The primary goal of this webpage is to provide regulators with science-based information to facilitate their decision making on whether a given transgenic crop is as safe as what is currently on the market.

Learning from their experience in regulating transgenic crops over the last decade, regulators have made considerable progress in designing logical frameworks. Risk Assessment is a key component of these frameworks and it is designed to help characterize the potential risks associated with the utilization of transgenic crops. The Australian Government, through its Office of Gene Technology Regulator (OGTR) (2007) has suggested that Risk Assessment is a process that should deal with the questions: “1) what might happen? 2) how might it happen? 3) will it be serious if it happens? 4) how likely is it to happen? 5) what is the risk?”

The African Biosafety Network of Expertise (ABNE) initiated by the New Partnership for Africa’s Development (NEPAD) endeavors to facilitate the access and use of scientific knowledge to drive the regulatory decisions that will impact on Africa. The information presented in this webpage is specifically tailored to fit into a risk assessment and management framework.

This section is designed to provide the tools necessary for characterizing environmental risks and designing measures to manage them. First, an overview is presented on the impacts of current-day agriculture on the environment, and the benefits expected and observed from the utilization of transgenic crops. Major principles and methodologies used for risk assessment are then discussed. Risk management processes are also described, and key information on crop biology is presented for species of interest for Africa.

The content of this webpage section is derived from documents of international bodies such as Organization for Economic Co-operation and Development (OECD), Food and Agriculture Organization (FAO), Codex Alimentarius, Governmental bodies such as Office of Gene Technology Regulator (OGTR), and peer-reviewed articles as well as from other relevant websites.

This section will not deal with safety issues associated with laboratory and greenhouses uses of GM materials. For that information, you may refer to other sources such as the US National Institute of Health (NIH, http://www.nih.gov/) or OECD (http://www.oecd.org).

1. Introduction

Agriculture is facing the challenge of feeding an increasing global population while natural resources are shrinking due to a combination of factors. Many feel that biotech crops can contribute to meeting global food needs by improving agricultural productivity. Yet, the potential risks associated with the cultivation of biotech crops should be accurately evaluated and managed. The base line in evaluating those risks should be a good knowledge of the impacts and foot prints of the current agricultural systems. Current practices such as tillage, water use, intercropping, crop rotation, grazing and extensive usage of pesticides affect the biodiversity of agricultural fields as well as the environment outside of fields (Tilman, 1999, 2002; Robinson and Sutherland, 2002; Butler et al. 2007, Quemada, 2009).

This subsection gives a brief overview on these impacts and discusses the potential benefits expected from the utilization of the new biotech crops.

2. Land use

By occupying 40% of the land surface, agriculture is currently a major land-use and is the main factor contributing to losses in biodiversity (FAO, 2007). Approximately 13 million hectares of biodiversity-rich forests are lost in developing countries annually (James, 2009). This situation will likely worsen in the near future as the global population must be fed and biofuel feedstocks additionally produced. The pressure to increase the area of land under cultivation will grow more and more important. Climate change also is “expected to accelerate many pressures on the wild environment, as long-established production systems become destabilized and abiotic stress (such as water shortages, salinity, aridity and heat) are increased” (FAO, 2007).

Therefore, it is critical to increase agricultural productivity per unit of land in order to reduce land conversion and biodiversity erosion. Genetic engineering technology has a great potential to contribute increasing that productivity and help reduce deforestation and loss of biodiversity in forests. James (2009) estimates that during the period 1996 to 2007 biotech crops have already precluded the need for an additional area of 43 million hectares of crop land.

Many studies across the world have reported on yield increases after the deployment of biotech crops. For example, from 1996 to 2006, average yield increases in the areas planted to biotech insect resistant traits was +5.7% for corn and +11.1% for cotton (PG Limited Economics, 2008). James (2009) reported that in 2008 Bt cotton yield increased by 31% in India and by 9.6%, in China; Bt maize resulted in an 11% higher yield in South Africa in 2005. Other data indicate a 31% average yield increases with herbicide tolerant soybeans in Romania, 15% increase with herbicide tolerant corn in the Philippines and more than 50% with insect resistant cotton in India (http://www.pgeconomics.co.uk/).

3. Insecticide use

Pests cause a loss of 40% of agricultural production worldwide, despite strategies and measures carried out to control them (Pimentel, 1998). Insects alone destroy annually about 25% of food crops worldwide. For instance, the European corn borer (the larvae of Ostrinia nubilalis) can destroy up to 20% of a maize crop (www.gmo-compass.org). In Africa, losses in agricultural production due to pests can reach 100% depending upon the agro-ecological zones (Abate et al., 2000).

Thus, huge amounts of synthetic pesticides are used every year to control agricultural insects around the world. Those chemicals not only have serious impacts on the environment but also can cause harm to human health. Every year, thousands people are poisoned by agricultural pesticides worldwide, mostly in developing countries (Brodesser et al., 2006, CGIAR, 2008).

Biotech crop cultivation has been shown to help reduce insecticide use in agriculture. Bt cotton cultivation, for instance, reduced insecticide use in India by 39% in 2008 and by 60% in China (James, 2009). A cost reduction of about 60% in insecticide use has been observed in South Africa due to Bt maize. In the West African country Burkina Faso, Bt cotton has reduced number of pesticide treatments from six or eight a year to one or two, while boosting production by 30% (Manson, 2009).

Why are Bt crops resistant to insects?

Bt crops have been transformed by adding a bacterial gene from Bacillus thuringiensis that produces Bt toxins. Those toxins affect specific groups of insects including lepidopterans and coleopterans. Bacillus thuringiensis is a bacteria naturally living in the soil. The engineered plants can produce Bt toxin on their own, and this allows them to defend themselves against specific types of insects. Consequently, farmers can use less chemical insecticides to control certain insects.

4. Herbicide use

In intensively managed agricultural systems, large amounts of the herbicides are annually used by farmers to control weeds. Many groups of herbicides are available including the chlorophenoxy acid herbicides which are selective for the angiosperm plants, the triazines herbicides used mostly to protect corn, apple, grapes, wheat; the thiocarbamates which are generally used as graminicides applied to soil before emergence of crops, to protect maize, rice, sorghum, sugar beets, soybean (Nagy et al., 1994); and the organic phosphorus herbicides including glyphosate, a non selective herbicide (http://science.jrank.org/).

Broadcast spraying of herbicides can have negative consequences on the environment through different pathways. Drift of the herbicide beyond the intended spray site, for instance, can cause offsite damage to susceptible vegetation. Herbicides also result in reduced habitats and food for non-target organisms such as birds and mammals, especially when these herbicides are applied in forestry (http://science.jrank.org/). It has been reported in the USA that the residual herbicides commonly used for corn and soybean production have been detected in rivers, streams, and reservoirs at concentrations exceeding the U.S. maximum contaminant levels or health advisory levels for drinking water (Martin et al., 2008).

Among the herbicides, glyphosate is the one most widely used both in agriculture and forestry. Its values include a low toxicity to animals, a rapid adsorption to soil particles reducing movement in the environment and a low persistence due to a rapid degradation by soil microbes (Cerdiera and Duke, 2006). It has been used commercially in the United States for over 35 years (Combs and Hartnell, 2008).

Glyphosate is a nonselective herbicide that kills annual and perennial plants including both weed and crop species (Duke et al. 2003, Brown, 2006, Combs and Hartnell, 2008).

Through genetic engineering, biotech crops have been developed that carry a gene which allows plants to tolerate glyphosate so that they are no longer killed. The use of glyphosate resistant biotech crops has the potential to reduce the use of the other more harmful herbicides and thereby reduce their negative effects on the environment, particularly in developed countries where agriculture greatly relies on herbicide use. Such a reduction has already reached 17 million of kg per year in the United States (Gianessi, 2005). Also, replacing the more harmful herbicides with glyphosate can help reduce the amounts of dissolved herbicide concentrations in runoff (Martin et al., 2008).

Genetically modified herbicide resistant crops

Transgenic glyphosate tolerant crops have been transformed with EPSPS* gene (taken from Agrobacterium strain CP4) that produces an enzyme resistant to glyphosate (Brown, 2006). Farmers growing glyphosate resistant crops can therefore more effectively control weeds during the entire growing season and have more flexibility in choosing times for spraying. Herbicide resistant crops also facilitate low or no tillage cultural practices, which many consider to be more sustainable.

* EPSPS = Enolpyruvyl-shikimate-3-phosphate synthase.

5. Tillage practices

Tillage or plowing can increase soil erosion and cause soil loss all around the world. These practices also demand fuel consumption, which contributes to the increased carbon dioxide emissions which are responsible for the greenhouse effect and global warming.

Reduced tillage has been shown to be environmentally beneficial by reducing soil erosion, increasing its moisture content and nutrient richness, and leading to favorable conditions for soil organisms and wildlife. Reduced tillage also contributes to decreases in the level of pollution by smoke and carbon dioxide release through reduced consumption of fuel (Fawcett and Towery, 2002; Dale et al., 2002).

No-till hectarage has increased rapidly in the USA and Argentina through the cultivation of transgenic herbicide tolerant soybean and cotton since the beginning in 1996 (Trigo and Cap, 2003).

6. Water consumption

Agriculture consumes approximately 70% of the world’s fresh water that is withdrawn for human use (FAO, 2007). This demand for water is expected to increase dramatically as the world population is growing and will reach 9.2 billion by 2050 (James, 2009). At the same time, global climate change is predicted to result in increased risks of water shortage and desertification. Water shortages are already costing billions of dollars a year in crop shortfalls around the world, and are likely to grow more costly.

Given that drought is the single most important constraint to increased productivity for crops worldwide, efforts are underway to develop biotech drought tolerant crops. This drought tolerance trait is viewed as the most important biotech trait that will be used in the second decade of commercialization 2006 – 2015 (James, 2009). Maize will be the first biotech drought crop to be commercialized, likely in 2012 in USA and by 2017 in Sub Saharan Africa (James, 2009). “Drought is the most important constraint of African agriculture severely affecting maize, the most important African staple food crop” (http://www.aatf-africa.org/).

1. Introduction

This subsection provides a short review of the potential environmental impacts of GM crops. Some commonly expressed environmental concerns about GM crops are: 1) gene flow between the transgenic plants and their sexually compatible relatives, 2) changes in levels of weediness or invasiveness of the GM crops or their wild relatives; 3) horizontal transfer of engineered traits to other species, 4) non-target effects and 5) development of pest resistance or new secondary pests.

2. Genetically Modified Crops and Gene Flow

Gene flow refers to the transfer and incorporation of genes from one population into another through pollen or seed movement (Andow and Zahlen, 2005). Pollen movement is not by itself an environmental concern unless the GM crop can successfully hybridize with a relative.

Hybridization occurs frequently between crops plants and their sexually compatible relatives (both crops and wild species). Ellstrand et al. (1999) gave many examples of natural inter-hybridization among the most important crops cultivated in the world. These include rice (Oryza sativa and Oryzaglaberrima), maize (Zea mays), sorghum (Sorghum bicolor), cotton (Gossypium hirsutum and Gossypium barbadense), millets (Pennisetum glaucum), sugarcane (Saccharum officinarum) and groundnut (Arachis hypogea). These are of particular importance in Africa as staple and cash crops.While hybridization between crops and their relatives is a common phenomenon, a number of factors must take place for gene flow to occur. Plant species must 1) be sexually compatible, 2) close enough for the pollen to be moved between plants by wind or vectors such as insects or birds, and 3) be in flower at the same time (Ellstrand et al. 1999).

Similar to their non-GM counterparts, GM crops will probably hybridize with their sexually compatible species if they are grown in close proximity. For example, Lu (2008) showed that hybridization between transgenic and weedy rice (Oryza f. spontanea) is unavoidable where they co-exist and that the insect resistant Bt transgenes can be incorporated and persist in the weedy rice populations.

Hybridization between GM crops and their compatible relatives may or may not have environmental consequences. The impact of transgenes on natural populations is determined by the combination of the trait expressed by the gene itself and the invasiveness of the recipient species (Hancock, 2003, Nickson, 2008).

3. Changes in Levels of Weediness or Invasiveness

Concerns have been raised that 1) transgenic crops themselves could become weeds and invade agricultural or natural ecosystems, and 2) the engineered traits could be introduced into wild relatives via hybridization and increase the competitive ability and the weediness of those wild plants or their hybrid derivatives. Thus, the risk of generating increased weediness has been a major environmental concern associated with transgene flow.

What is considered a weed differs according to eco-geographical regions and whether agronomic fields or natural plant communities are considered. Ellstrand et al. (1999) defined weeds as “wild plants that interfere with human objectives”; and wild plants as “plants that grow and reproduce without being deliberately planted.” USDA APHIS (2008) considers a noxious weeds as “any plant or plant product that can directly or indirectly injure or cause damage to crops, (–), or other interests of agriculture, irrigation, navigation, the natural resources, the public health, or the environment.”

Different weed species share some common ecological and biological attributes such as preference for disturbed habitats (cultivated fields, roadsides, fields margins, soil dumps), phenotypic plasticity allowing for adaptation to changing environments, low seed dormancy, indeterminate growth, continuous flowering and high seed production (Conner et al., 2003). However, the most critical predictors of the weediness of a given plant are (OGTR, 2003): “1) its taxonomic affinity to other known weedy species and 2) its history of weediness elsewhere in the world.”

Many of the GM crops developed so far such as cotton, maize and soybean have been selected for thousands of years by humans on agronomic sites and have lost many of their weedy characters. In fact, only a small percentage of agronomic crops are important weeds outside of agro-environments (Hancock and Hokanson, 2004; Hancock, 2003). They rely on human disturbances to become established and rarely persist outside of specific habitats. This means that the addition of the transgenes now commonly deployed is unlikely to make them noxious weeds (Conner et al., 2003). Thus, GM crops are not more likely to become weeds than their non-GM counterparts (Conner et al., 2003).

The bottom line in assessing the environmental risk of GM crops is the nature of the transgene itself, that is, how significant impact it will have on the reproductive potential of native populations should it escape (see section above on risk assessment). In fact, it is much easier to predict the environmental risk of a GM crop, than that of an exotic plant introduction, as the level of risk in GM crops can be measured by evaluating the impact of a single engineered trait than a whole syndrome of potentially invasive traits found in an exotic species. The GM crop is already in the environment that it is traditionally grown, and we know the evasiveness of the traditional crop. What we need to worry about is whether the addition of a single GM trait will increase its existing level of invasiveness to problem levels.

Concerns have been expressed about crops that contain herbicide tolerance genes turning into uncontrollable weeds, particularly if several crops of the same species and tolerant to different herbicides are grown in close proximity. Multiple hybridizations can therefore occur and may result in multi-herbicide resistant hybrids which are difficult to eradicate (McHughen, 2000). Studies in Canada have already revealed the presence of oilseed rape plants that had become tolerant to three different herbicides following hybridization with two GM herbicide tolerant varieties and a conventional one (Dale et al., 2002). It is important to realize that while the emergence of herbicide resistance in weed species could add to farm management burdens, it will not impact on the natural environment where herbicides are not applied.

Hybridization between wild species and transgenic insect or disease resistant varieties may also result in populations with increased fitness, but only if the target pests were typically controlling the natural population size of the wild relatives. For instance, because of reduced effects of LepidopteransHelianthus annus) have been shown to produce 14 – 55% more seeds (Snow et al., 2003). However, it is important to remember that an increase in fitness of individuals will not necessary lead to change in the population size of a given wild plant species. Density dependent selection can still keep population sizes the same, by increasing the mortality rates of additional individuals (Raybould and Cooper, 2005).

4. Horizontal Transfer of Engineered Traits to Other Species

Horizontal gene transfer is defined as a stable transfer of genetic material from one organism to another without reproduction or human intervention (Kesse, 2008). This phenomenon can occur between bacteria and is considered a significant source of genome variation (Conner et al., 2003). Concerns have been raised that engineered traits could be transferred to non-target organisms via horizontal transfer and thereby threaten environmental and animal safety (Droge et al., 1998, 1999).This topic has received considerable attention from numerous expert panels held under the auspices of various national and regional regulatory systems as well as international bodies such as OECD, WHO or FAO. Based on scientific evidence, they report that horizontal gene transfer from GM plants to other organisms is, at most, an extremely rare phenomena and to date no environmental harm has been reported from it (Nielsen et al., 1997; 1998, 200b; 2007, Schluter et al., 1995; Anderson, 2005, Gebhard and Smalla, 1998; Mercer et al., 1999, Lambert and al., 1999; Bird and Koltai, 2000, Simpson et al., 2007, Kesse, 2008). The European Food Safety Authority (EFSA) has confirmed that the nptII gene commonly used as a selectable marker in GM crops does not pose a risk to humans or the environment: http://www.efsa.europa.eu/EFSA/efsa_locale-1178620753812_1178620789287.htm, http://www.efsa.europa.eu/EFSA/efsa_locale-1178620753812_1178620775641.htm

5. GM Crops and Biodiversity

Biodiversity refers to “the variability among living organisms from all sources, including, among others, terrestrial, marine and other aquatic ecosystems and the ecological complexes of which they are part; this includes diversity within species, between species and of ecosystems” (Convention of Biological Diversity, 1992). Biodiversity is represented by both numbers of species and genetic variability within the species.Concerns have been raised that the wide spread of GM crops could adversely affect the levels of natural diversity through: 1) replacement of traditional varieties (land races), 2) hybridizations between GM crops and land races or wild relatives, particularly in “centers of diversity” and 3) interactions with non-target organisms. Land races and centers of diversity are of particular importance both to farmers and plant breeders as sources of food and storehouses for genetic diversity to improve crop varieties.Hybridization between GM crops and wild relatives could reduce levels of genetic variability through two routes: 1) a native population could go extinct due to high levels of pollen flow from a much larger commercial field, or 2) selectively advantageous crop genes could replace native ones through hybridization.

Extinction could occur if such high numbers of fertile hybrids are formed that they ultimately replace the pure native types. While there are no reports of crop/wild hybridizations leading to the extinction of a whole species, there have been instances where hybridization has resulted in the extinction of local populations (Ellstrand et al., 1999).

Genes in native populations could be replaced by transgenes through gene flow, if hybrids are formed and the transgene is gradually incorporated into the recipient population through backcrossing. Whether the transgene persists in the native population and has an impact on the range or fitness of a native population is associated with the phenotypic effect of the gene itself and the invasiveness of the recipient species (see next section). The transgenes most likely to be incorporated are those that confer a selective advantage.

The risk of altering native levels of genetic diversity via pollen flow is not unique to GM crops and is associated with the large scale cultivation of any new crop species. The presence of the transgene is the only difference associated with the GM crops (Hancock, 2009). While traditional varieties have been replaced in some areas by commercial varieties, traditional varieties have been maintained in many others, suggesting that farmers can keep land races pure if they so desire (Gepts, 2004; Perales et al., 2003; Bellon and Berthaud, 2004).
GM Crops and Centers of Diversity

Concerns have been raised that the planting of transgenic crops in “centers of species diversity” could result in losses of genetic variability (Rissler and Mellon, 1996; Gepts and Papa, 2003; Gepts, 2005). The most often expressed concern relates to sexual hybridization between the transgenic crop and its relative, and the possibility that levels of genetic diversity will be reduced from that interaction, and in the worst case senerio, that natural populations or land races might go extinct.

The potential ramifications of crop/species gene flow are of concern because landraces are important food sources to many indigenous peoples, and the centers of diversity are considered to be important storehouses of the raw genetic material necessary for breeding new characteristics into crops. Of course, the risk of losing a native population to gene flow from a transgenic crop is fundamentally no different than that possessed by conventionally bred crops. The only difference is the presence of the transgene.

In reality, most crop progenitors are dispersed over a wide geographical range and their numbers are large, so the likelihood of their species extinction is minimal. In many cases, it is also difficult to find single centers of crop diversity. Some crops have multiple centers diversity (wheat, sorghum, pearl millet, barley, pea, lentil, chickpea, flax, maize and lima bean), and some have no discernible centers of diversity at all (radish, sorghum, pearl millet, cole crops and bottle gourd) (Harlan 1995, 1976 and 1992).

Paradoxically, hybridization between crops and wild species, could actually result in higher levels of genetic diversity in the wild relative as the genes of the wild species and the crop are blended together. As the new taxon evolves, many alleles will be lost, but the hybrid population is likely to carry high levels of diversity for many generations.

6. Impact on Non-Target Organisms

Non-target organisms are any organisms that may be adversely affected by the GM crops that were not the intended targets. For example, a crop may be engineered for resistance to a specific insect pest (the target) and any other insects would be non-target organisms. These include various taxa that “fulfill important ecological functions such as biological control, pollination and decomposition” (Romeis et al., 2008). Craig et al, (2008) classifies the non-target organisms into “1) pollinators and natural enemies of pest and the wider category of beneficial species, 2) soil organisms, 3) non-target herbivores, 4) endangered and other species of conservation concern, 5) species which contribute to local biodiversity.”

Non-target organisms are a concern in risk assessment, only if the crop has been engineered with a toxin that makes it insect resistant and there are other organisms that might be sensitive to it. Non-target organisms are not of concern if the crop has been engineered for traits that will not affect organisms other than the crop, such as herbicide resistance, virus tolerance or drought tolerance.

Potential direct and indirect effects of GM plants on non target organisms have been described (Conner et al., 2003). Direct effects are due to toxicity through ingestion by the non-target organisms of a toxin produced by the GM plant, e.g. other insects feeding on Bt plants. Indirect effects occur via multi-trophic food chains, involving, for example, organisms not directly consuming the GM plant but feeding on other insects associated with the transgenic plants (Conner et al., 2003, Craig et al., 2008).

Of the GM crops currently on the market, Bt crops are those with potential for non-target impacts; so the effects of Bt proteins have been extensively evaluated. Results from a number of those studies have been summarized by OECD (2007) in a consensus document on “Safety Information on Transgenic Plants expressing Baccillus thuringiensis – derived insect control protein.”

The broad consensus is that few non-target organisms are negatively impacted by exposure to Bt proteins. Poultry, for instance, was not adversely affected when fed with maize expressing Cry34Ab1 and Cry35Ab1. Likewise, the freshwater daphnia (Daphnia magna) exposed to the Cry34Ab1/ Cry35Ab1 was not negatively affected as well as soil micro-organisms and macro-organisms (protozoa, fungi, bacteria, nematodes, earthworms) in Cry1Ab maize field. No effects were seen on honeybee (Apis mellifera) larvae and adults fed pollen containing proteins of Cry1Ab, Cry1Ac, Cry9C Cry3A, Cry2Ab2, Cry3Bb1, Cry34Ab, Cry35Ab1.

In another study, aphids that play an important role in agricultural systems, as prey or host to a number of predators and parasitoids were shown to be unaffected by Bt cotton in India (Lawo et al. 2009). Larvae of the Monarch butterfly (Danaus plexippus) can be adversely impacted when they ingest high amounts of Bt pollen ( they belong to the same Lepidoptera order targeted by some of the Bt crops); however, native population sizes of this insect are little affected by the widespread growing of Bt corn, as few insects receive toxic levels in the field. Mammals are not affected by the consumption of Bt, as they lack the specific receptors to which Bt endotoxins bind.

It should be noted that the higher yield gained from GM crops could actually reduce the conversion of rich forest into farmlands, and thereby protect the biodiversity hosted in those ecosystems (Clive, 2009). Also, the specificity of certain GM proteins such as Bt proteins has reduced the use of hasher pesticides that are toxic to many more species. In fact, earthworms, bees and beetles have been found more abundant in Bt cotton fields than in conventional sprayed fields (Marvier et al., 2007).

7. Other Agro-Ecological Impacts

Resistance to Bt Toxins

The widespread cultivation of pest-resistant GM crops might lead to resistance developing in the targeted pests. This is, of course, a risk to product performance, not the environment.Resistance to Bt has been reported in three Lepidopteran species in three countries: 1) South Africa in 2006 for resistance of stem borer (Busseolafusca) to Cry1Ab corn, 2) Puerto Rico in 2006 for resistance of fall armyworm (Spodoptera frugiperda) to Cry1F corn, and 3) USA in 2003 for resistance of bollworm (Helicoverpa zea) to Cry1Ac cotton (Tabashnik et al., 2008, Biosafety Assessment Tool, Centre for Biosafety, http://bat.genok.org/bat/).

Management strategies have been developed to delay the evolution of resistance in pest populations. Typically, these involve high doses of the pesticidal protein expressed by the transgenic plant, in conjunction with a refuge that is planted alongside the transgenic crop field. A refuge is an area planted with a similar non-transgenic crop which does not express the specific toxin. Its purpose is to maintain a population of the target insects that possess alleles for susceptibility to the transgenic protein. When the susceptible insects mate with resistant individuals, the level of resistance is reduced in the progeny. The high dose/refuge strategy is based on certain assumptions about the target pest genetics, including 1) resistance is a susceptible trait, 2) resistance is rare, 3) mating is random.

Overall, the evolution of resistance remains an important concern that must be continually addressed. The emergence of resistance will impact primarily on the control of pests in farmers’ fields and may cause a return to control strategies currently employed in non GM crops fields which are less friendly to the environment.

Resistant Weeds to Glyphosate Herbicide

In recent years, glyphosate resistant weeds have also been reported in some countries (Duke and Sanderman, 2006; Service, 2007). In the USA for instance, corn and soybean growers recently reported on weeds that had developed resistance to Roundup (Kruger et al., 2009). These weeds include giant ragweed (Ambrosia trifida) which is one of the weeds that drove the adoption of Roundup (Kruger et al., 2009). Other reports of resistance to glyphosate include Sorghum halepense (Johnsongrass) in Argentina and USA; Lolium spp (ryegrass) in USA and Australia, Amaranthuspalmeri, Amaranthus rudis and Amaranthus tuberculatus (waterhemp) in USA; Euphorbia heterophylla in Brazil; Conyza canadensis (horseweed) in both Brazil and China (Biosafety Assessment Tool, – Centre for Biosafety, http://bat.genok.org/bat/).

Losing the use of glyphosate would be a critical issue if it brought back the use of herbicides that are less friendly to the environment. However, management strategies have been developed to delay the development of resistance, including rotating the use of different herbicides (Kruger et al., 2009), and using “stacked” GM plants that have tolerance to more than one herbicide. However, gene flow from stacked plants to native relatives could lead to the development of multi-resistant weeds (Biosafety Assessment Tool, – Centre for Biosafety, http://bat.genok.org/bat/). This raises agricultural concerns; however, it is not an environmental issue as herbicides are not spayed in natural environments.

Risk assessment aims at identifying and characterizing the risks associated with GM plant cultivation; while risk management determines the necessary measures to be taken in order to mitigate the risks that have been identified through the assessment process that are unacceptable. Risk communication informs the public on the decisions that have been taken, and allows for collecting public suggestions relevant to the decision-making process. Pursuant to the Cartagena Protocol on Biosafety, risk assessment will be carried out in a scientifically sound manner (Article 15), considering all the adverse effects that the cultivation of a GM crop may have on the receiving environment.

Risk assessment involves the identification of the potential adverse effects and an estimation of the probability of such effects occurring (OECD, 2006). It is a process that “uses scientific evidence to estimate the level of risk based on a combination of both the likelihood (= exposure) and consequences of potential harm (= hazard)” (OGTR, 2007). Harm is determined in relation to a country’s protection goals. Risk assessment is not the consideration of all effects, but only those effects that are considered harmful by the jurisdiction in question

Risk assessment is conducted, taking into consideration the data that are collected that are relevant to regulators. In assessing environmental risk, regulators should 1) decide on what needs protection from harm, 2) evaluate how the utilization of GM material might cause harm and what needs to be protected, 3) collect data from available literature or if necessary from new studies, to evaluate the likelihood of and the magnitude of harm resulting from the utilization of that GE material (Raybould and Cooper, 2006)

Note that uncertainty is a key aspect in any risk assessment and “absolute certainty or zero risk in safety assessment is not achievable” (OECD, 2006). Regulators have to make decisions on what nature and level of uncertainty is acceptable for the approval of a GM crop, taking in account social, economic and political aspects.

Relevant comparators must be used to clearly describe levels of hazard, probability of that hazard, and the subsequent risk.

Conducting environmental risk assessment

Despite the critical importance of risk assessment in transgenic products regulation, there is as yet no global, consensus model for conducting risk assessment, even though guidelines have been suggested by several international bodies such as OECD, UNEP-GEF, Biosafety Protocol and EU.

The guidelines suggested by the Biosafety protocol (Annex III) include, among others:

• characterization of the GM traits that could have potential adverse effects on the environment

• an evaluation of the likelihood of the adverse effects, considering the exposure of the receiving environment of the GM material

• an evaluation of the consequences of the adverse effects

• characterization of the overall risk, considering the likelihood and the consequences of the adverse effects

The European Union directive and guidelines have proposed the following steps in environmental risk assessment (Commission of the European Communities, 2002; European Economic and Social Committee, 2004)

• identification of the potential adverse effects

• evaluation of the magnitude of the consequences derived from the adverse effects

• evaluation of the likelihood of the adverse effects

• estimation of the risks that may be posed, by combining the likelihood and the magnitude of the potential adverse effects

• identification of the strategies appropriate to manage the estimated risks

• identification of the risks, after taking in account the management strategies

Critically, the receiving environment first must be well characterized so that a “baseline” context can be defined to allow properly identify the potential hazards and measure the related impacts. Information needed for such characterization include location, geographical, climatic, and ecological characteristics, information on the biological diversity features and the status of centres of origin.

Lessons learnt in the past decade of experience in risk assessment have led regulators and researchers to emphasize the necessity of having a streamlined approach. To facilitate decision making, the “problem formulation” approach has been suggested as a critical first step in the risk assessment process (Raybould and Cooper, 2005; Raybould, 2006, 2007; Nickson, 2008). It is based on principles that have been successfully applied to assess risks for chemical products and GM crops (Raybould, 2007).

This approach includes the three following steps (Nickson, 2008):

• definition of the assessment endpoints

• development of a conceptual model

• development of an analysis plan

The assessment endpoints have been defined as the measurable entities that need to be protected. They derived from the protection goals set by governments or international bodies. The assessment endpoints should be “clearly and operationally defined, comprising an entity (example a population of a particular species in a particular area) and a measurable property of this entity (example the population size of that entity)”.

The conceptual model utilizes available information to describe the relationships between the valued entities (e.g. population of honeybees) the stressor (e.g. Bt toxins), the pathways of exposure (e.g. presence of Bt toxins in pollen) and the potential effects in the environment (e.g. decreased levels of pollination). The conceptual model describes the risk scenarios and formulates the risk hypotheses, using information on the characteristics of the conventional crop varieties, the nature and the characteristics of the GM traits introduced in the crop, the characteristics of the likely receiving environment and the likely interactions between these elements.

The analysis plan presents the data needed in risk assessment and describes how to collect it (Nickson, 2008). Thus, an environmental risk assessment will typically require:

• information on the characteristics of the product (transgene) obtained through molecular and expression analysis

• agronomic and phenotypic characteristics of the plant

• compositional characteristics of the plant products collected from fields

Tiered approaches have been found to make the best use of available information in the decision-making process (Romeis et al., 2008, Hancock, 2009). In tiered approaches, decisions are made at several stages using available information whenever possible. The process starts at the lowest tier with laboratory studies and then progresses, if necessary, towards higher tiers through contained experiments and open field tests (Raybould and Cooper, 2006).

In general, the lower tier tests are conducted under worst case scenarios, such as an unavoidable contact between a given organism and high doses of an active ingredient. Studies and tests in tier 1 are not intended to be realistic, but rather to aid in early decision making and thereby minimize unnecessary costs of testing of products that present very low hazard (Raybould, 2006). Lower-tier activities require less complex experimentation than those at higher tiers.

Four tiers are commonly used in insecticidal studies. In Tier # 1, the toxicity of microbially derived protein mixed with diet is analysed in the laboratory at elevated levels. If potential adverse effects are noted, Tier # 2 is conducted where the toxicity of more natural routes of exposure are analysed, such as transgenic leaf material and exposed prey. In Tiers # 3 and # 4, long-term, semi-field (greenhouse or cages) and open field studies are utilized to determine population level effects with more realistic environmental exposure levels. The types of monitoring done in Tier # 4 include aerial/flying invertebrates, surface dwelling organisms and below ground invertebrates and microorganisms.

The list of potentially affected organisms in natural and agroecosystems is very wide and for this reason, risk analysis of non-target effects has to be focused on organisms closely related to the target group and representatives of other classes of organisms such as pollinators (exp. honey bees), natural enemies (predators and parasitoids), endangered or insects of cultural and aesthetic value (exp monarch butterfly) and soil dwelling organisms (exps. earthworms, Collembola and microorganisms and microarthropods, protozoans).

To do tiered analysis of potential weediness/gene flow issues, Hancock (2003) suggested that only three main categories of information need to be assessed: 1) the geographical range of compatible relatives; 2) the invasiveness/weediness of the crop and its relatives and 3) the phenotype of the transgene. The most critical factor in assessing risk is the phenotypic trait conferred by the transgene, whether it is selectively neutral, detrimental, or beneficial to fitness in the native environment. Fitness represents the reproductive success of a plant.

The various crop/transgene combinations can be grouped into four tiers. In tier # 1 are combinations involving neutral selectable marker genes transgenes in all crops, deleterious genes in crops with no compatible relatives and herbicide resistance in crops with no native relatives or their relatives are not invasive. In tier # 1, none of the combinations would require any further experimentation. In tier # 2 is the combination of pest resistance into a crop with a compatible native relative. In tier # 2, the necessary experiment would be to show that a similar phenotype exists in wild populations; if it does not, then it must be determined whether the target pest does not significantly impact on natural populations. In tier # 3 is the combination of an advantageous trait into any species with a compatible relative. In tier # 3, the necessary experiment would be to show whether a similar phenotype exists in wild populations. If it does not, the fitness of the transgenic crop will need to be measured in representative native environments. In tier # 4 is the combination of a advantageous trait into any species with a compatible, highly invasive relative. Tier # 4 products should not be released unless containment systems are developed preventing gene flow.

Using tiered approaches in risk assessment allows regulators to structure their data requirements and to prioritize the relevant information needed. Most of the data and information obtained through the tiered approach, especially at the lowest tiers (laboratory studies, toxicity to particular groups of organisms, food toxicology, crop species distribution, etc.) should be accepted and shared among regulatory bodies, so that unnecessary costs can be avoided. The tiered approach is critical for improving the efficiency of regulatory decisions as it allows available resources to be focused on higher risk GE materials.

Click here Risk estimate by USDA-APHIS for an explanation of the new USDA APHIS proposed assessment procedure.