Benefits and risks of the use of herbicide-resistant crops - Kathrine Hauge Madsen & Jens Carl Streibig

INTRODUCTION

The discovery of herbicide-resistant weeds in the early 1970s triggered an interest in mimicking this unintentional development for use in crop breeding. The concomitant progress in molecular genetics made it possible to incorporate resistance genes from unrelated organisms into an otherwise susceptible crop. In other words, we were now able to adapt the biology of the crop to the chemistry of a herbicide, whereas we previously had to adapt chemistry to biology. It must, however, be noted that herbicide-resistant crops (HRCs) were first produced by methods of traditional breeding, whereas the major current HRCs have been produced by genetic engineering, the technology which has unintentionally placed these crops in a fierce debate between those in favour, and those against, the introduction and commercial use of genetically modified (GM) crops.

HRCs have been grown commercially since 1984, when the first triazine-resistant oilseed rape cultivar (OAC Triton) was introduced on the Canadian market. This cultivar was developed by methods of traditional breeding. Triazine resistance from Brassica rapa L. had been backcrossed into a commercial variety of oilseed rape (Hall et al. 1996).

On a world scale GM-HRCs constituted 85 percent (including stacked Bt and HR genes) of the total area of 52.5 million ha grown with GM crops in 2001 (James, 2001). GM-HRCs are often referred to as ‘first generation crops’ and questions have been raised as to their usefulness and putative risks to the environment and to consumers.

ASSESSMENT OF BENEFITS

Regarding GM organisms, most regulatory systems only touch indirectly on the issue of whether or not the GM products are useful. According to European regulation (2001/18/EC) it is implicitly assumed that if a product is commercially viable it is useful. However, a second definition to usefulness could be that the product must fulfil important community needs (Madsen et al. 2002a). Herbicide resistance has primarily been developed to benefit farm-management, whereas benefits to consumers are less obvious.

The general advantages seem to be connected with the fact that HRCs enable farmers to employ a flexible and easy management strategy. And, for some HRCs, we can replace herbicides with a less favourable environmental profile. Furthermore, in glyphosate-tolerant soybean, for example, costs in weed control programmes have decreased in both conventional and HRCs because of the reduced prices of herbicides.

Glyphosate-resistant soybean has been adopted principally because it simplifies weed control to the use of a single herbicide and with a more flexible timing than that required for conventional herbicides. Because glyphosate is strongly adsorbed to the soil there is a negligible threat of residual effects on succeeding rotational crops. The number of herbicide applications in soybeans is estimated to have dropped by 12 percent for the period 1995-1999. However, when this is measured in terms of the total amount of active ingredients used, there seems to be an increase. Increasing herbicide use in soybean in the United States may partly be explained by the increased area sown with this crop (Carpenter and Giannessi, 2001). It is, however, difficult to isolate the effects of adoption of GM crops from other factors which may affect pesticide use (Heimlich et al. 2000). The American Soybean Association states that glyphosate-resistant soybean protects the environment through changes in tillage practices and herbicide application, and by improved weed control. Additionally, farmers are producing cleaner crops containing fewer non-grain materials (Anderson, 2001).

Another example is HR-rice varieties, which are becoming commercially available. From an agronomic viewpoint, two main reasons are frequently put forward to justify the development and introduction of herbicide-resistant rice. The first reason is to improve control of the weed flora associated with this crop, especially of red rice and other weedy rice species (Olofsdotter et al. 2000; Gealy and Dilday, 1997). The second reason is to provide an alternative tool for the management of weeds that have already evolved resistance to particular herbicides, especially grasses such as Echinochloa spp. (Olofsdotter et al. 2000; Wilcut et al. 1996). HR-rice, furthermore, allows for the substitution of some of the currently used herbicides by others less detrimental to the environment (Olofsdotter et al. 2000).

In many parts of the world, soil erosion resulting from tillage practices is a problem. In general, HRCs may be favourable to the environment by allowing for flexible weed management compared to conventional systems. This may permit farmers to practice conservation tillage, e.g. no-till or reduced tillage, and thereby reduce soil erosion (Duke, 2001).

RISKS

For HRCs, risks can be considered qualitative estimates which combine the likelihood and severity of both immediate and delayed adverse effects to human health, the environment and the farmer’s economy. The likelihood and severity of each unwanted effect associated with HRCs depends on the crop, the HR trait, the local weed flora, climatic conditions and farm management practices and can only be estimated on a case-by-case basis (Madsen et l. 2002b).

In glyphosate-resistant crops, optimal weed control often requires sequential applications with glyphosate, and the timing relative to weed emergence is important (Swanton et al. 2000). When glyphosate is sprayed 2-3 times annually at high rates it imposes a high selection pressure on the weed flora. In 5-8 years this may cause shifts in weed composition towards species that naturally tolerate glyphosate (Benbrook, 2001; Shaner, 2000) and other herbicides may be needed to control these weeds (Shaner, 2). Payne and Oliver (2000) suggest implementing conventional, post-emergence herbicides in the weed control programme in glyphosate-resistant soybean to assist the control of glyphosate tolerant species, such as Sesbania exaltata (Raf.) cory, Ipomoea spp. or Amaranthus rudis Sauer. Furthermore, it may become difficult to control volunteer crops in subsequent years. If farmers grow glyphosate-resistant varieties of both corn and soybean in a soybean-corn rotation, then glyphosate cannot control the volunteer corn, which can be a serious weed problem in soybean (Shaner, 2000).

Gene-flow from crops to other crops or related species is another route to the development of resistant weed populations in the field. Once the resistance gene is present in crop volunteers or related weed species then it is expected that the same weed control practices (consistent sprayings with herbicides having the same mode of action), which cause herbicide resistance to occur in naturally tolerant/resistant weed biotypes, will lead to a rapid build-up of HR-weeds and volunteers.

Increased herbicide use is considered a risk in some parts of the world although the effects on human health or the environment are seldom specified in details, but derived effects from pesticide-use such as ground-water pollution and pesticide residues in vegetables, for example, have caused public concern. There seem to be two major explanations why herbicide use in HRCs may increase. One reason is that a high level of crop tolerance may enable the farmer to increase doses to achieve an improved weed control without harming the crop (see Figure 2). The other reason is problems with tolerant/resistant weeds and volunteers, which require farmers to increase dose or mix herbicides with different modes of action to maintain an acceptable level of weed control.

Biodiversity within the field may be influenced if the herbicide, to which the HRC is resistant, is used at a higher level of efficacy than hitherto in order to achieve an improved weed control. Furthermore, weed species respond differently to different herbicides or other weed-control measures and a shift in prevailing species is very likely. If the growing of an HRC is taking place at the centre of genetic origin, then changes in the diversity of the indigenous species and risks of diminishing the genetic diversity of these species is a hazard (FAO, 2001). It is, however, very unlikely that HR crops will cause erosion of genetic diversity of wild species outside the cultivated land, because the trait is associated with the use of herbicides which are not being applied in the wild, and a HR trait does not confer selective advantage unless the herbicide is used (Poulsen, 1995; Madsen et al. 1998). Therefore, there is a low risk of erosion of the genetic diversity of wild species growing in natural environments.

Despite these concerns, some uses of GM crops, e.g. herbicide resistant sugar beet, appear to be safe so far as ecological risks are concerned, when these are judged by ordinary scientific standards (Madsen and Sandøe, 2001).

Finally, it must be emphasized that risk assessment is only one of the steps leading to the final approval of an HRC, which implies a political decision. First, an ‘acceptable level’ of risk is not an objective term and second, risk assessment of HRCs is associated with a high degree of uncertainty with regard to the magnitude and likelihood by which potential hazards associated with the HRC may occur (Madsen et al. 2002b).

CORE ISSUES TO BE ADDRESSED WHEN ASSESSING RISKS FROM HRCS

The first step is to determine which unwanted effects are relevant in the particular scenario. In an earlier publication we proposed the use of decision-keys for identification of hazards/unwanted effects. These keys were developed to assess the likelihood that a new type of arable weed will be produced by gene flow between the transgenic crop and its relatives; the likelihood that the transgenic crop will become a volunteer problem on arable land or wild areas and the likelihood of a build-up of HR-resistant weeds (FAO, 2001).

GENE FLOW

Transfer of genes from one population to another may lead to unwanted effects for weed management and the environment. Gene flow may enable the resistance genes to move between HR and non-HR varieties and thus pollute a crop which is considered GM-free. Or HR-genes may be stacked from years of cross-pollination of HRCs, which may result in problems for the farmer in controlling volunteer crops in the field. Multiple herbicide-resistant volunteer oilseed rape has been observed in Canada where oilseed rape with resistance to different herbicides was grown on neighbouring fields (Hall et al. 2000). Gene flow between related species e.g. the crop and certain weeds in the field may, furthermore, result in the development of HR weeds.

If gene flow is a relevant process to consider for the particular HRC then the next step is often to quantify the level of gene flow within and between species in time and space. Over the past 10- 15 years there has been a range of studies on this issue, many of them focusing on hybridisation within and between crops and wild relatives (see Table 1 at the end of the chapter, from Madsen and Jensen, 1998). Other studies have aimed at determining the distance of gene flow (Figure 1), and some have evaluated the ability of the hybrids to survive and reproduce seed over a number of generations; e.g. Hauser et al. (1998) showed that biomass and seed production appear to be reduced in second generation hybrids (F₂ or BC₁) between oilseed rape and B. rapa L.

Figure 1 Pollen dispersal from Beta vulgaris var. conditiva to B. maritima - crossing frequency (percent) as function of distance from B. vulgaris var. conditiva. E=east, ENE=east-northeast, ESE=east-southeast (Source: Madsen, 1994).

COMPETITIVE ABILITY

Competition experiments have shown that herbicide-resistant biotypes of both crops and weeds may have similar competitive abilities as the non-resistant biotypes when they are not sprayed with the herbicide to which they are resistant (Jensen, 1993; Poulsen, 1995; Fredshavn et al. 1995; Madsen et al. 1998). It is, therefore, unlikely that fitness is increased by the mere presence of a HR-gene. Reduced fitness has, however, been observed, e.g. the triazine-resistant oilseed rape variety OAC Triton yielded significantly less than non-resistant varieties. When the HR weed biotype is sprayed with the herbicide then not only is it undamaged by the spray, it is, furthermore, released from competition with all the non-resistant weeds and volunteers, which perish. This is to the advantage of the HR biotype, which rapidly builds up vegetative biomass in the field and produces large numbers of seed or propagules that enter into the soil seed bank. There are several experimental designs available to assess competitive ability over one cropping season. However, in experimentation it is complicated to assess the long-term effects on population dynamics from an altered competitive ability.

DOSE-RESPONSE CURVES

The dose-response relationship for a particular combination of herbicide and plant species under specific growth stage and climatic conditions may be described as shown in Figure 2. The dose corresponding to a predetermined efficacy level of 90 percent control level (ED₉₀) can be used as a measure of weed sensitivity and the dose resulting in a 10 percent temporary yield depression (ED₁₀) at time of recording may be used to rank the crops according to tolerance to the herbicide. Note that the doses are on a logarithmic scale. Dose-response curves may be used to determine the crop tolerance vs the control level of the weeds. This procedure is often used in the development of new herbicides or HRCs. Figure 2.A shows an intrinsic selective herbicide, the index of selectivity, ED₉₀/ED₁₀, is greater that 1.0, preferably it should be around 3.0; Figure 2.B illustrates a non-selective herbicide where the ED₁₀ is far to the left of the desired control level for the weed; and Figure 2.C shows what happens when the crop in Figure 2.B is made tolerant to an otherwise non-selective herbicide, The ED₁₀ for the crop is far displaced to the right of the ED₉₀ for the weed.

Another application of dose-response curves is to rank the weed species after ED₉₀. This ranking may indicate which species are likely to become dominant and which may perish when growing a newly released HRC and spraying with its associated herbicide. We may thus get a first crude indication of shifts in the weed flora. However, uncritically extrapolating from dose-response relationships derived from experiments with monoculture of weeds without appreciating the growth habit of the weeds relative to the crop may, under some conditions, be misleading (Madsen et al. 2000). Therefore, experiments should be based upon realistic mixtures of crop and weeds preferably in the field, because crop coverage may affect deposition of droplets on the target weeds and climatic conditions may have a strong influence on uptake and translocation, depending on the herbicide and plant species.

SIMULATION MODELS

Many of the concerns over HRC can best be addressed experimentally by multiyear experiments, but these experiments are costly. Furthermore these crops are already being grown on large areas, which mean that management strategies to prevent/delay problems must be developed soon to be effective. Combining simulations with selected mid- and long-term experiments may provide a better understanding of the suitability of available weed management strategies that may prevent or delay the selection of HR-weeds. Furthermore, development of a model system would reveal where relevant information is lacking or scarce. Simulation may thus be able to anticipate problems with injudicious herbicide use such as development of resistant weeds and volunteers, and/or major shifts in the weed flora. Of course the pre-requisites for the model must be clearly stated to arrive at a ‘sound judgement’ of the predictions.

Madsen et al. (2002) developed a simulation model of growing HR varieties of rice in a rainfed Central American production system to investigate the following question: Is there a risk of increased weed problems, which is derived from gene flow and rice volunteers? Simulation with glufosinate resistant rice enabled a prediction of potential long-term effects and allowed for testing of different scenarios including contrasting weed management practices, hybridization levels between the commercial HR cultivated and weedy rice, and seed predation rates. Because risks may only become conspicuous after long-term cultivation of HR rice, simulations were run for a 10-year period. In a cropping system relying on glufosinate-resistant rice for weed control, the model predicted that resistance to glufosinate would occur after 3-8 years of monoculture. Increasing the hybridisation level from 1-5 percent decreased the time for resistance to occur by 1-3 years. Increasing annual rate of weedy rice seed predation at the soil surface delayed development of resistance. Tillage as a weed control tactic also delayed the occurrence of resistance compared to a no-till situation. It must, however, be emphasized that the model was a first attempt to simulate a production system with HR rice and many of its parameters were highly uncertain. Furthermore, the presented model had not been validated with field data, which is a prerequisite to make reliable predictions about long-term consequences of growing HR rice.

Figure 2. Dose-response curves:

A. Inherent selective herbicide that controls the weed without harming the crop (high index of selectivity ED₉₀/ED₁₀>>1.0).

B. Non-selective herbicide, both weed and crop are susceptible.

C. The crop in B has been made tolerant to the herbicide giving a very high index of selectivity.

Figure 3. Component of a model simulating a cropping system with HR rice, weedy rice and Echinochloa colona (L.) Link (Madsen et al. 2002b)

SOCIO-ECONOMIC RISKS

If we apply a broader interpretation of risks, a familiar argument draws attention to the dominant market position enjoyed by multinational agrochemical corporations. These companies have purchased several plant-breeding companies in order to link seed production to agrochemicals. Three of the five top agrochemical companies were also among the five top seed companies in 1997 (EU, 2000), and this monopoly may lead to the loss of both traditional crop varieties and long-standing farm management methods. If this were to occur, then despite the obvious benefits of new GM crops, the food supply could become vulnerable to slight changes in the behaviour (e.g. the resistance) of pests such as pathogens, insects and weeds (Madsen et al. 2002a). The HR-genes may, however, be transferred to different varieties; e.g. in 2000, 1100 soybean varieties in 3000 were glyphosate-resistant (Anderson, 2001).

When adopting GM crops, farmers also meet a number of constraints, namely: seed-saving is illegal, the cost of seed or of fees for technology agreement is increased, and fewer suppliers provide the input for crop production (EU, 2000). Furthermore, if the HR crop readily cross-pollinates with adjacent non-GM crops, then it raises questions both about property rights of these unintentionally resistant plants and threshold levels of GM seed in harvests from non-GM crops. Furthermore, public concern over GM food may lead to demands for the segregation of GM crops and non-GM crops, and the introduction of identity-preservation and trace ability in connection with exports and/or domestic consumption. Such measures increase costs. They may therefore result in different prices for GM crops and non-GM crops, and this differentiation may lead to crops being grown under contract. Identity-preservation has been estimated to represent a cost of 6-17 percent of farm-gate price (EU, 2000).

CONCLUSIONS

HRCs have a great potential in the simplification of weed management. Handled judiciously, these crops may be beneficial to the environment by enabling no-till systems, thus reducing erosion or allowing for later weed control, which may increase biodiversity in the field. However, it must be emphasized that the risk from HRCs should be carefully evaluated prior to releasing the HRC into a cropping system, especially when the HRCs possess weedy characters or may outcross to related weeds. If this is the case, and the HRC is grown commercially, then precautions need to be taken, similar to the management strategies adopted to prevent the development of naturally-resistant weeds. Furthermore, precautions must, in particular, be taken before release into the genetic origin of the species.

BIBLIOGRAPHY

Anderson, T. 2001. Biotech soybean seed helps growers produce safe and profitable crops. American Soybean Association. (available at http://www.monsanto.co.uk/news/ukshowlib.phtml?uid=5063).

Benbrook C.M. 2001. Trouble times amid commercial success for Roundup Ready soybeans. AgBioTech InfoNet technical paper number 4. May 3, 2001, 6 pp.

Carpenter, J.E. & Gianessi, L.P. 2001. Agricultural biotechnology: Updated benefit estimates. Report from the National Center for Food and Agricultural Policy. Washington DC, 46 pp.

Duke, S.O. 2001. Herbicide-resistant crops. In Pimentel, D., ed. Encyclopedia of Pest Management. Marcel Dekker, Inc., New York (in press).

EU. 2000. Economic impacts of genetically modified crops on the agri-food sector - A synthesis. Working document Directorate-General for Agriculture (available at http://europa.eu.int/comm/agriculture/res/index_en.htm). 43 pp.

FAO. 2001. Draft of guidelines for assessment of ecological hazards of herbicide- and insect-resistant crops. Plant Protection Division, Rome. (In collaboration with Kathrine H. Madsen, Valverde, Bernal E. & Streibig, Jens C., of the Royal Veterinary and Agricultural University, Denmark). 18 pp.

Fredshavn, J.R., Poulsen, G.S., Huybrechts, I. & Rudelsheim, P. 1995. Competitiveness of transgenic oilseed rape. Transgenic Res. 4: 142-8.

Gealy, D.R & Dilday, R.H. 1997. Biology of red rice (Oryza sativa L.) accessions and their susceptibility to glufosinate and other herbicides. Weed Sci. Soc. Am. Abstr. 37: 34.

Hall, J.C., Donnelly-Vanderloo, M.J. & Hume, D.J. 1996. Triazine-resistant crops: The agronomic impact and physiological consequences of chloroplast mutation. In Duke, S.O. ed., Herbicide-resistant crops. Agricultural, Environmental, Economic, Regulatory and technical aspects. USA, CRC Press. pp. 107-126.

Hall, L., Topinak, K., Huffman, J. & Davis, L. (2000) Pollen flow between herbicide-resistant Brassica napus is the cause of multiple-resistant B. napus volunteers. Weed Sci. 48, 688-694.

Hauser, T.P., Jørgensen, R.B. & Østergård, H. 1998. Fitness of backcross and F₂ hybrids between weedy Brassica rapa and oilseed rape (B. napus). Heredity 81: 436-443.

Heimlich, R.E., Fernandez-Cornejo, J.F., McBride, W., Klotz-Ingram, C., Jans, S. & Brooks, N. 2000. Adoption of genetically engineered seed in U.S. agriculture. In Proc. of 6^th Intl. Symposium on the Biosafety of GMOs. eds. Fairbairn, C., Scoles, G. &
McHughen, A. Saskatoon, Canada. pp. 56-63.

James, C. 2001. Global GM Crop Area continues to grow and exceeds 50 million hectares for first time in 2001. Intl. Service for the Acquisition of Agri-biotech Applications. http://www.isaaa.org/press percent20release/Global percent20Area_Jan2002.htm.

Jensen, J.E. 1993. Fitness of herbicide-resistant weed biotypes described by competition models. In Proc. of the 8^th EWRS Symposium Quantitative approaches in weed and herbicide Res. and their practical application, Braunschweig. 25-32.

Madsen, K.H. 1994. Weed management and impact on ecology of growing glyphosate tolerant sugarbeets (Beta vulgaris L). Royal Veterinary and Agricultural University, Denmark. (Ph.D. thesis)

Madsen, K.H., Poulsen, G.S., Fredshavn, J.R., Jensen, J.E., Steen, P. & Streibig, J.C. 1998. A method to study competitive ability of hybrids between seabeet (Beta vulgaris ssp. maritima) and transgenic glyphosate tolerant sugarbeet (Beta vulgaris ssp. vulgaris). Acta Agriculturæ Scandinavica, Section B, Soil and Plant Science 48: 170-74.

Madsen, K.H. & Jensen, J.E. 1998. Meeting and training on risk analysis of HRCs and exotic plants. Course material for the UN Food and Agricultural Organization (FAO) in Piracicaba, Brazil 19-22 May 1998.

Madsen, K.H. & Streibig, J.C. 2000. Simulating weed management in glyphosate-tolerant crops: Greenhouse and field studies. Pesticide Management Science 56: 340-344.

Madsen, K.H. & Sandøe, P. 2001. Herbicide resistant sugar beets - What is the problem? J. of Agricultural and Environmental Ethics 14 (2): 161-168.

Madsen, K.H., Sandøe, P. & Lassen, J. 2002a. Genetically modified crops: A US farmer’s versus an EU citizen’s point of view. Acta Agriculturae Scandinavica (in press).

Madsen, K.H., Valverde, B.E. & Jensen, J.E. 2002b. Risk assessment of herbicide resistant crops: A Latin American perspective using rice (Oryza sativa) as a model. Weed Tech. 16 (1), 215-223.

Olofdotter, M., Valverde, B.E. & Madsen, K.H. 2000. Herbicide resistant rice (Oryza sativa L.) in a global perspective: Implications for weed management. Annals of Applied Biology 137, 279-295.

Payne, S.A. & Oliver, L.R. 2000. Weed control programs in drilled glyphosate-resistant soybean. Weed Tech. 14: 413-422.

Poulsen, G.S. 1995. Weediness of transgenic oilseed rape - Evaluation methods. The Royal Veterinary and Agricultural University, Dept. of Agric. Sciences (Weed Science), Denmark. (Ph.D. thesis)

Shaner, D.L. 2000. The impact of glyphosate-tolerant crops on the use of other herbicides and on resistance management. Pest Management Science 56: 320-326.

Swanton, C.J., Shrestha, A., Chandler, K. & Deen, W. 2000. An economic assessment of weed control strategies in no-till glyphosate-resistant soybean (Glycine max). Weed Tech. 14: 755-763.

Wilcut, J.W., Coble, H.D., York, A.C & Monks, D.W. 1996. The niche for herbicide-resistant crops in U.S. agriculture. In Duke, S.O., ed., Herbicide-resistant crops, agricultural, environmental, economic, regulatory, and technical aspects. CRC Press Inc., Boca Raton, Florida, USA. pp. 213-230.

Table 1. Out-crossing and gene flow within and between different crop plants and relatives. Out-crossing frequency is between closely situated plants. Isolation distance used in plant breeding. Exp. det. distance (experimentally determined distance) is the maximum distance at which gene flow was found. Gene flow is measured as potential (P) or actual (A) gene flow. F₂/BC (second generation/back cross) is whether or not the hybrids can produce seed (Source: Madsen & Jensen 1998, after data from many literature sources).

+ indicates that a F₂/BC generation can be produced, but that the experiment did not test this.