Part 2 of 4
PART 2: GM PLANTS
Environmental Risks
One of the most commonly perceived risks associated with GM crops concerns the possibility that transgenes will escape from the confines of agriculture with serious environmental consequences. Indeed, the US NRC report (1989, p. 3) considered that “the potential for enhanced weediness is the major environmental risk perceived for introductions of GM plants.” Two questions that relate to this issue are commonly asked. Will crops that have been GM become invasive? Could the transfer of genes from transgenic crops to their wild relatives through natural hybridization result in the origin of more aggressive weedy types? Under both scenarios, so called “superweeds” are predicted as the unintended results of biotechnology, resulting in the genesis of novel biological invasions. Such invaders would not only reduce crop yields but could also cause serious disruptions in the functioning of natural ecosystems and losses in biodiversity. Because of the potential ecological hazards posed by transgene escape, a considerable literature on this topic has developed during the past decade (reviewed in Tiedje et al., 1989; Crawley, 1990; Ellstrand and Hoffman, 1990; Raybould and Gray, 1994; Snow and Palma, 1997; Warwick and Small, 1998). Recent theoretical and empirical work concerning the potential escape of transgenes from the crop environment enables some assessment of the likely environmental risks posed by GM crops. In this section, we review this work and also consider other ways in which GM crops may have undesirable environmental consequences.
Could GM Plants Become Invasive?
The likelihood that GM crop plants will become invasive and constitute serious weed problems is often considered fairly remote. This is because most of today’s major crop species (e.g. corn, rice, wheat, beans) have been subjected to intense artificial selection over long periods of time for traits with low survival value under most natural conditions. Traits such as nonshattering of grain in cereals, weakly developed chemical defences, lack of seed dormancy, and high fertilizer requirements restrict the ability of most domesticated species to thrive outside the crop environment. Indeed, although crops are grown over vast areas of the globe today there are relatively few cases in which they persist without deliberate cultivation for more than a few seasons. Such volunteer plants are usually confined to agroecosystems and rarely if ever invade undisturbed natural plant communities. Domesticated crop plants are not represented among the world’s serious plant invaders. This is because persistence in wild communities results from the combined effects of many genes working in cooperation to produce a functioning phenotype adapted to local ecological conditions. Therefore, in most cases insertion of specific transgenes into a crop species already possessing a syndrome of domesticated traits is unlikely to alter its ecology so that it becomes converted into an aggressive invading species. Such targeted genetic modifications are unlikely to nullify many generations of artificial selection involving countless genetic loci.
This argument clearly depends on the extent to which a particular domesticated species has been subjected to artificial selection. Many cultivated species, especially those involved in horticulture, forestry and rangeland agriculture, have only recently been brought under cultivation and consequently have been subjected to relatively little genetic alteration through conventional breeding. In this case, the degree of domestication may be quite minor and cultivated genotypes may resemble their wild ancestors in many respects. In these cases, cultivars are more likely to persist outside of cultivation, and under certain circumstances could become invasive (see below). GM species with a short domestication history are more likely to pose environmental problems than our major crop plants. However, invasiveness will occur only if the genetic modifications increase the survival and reproduction of cultivars in natural ecosystems. Little work in this area has been conducted. In the future, as the range of target organisms for genetic modification widens, it may not be safe to assume that all cultivated species have been genetically crippled through intense artificial selection. Indeed, recent experience in Canada with herbicide-tolerant canola (oil seed rape), discussed next, provides a warning that some crop plants have the potential to become serious weeds of agriculture.
Canola is a relatively recent plant domesticate compared with many of our major cereals (e.g. corn, wheat, rice). Unfortunately, two wild traits that persist in many canola cultivars are weak seed dormancy and a degree of seed shattering. As a result of these traits, large numbers of seeds enter the soil after cropping and can persist in the seed bank to emerge in subsequent seasons as volunteer plants (Pekrun et al., 1998; Derksen and Watson, 1999; Downey, 1999). Traditionally, volunteer crop plants occur at relatively low densities and are eliminated from crops by selective herbicides. However, this management tool is complicated if volunteers are herbicide resistant. Unfortunately, herbicide-resistant volunteer canola plants are beginning to develop into a major weed problem in some parts of the Prairie Provinces of Canada. Indeed, some weed scientists predict that volunteer canola could become one of Canada’s most serious weed problems because of the large areas of the Prairie Provinces that are devoted to this crop. Of particular concern is the occurrence of gene exchange via pollen among canola cultivars resistant to different herbicides. This can occur through crosses between volunteer plants and the crop, or between different volunteer plants. Three classes of herbicide-resistant canola (resistant to glyphosate, glufosinate and imidazolinone) are currently grown in western Canada. Recent evidence indicates that crosses among these cultivars have resulted in the unintentional origin of plants with multiple resistance to two, and in some cases three, classes of herbicide (Derksen and Watson, 1999; Downey, 1999; Topinka et al., 1999). Such “gene stacking” represents a serious development because, to control multiple herbicide-resistant volunteer canola plants, farmers are forced to use older herbicides, some of which are less environmentally benign than newer products. This example involving the origin of multiple herbicide-resistant canola serves to illustrate the dynamic nature of weed evolution within managed agroecosystems. It also demonstrates that crops plants are not immune from becoming weeds of agriculture under the appropriate selection regimes.
Because of the large areas devoted to herbicide-resistant canola in the Prairie Provinces, it is not surprising that opportunities for the genetic mixing of different varieties occur. Despite the best efforts of growers, seeds may often be transported accidentally between fields containing different herbicide-tolerant canola varieties by farm machinery, or simply be blown from trucks transporting seeds to and from fields (Gray and Raybould, 1998). Indeed, it has been argued that seed spillage, a form of gene dispersal, may be a much more common mechanism resulting in hybridization between varieties than is likely by long-distance pollen flow by animal pollinators (McHughen, 2000, p. 166). Regardless of the mechanisms giving rise to multiple herbicidetolerant canola varieties, this example illustrates the problems in trying to predict the likelihood of gene flow from small-scale test plots involving relatively small numbers of plants. In addition, it emphasizes the inherent difficulties in the containment of genetic material in the context of normal farming practices in which literally millions of small seeds are produced and harvested over large areas of the landscape. Industry argues that as long as “good farming practices” are followed, these problems should not occur. This perspective may be unduly naïve. Environmental assessments associated with the release of GM crops should take account of the fact that in the real world human error and expediency may often compromise guidelines for the growing of such crops.
Gene Flow Between GM Crops and Wild Plants
In contrast to many animal species, reproduction in plants can be quite promiscuous. Individuals can mate simultaneously with many partners including themselves and, in addition, hybridization with related taxa commonly occurs. Mating complexity is promoted by a fundamental feature of plant reproduction — plants are immobile and therefore require vectors (e.g. mostly animals or wind) to transport their gametes from plant to plant to ensure crossfertilization. The process of pollen dispersal is inherently imprecise and only a small fraction of the large number of male gametes produced by a plant (usually < 1%) reach conspecific stigmas resulting in successful pollination. The majority of gametes are lost to the vagaries of the pollination process while a small fraction is dispersed to stigmas of other plant species. If the pollen donor and recipient are related, an opportunity is provided for inter-specific hybridization. Most inter-specific hybrids are genetically sterile or possess maladapted trait combinations and are soon eliminated by natural selection. Others persist through clonal propagation, while a small minority can become successful new forms because they possess novel phenotypes. Historically, inter-specific hybridization has played an important role in the evolution of flowering plants, with a significant proportion (estimates range from 30% – 50%) of all species arising in this manner.
A major environmental concern associated with agricultural biotechnology is that gene flow from GM crops to related weeds will result in the formation of novel weed phenotypes that have the potential to become highly invasive. Considerable effort in recent years has been directed toward understanding how likely this process is to occur for particular crops, and how to mitigate any negative environmental consequences that might result from such accidental gene transfer (Ellstrand and Hoffman, 1990; Jorgensen and Andersen, 1994; Kareiva et al., 1994; Raybould and Gray, 1994; Snow and Palma, 1997; Lavigne et al., 1998; Rieseberg et al., 1999). Indeed, one of the primary motivations for the use of maternally inherited genetic constructs has been that these technologies will reduce transgenic escape routes through pollen (Daniell et al., 1998; Gray and Raybould, 1998).
It is worth recognizing at the outset that gene flow between crops and weeds has been known for over a century and is not a unique characteristic of the technique of genetic modification per se. Inter-specific or inter-racial hybrids between crops and weeds are commonplace and have been well studied by weed scientists (e.g. carrots, oats, rice, oilseed rape, sorghum, sugarbeet, sunflower; see Table 2.2 in the US NRC report (2000) and similar tables in Snow and Palma, 1997; Rieseberg et al., 1999). Indeed, this phenomenon has resulted in the evolution of a special class of agricultural weeds known as crop mimics that resemble crops in appearance and/or behaviour and thereby evade detection (Barrett 1983, 1988).
The experimental study of pollen dispersal and gene flow in plants generally indicates that the distribution of pollen dispersal distances is highly leptokurtic (most pollen is dispersed short distances with a steadily declining fraction involved with long-distance dispersal). For example, most pollen in herbaceous plants is dispersed within two to three metres of source plants, with a small fraction being transported up to one kilometre or more (Levin and Kerster, 1974; Lavigne et al., 1998). The two most important determinants of pollen dispersal are the mating system of the plant and dispersion of pollen by wind or animals. In general, predominantly selfing species produce far less pollen than outcrossers, and little of this finds its way to other plants. In contrast, outcrossing species produce significantly more pollen and maximum dispersal distances can be considerable, especially in wind-pollinated species. There is no a priori reason why these general principles of pollen dispersal should be different for GM crops; so transgenic pollen should not behave in a manner different from the pollen of non-GM plants. However, comparative studies on this issue need to be conducted to confirm this assumption.
Crops can be roughly divided into three groups with respect to the likely incidence of the natural transfer of genes. 1) No possibility — where wild relatives are absent from the region where the crop is grown (e.g. GM maize, soybean, tomato in Canada). 2) Low possibility — GM crops that are either predominantly autogamous (many cereals) or propagated largely by asexual reproduction and flower rather infrequently (sweet potato, sugar cane). 3) Moderate to high possibility — where the crop is an outbreeder and is being grown in an area where cross-compatible wild relatives occur (e.g canola in many parts of Europe and North America; rice in South East Asia). Current guidelines for the field testing of GM crops recognize these distinctions and recommendations for the size and isolation of test plots reflect the likelihood of gene exchange with wild relatives. A major issue here concerns the issue of scale. Opportunities for gene transfer will be considerably greater for large-scale commercial plantings of GM crops than for small test plots. Hence, generalizations about pollen dispersal distances of commercial planting based on experimental studies of small plots should be treated with some caution.
How is hybridization between cultivated and wild plants studied, and is there evidence for the natural transfer of transgenes from GM crops to weeds? The frequency of hybridization events between crops and weeds has usually been detected by simply observing the occurrence of putative hybrids in close proximity to agricultural fields. The presence of plants with “intermediate phenotypes”, or character combinations predicted from hybridization, signal the occurrence of gene transfer. However, there are two reasons why this approach may greatly underestimate the true frequency of gene transfer. First, many products of hybridization are selected against during the establishment phase and hence do not give rise to viable offspring (see below). Second, hybrids can go undetected because of similarities in phenotype to parental forms. This is especially likely where backcrossing and advanced generation crosses result in hybrid swarms composed of plants spanning the entire spectrum of phenotypic variation encompassed by crop and weed. To avoid these difficulties in estimating the true frequency of gene transfer, researchers have recently used simply inherited genetic markers diagnostic for the parental forms to detect hybridization between crops and weeds (e.g. Luby and McNicol, 1995; Whitton et al., 1997, Wilkinson et al., 2000). Assays of seed families collected from individual plants enable the quantitative analysis of gene transfer.
While there is considerable evidence for crop–weed hybridization, only a few cases have been reported involving experimental trials of GM crops. To our knowledge, there are no known cases involving the escape of a transgene into weed populations from commercial scale plantings. To date, most work has involved the insect-pollinated outbreeder, Brassica napus (oilseed rape or canola), which can hybridize with several congeneric species (B. rapa, B. oleracea) as well as the related wild radish (Raphanus raphanistrum). It has been suggested that this species can potentially hybridize with up to nine related taxa (Stewart et al., 1997). Since several of these are also cross-compatible with other wild Brassica species, the pool of species that transgenes could potentially infiltrate is quite large. Chèvre et al. (1997) produced F1 inter-specific hybrids between oilseed rape and radish, and after four generations in field plots, herbicide-resistant plants with a similar morphology and chromosome number to the weed were established. The authors concluded that under normal agricultural conditions this process is likely to occur only rarely. Wilkinson et al. (2000) used remote sensing to identify areas of sympatry between non-GM oilseed rape and wild B. rapa over an extensive area (15,000km2) of South England. Flow cytometry and molecular markers were used to screen for hybrids. Only one naturally occurring hybrid was found. This was a much lower rate of hybridization than anticipated based on earlier predictions from hybridization rates in adjacent populations of the two species and presumed areas of sympatry (Scott and Wilkinson, 1998).
While the available data are sparse and limited to experimental plots of a single GM crop (oilseed rape) it does indicate that transgenes, not unexpectedly, can be transferred to wild plant species, albeit at a low frequency. Other crop–weed systems in which hybridization occurs more frequently (e.g. rice, Langevin et al., 1990) could pose greater risks. Where crops and interfertile wild plants co-exist in the same area, it is probably safest to assume that some degree of gene transfer will occur over time. It is important to recognize, however, that the process of gene flow from GM crops to weeds by itself does not pose an environmental risk. It is the potential consequences of such an event that is the cause for concern. The ecological outcome of hybridization will depend entirely on whether wild plants with newly acquired transgenes have sufficiently enhanced fitness to cause their numbers to increase in frequency. We address this issue below.
Finally, our focus in this section has been on the transfer of transgenes from GM crops to wild plants. As discussed above for canola, another potential escape route involves the transfer of transgenes to other crops of the same species that are not GM. Where GM crops are grown in the same region as non-GM cultivars, opportunities for cross-pollination exist. Indeed, the likelihood of this process occurring is likely to be higher than for most crop–weed transfers because of the very large population sizes involved in crop plantings and the complete absence of breeding barriers that are likely between conspecific cultivars. Recent reports from various European countries of the contamination of canola originating from Canada with small quantities of GM DNA seem likely to have arisen in this manner. Both GM and non-GM canola are grown over extensive areas of western Canada, facilitating insect-mediated cross-pollination between cultivars. While such cross-contamination is unlikely to pose environmental hazards to wild plant and animal communities, it does raise economic and political problems because of concerns in Europe over the food safety of GM crops discussed elsewhere in this report. In addition, the contamination of non-GM crops with transgenes represents a serious problem for low-input farming (organic agriculture) and may require much larger isolation distances than have been used to now to ensure the purity of non-GM produce (Moyes and Dale, 1999).
Spread of Transgenes in Wild Plants
Predicting the fate of transgenes in wild plant populations is considerably more difficult than determining whether gene flow between crops and weeds is likely to occur. This is because diverse ecological and evolutionary processes will govern the survival and spread of transgenes once they are incorporated into the genetic backgrounds of wild plants. Determining the ecological and evolutionary consequences of transgene spread in wild populations is one of the central issues in assessing the environmental impact of GM crops. While analytical tools have been developed by evolutionary biologists to measure the strength and direction of natural selection (reviewed in Endler, 1986), these approaches have yet to be applied to GM traits. Our ability to predict the spread of transgenes into wild plant communities is hampered by a lack of empirical data on the fitness costs and benefits of transgenic traits in non-crop species. Moreover, it is important to stress that such information is meaningful only when obtained from diverse ecological contexts. Because weedy plants, the likely first recipients of transgenes, have the potential to migrate to diverse habitats through natural dispersal, genotypes containing engineered traits have the opportunity to be tested by natural selection in countless environmental settings. While in many situations weedy genotypes are likely to be poorly adapted, it would be foolhardy to suggest that appropriate conditions do not exist in nature for successful spread. Indeed, experience suggests that novel phenotypes often succeed in circumstances not predicted based on simple demographic models that do not incorporate ecological variation.
Once transgenes are transferred to wild gene pools, their subsequent fate will be strongly influenced by population size. Because transgenes will initially be present at low frequency, they may often be lost from populations through stochastic processes such as genetic drift. Weed populations are especially vulnerable to stochastic processes since population sizes are often small and frequent colonizing episodes lead to an erosion of genetic diversity (Barrett, 1992). Repeated reintroduction through gene flow from GM crops may be necessary for establishment in some weed populations that are subject to frequent fluctuations in population size. The mating system of the weed species will be critical for determining how frequent the introgression of transgenes into wild gene pools is likely to be. Many weeds of agricultural land are predominantly selfing, reducing the likelihood of gene transfer. However, Bergelson et al. (1998) and Bergelson and Purrington (2002) reported that some herbicide resistant transgenic lines of the annual selfing weed Arabidopsis thaliana were roughly 20 times more likely to outcross than mutant plants of the same species. Therefore it may not be safe to assume that all selfing plants are immune from genetic contamination since even predominant selfers usually exhibit low levels of outcrossing.
The spread of transgenes into wild populations will be governed by the benefits that they confer to their carriers in terms of enhanced survival and reproductive success. This will depend on the types of transgene under consideration and their effects on plant phenotype. The first generation of GM crops largely involved genes conferring resistance to herbicides and to various pests and diseases, but new genes associated with stress tolerance (e.g. salt, drought and temperature tolerance) are also likely to become commercially available in the near future. It is not difficult to imagine that the escape of such genes could have potential influences on the ecology of wild plant communities. However, the potential ecological impacts of other targeted genes in the second generation of GM crops (e.g. vitamin-rich rice and increased floral longevity in ornamental species) are more difficult to assess.
Most engineered genes are likely to be ecologically neutral and some may carry fitness penalties to their carriers. In these cases, they are likely to be lost from populations quite rapidly through genetic drift or natural selection. Alternatively, some transgenes may provide a selective advantage within wild populations but predicting which constructs these are likely to be is not an easy task. To assess the ecological impacts of transgene escape, recent attempts (reviewed below) have been made to measure the fitness of GM varieties by assessing the costs and benefits of various transgenes in comparison with unmodified varieties. It is particularly important to determine whether transgenes persist in wild plant populations in the absence of selection to maintain engineered traits (e.g. continued herbicide sprays or pest and disease outbreaks). Alternatively, such genes may be selected against because of the costs that they can exert on plant fitness. Such costs can be caused by pleiotropy, linkage to deleterious genes, disruption of coding regions during insertion or the physiological costs associated with maintaining engineered traits.
Not surprisingly, the results of comparative studies of GM versus non-GM plants have been mixed. Some investigators have found no significant differences in performance, whereas others have demonstrated both costs and benefits to the possession of GM traits. For example, Snow et al. (1999) found no significant differences between transgenic herbicide resistant and non-transgenic plants of Brassica napus x B. rapa hybrids in both survival or seed production in growth chamber experiments. They concluded that the costs associated with herbicide resistance in the hybrids were probably negligible. Similar conclusions were also reached by Lavigne et al. (1995) following competition experiments conducted under field conditions between herbicideresistant and non-resistant lines of white chicory (Cichorium intybus). In contrast, Bergelson et al. (1996) demonstrated a strong cost to herbicide resistance in the weed Arabidopsis thaliana, with a 34% reduction in seed production of transgenic plants compared to susceptible genotypes sown into field plots. One of the only studies demonstrating increased fitness of transgenic plants involves Brassica napus containing Bt cryIAc, an insecticidal transgene that confers resistance to various caterpillars (Stewart et al., 1997). Insect attacks causing defoliation of non-transgenic plants favoured Bt plants in plots that were initially cultivated but were allowed to naturalize. This study is particularly significant because it involved fitness comparisons in natural vegetation.
To fully understand the dynamics of transgenic escape, large-scale demographic studies are required in which the complete life histories of populations are monitored over several successive years. Crawley et al. (1993) estimated demographic parameters of transgenic and nontransgenic Brassica napus in a variety of habitats and climatic conditions over a three-year period in the UK. Despite considerable variation in performance among sites and treatments, they found no evidence that transgenic lines were more or less likely to persist in disturbed habitats than plants that were not GM [and see Linder and Schmitt (1995) and Hails et al. (1997) for additional studies demonstrating similar results in B. napus]. Ecological comparisons of other GM crops and their associated weed complexes are urgently needed to assess the likelihood that transgene escape could result in negative environmental consequences.
While most studies to date have failed to demonstrate any strong ecological advantage to transgenic plants in comparison with conventional varieties, this should not be taken as evidence that the ecological risks associated with transgene escape will be always be minimal. Too few GM species have been examined for any broad generalizations to be made. Indeed, given the complex nature of many ecological interactions it may not be easy to make firm predictions in this area. Most workers that have considered the problem of transgene escape in any depth agree that each GM crop and transgene combination has to be considered separately, taking into account both the life history attributes of GM crop–weed complexes and the ecological context in which they occur.
GM Crops and Biodiversity
One of the least understood issues associated with GM organisms is their potential impact on biodiversity. For GM plants we have already considered the escape of transgenes into wild populations resulting in the origin of potentially aggressive weeds. While these weeds would impact agroecosystems first, causing yield reductions and economic losses, they could also lead to losses in biodiversity if they subsequently invaded natural plant communities. Most agricultural weeds are rather poor colonizers of undisturbed vegetation and seem unlikely candidates to invade plant communities. However, as discussed above, future genetic modification of a broader range of plant species, including trees, shrubs and clonal perennials, could potentially lead to the transfer of transgenes into plants with competitive life histories. These types of plants are more likely to be capable of moving into communities that up to now have resisted invasion.
While biodiversity is commonly thought to signify the number and kinds of species in a community, it is important to also recognize that biodiversity includes an intra-specific component, specifically the genetic diversity within species. How might large-scale introduction of GM crops influence this component of biodiversity? One potential impact involves genetic alterations in wild plant populations associated with changing agricultural practices. For example, the widespread introductions of herbicide-resistant crops (HR crops) will undoubtedly influence the spectrum of weeds occurring in and around arable fields. If herbicide usage increases because of HR crops, we might also expect more cases of the evolution of weeds that are genetically resistant to herbicides (HR weeds). Warwick et al. (1999) review this issue for the Canadian situation and point out that selection of HR weed biotypes is highest when a single class of herbicide is used repeatedly and is highly efficacious. There are now over 200 reported cases worldwide of herbicide-resistant weed biotypes since resistance was first reported in 1968 (reviewed in Heap, 1999); of these, nearly 30 have been reported in Canada (see Table 6 in Warwick et al., 1999).
Another serious biodiversity concern is the contamination of wild gene pools of the world’s major crop plants by genetic constructs engineered through biotechnology. Because, as discussed earlier, many crops have wild and weedy relatives with which they are fully interfertile, the potential for gene transfer into crop gene pools is a serious possibility. This is especially worrisome where crops are grown in regions of the world where they have originated and are thus in contact with a range of close relatives. For example, in South East Asia where rice originated, several cross-compatible wild and weedy species of rice inhabit wetland environments in and around rice fields. In contrast, the majority of crops that are grown in Canada were domesticated elsewhere and the number of cross-compatible relatives vulnerable to this form of so-called “genetic pollution” is more restricted than in many other regions of the world. Nevertheless, the weedy relatives of several crops grown in Canada, such as canola, carrot, sunflower and sorghum, occur in agricultural fields, a situation that creates opportunities for transgene escape.
Agriculture has resulted in the large-scale global destruction of natural ecosystems with concomitant losses in biodiversity. For example, it has been estimated that 70% of the land surface in the UK is under some form of agriculture, and in Canada and the US equivalent figures are 11% and 52%, respectively (reviewed in Maguire, 2000). A question that is often asked is whether the introduction of GM crops will exacerbate the problem of biodiversity loss or alternatively whether the impacts will be minimal. In Europe, this concern has received much greater public attention than in other parts of the world, presumably because farming and wildlife have co-existed for a much longer period and this has resulted in the development of a distinctive flora and fauna associated with farmland. In Europe, many species are adapted to the habitats associated with agricultural practices such as hedgerows, ditches, hayfields and meadows. The widespread use of broad-spectrum herbicides associated with herbicide-resistant crops could potentially reduce plant biodiversity with direct and indirect influences on vertebrate and invertebrate species. For example, a recent report by Watkinson et al. (2000) drew attention to the possibility that the use of GM herbicide-tolerant crops could result in severe reductions in weed populations with subsequent negative effects on seed-eating birds. Agricultural land in North America is also important for wildlife (Best et al., 1995; Boutin et al., 1999) and detailed studies are urgently needed to assess the impact of the large-scale growing of GM crops on the maintenance of biodiversity in agricultural ecosystems. We support the view taken by Maguire (2000) that conserving biodiversity is an essential part of sustainable agriculture that is beneficial from both an economic and ecological perspective. Agroecosystems that are sterile wastelands not only have little aesthetic appeal but are unlikely to be ecologically sustainable over the long term.
Perhaps the least appreciated way in which the biodiversity of natural plant and animal communities could be threatened by biotechnological change is through genetic alterations in the ecological amplitude of domesticated plants. As discussed above, one of the most potent forces resulting in the erosion of biodiversity is the replacement of natural ecosystems by agriculture and forestry. In the future, because of increasing pressures on land for food, it may be possible to engineer crops to grow in environments that up to now have been considered unsuitable or at best marginal for arable cropping systems (e.g. salt marshes, deserts, rainforests, mangrove swamps). The expansion of the range of conditions in which agriculture can be practised because of advances in genetic engineering could potentially lead to the extensive loss of wildlands and their constituent biodiversity.
Regulatory Implications
Predicting the environmental risks associated with GM crops is difficult because of the diverse ecological interactions that can potentially occur in agricultural and natural plant communities. Serious ecological impacts could arise following very rare events that would be hard to predict from data collected in conventional ecological experiments conducted at restricted spatial and temporal scales. The sparse knowledge base available concerning the ecology and genetics of GM crops is a major hurdle for sound risk assessment, with important regulatory implications. We recommend that before GM crops are released they should be subjected to a more thorough ecological risk assessment than has been conducted to date. In particular, more effort should be given to following the intent of the current Canadian Environmental Protection Agency guidelines with respect to potential adverse environmental impacts. Industry submissions often satisfy current guidelines through reliance on literature reviews without collecting their own experimental data on ecological impacts. Moreover, the whole focus of environmental assessment occurs within the context of agroecosystems only, with little effort paid on assessing likely impacts on the biodiversity of natural ecosystems. In future, we suggest a staged approach in which any new GM variety is subjected to a series of experimental comparisons with conventional varieties, including the unmodified variety from which it originated. These comparisons should be conducted under various sets of conditions (e.g. glasshouse, growth chamber, field plots, disturbed natural habitats) reflecting increasing ecological realism. The basic goal of these experimental comparisons is to determine whether the new GM crop differs from conventional varieties in any life-history attribute that is likely to have fitness implications for survival in the wild.
These experimental comparisons should provide the necessary information to make informed judgments related to regulation on whether a new GM variety, or its transgenes, are likely to pose an environmental threat by resulting in an invasion scenario. Another series of experiments is also required to determine the likelihood of pollen-mediated gene flow to related species. In these experiments, plots of various sizes of the GM crop should be established at different distances from target colonies (also of different size) of cross-compatible relatives. This will allow quantitative assessment of how gene flow interacts with distance. These experiments differ from those currently used to determine isolation distances of crop varieties from one another. This is because in the proposed experiments the target organism is a related species, not the crop. The wild species chosen should include all related species that are known to be cross-compatible with the cultivar and occur in the area in which the GM crop is likely to be grown.
Future Research
To address public concern about the potential environmental impacts of GMOs, a sound body of ecological research on this topic is required. While Canadian ecological research ranks highly by international standards, very little research currently being conducted by leading ecologists and evolutionary biologists in the country concerns GMOs. Moreover, in our opinion the quantity and the quality of research on the potential environmental impacts of GMOs is not sufficient to address many of the pressing questions that concern the environmental impacts of GMOs. The reasons for limited study in this area are complex and involve a variety of factors. These include: 1) limited funding from government agencies and industry for basic research on the ecology of GMOs (see our discussion of this in Chapter 9); 2) a failure by industry to recognize and take seriously potential environmental problems; 3) an early lack of interest by academic ecologists in what was seen as an uninteresting and perhaps even trivial research question; 4) the reluctance of the research community to commit limited research dollars to the kind of long-term ecological monitoring required in this area. For these and other reasons, the research capability required to answer satisfactorily the questions that are repeatedly raised by the environmental community and the general public is at present severely compromised.
The initial types of investigation that should be conducted on the environmental impacts of GMOs should grow out of research associated with regulation (see previous section). However, it is hoped that once routine protocols are in place for these environmental assessments more basic research will be addressed. It seems more likely that novel insights will be obtained if scientists are not constrained by the regulatory framework and are free to ask novel questions about the ecology and evolution of GMOs. Below are suggestions for future research on the potential ecological impacts of GMOs.
1. Glasshouse and growth chamber studies: The experimental material (GM crop and immediate ancestor) should initially be compared under uniform growing conditions in the glasshouse and in growth chambers using standard randomized block designs with sufficient replication. An important philosophy behind these comparisons not evident in the current guidelines is that investigators look beyond the normal agronomic traits associated with productivity and consider traits likely to have ecological significance in the context of potential escape scenarios. These comparisons could include several experimental treatments to simulate environmental variation, for example, various nutrient, light or temperature treatments. Traits measured should involve standard life-history variables, including growth rate, timing of reproductive events, reproductive and vegetative allocation, seed production, dispersal potential and seed dormancy. Of particular significance is the search for any unanticipated pleiotropic effects of transgene insertion on fitness traits.
2. Field trials: Field trials at several locations within Canada with contrasting climate and environmental conditions should be conducted. Once again, it is important that not just agronomic traits are compared but that a range of life-history variables related to fitness are assessed. Of particular interest for these comparisons is the detection of Genotype X Environment interactions in which the life-history traits of the GM crop vary depending on location. Evidence of these interactions can provide useful information on the plasticity of traits and how they might respond to novel environments. More ecological realism could be built into these field trials by introducing biotic interactions with competitors, parasites, predators and mutualists (see section on insect interactions).
3. Wild communities: There are obviously inherent dangers in introducing GM plants into wild communities because of the possibilities of escape. However, unless these types of experiments are undertaken it will not be possible to provide an answer to the question of whether GM plants could invade natural ecosystems. We suggest that in the future researchers consider how these types of comparison could be conducted with safety using isolated sites, quarantine procedures and restricted access to the general public to prevent inadvertent escape. Seeds of the GM plants could be sown into a range of disturbed and undisturbed plant communities and their demography monitored for as long as colonies persist. This approach was used by Crawley et al. (1993) in their studies of transgenic Brassica napus in habitats throughout the UK. By measuring standard demographic parameters, projection matrices can be used to predict population growth and likely invasiveness. It is particularly important to determine whether the GM crop has any capacity for persistence through seed dormancy and the maintenance of a seed bank. Most crops possess no dormancy (although see earlier discussion of volunteer canola) so this seems unlikely. However, since dormancy can have an environmental component, and this trait is critical for survival in most wild communities, it is important that this feature of the life history is subjected to the closest scrutiny.
We also recommend that researchers consider conducting a parallel series of comparative experiments on selected cross-compatible relatives containing transgenes. Here the idea is to introduce the transgene artificially into the wild relative and then observe how these plants differ from other individuals of the same species without the gene. Once again, to ensure ecological realism these fitness comparisons should be made under field conditions in a range of disturbed and undisturbed wild communities that would be carefully monitored to prevent escape. These comparisons are time consuming and clearly cannot be conducted on all cross-compatible relatives of GM crops. The choice of which species to examine should be based on their distribution relative to the potential range of the GM crop and the likelihood of escape.
RECOMMENDATIONS
6.5 The history of domestication, and particularly the time period and intensity of artificial selection, of GM plants should be taken into account when assessing potential environmental impacts. Species with a short history of domestication should receive particularly close scrutiny because they are more likely to pose environmental risks.
6.6 Environmental assessments of GM plants and their particular genetic constructs should pay particular attention to reproductive biology, including consideration of mating systems, pollen flow distances, fecundity, seed dispersal and dormancy mechanisms. Information on these life-history traits should be obtained from specific experiments on the particular GM cultivar to be assessed, not solely from literature reports for the species in general.
6.7 Environmental assessments of GM plants should not be restricted to their impacts on agroecosystems but should include an explicit consideration of their potential impacts on natural and disturbed ecosystems in the areas in which they are to be grown.
6.8 Research data from experiments conducted by industry on the potential environmental impacts of GM plants used in Canadian Environmental Protection Agency assessments should be made available for public scrutiny.
6.9 The Panel recommends that a federally funded multidisciplinary research initiative be undertaken on the environmental impacts of GM plants. Funds should be made available to scientists from all sectors (industry, government and university) with grant proposals subject to rigorous peer review.
REFERENCES
Barrett, S.C.H. 1983. Crop mimicry in weeds. Econ. Bot. 37: 255–82.
Barrett, S.C.H. 1988. Genetics and evolution of agricultural weeds. In M.A. Altieri, M. Liebman (eds.), Weed Management in Agroecosystems: Ecological Approaches, 57–75. Boca Raton: CRC Press.
Barrett, S.C.H. 1992. Genetics of weed invasions. In S.K. Jain, L.W. Botsford (eds.), Applied Population Biology, 91–119. Dordrecht, The Netherlands: Kluwer Academic Publishers.
Bergelson, J., C.B. Purrington, C.J. Palm, J.-C. Lopez-Gutierrez. 1996. Costs of resistance: a test using transgenic Arabidopsis thaliana. Proc. Royal Soc. London B. 263: 1659–63.
Bergelson, J., C.B. Purrington, G. Wichmann.1998. Promiscuity in transgenic plants. Nature 395: 25.
Bergelson, J., C.B. Purrington. 2002. Factors influencing the spread of resistant Arabidopsis thaliana populations. In D. Letourneau, B. Burrows (eds.), Genetically Engineered Organisms: Assessing Environmental and Health Impacts. (In Press). Boca Raton: CRC Press.
Best, L.B., K.E. Freemark, J.J. Dinsmore, M. Camp. 1995. A review and synthesis of habitat use by breeding birds in agricultural landscapes of Iowa. Am. Midland Nat. 123: 1–29.
Boutin, C., K.E. Freemark, D.A. Kirk.1999. Farmland birds in southern Ontario: field use, activity patterns and vulnerability to pesticide use. Agric. Ecosyst. Environ. 72: 239–54.
Chevre, A.-M., F. Eber, A. Baranger, M. Renard. 1997. Gene flow from transgenic crops. Nature 389: 924.
Crawley, M.J. 1990. The ecology of genetically engineered organisms: assessing the environmental risks. In H.A. Mooney, G. Bernardi (eds.), Introduction of Genetically Modified Organisms into the Environment, 133–50. Chichester, UK: John Wiley & Sons.
Crawley, M.J., R.S. Hails, M. Rees, D. Kohn, J. Buxton. 1993. Ecology of transgenic oil seed rape in natural habitats. Nature 363: 620–23.
Daniell, H., R. Datta, S. Varma, S. Gray, S.B. Lee. 1998. Containment of herbicide resistance through genetic engineering of the chloroplast genome. Nat. Biotechnol. 16: 345–48.
Derksen, D.A., P.R. Watson. 1999. Volunteer Crops: The Gift That Keeps on Giving. Poster, Expert Committee on Weeds. Ottawa: ECW.
Downey, R.K. 1999. Gene flow and rape – the Canadian experience. In Gene Flow and Agriculture: Relevance for Transgenic Crops, 109–16. Farnham, Surrey, UK: British Crop Protection Council.
Ellstrand, N.C., C.A. Hoffman. 1990. Hybridization as an avenue of escape for engineered genes. BioScience 40: 438–42.
Endler, J.A. 1986. Natural Selection in the Wild. Princeton, NJ: Princeton University Press.
Gray, A.J., A.F. Raybould. 1998. Reducing transgene escape routes. Nature 392: 653–54.
Hails, R.S., M. Rees, D.D. Kohn, M.J. Crawley. 1997. Burial and seed survival in Brassica napus subsp. oleifera and Sinapsis arvensis including a comparison of transgenic and nontransgenic lines of the crop. Proc. Royal Soc. London B. 264: 1–7.
Heap, I.M. 4 Feb 1999. International Survey of Herbicide Resistant Weeds. At: <www.weedscience.com>
Jorgensen, R.B., B. Andersen. 1994. Spontaneous hybridization between oilseed rape (Brassica napus) and weedy B. campestris (Brassicaceae): a risk of growing genetically modified oilseed rape. Am. J. Bot. 81: 1620–26.
Kareiva, P.R., W. Morris, C.M. Jacobi. 1994. Studying and managing the risk of cross-fertilization between transgenic crops and wild relatives. Mol. Ecol. 3: 15–21.
Langevin, S.A., K. Clay, J. Grace. 1990. The incidence and effects of hybridization between cultivated rice and its related weed rice. Evolution 44: 1000–08.
Lavigne, C., H. Manac’h, C. Guyard, J. Gasquez. 1995. The cost of herbicide resistance in whitechicory: ecological implications for its commercial release. Theor. Appl. Genet. 91: 1301–08.
Lavigne, E., K. Klein, P. Vallee, J. Pierre, B. Godelle, M. Renard. 1998. A pollen-dispersal experiment with transgenic oil seed rape. Estimation of the average pollen dispersal of an individual plant within a field. Theor. Appl. Genet. 96: 886–96.
Levin, D.A., H.W. Kerster. 1974. Gene flow in seed plants. Evol. Biol. 7: 139–220.
Linder, C.R., J. Schmitt. 1995. Potential persistence of escaped transgenes: performance of transgenic, oil-modified Brassica seeds and seedlings. Ecol. Appl. 5: 1056–68.
Luby, J.J., R.J. McNicol. 1995. Gene flow from cultivated to wild raspberries in Scotland: developing a basis for risk assessment for testing and deployment of transgenic cultivars. Theo. Appl. Genet. 90: 1133–37.
Maguire, R.J. 2000. Report of the Environment Canada Workshop on the Potential Ecosystem Effects of Genetically-Modified Organisms. NWR Contribution No. 00-034. Burlington, ON.
McHughen, A. 2000. Pandora’s Picnic Basket: The Potential and Hazards of Genetically Modified Foods. Toronto: Oxford University Press.
Moyes, C.L., P.J. Dale. 1999. Organic Farming and Gene Transfer from Genetically-modified Crops. MAFF Research Project OF0157. Norwich, UK: John Innes Centre.
NRC (National Research Council). 1989. Field Testing Genetically Modified Organisms: Framework for Decision. Washington, DC: National Academy Press.
NRC. 2000. Genetically Modified Pest-Protected Plants: Science and Regulation. Washington, DC: National Academy Press.
Pekrun, C., J.D.J. Hewitt, P.J.W. Hewitt. 1998. Cultural control of volunteer rape. J. Agric. Sci. 130: 150–63.
Raybould, A.F., A.J. Gray. 1994. Will hybrids of genetically modified crops invade natural communities. Trends Ecol. Evol. 9: 85–89.
Rieseberg, L.H., M.J. Kim, G.J. Seiler. 1999. Introgression between the cultivated sunflower and a sympatric wild relative, Helianthus petiolaris (Asteraceae). Int. J. Plant Sci. 160: 102–08.
Scott, S.E., M.J. Wilkinson. 1998. Transgene risk is low. Nature 393: 320.
Snow, A.A., P.M. Palma. 1997. Commercialization of transgenic plants: potential ecological risks. BioScience 47: 86–96.
Snow, A.A., B. Andersen, R.B. Joegensen. 1999. Costs of transgenic herbicide resistance introgressed from Brassica napus into weedy B. rapa. Mol. Ecol. 8: 605–15.
Stewart, C.N. Jr., J.N. All, P.L. Raymer, S. Ramachandran. 1997. Increased fitness of transgenic insecticidal rapeseed under insect selection pressure. Mol. Ecol. 6: 773–79.
Tiedje J.M., R.K. Colwell, Y.L. Grossman, R.E. Hodson, R.E. Lenski, R.N. Mack, P.J. Regal. 1989. The planned introduction of genetically engineered organisms: ecological considerations and recommendations. Ecology 70: 298–315.
Topinka, K., J. Huffman, L. Hall. 1999. Pollen flow between herbicide tolerant canola (Brassica napus) is the cause of multiple resistant canola volunteers. Poster, Expert Committee on Weeds, ECW, Ottawa.
Warwick , S.I., E. Small. 1998. Invasive plant species: evolutionary risks from transgenic crops. In L.W.D. van Raams donk, J.C.M. den Nijs (eds.), Plant Evolution in Man-Made Habitats, 235–56. Proceedings of the VIIth International Symposium of the International Organization of Plant Biosystematists. University of Amsterdam, Hugo de Vries Laboratory.
Warwick, S.I., H.J. Beckie, E. Small. 1999. Transgenic crops: new weed problems for Canada? Phytoprotection 80: 71–84.
Watkinson, A.R., R.P. Freckleton, R.A. Robinson, W.J. Sutherland. 2000. Predictions of biodiversity response to genetically modified herbicide-tolerant crops. Science 289: 1554–57.
Whitton, J., D.E. Wolfe, D.M. Arias, A.A. Snow, L.H. Rieseberg. 1997. The persistence of cultivar alleles in wild populations of sunflowers five generations after hybridization. Theor. Appl. Genet. 95: 33–40.
Wilkinson, M.J., I.J. Davenport, Y.M. Charters, A.E. Jones, J. Allainguillaume, H.T. Butler, D.C. Mason, A.F. Raybould. 2000. A direct regional scale estimate of transgene movement from genetically modified oilseed rape to its wild progenitors. Mol. Ecol. 9: 983–92.