 2006. Savings for farmers in Europe if GR sugarbeet were adopted are estimated to be 220 ha^–1 yr^–1 (Pidgeon et al., 2004).

Glyphosate-resistant alfalfa (Medicago sativa L.) was deregulated in 2005, and there is a petition for deregulation of GR creeping bentgrass (Agrostis stolonifera L.). Glyphosate-resistant onions (Allium cepa L.), Kentucky bluegrass (Poa pratensis L.), and peas (Pisum sativum L.) have approval for field testing. However, as was the case with wheat, this does not ensure that these GRCs will be commercialized. There is already concern about gene flow from creeping bentgrass to weedy relatives (Watrud et al., 2004).

	EFFECTS ON HERBICIDE AND FOSSIL FUEL USE

TOP ABSTRACT INTRODUCTION GLYPHOSATE AND GRCs EFFECTS ON HERBICIDE AND... EFFECTS ON SOIL EFFECTS ON AIR EFFECTS ON WATER EFFECTS ON OTHER NON-TARGET... FOOD AND FEED SAFETY GLYPHOSATE-RESISTANT WEEDS... GENE FLOW (INTROGRESSION)... CONCLUSIONS REFERENCES

 
Effects on Herbicide Use
Since the mid-twentieth century, herbicides have been the primary means of weed management in developed countries. In North America, for the past two decades, herbicides have accounted for about 70% of pesticide use in crops (Agrow World Crop Protection News, 1998). Before herbicides, extensive tillage and manual weeding were the primary means of weed management. There has been controversy about whether GRCs have increased herbicide use or not. Glyphosate is normally often used at rates of a kg or more per ha, whereas some of the newer herbicides such as carfentrazone-ethyl are effective at only about 10–50 g per ha (Vencill, 2002). The controversy over increased volume of herbicide use has been fueled by others' assumption that an increased amount of chemical use equals increased environmental damage and toxicological risk. This assumption does not take into account the facts that the potential environmental damage and toxicological risk can vary considerably between different herbicides as determined by many factors, including use rate and toxicity. Thus, comparing herbicide use rates has relatively little bearing on potential environmental damage or toxicological risk to humans. A very few studies, such as those by Nelson and Bullock (2003), have compared toxicological risk, rather than herbicide active ingredient used per unit area.

Even though glyphosate is not a low use rate herbicide, it is considered to be a low risk herbicide in terms of toxicity and environmental effects. Nevertheless, we will discuss some of the literature that addresses the question of use rate. A few studies have claimed that the volume of herbicide use is greater with GRCs (Benbrook, 2001b, 2003). However, others such as Heimlich et al. (2000) have concluded that no significant change in the overall amount of herbicide has been observed with the adoption of GRCs in the U.S.A. Heimlich et al. (2000) noted that substitution of glyphosate for other herbicides resulted in the replacement of herbicides that are at least three times more toxic, and that persist nearly twice as long as glyphosate. Gianessi and Carpenter (2000) came to similar conclusions. An analysis by Trewavas and Leaver (2001) showed that 3.27 million kg of other herbicides have been replaced by 2.45 million kg of glyphosate in U.S. soybean fields. Carpenter and Gianessi (2003) concluded that introduction of GR soybeans in the U.S.A. resulted in a decrease in the total volume of herbicides used. Gianessi (2005) claims that GRCs generally require less herbicide than non-transgenic crops. Furthermore, he estimates that averaged over all GRCs, glyphosate-resistance technology has reduced herbicide use by 17 million kg per yr in the U.S.A. In cotton, the amount of herbicide used per unit area in the U.S.A. stayed about the same between 1996 and 2000 (Carpenter and Gianessi, 2003), a period during which adoption of GR cotton grew from 0 to about 50% (Fig. 2). Among the herbicides replaced in cotton were arsenic-containing compounds. Gianessi's (2005) calculations indicate that if GR sugarbeets were adopted, herbicide use reduction would not be as great as for combined GRCs, as the herbicides now used in non-transgenic sugarbeets are mostly low use rate compounds in the U.S.A. Coyette et al. (2002) estimated that introduction of GR sugarbeet to Europe would result in a decrease of herbicide use.

Weed control could be achieved with very low use rate herbicides (Benbrook, 2001a, b), reducing the volume of chemicals used for weed management to levels below that currently used in non-glyphosate HRCs or with GRCs. If this were more economical and efficacious, farmers would probably adopt such a strategy. But, again, simply reducing the volume of chemical used does not assure that risks are reduced. For example, many of the low use rate herbicides have long persistence in soil compared to glyphosate.

Shiva (2001) claims that introduction of GRCs to underdeveloped countries, where hand weeding is the primary means of weed management, will increase herbicide use (Shiva, 2001). At this time, there is no evidence that this has occurred. The economic constraints that prevent these farmers from using selective herbicides will be similar for GRCs. However, should weed management with GRCs become economically viable for poor farmers in underdeveloped areas, herbicide use will increase, displacing tillage and hand labor. Hand labor is rarely used with canola or soybeans, even in developing countries.

As discussed below in this review, the amount of glyphosate used with a GRC year after year may increase with time, as naturally-resistant weed species and biotypes invade fields and resistance to glyphosate evolves. Both increased amounts of glyphosate and other herbicides will be used in these cases. Benbrook (2003) claims that this biologically-driven increase in herbicide use has already occurred with GRCs. Owen and Zelaya (2005) also report that this is already happening with GRCs in some locales.

The worldwide decrease in cost of glyphosate due to loss of patent protection (Woodburn, 2000) also makes higher application rates economical in some cases. The heavy adoption of GR soybeans in the USA contributed to the dramatic (as much as 80%) reductions in the costs of most other soybean herbicides due to competition (Nelson and Bullock, 2003). Thus, indirectly, GR soybeans have, in some cases, helped make it more economical for farmers to use higher rates of other herbicides, sometimes with less desirable toxicological or environmental profiles than glyphosate. Another factor contributing to the reductions in herbicide costs have been patent expirations of other herbicides. Despite the more competitive prices of competing herbicides, adoption of GRCs increased dramatically during this period (Fig. 2).

In a study using the environmental impact quotient method of Kovach et al. (1992), Kleter and Kuiper (2003) calculated total environmental impact of the herbicides, farm worker exposure impact, consumer impact, and ecology impact associated with the herbicides used with various GRCs versus those used with the same non-transgenic crops. Their calculations were based on herbicide use data of Gianessi et al. (2002). The amount of herbicide used was reduced for all crops. All impacts are reduced in all crops by adoption of HRCs (primarily GRCs). With canola, cotton, and soybean, farm worker and consumer impact are reduced more than ecological impact in their study. Brimner et al. (2005), using the same method, found the environmental impact of herbicide-resistant canola in Canada markedly lower than that of conventionally grown canola.

Effects on Fossil Fuel Use
A major expense and source of pollution in weed management are the fossil fuels used in tillage and herbicide application. This factor is seldom considered in evaluations of environmental impacts of herbicide use. In some countries (e.g., Denmark), mandated herbicide reduction programs have also required fewer applications of herbicides. Certainly, GRCs have greatly reduced tillage (discussed below) and, in some cases, the number of herbicide applications (Gianessi, 2005). Energy use with tillage is much higher than that with herbicide spraying.

Few studies have carefully evaluated the impact of GRCs on reduced fossil fuel use in weed management, although this is generally recognized as a beneficial aspect of GRCs (Olofsdotter et al., 2000). In a recent study in Europe using a life-cycle assessment approach, Bennett et al. (2004) concluded that the major environmental advantage of growing GR sugarbeet would be much lower emissions from herbicide manufacturing, transport, and field operations (overall, approximately 50% less energy required), thus reducing contributions to global warming, smog, ozone depletion, ecotoxicity of water, and acidification and nitrification of soil and water. Some of these effects are illustrated in Fig. 3 . They qualified their conclusions by stating that the environmental and health impacts of growing GRCs should be assessed on a case-by-case basis, using a holistic approach.

View larger version (21K): [in this window] [in a new window]   Fig. 3. Comparison of impacts of typical herbicide regimes for conventional, compared with glyphosate-resistant sugarbeet in the UK and Germany in terms of energy requirements (MJ), global warming potential [kg carbon dioxide (CO2) equivalent], and ozone depletion [kg chlorofluorocarbon (CFC) 11 equivalent] per functional unit. UKA and UKB are two different herbicide regimes used with non-transgenic sugarbeet in the UK, GER is a typical herbicide regime used with non-transgenic sugarbeet in Germany, and Ht is with use of glyphosate only with glyphosate-resistant sugarbeet (Bennett et al., 2004).

	EFFECTS ON SOIL

TOP ABSTRACT INTRODUCTION GLYPHOSATE AND GRCs EFFECTS ON HERBICIDE AND... EFFECTS ON SOIL EFFECTS ON AIR EFFECTS ON WATER EFFECTS ON OTHER NON-TARGET... FOOD AND FEED SAFETY GLYPHOSATE-RESISTANT WEEDS... GENE FLOW (INTROGRESSION)... CONCLUSIONS REFERENCES

Soil Contamination
Glyphosate is not considered to be a significant soil contaminant when used in recommended doses. It is applied as foliar sprays, so that contamination of soil is from direct interception of spray by the soil surface or from runoff or leaching of the herbicide and/or its breakdown products from vegetation. Glyphosate can be translocated to roots from foliar tissues and exuded by the roots into the soil (Coupland and Caseley, 1979). Glyphosate strongly adsorbs to soil particles and is rapidly degraded by soil microbes (Duke, 1988; Duke et al., 2003a).

After long-term use of glyphosate in Canadian soils, no residues were detected (Miller et al., 1995). Haney et al. (2000) found a cumulative soil carbon mineralization with increasing glyphosate rate. The CO₂ flush 2 d after application suggested that glyphosate was either readily and directly utilized by soil microbes or made other resources available (Fig. 4 ). Glyphosate has a moderate half-life in soils with an avg value of approximately 47 d, but reaching 174 d in some soils under some conditions (Vencill, 2002; Wauchope et al., 1992). Gimsing et al. (2004) indicated up to 40–50% of glyphosate is mineralized in 90 d, with AMPA accumulating as the major metabolite. They hypothesized that the soil pseudomonad population is responsible for most of the degradation. Abiotic oxidative degradation of glyphosate and AMPA by manganese oxide in the form of birnessite, a common component of soils, was recently reported (Barrett and McBride, 2005). Manganese oxide caused breakage of both the C-P and C-N bonds.

View larger version (15K): [in this window] [in a new window]   Fig. 4. Effect of glyphosate addition dose on soil carbon mineralization. Carbon mineralized from basal microbial respiration in control samples has been subtracted. The 1, 2, 3, and 5X represent glyphosate addition doses of 47, 94, 140, and 234 mg ai kg–1 soil, respectively. Error bars indicate ± 1 standard deviation (Haney et al., 2000).

Effects on Soil Biota
The potential direct effects of GRCs and their management on soil biota include changes in soil microbial activity due to direct effects of glyphosate, differences in the amount and composition of root exudates of GRCs vs. non-GRCs, changes in microbial functions resulting from gene transfer from the transgenic crop, and alteration in microbial populations because of the effects of management practices for transgenic crops, such as changes in other herbicide applications and tillage (Dunfield and Germida, 2004). Most of the available literature addresses the first effect.

In general, there is little effect of glyphosate on soil microflora. For example, Gomez and Sagardoy (1985) found no effect of glyphosate on microflora of soils in Argentina at twice the recommended rates of the herbicide. Studies on the effect of glyphosate on microbial activity of typical Hapludult and Hapludox Brazilian soils measured by soil respiration (evolution of CO₂) and fluorescein diacetate (FDA) hydrolysis revealed a transient increase of 10–15% in the CO₂ evolved and a 9–19% increase in FDA hydrolysis in the presence of glyphosate (Araujo et al., 2003). Soil which had been exposed to glyphosate for several years had a strong response in microbial activity. After 32 d incubation with glyphosate, the number of actinomycetes and fungi had increased, while the number of bacteria was slightly reduced. After the incubation period, HPLC detected AMPA, indicating glyphosate degradation by soil microorganisms. Haney et al. (2000, 2002) generated data suggesting that glyphosate causes enhanced microbial activity directly. An increase in the carbon mineralization rate occurred the first day following glyphosate addition and continued for 14 d (Fig. 4). Glyphosate appeared to be rapidly degraded by soil microbes regardless of soil type or organic matter content, even at high application rates, without adversely affecting microbial activity. Liphadzi et al. (2005) found no differences in the effects of glyphosate and conventional herbicides in GR maize and soybean on nematode densities, soil microbial biomass, or substrate-induced respiration in soil, suggesting that soil health was similar under the two herbicide regimes.

Siciliano and Germida (1999) found differences in rhizosphere-associated microbes between a GR and two non-transgenic canola cultivars. Endophytic bacterial populations also varied between cultivars. In a later study, Dunfield and Germida (2001) concluded that there were differences in bacterial communities in the rhizosphere of GR canola varieties compared to non-transgenic varieties. However, the changes were temporary and did not persist until the next field season (Dunfield and Germida, 2003).

The soybean nitrogen-fixing symbiont Bradyrhizobium japonicum Kirchner possesses a glyphosate-sensitive EPSPS, and on exposure to glyphosate accumulates high concentrations of shikimate and certain benzoic acids that could be plant growth inhibitors (Moorman et al., 1992). These effects are accompanied by growth inhibition and/or death of the microbe, depending on the glyphosate concentration. This paper suggested that there could be effects of glyphosate on nitrogen metabolism in GR soybeans. Furthermore, glyphosate is preferentially translocated to nodules of soybeans (Reddy and Zablotowicz, 2003). Related research (King et al., 2001; Reddy et al., 2000), summarized by Zablotowicz and Reddy (2004), indicated that such effects as reduction in nodulation, nodule size, and leghemoglobin content of nodules can be caused in GR soybeans sprayed with glyphosate. However, the effect of glyphosate on nitrogenase activity in nodules from GR soybeans in field studies was inconsistent. Greenhouse studies indicated that adverse effects should be maximal under moisture stress. Sensitivity of B. japonicum strains varied. In the field, effects are transient, and there is no evidence that crop yield is affected. In GRCs, amino acids from the plant may prevent significant effects on B. japonicum, since aromatic amino acids added to B. japonicum grown in culture can reduce growth inhibition by glyphosate (dos Santos et al., 2005).

Kremer et al. (2005) reported that root exudates from GR soybean treated with glyphosate contained glyphosate at concentrations higher than 1000 ng per plant and about twice the amino acid content of non-treated plants over a period of 16 d after treatment. Carbohydrate content of the root exudate of GR plants was unaffected by glyphosate treatment, although carbohydrate root exudation of GR soybean was about twice that of non-transgenic soybean. In vitro bioassays showed that root exudates of glyphosate-treated GR soybean stimulated growth of selected rhizosphere fungi (Fusarium spp. and Pseudomonas spp.) that could adversely affect plant growth and biological processes in the soil and rhizosphere. With Fusarium spp., microbial growth was generally better with glyphosate-treated GR soybean root exudate than with exudate from non-transgenic soybean either treated or not treated with glyphosate. These results would predict that there could be more problems with Fusarium spp. in GR soybeans than non-transgenic soybeans. In a study in which glyphosate was sprayed on soil, the bacterial endophyte community associated with subsequently planted soybeans was altered (Kuklinsky–Sobral et al., 2005).

Motavalli et al. (2004) concluded in a review that there is so far no conclusive evidence that those GRCs and other transgenic crops, which have been deregulated and used in many cropping situations, climates, and soil types over the past 10 yr, have had any significant effect on nutrient transformations by microbes. However, they point out that this topic needs further study, as not every situation has been adequately researched. In another recent review, Dunfield and Germida (2004) stressed that the effects shown are field site- and season-dependent and that the method of analysis can affect the results. They point out that the changes in microbial communities associated with GRCs are more variable and transient compared to those caused by other agricultural practices such as crop rotation, tillage, use of certain other herbicides, and irrigation. Nevertheless, they stated that minor alterations in the total diversity of the soil microbial community, such as removal or appearance of certain microbes (e.g., rhizobacteria or plant pathogens) could affect soil health and ecosystem function. Of course, other herbicides could also cause such minor effects, and soil-applied herbicides might cause more than minor temporary effects. Kowalchuk et al. (2003), in a review of the effects of all transgenic crops on soil microbes, states that observed effects have generally been minor and that they are very small in comparison with other sources of variation. They propose case-by-case approaches that target both potentially vulnerable microbes, as well as community parameters in evaluating the impact of transgenic crops on soil microorganisms. So far, however, no agriculturally significant effect of glyphosate on soil microorganisms has been documented.

Soil Loss and Compaction
A positive impact of the use of GRCs is that they facilitate reduced- or zero-tillage systems, which contribute to reductions in soil erosion from water and wind, fossil fuel use, air pollution from dust, loss of soil moisture, and soil compaction (Holland, 2004). Reduced tillage also improves soil structure, leading to reduced risk of runoff and pollution of surface waters with sediment, nutrients, and pesticides.

Considering the relatively high level of potential environmental improvement that can be gained by reducing tillage, there is a remarkable paucity of refereed publications on the influence of GRCs on tillage practices and associated environmental effects. Loss of top soil due to tillage is perhaps the most environmentally destructive effect of row crop agriculture. Even taking land out of its natural state for agriculture is more rapidly reversible than the loss of top soil, which, once lost, can take centuries to replace.

A survey by the American Soybean Association (2001) found that 53% of U.S. soybean farmers made an average of 1.8 fewer tillage passes per yr through their soybean fields since GR soybeans were introduced. This translates to a savings of $385 million per yr in reduced tillage costs. In a 5-yr period in the U.S.A., during which the planting of GR soybeans increased from only a few percent to about 70% (Fig. 2), there was a dramatic increase in the adoption of no-tillage and reduced-tillage management (Fig. 5 ). Most of this change was associated with the growing of GR soybeans (Fig. 6 ). We have not found documentation as to whether this trend has continued; however, weed changes (as discussed later in this review) in GRCs have caused some farmers to return occasionally to tillage as a weed management tool. Similarly, there has been a rise in no-tillage agriculture in soybeans in Argentina with the adoption of GR soybeans, where there is a yearly loss of 10 tons of topsoil per ha in soybeans produced with conventional tillage, compared to 2.5 tons annually with no-tillage agriculture in GR soybeans (Penna and Lema, 2003).

View larger version (24K): [in this window] [in a new window]   Fig. 5. Soybean tillage methods by hectares farmed in the U.S.A. in 1996 and 2001. In 1996 and 2001, there were 19.2 and 23 million ha, respectively, of soybeans grown (American Soybean Association, 2001).

View larger version (29K): [in this window] [in a new window]   Fig. 6. Tillage practices and glyphosate-resistant soybean use by hectares in the USA in 1996 and 2001. (American Soybean Association, 2001).

	EFFECTS ON AIR

TOP ABSTRACT INTRODUCTION GLYPHOSATE AND GRCs EFFECTS ON HERBICIDE AND... EFFECTS ON SOIL EFFECTS ON AIR EFFECTS ON WATER EFFECTS ON OTHER NON-TARGET... FOOD AND FEED SAFETY GLYPHOSATE-RESISTANT WEEDS... GENE FLOW (INTROGRESSION)... CONCLUSIONS REFERENCES

 
One study has focused on impacts of GRCs on the air. Bennett et al. (2004) used a life-cycle assessment model comparing the environmental and human health impacts of conventional sugarbeet-growing regimes in the United Kingdom and Germany with those that might be expected if GR sugarbeet were grown. The results suggested that growing this GRC would be less harmful to the environment and human health than growing the conventional crop, largely due to lower emissions from herbicide manufacture, transport, and field operations. Most of this analysis dealt with air and water pollution. Emissions contributing to negative environmental impacts, such as global warming, ozone depletion, toxic impacts on aquatic water ecosystems, and acidification and nutrification of soil and water, were much lower for the GRC than for the conventional crop. Emissions contributing to summer smog, toxic particulate matter and carcinogenicity, which have negative human health impacts, were also substantially lower for the HRC (Fig. 3). Herbicides can pollute the air by drift (air movement of sprayed herbicide-containing droplets to unintended sites) or volatility. Glyphosate is essentially not volatile at 25°C (Vencill, 2002) and has not been reported as an atmospheric contaminant (Van Dijk and Guicherit, 1999).

Most herbicides are applied by spraying, resulting in movement to non-target sites and organisms through the air. Herbicide drift effects on unintended crops and other vegetation have been a problem since the use of potent, synthetic herbicides began. After GR soybeans were introduced, Owen (1998) reported that complaints of herbicide drift problems increased in Iowa. Growing a GRC next to a non-GRC of the same species may exacerbate such problems, as there is no visual difference between the two crops to the herbicide applicator. Furthermore, with GRCs, the herbicide can be used during later crop development by aerial application, further increasing the risk of drift.

	EFFECTS ON WATER

TOP ABSTRACT INTRODUCTION GLYPHOSATE AND GRCs EFFECTS ON HERBICIDE AND... EFFECTS ON SOIL EFFECTS ON AIR EFFECTS ON WATER EFFECTS ON OTHER NON-TARGET... FOOD AND FEED SAFETY GLYPHOSATE-RESISTANT WEEDS... GENE FLOW (INTROGRESSION)... CONCLUSIONS REFERENCES

 
Water Contamination
Glyphosate is strongly adsorbed to soil particles, and, even though it is highly water soluble, it does not leach into ground water in most soils. Soil and sediments of bodies of water are the main sinks for glyphosate residues from surface water, greatly reducing further transport (Franz et al., 1997; Goldsborough and Brown, 1993). Once in surface water, it dissipates more rapidly than most other herbicides. Various studies have shown that glyphosate appears in surface water less than several alternative herbicides (Carpenter et al., 2002). For example, in long-term studies conducted in Canada, no residues of glyphosate were found in groundwater (Miller et al., 1995). Several studies have found no glyphosate in ground water in the U.S.A. where it is applied in no-tillage cropping systems (Kolpin et al., 1988) and in Brazil in various cropping systems (Bonato et al., 1999; Cerdeira et al., 2005; Lanchote et al., 2000; Paraíba et al., 2003). Similar results were found for surface waters (Clark et al., 1999). Even in a worst case scenario of preferential flow, where compounds that bind soil strongly might leach more readily, glyphosate at recommended use rates did not leach past 2.4 m at concentrations approaching environmental concern (Malone et al., 2004).

A recent study in Denmark found that both glyphosate and AMPA can leach to a depth of 1 m through structured soils with pronounced macropore flow (Kjaer et al., 2005). Considering the results of many other studies, this must be an unusual situation. Spraying onto vegetated soil vs. bare soil in vineyards reduced the potential for movement of both glyphosate and AMPA to ground water (Landry et al., 2005). Kolpin et al. (2006) detected AMPA almost four times as often as glyphosate downstream from waste water effluent from treatment plants treating water from urban environments where glyphosate was used.

In the intensely farmed maize-growing regions of the mid-western USA, surface waters have often been contaminated by herbicides, principally as a result of rainfall runoff occurring shortly after application of herbicides to maize and other crops (Wauchope et al., 2002). A model was used to predict maize herbicide concentration in reservoirs as a function of herbicide properties, comparing broadcast surface pre-plant atrazine [6-chloro-N-ethyl-N'-(1-methylethyl)-1,3,5-triazine-2,4-diamine] and alachlor [2-chloro-N-(2,6-diethylphenyl)-N-(methoxymethyl)acetamide] applications with glyphosate or glufosinate [2-amino-4-(hydroxymethylphosphinyl)butanoic acid] with both GR and glufosinate-resistant maize (Wauchope et al., 2002). Because of their lower post-emergent application rates and greater soil sorptivity, glyphosate and glufosinate loads in runoff were generally one-fifth to one-tenth those of atrazine and alachlor, indicating that the replacement of pre-emergent maize herbicides with these post-emergent herbicides, allowed by use of transgenic crops, would dramatically reduce herbicide concentrations in vulnerable watersheds. Similarly, Estes et al. (2001) predicted in a higher tier modeling examination, that various herbicide-use regimes employed in the U.S.A. in maize caused more ground and surface water contamination than did glyphosate when used with GR maize, thereby reducing risk to drinking water and related ecosystems. By modeling, Peterson and Hulting (2004) predicted less risk to groundwater and aquatic plants by a GR wheat/herbicide system than from a non-transgenic wheat/conventional herbicide system.

Effects on Aquatic Life
Because of its relative safety, glyphosate is one of only nine synthetic herbicides approved for use in aquatic sites in the U.S.A. (Getsinger et al., 2005). Glyphosate was not found to bioaccumulate, biomagnify, or persist in an available form in the environment (Solomon and Thompson, 2003). This study also showed that the risk to aquatic organisms is negligible or small at application rates < 4 kg ha^–1 and only slightly greater at application rates of 8 kg ha^–1. As we mentioned earlier, glyphosate is less likely to pollute ground and surface waters than many of the herbicides that it replaces. Cedergreen and Streibig (2005) found glyphosate to be the least toxic herbicide of ten tested to the aquatic plant Lemna minor L. and the third least toxic of the ten herbicides to the green alga Pseudokirchneriella subcapitata (Korschikov) Hindak. The formulated version of glyphosate was four-fold more phytotoxic than unformuated glyphosate, but this did not change the rankings. A life-cycle assessment technique used to compare conventional sugarbeet agricultural practices with risks that might be expected if GR sugarbeet were grown suggested that growing this GRC would be less harmful to the ecology of water for the GRC than for the conventional crop (Bennett et al., 2004). Relyea (2005) reported that a commercial formulation of glyphosate sprayed directly into aquatic mesocosms (in 1200-L polyethylene tanks) caused a 22% reduction in species diversity with particularly severe impacts on amphibians. The control mesocosms were not sprayed with a formulation blank, so whether the effects were due to glyphosate or the other formulation ingredients is unclear.

Finally, glyphosate was recently found to reduce toxicity of some metal ions (Ag, Cd, Cr, Cu, Ni, Pb, and Zn), but not ions of Hg or Se, to a freshwater claderceran by binding their ionic forms and reducing their bioavailability (Tsui et al., 2005). Thus, under circumstances of metal ion toxicity, glyphosate could benefit some organisms. Such effects would be transient due to the relatively short environmental half-life of glyphosate. One would assume that these results would extrapolate to other aquatic organisms, but further work should be done. In general, what we know about glyphosate's movement to surface water suggests less impact of GRCs on aquatic vegetation than conventionally-grown crops.

	EFFECTS ON OTHER NON-TARGET ORGANISMS

TOP ABSTRACT INTRODUCTION GLYPHOSATE AND GRCs EFFECTS ON HERBICIDE AND... EFFECTS ON SOIL EFFECTS ON AIR EFFECTS ON WATER EFFECTS ON OTHER NON-TARGET... FOOD AND FEED SAFETY GLYPHOSATE-RESISTANT WEEDS... GENE FLOW (INTROGRESSION)... CONCLUSIONS REFERENCES

 
In Europe there has been some controversy about the effects of HRCs on biodiversity in farm-scale evaluation studies. There is a desire in Europe to incorporate the maintenance of some weed species within crops to maintain ecological diversity. Studies have linked the presence of weeds to biodiversity of invertebrates, wildlife, and birds, since weeds provide shelter and a food supply for these animals. Marshall (2001) speculated that the currently available HRCs, including GRCs, seem unlikely to provide the required flexibility of management for leaving sufficient weeds for these purposes. Perry et al. (2004) found larger weed abundance in GR fodder maize than with conventional weed management. Dewar et al. (2003) devised a strategy to use band spraying in a GRC to increase biodiversity within the crop while providing habitat for birds and wildlife. Their method of leaving weeds between crop rows could, in some cases, be used without compromising crop yield. Petersen and Röver (2005) used such a method in GR sugarbeet and found no loss in yield. In a similar study using such a strategy, May et al. (2005) found that GR sugarbeets offered more flexibility in devising management strategies to enhance wildlife than conventional weed management methods.

Some have suggested that biodiversity would be increased with HRCs that use a broad-spectrum, foliar-applied herbicide like glyphosate, since the farmer can wait to spray weeds after there has been some weed growth, providing habitat and/or food for birds, arthropods, or other herbivores. However, Freckleton et al. (2004), using weed phenologies and a population model, predicted that such effects would probably be transient. The study suggested that if herbicide application could be ceased earlier, a viable population of late-emerging weeds could be maintained. In a multi-year study in Argentina on GR soybeans in which glyphosate was used yearly, weed species diversity decreased or remained stable early in the growing season and increased by harvest time as a result of this weed management choice (Vitta et al., 2004).

Perhaps an even greater effect on non-target plant life than that caused by glyphosate is the effect of tillage on vegetation. No-tillage agriculture, which is used more with GRCs, results in weed species shifts (as discussed later in this review) and more vegetation on the field before and after the period of crop production, which in turn results in improved habitat for other organisms.

Non-Target Plants
For most farmers, the goal of using herbicides is to kill all unwanted vegetation with as little damage as possible to the crop. Even a small infestation of weeds can reduce crop yield by competing for resources, allelopathy, and/or by harboring pests and diseases. The farmer often also wants to kill or reduce most vegetation within a meter or two of the crop, to prevent the spread of weed seeds and/or other propagules to crop land. Thus, for the farmer in most parts of the world, non-target plants are usually those more than a couple of meters from the field. Drift of herbicides to non-target plants has been a problem since synthetic herbicides were introduced. As mentioned above, drift to non-transgenic crops of the same species is a new problem with HRCs, although effects of herbicides on non-target crops of a different species are not a new problem. There are several studies on such effects with glyphosate.

Ellis and Griffin (2002) evaluated the response of non-transgenic soybean and cotton to simulated drift of glyphosate. Soybean and cotton injury and height reductions occurred in most cases. Soybean height was reduced by no more than 11%, regardless of herbicide rate or timing. Injury by 12.5% of the recommended rate was minimal when applied at first flower and was up to 35% injury at 14 d after application when applied at the 2- to 3-trifoliolate stage (Table 2). However, by 28 d after application, glyphosate did not cause injury, regardless of application rate and timing, except for the highest rate at the early application time, and, even then, the injury was only 8%. A similar study was done with rice (Oryza sativa L) and maize with similar results (Ellis et al., 2003). Injury to both crops was observed, particularly when glyphosate was applied early at the highest rate (12.5% of the recommended rate for weed management).

Temperature and water stress can affect the injury due to glyphosate application to plants. Higher temperatures, light intensity, and water stress can decrease the resistance of GR soybean to glyphosate (Pline et al., 1999; King et al., 2001). Glyphosate translocates from source to sink tissue, such as reproductive tissues. In GR cotton, accumulation of glyphosate in reproductive tissues and insufficient expression of the CP4 EPSPS can affect the boll retention under some conditions (Pline et al., 2002; Pline–Srnic, 2005). Impact on yield and fruit set varies greatly with the environment and timing of glyphosate applications. In general, applications made near the time of pollen development in GR crops cause more reproductive damage than early applications (Pline–Srnic, 2005). These effects and others are summarized by Pline–Srnic (2005).

Glyphosate-resistant turf grass may present another method of affecting non-target plants. Grass clippings from GR creeping bentgrass that had been sprayed with glyphosate for weed control possessed enough residual glyphosate for 3 d after treatment to cause injury to other species when used as mulch (Goss et al., 2004). The indirect effect of glyphosate on plants through their influence on plant pathogens is discussed later in this review. Potential subtle effects of sublethal concentrations of glyphosate on non-target vegetation through this mechanism have not been studied in the field.

At subtoxic concentrations, glyphosate can be a growth stimulant (Schabenberger et al., 1999; Wagner et al., 2003; Belz et al., 2006). Thus, at low concentrations glyphosate drift could stimulate growth of non-target plant species. This type of effect (hormesis) has been a controversial aspect of environmental toxicology (Calabrese and Baldwin, 2003). Sugarcane farmers in several parts of the world use glyphosate at low doses as a ripener to increase sugar content (McDonald et al., 2001).

Plant Pathogens
The effect of pesticides on plant pathogens that affect crops has been an inadequately studied and controversial topic (Altman, 1993). Glyphosate inhibits the biosynthesis of the aromatic amino acids affecting biosynthesis of proteins, auxins, phenolic phytoalexins, folic acid, lignins, flavonoids, plastoquinone and hundreds of other phenolic and alkaloid compounds and might be expected to increase the susceptibility of sensitive plants to pathogens or other stresses (Pline–Srnic, 2005).

Glyphosate is toxic to many microorganisms, including plant pathogens (Toubia–Rahme et al., 1995; Wyss and Muller–Scharer, 2001) and even some animal pathogens, such as apicomplexan parasites (e.g., Plasmodium spp.) containing the apicoplast (Roberts et al., 1998). Not all fungi are susceptible to glyphosate. For example, Morjan et al. (2002) found that glyphosate alone was not fungicidal to the entomopathogenic fungi Beauveria bassiana (Balsamo) Vuillemin, Metarhizium anisopliae Metschnikoff, Nomuraea rileyi (F) Samson, and Neozygites floridana F. However, when formulated, N. floridana and M. anisopliae were susceptible to all glyphosate formulations. The four fungi were susceptible to various glyphosate formulations when exposed to field concentrations. In a laboratory study, growth of Pythium ultimum Trow and F. solani could be stimulated or inhibited, depending on glyphosate concentration (Kawate et al., 1992).

The influence of glyphosate on plant diseases in GR crops is variable, sometimes reducing and other times increasing disease. In GR cotton, glyphosate application at the four-leaf stage reduced disease on hypocotyls in field studies (Pankey et al., 2005). Glyphosate was recently reported to inhibit rust diseases in both glyphosate-resistant soybean and wheat (Anderson and Kolmer, 2005; Feng et al., 2005). Glyphosate inhibits the growth of the plant pathogen that causes red crown rot [Calonectria crotalariae (Loos) Bell] on non-GR soybean (Berner et al., 1991). Field trials showed a reduction in red crown rot incidence with preplant applications of low rates of glyphosate.

Kremer et al. (2001) compared the effects of glyphosate, a conventional herbicide mix of pendimethalin [N-(1-ethylpropyl)-3,4-dimethyl-2,6-dinitrobenzenamine] plus imazaquin {2-[4,5-dihydro-4-methyl-4-(1-methylethyl)-5-oxo-1H-imidazol-2-yl]-3-quinolinecarboxylic acid)}, and glyphosate plus the conventional herbicide mix on soil microbes in four GR soybean varieties at eight sites. Frequency of Fusarium spp. on roots increased 0.5–5X at 2 or 4 wk after application of glyphosate or glyphosate plus conventional herbicides compared with the conventional herbicide alone. This effect was due to both higher root exudate levels of GR than non-GR soybeans, as well as to glyphosate stimulation of root exudation of microbial substrates (Kremer et al., 2005). In another study, Fusarium spp. populations increased after glyphosate treatment of weeds in the field, but crops subsequently grown in these fields were not affected by Fusarium spp. (Levesque et al., 1987).

Dead or dying weeds can provide a good microenvironment for plant pathogens. P. ultimum and F. solani populations increased in soils containing glyphosate-treated weeds (Kawate et al., 1997). Smiley et al. (1992) found that the incidence of Rhizoctonia root rot was more severe and yields lower when intervals between glyphosate treatment and crop planting were short, which they attributed to greater availability of nutrients from dying weeds for pathogen populations.

Glyphosate can also affect how a plant responds to a pathogen. In non-GR plants, glyphosate can make both crops and weeds more susceptible to plant pathogens (Johal and Rahe, 1988; Liu et al., 1997; Sharon et al., 1998), largely or at least partly by inhibiting the production of defense-related compounds derived from the shikimate pathway, such as phytoalexins and lignin. Low doses of glyphosate can sometimes make pathogen-resistant cultivars susceptible to plant disease (Brammall and Higgins, 1988). Colonization of root tissues in tomato seedlings genetically resistant to Fusarium oxysporum F. occurred following exposure to a sublethal concentration of glyphosate (Brammall and Higgins, 1988). Glyphosate was even patented as a synergist for a plant pathogen that controls weeds (Christy et al., 1993).

Theoretically, reduced resistance to plant pathogens caused by glyphosate through these processes should not occur in GRCs. Nevertheless, there have been reports of increased susceptibility of GRCs to plant pathogens. Farmers in Michigan have reported increased susceptibility of GR soybean to S. sclerotiorum (Lee et al., 2003). Glyphosate, its formulation components, and the glyphosate resistance transgene were not implicated in the increased susceptibility (Lee et al., 2000). In this case, there was no effect of glyphosate or the shading from the narrower rows that farmers use with this crop on the plant's defense. In a wider study, Harikrishnan and Yang (2002) concluded that GR and susceptible soybeans reacted similarly to most herbicide treatments with respect to root rot and damping off diseases cause by Rhizoctonia solani Kuehn. Similarly, the response of GR soybeans to F. solani-caused sudden death syndrome (SDS) was not different than that of conventional soybeans in which disease symptoms were increased by application of glyphosate (Sanogo et al., 2000, 2001). Njiiti et al. (2003) had similar results with F. solani-caused SDS in soybeans as influenced by glyphosate and the glyphosate resistance trait. Nelson et al. (2002) had mixed results with different GR soybean cultivars and application of different herbicides to these cultivars with respect to susceptibility to S. sclerotiorum-caused stem rot. Thifensulfuron {3-[[[[(4-methoxy-6-methyl-1,3,5-triazin-2-yl)amino]carbonyl]amino]sulfonyl]-2-thiophenecarboxylic acid} treatment resulted in lower disease severity in isogenic glyphosate-susceptible cultivars than with GR cultivars. Sulfonylurea herbicides, such as thifensulfuron, have been reported to stimulate production of products of the shikimate pathway (Suttle et al., 1983), from which some phytoalexins are derived.

When evaluating pest management implications of glyphosate resistance in wheat, Lyon et al. (2002) considered that a lack of an equally effective and affordable herbicide to control GR volunteer wheat could increase wheat diseases such as wheat streak mosaic virus and Rhizoctonia root rot. Field observations in Ohio suggested a possible interaction between soybean cyst nematode (Heterodera glycines Ichinohe SCN) and glyphosate in a transgenic GR variety that also expresses SCN resistance derived from the PI88788 soybean line (Yang et al., 2002). To investigate this possible interaction under controlled conditions, greenhouse experiments were conducted. Inoculation with SCN reduced shoot fresh weight of GR soybean 8 to 29% across all experiments, but there was no interaction of glyphosate and SCN in GR soybean.

One unexpected aspect of the interactions between GRCs and pathogens is that if the expression of the herbicide resistance transgene is promoted by cauliflower mosaic virus 35S promoter, the expression of the resistance gene can be reduced if the crop is infected with cauliflower mosaic virus, leaving the crop susceptible to the herbicide (Al-Kaff et al., 2000). This has not yet been reported in the field. In summary, there are cases in which the herbicide or the GRC itself may influence plant pathogens either negatively or positively.

Arthropods
Glyphosate has not been reported to have insecticidal or other activities against terrestrial arthropods. However, any herbicide can indirectly affect arthropod populations and species compositions in an area by its effects on vegetation. Furthermore, changes in cropping systems (e.g., changing from tillage to no-tillage) can drastically influence arthropod populations. Virtually all studies show no significant direct effects of glyphosate on arthropods. For example, Haughton et al. (2001), in a study of the effects of glyphosate on spiders, stated that "their results support other limited data which suggest that glyphosate is harmless to non-target anthropods." Gomez and Sagardoy (1985) found no effects of glyphosate on microarthropods in soil at double the recommended application rates.

An indirect effect of the herbicide through effects on weed species compositions and densities is more likely. For example, Jackson and Pitre (2004) found that populations of adult Cerotoma trifurcate Forster, adult Spissistilus festinus Say, larvae of Plathypena scabra F., and the caterpillar of Anticarsia gemmatalis Huebner were unaffected by GR soybeans or by glyphosate at recommended or delayed doses. But, adult Geocoris punctipes Say populations were decreased by the herbicide. The authors concluded that this effect was due to reduced weed densities after glyphosate treatment.

No effects of glyphosate resistance transgenes in plant material have been found on insects. Host plant suitability to green cloverworm (Hypena scabra F.) was evaluated on two conventional soybean varieties and two GR varieties, with and without exposure to glyphosate (Morjan and Pedigo, 2002). Treatments did not affect developmental time and survivorship. No sex bias or morphological effect was detected among treatments. Soybean genetic differences (between conventional varieties and analogous transgenic varieties) or plant stress (induced by glyphosate) did not affect the plant suitability to H. scabra. Huang et al. (2004) found that pollen from GR canola had no adverse effects on honeybees.

Weed management systems that allowed more weeds generally had higher insect population densities (Buckelew et al., 2000). However, some species did not fit this generalization, as systems with fewer weeds appeared to be preferred by potato leafhoppers (Empoasca fabae Harris). Bean leaf beetles (Cerotoma trifurcate Forster) and potato leafhoppers preferred certain soybean varieties, but these effects were attributed to soybean plant height. Their findings (Table 3) indicate that although the GR soybean varieties did not strongly affect insect populations, weed management systems can affect insect populations in soybean.

Birds and Wildlife
The environmental effect of the use of GR soybeans was compared to the effects with non-trangenic soybeans for over 1400 midwestern U.S. farms by Nelson and Bullock (2003). They used the LD₅₀ doses of glyphosate and other herbicides for rats in making their environmental effect estimates. Unlike most previous studies, this one considered the relative toxicity of herbicides available to farmers. The simulation model results suggested that glyphosate resistance soybean technology is more environmentally friendly, especially with regard to mammalian toxicity, than other herbicide technologies for all farms in the midwestern USA. The effect was generally more pronounced in the southern part of the Midwest, where a longer growing season makes overall weed pressure more serious, resulting in more herbicide use.

Peterson and Hulting (2004) compared the ecological risks of glyphosate used in GR wheat with those associated with 16 other herbicides used in spring wheat in the northern U.S. Great Plains. A Tier 1 quantitative risk assessment method was used. They evaluated acute dietary risk to birds and wild mammals and acute risk to aquatic vertebrates, aquatic invertebrates, and aquatic plants, and effects on seedling emergence and vegetative vigor to non-target terrestrial plants. They also estimated groundwater exposure to the herbicides. They found that the ecological risks for the 15 herbicides relative to glyphosate were highly variable, with glyphosate having less relative risk to non-target terrestrial and aquatic plant life and groundwater than than most other active ingredients. The study predicted that glyphosate use in GRCs will be less toxic to terrestrial and aquatic wildlife than several of the herbicides which they replace.

GRCs can affect birds and wildlife indirectly by altering habitat and food sources through effectively reducing weed biomass and/or changing weed species composition within the agricultural field. We mentioned above the studies by Dewar et al. (2003) and Perry et al. (2004) that have shown or predicted indirect effects of GRCs on wildlife through effects on habitat. Glyphosate can be applied to GRCs later in the growing season because it is generally effective against most weed species at later growth stages. If desired,