1. Introduction
The theme of this conference, the seventh one in the FAO Biotechnology Forum, was the potential importance and impact of gene flow from genetically modified (GM) crops, forest trees, fish or animals to non-GM populations, with particular focus on developing countries. It was the first conference to deal exclusively with genetic modification and not include other biotechnologies, such as those based on molecular markers or vegetative reproduction.
From 31 May to 6 July 2002, a total of 118 messages were posted and they were numbered in the order of posting. In this document, specific references to messages posted, giving the participant’s surname and the message number are included. All individual messages can be viewed at the Archives of Conference 7.
Before summarising the main arguments and concerns raised in the 118 messages, some general observations can be made about the conference.
Firstly, the topic of gene flow involving GMOs was clearly of great interest. Although running for a shorter time period (5 weeks) than any of the other conferences (which lasted, on average, 9 weeks), more people joined (382) and sent messages (61) than in any other conference. Only conference 1, which lasted 10 weeks, had more messages (138). The large interest in the topic may be explained by the importance given to it by people like Choudhary (message message 20), saying that "cross pollination between GM and non-GM compatible species is a vital issue as far as the future of GM crops are concerned" and/or by the urgency expressed by Stamp (42) i.e. "we believe that the advent of GMOs cannot be turned back, therefore any potential risks linked to the release of pollen from GMOs should be reduced or prevented NOW". The recent controversy about GM gene flow in the centre of origin and diversity of maize may also be a factor.
Secondly, that widely differing opinions are held regarding GMOs and the potential impacts of gene flow from GM to non-GM populations. In the past, the debate on GMOs has tended to be highly controversial and polarised. From this conference, there is no evidence of any change. The divergent views exchanged on the term "genetic pollution" and on gene flow terminology (e.g. Cummins, 31; Ghislain, 37; Redenbaugh, 39; Ashton, 47) illustrate this clearly. However, it can be hoped that, by providing a neutral, moderated platform for the airing of the wide range of views on this subject, the conference may have at least increased the understanding of the other arguments. Comments of two participants on the last day are positive in this regard i.e. "participants of this debate in my opinion have mainly focused on geneflow in what I consider a valuable and rather respectful exchange of thoughts" (De Lange, 111) and "many of the contributions I read gave me cause to reflect on my own fundamental beliefs" (Nickson, 115).
Thirdly, discussions focused overwhelmingly on the issues concerning gene flow in the crop sector with only a small minority of messages dedicated solely to these issues in forest trees (Lindgren, 100; Heinze, 103; Cummins 104, 106, 107), fish (Cummins, 97) or animals (Blair, 81). This tendency was also noted in previous Forum conferences and seems to be a consequence of the fact that GM crops are already a reality, having been cultivated now for a number of years, and the complex of issues that they raise are already been dealt with by policy-makers, consumers and researchers etc. In contrast, there is still no commercial-scale planting of GM trees and no GM animals or fish are currently produced for human consumption.
Finally, Muir's comment (110) that "the primary impact of gene flow [from GM to non GM organisms] is ecological, secondary impacts may be political, economic, social, and pathological" was reflected in the conference, where participants gave far greater emphasis to the ecological impacts (and to ecological risk assessment) of gene flow than to the other potential impacts.
In Section 2 of this document, the main elements of the discussions are summarised under eight main themes (2.1-2.8). Section 3 provides some information about participation in the conference and Section 4 gives the name and country of the people that sent referenced messages. Section 5 provides an explanation of abbreviations used. Note, unless otherwise stated, the term gene flow used here refers to gene flow from GM to non-GM populations.
2. Main themes discussed in the conference
2.1 The ecological impacts of gene flow from GM to non-GM populations
Participants focused on two main areas. The first was the impacts of GM gene flow on biodiversity (and the relative importance of these impacts compared with those caused by other factors). The second was the specific ecological impacts of known transgenes - either those affecting herbicide tolerance and insect resistance traits (present in the majority of GM crops currently cultivated) or those that might be inserted in future commercialised GMOs.
a) Impacts on biodiversity
As Muhunthan (2) dramatically put it: "the million dollar question is whether gene flow from GM to non-GM populations will affect genetic diversity and pollute its purity?". As a precaution, he proposed carrying out studies prior to GMO release and setting up buffer zones to prevent gene flow. Menne (84) argued that gene flow could lead to loss of traditional plant varieties and agricultural diversity.
Muir (24) tried to answer Muhunthan's (2) question by analytically considering the potential implications of a GM plant or animal escaping into an ecosystem on genetic diversity at three levels - within a species, between species within a community and, thirdly, between communities. For the first level, he argued that if the species is wild and its population size is large, spread of the transgene should not reduce genetic diversity at other genes. If the population size is, however, small and the transgene spread due to selection and was eventually fixed, then genes linked to the transgene may also become fixed, leading to some reduction in genetic variability. If, on the other hand, the species is domesticated then farmers might choose to preferentially use and breed from the GM plant or animal and neglect other plant varieties or animal breeds, resulting in a large loss of genetic variation. He pointed out that this could also happen with conventional breeding, if farmers had a strong preference for one kind of genetic material (as had already happened with Holstein dairy cattle, an example also used by Blair (81)). At the next two levels, Muir (34) argued that if the transgene allowed a species to expand its niche it could result in loss of species within a community and that, at the third level, if this happened in several communities they would become more homogenous.
Both Knibb (51) and Mettler (53) argued that if considering the risk of transgenes to biodiversity, then it should be compared with risk to biodiversity from natural mutations in non-GM populations, as they lead to genes being altered and rearranged each generation.
Some participants argued that, compared to transgene flow, other factors had greater impacts on biodiversity. Livermore (18) suggested that farmers selecting seed for the next generation was "the only potential threat to genetic diversity" of landraces, while Mettler (53), similar to Dusi (9), suggested that "by far, the real threat to biodiversity is the extent of land that is devoted to agriculture and the simple displacement of existing ecosystems by farms". Uijtewaal (4), supported by Muhunthan (8) and Dusi (9), pointed out that introduction of a species to a new area where it could mate with wild relatives could be a "disaster". Claparols (54) also highlighted the negative impacts that invasive alien species had on biodiversity, but argued that this example should encourage more prudence with GMOs, as scientific consideration of the risks of gene transfer are still unclear. This point was echoed by Ashton (55) who wondered, given that it had taken a long time to recognise the risks and costs of inter-continental transfer of invasive alien species, "how long must it take us to realise the dangers" of genetic modification.
In contrast to these discussions about the potential detrimental impacts of gene flow on biodiversity, Halos (14) suggested it could have a positive impact in developing countries with limited resources. She argued that in situations where local domesticated plant varieties are sensitive to a disease, then gene flow from nearby disease resistant GM plants would be a cost-effective way of transferring disease resistance to the local varieties and ensuring their survival.
b) Spread of herbicide tolerance and insect resistance genes to non-target plants
As explained in the Background Document to this conference, the majority of commercially available GMOs are crops modified for just two traits - herbicide tolerance or insect resistance. In 2001, it was estimated that 77% of GM crops were modified for herbicide tolerance; that 15% included one of the toxin-producing genes from the soil bacterium, Bacillus thuringiensis (Bt), to confer insect resistance, while 8% had both herbicide tolerance and insect resistance characteristics. It was argued that spread of these transgenes to non-target plants already had or could have negative ecological impacts.
i) Herbicide tolerance genes
Regarding the spread of herbicide tolerant genes, participants focused on crosses of herbicide tolerant canola in Canada with canola that was not herbicide tolerant or that was modified for tolerance to different herbicides. A range of herbicide tolerant GM canola varieties are commercially available in Western Canada, including tolerance to glyphosate, glufosinate or imazethapyr (Cummins, 12). Although mainly self pollinating, canola may cross with plants of the same species, and with weedy relatives. In a paper published in the journal "Weed Science" (2000, volume 48, pages 688-694), Hall and co-authors presented results from a farm in Alberta, Western Canada, where canola plants with resistance to two and three herbicides had been found. They concluded that gene flow between different GM varieties was the most likely explanation for development of multiple resistance. Both Cummins (12) and Jenkins (27) emphasised the gravity of this situation, as herbicide tolerant volunteer plants can become a major weed problem (Jenkins, 27) and as weedy relatives exist with which the multiple resistant plants could form hybrids (Cummins, 12). Nickson (62), responding to their concerns, emphasised instead that "this study constituted one grower in Alberta", that practices on the farm were atypical and that "this example does not constitute a measurable ecological risk". He concluded, instead, that "it is a good example of how agricultural systems have to adapt to new technology".
Because of extensive gene flow, Cummins (12) and Jenkins (93) also stated that conventionally bred canola in Canada could no longer be guaranteed GM-free, a situation with detrimental consequences for organic farming (Cummins, 12; Di-Giovanni, 23). This is discussed further in Section 2.8.
ii) Insect resistance genes
Regarding the so-called Bt-crops (containing Bt genes), Menne (84) warned that gene flow from Bt cotton and maize represented an ecological threat to their closely related wild species (such as Gossypium herbaceum in cotton) because of "contamination". For Bt-cotton, Wozniak (87) stated that the United States Environmental Protection Agency had taken a cautious approach to the question of gene flow "since it is not possible to say with any certainty what the impacts on these wild populations might be if this novel Bt insect resistance trait were to introgress". He pointed out that recent preliminary research results on Bt-sunflowers in the United States indicated that there were potential weediness impacts, as gene flow to weedy relatives could increase their fecundity and fitness under natural conditions. Nickson (71) also raised this issue, indicating that in canola as well as sunflower, field trials were showing that Bt genes could confer a fitness advantage to wild relatives and that "a thorough risk assessment would have to carefully evaluate the potential for this altered property of the transgenic plant to confer a hazard". Knibb (76) was sceptical about claims that GMOs had increased fitness in a natural setting, although Muir (69, 79) argued that it had been clearly demonstrated in natural environments for both canola (Muir, 69) and papaya (Muir, 79).
Verzola (78) also suggested that spread of Bt genes to non-target varieties, and hence the increased levels of expression of the Bt genes in the field, would hasten the development of Bt resistance among pests, resulting in the eventual loss of an important tool for pest control.
c) Ecological impacts may depend on the transgene
Different transgenes may raise different ecological issues in different environments. For example, Halos (14) hypothesised that, in the Philippines where coconuts are threatened by a viroid disease, gene flow from disease resistant GM varieties would be beneficial. Nishio (10) asked rhetorically whether flow of a transgene conferring high phosphorous uptake efficiency, which could enable plants to outcompete others, from crops to native populations should be considered "good" or "bad". Cummins (106) considered the potential impacts of developing transgenic trees to convert highly toxic ionic or organic mercury to less toxic elemental mercury. He argued that such phytoremediation of mercury pollution would merely relocate soil mercury from contaminated soil sites in the South and redistribute the mercury to the North and that if gene flow occurred, resulting in large expanses of transgenic trees, it "could lead to a global catastrophe". He also signalled potential ecological risks from growing GM forest trees modified for growth (Cummins, 107) or low lignin traits (Cummins, 104).
Because of the wide range of potential ecological issues that can be raised, it was proposed that the ecological impacts of gene flow from GMOs should be considered on a case by case basis rather than as a whole (e.g. Muhunthan, 2; Valdivia-Granda, 40; Aniol, 58).
2.2 The economic impacts of gene flow from GM to non-GM populations
Discussions on the economic impacts of gene flow dealt with two main aspects: the impacts on trade and exports if gene flow occurred and, secondly, the impacts on individual farmers due to liability arising from intellectual property issues.
a) Trade
In the words of Louwaars (50), "where farmers intend to sell their product at premium prices in certified non-GMO markets, unintended introgression of transgenes may pose a threat to the commercial position of these farmers". Menne (84) argued that gene flow could result both in loss of markets and in additional costs associated with labelling and separation of GM-free produce. Verzola (82) emphasised that "because developing countries are usually dependent on a few agricultural products for export", they could not afford market loss through GM gene flow and, given that there was also a current preference for GM-free products, he advocated that developing countries, like the Philippines, should aim "to keep their entire territory GE-free, by avoiding field releases including field-testing" (Verzola, 105). Gallego-Beltran (108) supported this strategy, saying it would offer a potential advantage for reaching selective markets.
b) Liability
As mentioned in the Background Document to the previous Forum conference, which dealt with the issue of intellectual property rights, patents have been granted in the field of agriculture "on a wide range of biotechnology processes and products, involving genes, viruses, bacteria and even living higher organisms". Ownership of genes or seeds thus introduces additional potential impacts of gene flow.
For example, if gene flow has negative consequences, then someone may have to pay. According to Nishio (11) "if one owns the gene, and it escapes and causes economic and social damage, then the owner should be held responsible. It is the risk of ownership". However, Di-Giovanni (23) felt that this was not the reality with gene flow i.e. "in the environmental field, the concept of "polluter pays" is well established. However, in agriculture, the onus has generally been on the producer of the crop (products) to sufficiently isolate their fields so as to produce a "pure" product". Ashton (98) argued that in cases of gene flow in Mexico, United States and Canada, involving conventional and organic farms, "supporters of transgenics have shown remarkable reluctance to accept responsibility".
Another impact is that patent owners may enforce intellectual property legislation if gene flow has taken place i.e. that "introgression of patented genes may sooner or later lead to claims by the holder of the patent, even where the genes were introduced unintentionally" (Louwaars, 50). Muhunthan (15) argued that in developing countries, because of small farm sizes, it would not be possible to prevent gene flow from GM to non-GM crops and that if farmers were sued by seed companies for breach of patents following gene flow, they would suffer serious economic consequences. Verzola (78) pointed out that in developing countries, such claims would be especially contentious as the majority of patents are owned by companies in the developed world.
Wozniak (33) and Namai (114) provided a reminder that, although not the subject of this conference, gene flow can also take place from non-GM to GM populations and that for farmers willing to pay extra for GM seed there might be economic losses due to "contamination" with non-GM pollen.
2.3 Assessing the ecological risk of gene flow from GM to non-GM populations
As seen in Section 2.1 (and elsewhere), there is much concern about the current and/or potential ecological impacts of gene flow from GM to non-GM populations. Assessment of these risks, prior to GM release, was therefore considered highly important (e.g. Muhunthan, 2; Wuerthele, 80). In addition, Wuerthele (80) suggested that risk assessment of GMOs was especially important for developing countries, as they "are least able to afford additional environmental problems". There was much discussion in the conference about how a science-based risk assessment framework might be applied to gene flow and how the results might eventually be used.
Nickson (24) began the discussion by briefly describing the fundamental principles of the ecological risk assessment framework used for GM plants produced by his company and by explaining that to conduct an appropriate risk assessment for gene flow it was critical to have "clearly defined and operational terms" and that the two most important terms in risk assessment are hazard and exposure, where risk = hazard x exposure. Stated simply, this key formula relates risk (i.e. in our case, the ecological risk resulting from gene flow from GM to non-GM populations) to hazard (undesired/injurious events or harm caused by gene flow to the environment) and exposure (the frequency of gene flow or the probability of the transgene spreading in the environment) (Nickson, 24; Muir, 69). The assessment of ecological risk is carried out by a consideration of these two key components.
a) Hazard
According to Nickson (24), "the challenge that faces scientific risk assessors studying gene flow is having an accurate and testable definition for hazard. Given that hazard is a property that has undesired or injurious consequences, the challenge for scientists is to develop risk assessment experiments that can quantitatively or qualitatively assess the nature and magnitude of an injurious event associated with gene flow". He argued that there were "broad characterizations of the hazards associated with gene flow from GM crops such as: impacts on biodiversity, impacts on population dynamics, genetic swamping, and alterations of gene pools; all of which are inoperative in terms of science based hypothesis testing". The hazard that his company focused on was "the potential for the transgene to confer increased weediness to the crop or its sexually compatible wild relative", which could be scientifically assessed.
Jenkins (27) wrote that, on the contrary, the broad potential hazards mentioned by Nickson (24) were amenable to scientific testing but that the difficulty in testing for them lay in trying to draw conclusions about the real world in which GM crops would be applied from small-scale field tests. Raybould (61) disagreed with this point, arguing that several of the phenomena described by Nickson (24) were not operational "because we are not agreed on the variables that specify them. Unless we can agree on the changes in measurable variables that constitute 'genetic swamping', 'impacts on biodiversity', 'alterations of gene pools' etc., even very large experiments will fail to advance scientific risk assessment" and that ecologists needed to be able to predict "how agreed variables would change after the release of a GM plant".
Wuerthele (80) emphasised that in risk assessment it was first necessary to identify what potential adverse outcomes should be considered. She argued that since GMOs are fundamentally different from conventionally bred organisms, they might present new hazards (discussed in more detail in Section 2.5). Muir (69) said it was very difficult (perhaps impossible) to address the issue of potential environmental harms resulting from gene flow as "all potential harms may not be known a priori". He preferred instead to focus on the second term, exposure, which could be more easily addressed, arguing that if there was a low probability of the transgene spreading then the issue of potential harms became irrelevant.
b) Exposure
In real life, a large number of factors can influence the frequency of gene flow and spread of a transgene in the environment and these should be included in the risk assessment analysis. They include "fitness in a specific environment; gene flow based on characteristics of the inserted gene elements; distance of pollen movement; presence of pollinators; crop rotation; intercropping systems, as well as volunteer plants and their removal" (Valdivia-Granda, 40). Transgenes may also be spread through human intervention e.g. by road transport of GM seeds (De Lange, 91), by whole-grain GM food provided as aid (Ashton, 98) or, simply, through "brown bag" seed being passed from one farmer to another (Morris, 74).
In the conference there was much discussion about the potential value of population genetic mathematical models (described by Muir (69, 73, 77)), which, given that gene flow has first taken place, try to predict whether the transgene will be eliminated or increase in frequency in a natural setting. The models assume that the ultimate fate of the transgene can be predicted based on estimates of fitness components (e.g. adult viability, mating success) of the GM individuals and they can be applied regardless of whether the initial release of the transgene occurs by pollen spreading, animals escaping or intentional release by humans (Muir, 73).
Knibb (72) was not convinced by the value of such models, arguing that "because of simplifying assumptions required to operate population genetic models,....these mathematical models inherently are of little or no predictive use in real world situations". Muir (73) replied that two potential weaknesses of the models, raised by Knibb (72) in his message, could be dealt with and that the models had already been applied successfully in real situations. Trus (75) maintained that these kind of models were "essential", although it was important that their assumptions be stated clearly. Nielsen (95) argued that they "are the best tools available today to evaluate the consequences of such [GM] gene flow into wild relatives".
However, as Muir (77) wrote, "a model is only as good as the estimates of fitness components put into it" and to get accurate estimates, appropriate testing facilities would be required which "could be exceedingly expensive to build, depending on the type of GM organism examined". For forest trees this might be an important issue as Lindgren (100) pointed out that, compared to crops, forest trees are more expensive to field test and the test results may be less reliable. Morris (74) questioned whether such models would be helpful in developing countries because the environmental information is often poorly documented and the spread of GMOs will often occur through human intervention, which is hard to document or model.
c) Risk assessment in developing countries
As the conference had a special focus on developing countries, a number of messages dealt with some specific aspects of ecological risk assessment in these countries.
A common theme raised was the dearth of key information on the ecology of native plant species. Morris (6) noted that there was insufficient information available on the potential for crops to cross-pollinate with African wild relatives and, secondly, on insect pollinators in Africa and their habits. She emphasised the huge potential for research projects in these areas, while Bothma (48) confirmed that little research has been done in Africa regarding gene flow in GM crops. Di-Giovanni (23) agreed with Morris (6), adding that for assessing the probabilities of pollen dispersal and gene flow from wind pollinated outcrossing plants "scientific information on pollination mechanisms of many tropical plants is not as well developed as we would like, and for certain plant-types pre-requisite pollination ecology studies may be required". Badr (21) suggested that a lot of relevant research material is available in libraries, but not electronically.
Louwaars (19), agreeing with Morris (6), stated that "information on possibilities for cross fertilisation is basic to analysis of environmental safety" and continued "whereas food safety research from the North can be used for risk analysis in any other country, cross fertilisation needs to be researched taking the local plant populations into account". He proposed that "botanical files" be built up on local species, whereby "cultivated, weedy, feral and wild populations, can be plotted. Combined with the knowledge on reproductive biology of the species and its relatives in the country/region, this provides exactly the information that Ms. Morris is looking for". The need for local ecological information was also emphasised by Valdivia-Granda (40) who concluded that "gene movement between transgenic crops to other crops and wild species should be examined on a case-by-case basis considering eco-geographical characteristics". The reproductive biology of the species involved may be quite complex, as shown by Badr's (29) description of the flowering system of the papaya and by Namai's (94, 114) conclusion that breeding systems are highly variable between and within plant varieties, where even species normally considered to be self-pollinating can cross-pollinate.
Risk assessment in developing countries should also be based on the realities of their farming systems. Verzola (105) argued that even if the initial levels of gene flow were low, the probability of a transgene spreading was higher in developing than in developed countries as farms tend to be smaller and closer together and farmers commonly save seed for subsequent planting. In addition, in developing countries "it is common to cultivate either several varieties and/or mix them with secondary crops" (Valdivia-Granda, 40).
Muir (110) emphasised that if one is to determine which GM species (plant or animal) might present a gene flow risk in developing countries, then we have to first consider that gene flow can only occur with species already found in developing countries. For the different agricultural sectors, he summarised thus: "many of the domesticated plants came from developing countries, including, but not limited to: rice, papaya, cassava, eucalypti trees, maize, and tomato. Many domesticated animals have their origin in developing countries but the species of perhaps greatest concern is tilapia, which come to us from Africa. The world market for tilapia is growing at record pace and GM tilapia have been developed. The concern is what happens if these GM species find their way back to their global centers of origin?". For forest trees, Lindgren (100) suggested that in order to avoid gene flow to native species, GM forests with exotic species may be proposed for developing countries.
d) Comparative risk assessment and risk/benefit analysis
Having assessed the ecological risk of gene flow from a certain GM variety, what then ? Nickson (24) underlined the importance of "comparative risk assessment" i.e. "where the risks associated with the GM plant are characterized and compared to those associated with the conventional system in which the GM crop will be introduced". Wozniak (87) agreed, writing that the ecological risk of GM gene flow to wild relatives needed to be compared "to the impacts already occurring from non-engineered cultivated varieties that hybridize with related species or wild populations". In this comparative context, Muir (110) pointed out that it is easier to determine the ecological risks from GM gene flow than from invasive alien species because the non-GM species already exists in the ecosystem and can act as a control, whereas for an introduced species there is no real control.
Ghislain (35) suggested that any ecological risks from gene flow needed to be put in the context of the environment in which they might be used. He argued that if risk assessment showed that there was a reasonable probability of a certain GM variety posing a threat to biodiversity, then the country would have to develop policies "considering the relevance of these threats for each region. By relevance, I mean in an area of intensive mining, deforestation or urban pollution, it is irrelevant to care about a remote event of gene flow in balance with all the other threats...".
It was also argued that potential ecological risks from gene flow should be weighed up against potential benefits of applying GM crops. Nickson's (115) viewpoint was that "the two basic elements required to conduct a risk/benefit assessment are scientific capacity to interpret experimental science and some form of public policy to assesses criteria for acceptability (i.e. define benefit and risk). It is inconceivable to me that a country in this world does not have people with the scientific and public policy capabilities. As such, I firmly believe that a scientifically based, risk assessment that integrates social aspects is the appropriate tool for decision making anywhere in the world". Morris (102), in the context of developing countries, similarly argued for a risk/benefit approach to the question and suggested that "if the potential benefits for developing countries can be clearly defined, then we should not deny the consumer those benefits in an environment when the risks are low or negligible". For developing countries, Lingareddy (99) also weighed up the potential increases in production from applying GMOs against the potential negative ecological consequences, but felt that for their long-term interests it was better to be cautious and not use them.
Jeggo (86) pointed out that many countries have committees and procedures in place to carry out evaluations involving the development, release or use of GMOs in terms of safety and benefits and proposed that, since the risks cannot be limited by national boundaries, an international committee be formed to carry out such evaluations.
2.4 Mechanisms to limit or prevent gene flow from GM to non-GM populations
As seen in the previous section, concerns about the current or potential ecological impact of gene flow from GM to non-GM populations, meant that there was much discussion about how the potential risks might be assessed prior to release. Potential ecological hazards need first to be identified. This may not be straightforward, as it depends on whether GMOs are considered to be fundamentally different from conventionally bred organisms (Wuerthele, 1) or whether potential hazards may be identified prior to release (Muir, 69). The probability of gene flow (exposure) needs then to be calculated and, as seen earlier, this can be quite complex and, if population genetics models are to be used, may require expensive testing facilities (Muir, 77). In addition, in developing countries, gathering ecological information can be difficult, as funding (Morris, 6) and capacity (Ashton, 47) may be limited.
An alternative approach is to simply prevent or limit gene flow from GM populations (Muir, 110). Gressel (43) pointed out that there are a variety of strategies that would "render gene introgression to other varieties, landraces and wild species nigh impossible" and suggested that "the use of such strategies should be a requirement prior to release when there is a crop at risk". The large number of different strategies available was highlighted by Choudhary (20) and, by coincidence, the journal "Nature Biotechnology" (June 2002, Number 6), containing a special section on the environmental impacts of GM crops and describing the current status of these strategies, was released while the conference was underway (Burke, 17; Smyth, 26). As Muir (110) suggested, these strategies could be combined to ensure that gene flow will not take place. A range of different strategies were discussed in the conference:
a) Temporal separation of GM and non-GM populations
According to Nishio (11), timing reproduction of GM crops to occur at different times from native varieties "does not seem feasible". However, Lindgren (100) suggested that growing GM forest trees with short rotation times so that they can be harvested in their juvenile stage (when they typically would have spread little pollen or seeds) might be useful for limiting gene flow.
b) Spatial separation of GM and non-GM populations
Cummins (97) proposed that cultivation of GM fish could be considered as long as it takes place in inland facilities rather than fish pens, where they could escape and mate with wild relatives. For crops, Muhunthan (2) proposed establishing "GM-free zones around the GM populations to prevent gene flow between GM and non-GM populations", although participants mentioned the problems of defining appropriate isolation distances. Cummins (12) reported that pollen from GM canola had been observed to spread far greater distances in Canada than previously considered. Di-Giovanni (23) highlighted the difficulties of setting standard isolation distances for GM populations as "pollen- and gene-flow are inherently variable phenomena" (potentially influenced by factors such as wind speed, atmospheric stability and turbulence, pollen viability and other biological factors) meaning that "it would be unwise to base decisions on a few field trials". He advocated the use of computer simulation models to assess the probabilities of pollen dispersal for wind pollinated outcrossing plants. Muhunthan (15) argued that because farm sizes are small in developing countries, "there will not be any space left to set up a refuge".
c) Sterility strategies
A number of different strategies to ensure the GMOs (or their pollen) are sterile were discussed. One of them is development of GM plants whose seeds are sterile, using so-called "terminator" technologies (e.g. Nishio, 11; Stuart, 26). Valdivia-Granda (40) warned, however, that the strategy represented a risk to subsistence farmers who might be unable to segregate the sterile seed.
Gressel (43) proposed that for vegetatively propagated GM crops, a gene causing infertility (no pollen) could be inserted in a tandem construct with the transgene, so that gene flow would not be an issue. He pointed out that many genes are already known that render pollen infertile.
For plant species that produce a lot of pollen (e.g. maize, canola), Stamp (42) proposed that dispersal of GM pollen could be eliminated or limited by "growing male sterile GM plants in a mixture with male fertile non-GM plants, which act as pollen donors for the GM plants". He proposed that the male sterile varieties could be based on systems of cytoplasmic male sterility.
For fish, Cummins (97) considered the proposal to use triploidisation (i.e. production of individuals with three chromosome sets instead of the normal two) to ensure that GM fish released in the environment are sterile. He argued, however, that triploidisation might be problematic as the technique could have physiological side effects and, secondly, triploids might be "leaky", allowing some fertile gametes to be produced. Given the potential risk that release of GM fish might have on the environment, he proposed that spatial separation of GM fish rather than triploidisation should be used and concluded that "extensive studies on sterile triploid leakiness to produce gametes should be done before any transgenic fish are exposed to the environment".
For forest trees, Lindgren (100) maintained that use of sterile trees could eliminate or reduce gene flow. He pointed out, however, that to "prove absolute sterility, long field-testing under variable conditions is often needed, and this is expensive, complicated and time consuming". He suggested that if tree sterility was a requirement then it "is likely to lead to increased use of vegetative propagation with a few well-tested sterile clones".
d) Chloroplast genetic engineering
Another strategy mentioned was to insert the transgene(s) into the chloroplast genome rather the nuclear genome (Valdivia-Granda, 40; Murphy, 89). As chloroplast genomes are maternally inherited in most crops, dispersal of transgenes through pollen would thus be prevented, although Valdivia-Granda (40) indicated that it would not be effective in all crops. Cummins (52) emphasised that some species might not show clear maternal transmission (i.e. it could also be paternal or both) and suggested that pollination of a chloroplast GM plant by other crops or weedy relatives might lead to altered chloroplast transmission. He concluded that "individual crop plants and weeds need full analysis of the mode of chloroplast transmission before it could be concluded that the transgenic chloroplast modifications eliminate transgene transmission through pollen".
2.5 Whether GMOs are fundamentally different from conventionally bred organisms (CBOs), thus raising new hazards regarding gene flow from GM to non-GM populations.
This was one of the most divisive topics raised during the conference, resulting from a dichotomy in the way that GMOs are viewed in relation to CBOs. Nielsen (95), agreeing with Wuerthele (1), felt that the essential issue regarding the consequences of gene flow from GMOs was whether transgenic organisms differ fundamentally in their genetic make-up from other traditionally bred organisms, "if the answer to this question is no, then no particular concerns are to be raised that would separate the assessment of GMOs as compared to traditionally bred organisms. If the answer is yes, then the unique features should be identified and the consequences of their dispersal by gene flow evaluated".
Some participants considered that the answer was no and, as a consequence, gene flow from GM-populations is not more of an issue than gene flow from non-GM populations. Thus, Mettler (53) argued that "the identification of one gene as being a transgene (for example for disease resistance) is no more of a threat [to biodiversity] than the already common use of conventionally developed traits for disease resistance", while Burke (17) questioned why herbicide tolerant canola developed using genetic modification or conventional breeding should be treated differently. A number of participants (e.g. Bradshaw 5; Wozniak, 25; Burke, 64) also emphasised that conventional breeding may use technologies with considerable impact on a plant's genetic material i.e. it "includes embryo rescue techniques, pistil/style modifications, colchicine-mediated chromosome doubling, bridging and wide hybrid crosses, phytohormone treatments to alter post-fertilization events, and chemical or irradiation induced mutations. This includes transfer from species through bridging crosses that bring gene combinations together that would otherwise not occur naturally" (Wozniak, 25), and that GMOs are therefore not fundamentally different from CBOs.
Other participants, instead, felt that GMOs are fundamentally different from CBOs in one or a number of ways and, consequently, there are "novel concerns about their effects on ecosystems at the genetic level and about their behavior in ecosystems at the agricultural level" (Wuerthele, 1). She argued that some of the world's most serious environmental problems came from the failure to identify new hazards raised by new technologies and thus, "if GMOs are fundamentally different,..., then it is wise to try to fully understand those differences, and use that understanding to consider what new hazards they might present before making conclusions about risk" (Wuerthele, 80).
The principal differences between GMOs and CBOs that participants mentioned in the context of potential impacts for gene flow may be roughly subdivided into the following two categories, that are presented here, together with the resulting new hazards they may entail:
a) GMOs may transfer exotic genes to the ecosystem
As described in the Background Document, GMOs typically contain one or more genes from another species. These exotic genes may then be transferred to individuals of the same population, to wild relatives or to different species.
For Wuerthele (1), "GMOs are unique because they are created by recombinant DNA techniques. These processes intentionally introduce into a host species genes from organisms with which the host could never breed. This makes GMOs conduits for the transfer of exotic DNA to the host's genetic ecosystem (the gene pools of all the organisms with which it can breed). In contrast, organisms created by conventional breeding cannot transfer exotic genes because conventional breeding merely rearranges genes already present among compatible species" and "this ability to transfer exotic genes across species is the essence of what makes GMOs unique: they are gene vectors" (Wuerthele, 13). Similarly, Ashton (55) argued that genetic modification runs independently of the evolutionary process "in that a construct that can never naturally occur, has been introduced to the gene pool". Bradshaw (5, 16) was not convinced about the enormity of this difference between GMOs and CBOs, arguing that conventional breeding also involved induced and spontaneous mutations combined with intense artificial selection i.e. more than simple rearrangement of native genes and, secondly, that CBOs could also contain exotic genes mentioning, as an example, that the pathogenic bacterium Agrobacterium tumefaciens inserts genes into the genome of host plants.
The resulting ecological hazards from gene flow mentioned by participants were primarily related to the evolutionary implications. Nielsen (95) argued that "unintended gene flow from GMOs has the potential to significantly change the evolutionary trajectories of their wild relatives. Whereas traditional breeding is largely based on artificial selection, modern gene technology introduces novel genetic variation that is naturally unachievable in the organism in question. Mechanisms providing genetic variability in higher eukaryotes do not combine DNA sequences from several organisms into a compact functional unit within the time scale achieved by genetic engineering". He suggested, however, that GMOs developed with simple intra-chromosome modifications (unlike those developed with species-foreign or novel genes) are likely to cause few concerns in this respect.
In a similar vein, Muir (57), suggested that, in contrast to natural mutations, which could result in formation of new species only over a long time period and which tended to involve small changes, be entirely random and have negative impacts on fitness of the organism, the "creation of new mutations by man (transgenes)" occurs rapidly, they do not occur at random and the transgene normally confers some advantage to the organism. As an illustration, he suggested that a natural mutation allowing goats to produce spider silk in their milk would be impossible. He concluded that the creation of transgenes "that result in formation of new species and elimination of others is clearly unacceptable, we do not have evolutionary time to adjust to the changes that we can bring upon ourselves through such actions. We can also bring about more changes too rapidly for any ecosystem to adapt to". Knibb (63) disagreed with this analysis, arguing that there was no evidence for species formation by genetic modification, that natural mutations could also have large visible effects and that "there is no empirical evidence that genetically engineered changes are more likely (than natural mutations) to be fit in the wild".
Another potential ecological hazard raised is that there might be more horizontal gene flow (gene flow by non-sexual means) to bacteria. Nielsen (95) argued that as transgenes often have DNA sequence homology to prokaryotes, this significantly increases the likelihood of transgene integration in bacteria. Verzola (82) suggested that there were some reports of horizontal gene flow occurring with GM crops. Valdivia-Granda (22) also raised the issue of GM crops containing the coat protein genes of specific viruses, arguing that, through horizontal gene flow, the transgene could be taken up by infecting viruses, leading to new viral genomes.
In addition, Valdivia-Granda (22) discussed the potential risk of antibiotic resistance genes being transferred to pathogens, because "a distinguishing characteristic of many transgenic plants is the presence of antibiotic resistance genes" used as markers to select transformed cells. He mentioned in particular the kanamycin resistance gene, as the antibiotic kanamycin is still used for treating human infections. Verzola (96) reported that there had been calls to phase out use of antibiotic resistance genes, although Valdivia-Granda (22) was concerned that the large investments in time and money made in developing a GM plant might make it difficult to withdraw products already available from the market.
b) The genetic modification process may create organisms that are unstable
Wuerthele (1) argued that when identifying the hazards of GMOs, consideration should also be given to the behaviour of the organism itself because recombinant DNA techniques "create organisms with inherently unstable and unpredictable behavior" and that the instability derives from the way GMOs are made (Wuerthele 1, 13). The potential new hazards this raises regarding gene flow were not discussed in detail.
One feature of the genetic modification process mentioned in this respect is the way transgenes are regulated. Wuerthele (1) argued that "transgenes are multiplied in number or are accompanied by promoters so that the products for which they code are expressed in high concentration. Often, transgenic products are not controlled temporally or anatomically, but are expressed throughout the host's tissues and life cycle. Moreover, the [promoters] used to activate transgenes may produce unintended effects by also activating host or retroviral DNA. In contrast, highly expressed traits in conventionally-bred organisms are under genetic controls characteristic of the organism". To this, Bradshaw (5, 16) replied that no general statement of that nature could be made about transgene stability or regulation - that each transgene and transgenic event must be characterised separately.
Nielsen (95) noted that the way transgenes are regulated could, however, increase the likelihood of transgene expression if gene flow occurs to wild relatives, since a feature of transgenes is that they "are often modified to allow broad expression and often require few interactions with the host cytoplasm for activity".
Another feature mentioned is that insertion of the transgene may cause mutations in the host DNA. Wuerthele (13) proposed that "mutations are an unintentional but necessary by-product of inserting foreign genetic material into the host genome" and that "transgene mutations may unexpectedly interfere with important gene function in the GMO as well as be passed to organisms with which it breeds". Bradshaw (5) agreed that transgene insertion could cause insertional mutations that abolish gene function in the organism, but said there was no evidence that this also leads to instability. Verzola (67) highlighted the high frequency of mutations introduced when transgenic events are made and argued that, through gene flow, the risk of damaging mutations would be passed on to other organisms (Verzola, 78). Stewart (68) emphasised that in genetic modification, large numbers of transgenic events are made but only a small minority of resulting plants are selected, based on expression of the gene of interest and on fertility and other characteristics of the plants. He also argued that similar procedures are followed for conventional breeding techniques such as tissue culture or wide-cross or mutation breeding.
Responding to Wuerthele (13), Wozniak (25) argued that there was no evidence that the genetic modification process produced inherently unstable or unpredictable organisms and, secondly, that instability in the plant genome (due to transposons, natural mutations or translocation etc.) was an everyday reality, a point also made by Bradshaw (5). Wozniak also added that the studies he had reviewed regarding stability of transgene inheritance had not provided any suggestion "that the inserted gene construct was unstable or in anyway altered following insertion". Like Datta (7), Burke (64) and Stewart (68), he argued that, as with conventional breeding, plants with undesirable characteristics would be identified and discarded through standard production and selection procedures. Uijtewaal (4) stated that characteristics like "stability of expression" and "unexpected side effects" are studied "for at least 3-5 years during an intensive selection program. The costs related to the development and registration of such a [GM] product are so high that a company can not afford to develop a product that will not last".
2.6 Philosophical/ethical aspects of gene flow from GM to non-GM populations
Apart from the ecological and economic impacts of gene flow, there is also a philosophical/ethical dimension to the question. Louwaars (50) said that one of the impacts of gene flow might be that "genes from foreign species may be regarded by local communities as a threat to the natural integrity of the local crops". De Lange (91) emphasised the importance of this aspect, arguing that in the conference the spiritual dimension had been overlooked, "which is very real for most, if not all, indigenous peoples. Maize, for example, is considered by Mexican and other meso-american indigenous peoples as sacred. Apart from all other dimensions (food safety, environmental safety, patents etc.) the transgenic contamination of Mexican indigenous varieties is considered as spiritual pollution".
Regarding naturalness and integrity of local populations, Trus (75) suggested that "in biological systems there is rarely such a thing as absolute purity. Similarly, the concept of "pollution" is a relative one only. Any apparent purity in a breeding population is really a function of the amount of time since the last novel genetic "migrations" (intentional or otherwise), the nature of the novel genetic contributions and the genetic stability of the resulting population". Heaf (66) addressed the issue of naturalness of genetic modification and any resulting gene flow by quoting from a passage in Shakespeare's play "A Winter's Tale", where Polixenes argues that since man is part of nature, anything he makes or does is also the work of nature (i.e. in our context, as man has developed GM varieties then gene flow from GM varieties is also natural).
2.7 Centres of origin
Following reports of GM maize in Southern Mexico (part of the centre of origin and diversification of maize), there has been considerable focus recently on the specific issue of the impacts of gene flow in centres of origin. For developing countries, the issue is especially important because, as Valdivia-Granda (22) pointed out, "many developing countries are the genetic centers of origin for cultivated plants modified by genetic engineering". Participants emphasised strong concern about this issue.
Ashton (47) and Rosset (83) expressed "alarm" about the "Mexican maize contamination saga", with Rosset calling for a moratorium on GM releases in such centres until more information is available. According to Ghislain (35), the issue of gene flow in centres of origin and diversity needs special attention "due to its complex mixture of scientific, social, and cultural issues", but warned that the scientific and socio-cultural aspects should not be confused. Krell (41) urged that the precautionary principle should be embraced regarding this issue and that, in centres of origin, conditions should be promoted that favour conservation of their natural genetic resources. He proposed that farmers in these areas should "receive subsidies or other motivation for not using introduced or genetically modified material, but using local varieties". Muir (110) argued that, although all GMOs introduced into their centres of origin will not result in harm, they have the potential to produce effects on the ecosystem that are just as devastating as introducing invasive species.
2.8 Organic agriculture
The issue of gene flow from GMOs to plants on organic farms is especially sensitive as organic agriculture does not permit use of GMOs. As Nickson (101) put it, "perhaps the most contentious place of detection [of GM material] would be in organic where transgenes have been designated as unacceptable based on personal preferences".
Cummins (12) and Di-Giovanni (23) referred to the specific case of organic farmers in Canada where, because of gene flow, "essentially no canola grown in western Canada can be claimed to be free of gene modification" (Cummins, 12) and where legal proceedings are consequently being taken by organic farmers against biotechnology companies (Di-Giovanni, 23). Wozniak (33), agreeing with Cummins (28) that the process of "double fertilisation" could result in detectable GM products in some fruits or grains pollinated by GM crops, noted that the rules governing organic production in most countries tend to be unforgiving regarding the presence of GM material. Redenbaugh (39) also noted that gene flow from CBOs to organic farms was not considered to render the crops non-organic. According to Burke (17), the rules disqualifying growers from organic certification if GM material is detected were "imposed, in essence, by organic farmers on organic farmers" and that such certification bodies were now trying to assert the rules over non-organic farmers.
3. Participation in the conference
The conference ran for five weeks, from 31 May to 5 July 2002. A total of 275 people had subscribed to the conference by the opening day and the numbers gradually increased to 382 by the final day of the conference. Of these 382 people, 61 of them (i.e. 16%) submitted at least one message. 48 of the 118 messages posted (i.e. 41%) came from participants in North America while the others came from Europe (21%), Asia (18%), Africa (12%), Latin America and the Caribbean (4%) and Oceania (4%).
People send messages from 25 different countries - the greatest proportion came from the United States (30%), Canada (11%), The Philippines (8%), South Africa (8%) and United Kingdom (8%). A total of 32% of messages were from participants in developing countries and 68% from developed countries. (Note that these figures are only an approximate indicator of the relative contributions of the developing versus developed world and of the different world regions to the conference - people from developing countries may be currently living in developed countries (and vice versa)).
The greatest proportion of messages came from people working in universities (32%), followed by those in research centres (24%), NGOs (17%) and private companies (13%). Note, again, that these results are only an approximation - people may have several roles at any one time (e.g. a participant with a university work address could also be on a governmental advisory board and/or a member of a NGO) and they may change over time.
4. Name and country of participants with referenced messages
Aniol, Andrzej. Poland
Ashton, Glenn. South Africa
Badr, Aisha. Egypt
Blair, Hugh. New Zealand
Bothma, Gurling. South Africa
Bradshaw, Toby. United States
Burke, Derek. United Kingdom
Choudhary, Bhagirath. India
Claparols, Javier. The Philippines
Cummins, Joe. Canada
Datta, Swapan. The Philippines
De Lange, Wytze. Netherlands
Di-Giovanni, Franco. Canada
Dusi, André. Brazil
Gallego-Beltran, Juan. Colombia
Ghislain, Marc. Peru
Gressel, Jonathan, Israel
Halos, Saturnina, The Philippines
Heaf, David. United Kingdom
Heinze, Berthold. Austria
Jeggo, Martyn. Austria
Jenkins, Peter. United States
Knibb, Wayne. Australia
Krell, Rainer. Italy
Lindgren, Dag. Sweden
Lingareddy, Tulasi. India
Livermore, Martin. United Kingdom
Louwaars, Niels. Netherlands
Menne, Wally. South Africa
Mettler, Irvin. United States
Morris, Jane. South Africa
Muhunthan, Rajaratnam. Sri Lanka
Muir, William. United States
Murphy, Denis. United Kingdom
Namai, Hyoji. Japan
Nickson, Thomas. United States
Nielsen, Kaare. Norway
Nishio, John. United States
Raybould, Alan. United Kingdom
Redenbaugh, Keith. United States
Rosset, Peter. Mexico
Smyth, Stuart. Canada
Stamp, Peter. Switzerland
Stewart, Neal. United States
Trus, David. Canada
Uijtewaal, Bert. Netherlands
Valdivia-Granda, Willy. United States
Verzola, Roberto. The Philippines
Wozniak, Chris. United States
Wuerthele, Suzanne. United States
5. Abbreviations
Bt = Bacillus thuringiensis; CBO = Conventionally Bred Organism; FAO = Food and Agriculture Organization of the United Nations; GE = Genetically Engineered; GMO = Genetically Modified Organism; NGO = Non-Governmental Organisation
6. Acknowledgements
We wish to give a very special and sincere thanks to all the individuals who submitted messages, for devoting their time and effort to sharing their views, insights and experiences with the Forum members.
