Ethics, Law and Science of Using New Genetic Technology in Medicine and Agriculture

Darryl R. J. Macer, Ph.D. Eubios Ethics Institute 1990

Copyright1990, Darryl R. J. Macer. All commercial rights reserved. This publication may be reproduced for limited educational or academic use, however please enquire with the author.

8. Environmental Safety of GMOs

pp.127-151 in Shaping Genes: Ethics, Law and Science of Using New Genetic Technology in Medicine and Agriculture, D.R.J. Macer (Eubios Ethics Institute, 1990).
Risks from Laboratory Applications

Safety of this technology is a serious concern of many people. It has been important since the first use of genetic engineering in the early 1970's. Public awareness of all forms of genetic engineering has recently increased again with the first field trials of GMOs. The public in general have an inflated fear of genetic manipulation, which we can say in mid-l990, is largely the result of a lack of and misinformation rather than realistic fears. A voluntary moratorium on several types of genetic experiment involving use and construction of genes and their insertion into vectors for their multiplication in bacteria was imposed in the 1970's (Wheale & McNally 1988). This has long been removed as the potential hazards have been assessed and it was decided that suitable physical and biological containment should be adequate. "Biological" containment advocated the use of "crippled" host cells and vectors, such that these would have no success in colonising any environment outside that of the contained laboratory even if they managed to escape from it. Since the initial categories of physical containment were decided on there has been widespread experience gained in the practise of these experiments, which has resulted in a decrease in the assessed hazards and thus the type of containment judged necessary. The principle of biological containment is still used for most laboratory experiments, especially when dealing with human genes and/or tumour-promoting agents. Physical containment is not so strict, but is still maintained for work on tumour or disease-promoting agents.

Before the appearance of GMOs there have been harmful effects from some of the accidental releases of organisms from laboratories. In 1958 tobacco blue mould (Peronospora tabacina) was brought into the UK for a research institute. In that year the mould spread to four other institutes, including one in the Netherlands, and to a commercial tobacco crop in England. In the following year the disease appeared in the tobacco fields of Belgium and the Netherlands, from where it spread quickly across the rest of Europe (advancing in Germany at the speed of 5-20 km per week). After several years of crop breeding resistance was increased, but it is a powerful example of the risks of accidental release of new organisms (Mantegazzini 1986).

The recent release of some agriculturally important bacteria into the environment, several years after it was first planned, highlights the growing ease with which scientists now regard some types of genetic manipulation. No major problems have arisen, but registration of work and containment levels are useful requirements. Some examples of the current concerns include (Mantegazzini 1986):

* spontaneous mutations in pure and mixed cultures when growth conditions are changed

* toxins produced in thermophilic systems

* modification of viruses during fermentation

* cloning of toxic genes and the introduction of antibiotic resistance genes into microorganisms not known to naturally acquire them.

The principle problem now in all work is not mechanical, but of laboratory discipline. Safety committee's exist in all major laboratories, but there is still room for laxity. Medical surveillance of laboratory workers should be more common in case there may be long term affects of exposure.

There are no known health hazards that are specific to genetic manipulation. However, some of the organisms used, especially microorganisms, are capable of infecting humans so can present health hazards. There could be several ways in which genetic manipulation could result in hazards to humans. A GMO which facilitates delivery of a biologically active gene product to a specific tissue could occur. Autoimmune disease resulting from production of antibodies to a GMO or gene product,. Or enhanced immune response to proteins that are fusion products of the proteins that one has become immune to. In Britain, the Health and Safety Executive is trying to implement a scheme for surveillance as recommended by the U.K. Advisory Committee on Genetic Manipulation. It advises that people working with certain categories of GMO should carry cards, and also have cards attached to their medical records, that would allow contact with a supervisory medical officer should they get sick, as one way of monitoring any effects. It also recommends the storage of serum samples before conducting such work, to allow comparison at a later date in the event of illness.

There is still controversy over the use of genetic engineering in industry in some countries. A West German court ruling in November 1989 forbid the use of these techniques in industry until there is a better legal basis for regulation. This forced Hoechst AG to stop the construction of a factory for two years, until recently when a law was introduced. The application will be considered now that Germany has a gene technology law. The West German law also fits with an EEC directive on genetic engineering that provides minimum protection in EEC countries without regulations,taking effect in 1991. The German's are particularly sensitive over the use of genetics. Industrial size means that any problems could be on a larger scale, and the large number of GMOs increases the chance of release. The sterilisation of waste products is costly, but very important.

Guidelines for Contained Genetic Engineering

The European Parliament has approved a new directive to regulate both the contained use of GMOs and deliberate release. It is a useful example of simplified regulations and is one model. The points to be considered in the safety assessment include characteristics of parental and modified organisms, health considerations such as pathogenicity, and environmental considerations. There should be methods available for decontamination of areas in the event of accidental releases. There are three classes of operation, depending on the organism. The required information is laid out in the legislation in detailed form (EP 1989), and is shown in Table 9-1. Member states have eighteen months from April 1990 to implement the directives through national legislation.

Many regulations are written principally in terms of microorganisms, but they need to include plant and animal cell culture also. They must also include the laboratory growth of transgenic plants and animals, and define what is contained or not. The disposal of such organisms should be performed in a similar way to microorganisms. Larger organisms do have the advantage that they are not spread as easily as microorganisms, but pollen from plants or animal insects could be potential vectors.

Table 9-1: European Specifications for Contained GMOs (EP 1989)

Format is specifications
Containment Categories - 1; 2; 3;

Viable microorganisms must be in a system which physically separates the process from the environment
Yes; Yes; Yes;
Exhausted gases from the closed system should be treated so as to:
Minimise release; Prevent release; Prevent release;
Sample collection, addition of media and transfer operations to another closed system should be performed so as to:
Minimise release; Prevent release; Prevent release;
Bulk culture fluids should not be removed from the closed system unless the viable microorganisms have been:
Inactivated by validated means; Inactivated by valid chemical or physical means; Inactivated by valid chemical or physical means;
Seals should be designed so as to:
Minimise release
Prevent release; Prevent release;
Closed systems should be located within a controlled area
Optional; Optional; Yes, purpose built;
Biohazard signs should be posted
Optional; Yes; Yes;
Area should be restricted to nominated personnel only
Yes, work clothing; Yes; Yes, via airlock;
Personnel should wear protective clothing
Yes; Yes; A complete change;
Decontamination and washing facilities provided
Yes; Yes; Yes;
Personnel should shower before leaving the area
No; Optional; Yes;
Effluent from sinks and drains should be collected and inactivated before release
No; Optional; Yes;
The controlled area should be adequately ventilated to minimise contamination
Optional; Optional; Yes;
The contained area should be maintained at an air pressure negative to the atmosphere
No; Optional; Yes;
Input air and extract air to the controlled area should be HEPA filtered
No; Optional; Yes;
The controlled area should be designed to contain spillage of the entire contents of the closed system
No; Optional; Yes;
The controlled area should be sealable to permit fumigation
No; Optional; Yes;
It must be ensured that before final disposal, waste containing living organisms or biologically active nucleic acids are:
Optionally inactivated by validated means; Inactivated by valid chemical or physical means; Inactivated by validated physical means;

Field Release of GMOs

Biotechnology is advancing into areas that depend on the introduction of genetically modified organisms into the environment. At the time of writing there have been over 250 known experimental releases of GMOs (OECD 1990). There are many possible uses (Marx 1989a). The deliberate environmental introduction of any new organism, including GMOs, should be only undertaken within a framework that maintains appropriate safeguards for the protection of the environment and human health. Natural habitats already contain their own indigenous populations of organisms, organised in a delicate web of nature, which needs to be maintained.

There are several activists, such as Jeremy Rifkin, who are opposed to genetic manipulation and have tried to prevent all environmental release experiments. Their objections did mean that scientists have had to prove beyond reasonable doubt that their experiments are safe, but now that has been done, the experiments should go ahead so that we can learn from them. There have been a growing number of major reports on the release question, often with similar conclusions (NAS 1989, SCOPE 1987, OTA 1988b, Tiedje et al. 1989, HMG 1989b). There are serious ecological concerns, and ecologists stress that the organism should be evaluated and regulated according to their biological properties, such as their ability to tolerate various environments, rather than according to the method of manufacture. Not every new genetic possibility has ever been tried, so there are concerns regarding the introduction of new organisms.

The first experiments are being conducted in as closed environmental situations as possible. Our past experience with GMOs is principally in controlled situations, and we must be careful about shifting to new environments (Sharples 1987). The initial experiments on plants and animals are in enclosed research areas, but when they have proved safe and are economically useful, then they will need to be grown in large quantities. It may be feasible to use enclosed farms for some animals, especially such as fish or chicken farms, or even pigs or cattle (assuming battery farming is acceptable).

To be of a major practical use to worldwide agriculture, any GMO must be released into the environment. Only small scale agriculture can be conducted in closed environmental systems, though some important products used today are produced in that way, such as eggs from battery farming of chickens. There have been many field trials since 1984 when the Canadians field tested a transgenic plant. There is widespread public concern about the free release of recombinant organisms into the environment, and questions about how many organisms constitute a significant environmental release (Dixon 1985, Strauss 1987). The degree of care required depends on the potential, or known, risk to the ecological balance and humans.

For plants, to be economically significant they will need to be grown over large areas, so that will mean free release on a large scale. If the food is to be grown where it is especially needed in the developing countries, both the economic situation and the social structure of small-scale peasant farmers, will mean completely free release is required. Serious problems have arisen from the unexpected results of the movement of weeds and insect pests into new environments. Some were deliberately introduced as pests into new environments, some introduced to solve one problem, but caused another (Brill 1985).

Some previous releases of organisms into new environments have proved to be beneficial, or at least harmless, but others have been deleterious. The question of environmental release of GMOs is applicable to the release of bacteria, plants, animals and humans. The possibility of a novel organism or virus being widely dispersed, and/or disruption of the ecosystem, is the main fear, and in view of the dramatic consequences possible, it is very serious. It has been considered serious enough for several countries to ban any introduction for testing already. It is a rapidly increasing problem as the technology is very cheap and becoming well known, so that many people can try different modifications. In view of the important benefits that new organisms can offer innovation should not be discouraged. A few examples are given below. A reasonably comprehensive list is produced by the OECD on a database (OECD 1990).

Degradation of Pollutants

Bacteria can be made to degrade environmental pollutants. There are a range of bacteria available that can metabolise, or literally live off substances while removing them from the environment, including many important pollutants such as polychlorinated biphenyls, dioxin, herbicides, pesticides and oil spills (OTA 1988b). The first patents obtained for a microorganism involved the engineering of Pseudomonas strains to degrade chemicals found in oil spills. The genes enabling degradation of environmental pollutants can be introduced into different bacteria. For instance the ability to degrade toluene has been transferred into bacteria that can live a zero degrees Celsius (Lindow et al. 1989). Many of the bacteria used to supply the genes are actually discovered at hazardous waste sites, after naturally evolving to cope with the chemicals present. This pollution problem is quite serious, so this could be a major use. A bacteria that can reduce the concentration of trichloroethylene by a thousand times has recently been made (Winter et al. 1989). Trichloroethylene is one of the most significant environmental pollutants, and a suspected carcinogen (cancer causing agent). The use of bacteria to reduce the level of it in drinking water could be very important.

There are other bacteria that can be used to extract and concentrate heavy metal containments from places such as land fills, mine tailings, or low grade mineral ores. There are organisms that can extract cobalt, and research to produce mercury concentrating bacteria. One study found that heterogeneous microbial populations may be better than single species systems. The organic matter needed to act as a metal-chelating biomass can be reused, with autodigestion of excess biomass. Acclimation to high levels of one metal, e.g. cadmium, did not confer acclimation to another metal, e.g. copper (Lowe & Gaudy 1989). Genetic engineering will be needed to insert multiple genes to allow recovery of the mixture of toxicants at many sites. These metals are also valuable to recover, so the work has dual benefits. High value products, like precious metals, provided some incentive to the process, and there are commercial products available.

There are also organic pollution problems which may be treated by using bacteria, one of the most important being wastewater treatment schemes. It is also possible to use fungi for some of these applications. The fungi are attached to woodchips which can be mixed with the polluted soil, to metabolise pollutants.

Bacterial Mining

Mining operations have long benefitted from the activities of naturally occurring microbes. From 1,000 BC mine workers in the Mediterranean basin recovered the copper that was leached into mine drainage waters by bacteria. The Romans in the first century, and the Europeans in the sixteenth century used microbial leaching. However, recognition of bacterial leaching did not occur until the 1920's, and the bacteria Thiobacillus ferroxidans, which leaches metal sulphide ores, was not identified until 1947 (Woods & Rawlings 1989). Bacteria are used to extract 10-20% of the world's copper supply. In Canada, bacteria that leach uranium are also used (Lindow et al. 1989). By the year 2,000, the microbial metal industry is estimated to be worth US$ 90 billion. Bioleaching also has the potential to remove sulphur from fossil fuels, which would decrease acid rain. It is certainly more energy efficient and less polluting than smelting of low grade ores. Genetic manipulation is being investigated, but is at an early stage.

There has also been research into the use of microorganisms to enhance oil recovery. Compounds produced from microorganisms may be added to reservoirs so that crude oil viscosity is reduced and it is easy to displace. Also microorganisms themselves may be released. Some field trials have been conducted with the injection of Bacillus and Clostridium species together with fermentable brew, to produce carbon dioxide, methane or other gases. The carbon dioxide makes crude oil less viscous, and the gases in general can repressurise the reservoir for further pumping (Mantegazzini 1986).

A wide range of fungi can solubilise coal, to make a liquid product. The liquid product has potential as a chemical feedstock, and as a substrate for further microbial modification to a fuel gas (Wainwright 1990).

Nitrogen Fixation

The main type of microbial inoculant in agriculture is currently Rhizobium. Certain crops require inoculation because of the low local population of the specific Rhizobium species which are required for particular crops. There have been release experiments with R.leguminosanum with a plasmid containing a marker transposon determining resistance to the antibiotic neomycin. Root nodules have been traced for two years since release with no horizontal gene transfer observed (Hirsh 1989). The release of genetically altered nitrogen-fixing bacteria R.meliloti, designed to increase its nitrogen fixing capacity with alfalfa was another trial (Van Brunt 1987a). This may lead to higher plant growth rate, and less need for nitrogen fertiliser, saving money and alleviating nitrate pollution problems that arise from nitrogen fertiliser application.

While we have only relatively few examples of the release of GMOs into the environment, we do have much potential information on the dispersal of bacteria already. Many pathogenic bacteria are continuously released into the environment in sewage, and millions of hectares of land are inoculated with Rhizobium each year to improve the growth of leguminous crops (Behringer & Bale 1988). It is generally difficult to make predictions about the potential of a given organism to become established and to maintain high populations in a given environment.

There have been other trials of microorganisms with marker genes. In 1987 there was a release of Rhizobium containing antibiotic resistant genes in Bavaria, Germany. The purpose was to determine the survival of the bacteria (found to be very low), and to investigate gene transfer (Dickman 1987). There was major public protest over this experiment, which resulted in tough guidelines.

Ice Nucleation and Microorganisms

One famous test case in the United States concerned the experiment of Steven Lindow and co-workers in Berkeley, California, who asked the National Institutes of Health Recombinant DNA Advisory Committee (RAC) for permission to field-test a recombinant strain of Pseudomonas syringae (McCormick 1985, Van Brunt 1987a). The normal bacteria, P.syringae, is present on the leaves of many crops, and it leads to frost damage of leaves in mild frosts, because it catalyses the crystallisation of water to form ice at temperatures below -1.5C. Annual costs of frost damage to crops in the USA is US$ 1.6 billion. A strain of P.syringae was constructed that was incapable of initiating ice formation, until the temperature dropped to about -5C (Hirano 1985). In the absence of any ice nuclei pure water can supercool to -40C before freezing. Frost injury in plants is proportional to the logarithm of the population size of ice-nucleating bacteria on plants at the time of freezing. The genetically modified bacteria were sprayed onto plants to provide protection against frost, by replacing the unmodified bacteria of the same species. Field trials tested if the non-nucleating strain (Ice-) will replace the normal ice-nucleating strain (Ice+), and thus prevent frost injury to plants under field conditions. The technique is applicable to many plant species, and different bacteria are being engineered. Potentially this bio-control system could save much of the economic loss caused by frost damage.

There have been many protests to prevent this research, and the field trials were delayed for several years. The first experiments began in securely protected areas, the main concern over containment is the accidental release of the organisms caused by the protesters and vandals who are attacking the research sites. Both strawberry and potato plants were involved in the first release of Pseudomonas species in 1987 (marketed under the name "Frostban") (Van Brunt 1987a). By May 1989 there had been four independent field experiments to evaluate the dispersal, environmental fate, competitiveness, the chances of the bacteria spreading into the ecosystem, and the effectiveness of the bacteria. The data suggest that the there is an extremely small likelihood of any survival of these strains outside the area of use (Suslow 1989). In these experiments, previous laboratory studies of bacterial behaviour predicted the observed environmental behaviour. Various pretests were made to determine the spread and survival of the bacteria (Lindow et al. 1988). No significant difference in the ecological behaviour of the new strains was anticipated, or found. They grew at the same growth rate both in the laboratory or when co-inoculated on plants. They were tested on 67 different species of plant in the laboratory (Lindow & Panopoulos 1988).

There are several other possible uses of bacteria with altered ice nucleation. Ice+ bacteria could be used in the water sprayers used for artificial snow-making, and might lower costs. The bacteria could be engineered to self-destruct after use. Frozen food often contains many small ice crystals, which during storage conditions, increase crystal size. It is desirable to stabilise such ice crystals, and keep their size small. It would be possible to use viable food grade microorganisms to provide the ice nuclei. There are also research uses (Warren 1988). These bacteria are important in precipitation of rain fall, but even with the large scale industrial use of them there is a negligible risk of any influence on this (OTA 1988b).


There has been a recombinant Hepatitis B vaccine approved for use on human beings in most countries, since 1986. There have also been small clinical trials of other human vaccines. In diseases with costly cures, or no cure, there are often compelling ethical arguments to test vaccines in volunteers. There is research on developing vaccines for Hepatitis A, dengue haemorrhagic fever, leprosy, leishmaniasis and respiratory synctial virus, as a few examples (Bloom 1989).

Vaccines against animal disease also come in this category. Recombinant virus-based vaccines have some advantages for controlling disease (Fields et al. 1986). A single vaccine can express the antigenic determinants for more than one infectious agent, reducing the costs of administering the vaccines. Inactivated vaccines are prepared by the inactivation (e.g. with formalin) of large quantities of the virus. Attenuated vaccines are made by growing the virulent organism in an unnatural host, or under conditions such that the product will then proliferate in the natural host without causing the disease (Brown 1989). Recombinant vaccines should be safer than the attenuated vaccines as only a portion of the pathogen is expressed, so there is no danger of the virus reverting to a virulent form (Finkelstein & Silva 1989). Some safe attenuated viruses can be used for vectors for the antigens of chosen diseases. The new regulations will mean that new live vaccines are likely to be safer than the old vaccines obtained by attenuation of pathogenic viruses that were brought into use on a trial and error basis. They will have to satisfy much more stringent safety criteria before being accepted.

There was an unauthorised test of a vaccine, in an attempt to vaccinate Elm trees against Dutch Elm disease. It was conducted at Montana State University, and involved inoculation of 14 trees with a genetically engineered bacteria (Pseudomonas syringae) designed to fight the fungus that causes Dutch Elm disease. The bacteria was the product of mating a recombinant DNA modified bacteria with a strain that was not. Technically, the end product was called nonrecombinant, under the RAC rules. The researcher notified authorities, but did not wait for approval. Shortly after the trial was rejected, the trees were cut down and destroyed by the researcher. The bacteria had prevented appearance of the disease up to that stage (OTA 1988b). The real danger was the deliberate release of Dutch Elm disease, a very harmful disease, rather than the novel bacteria.

There have been several trials involving recombinant vaccines. Rabies is important worldwide, and in Europe about 1.3 million foxes are killed annually in attempts to control it, and there are 1 to 4 human deaths. Several vaccines have been successful using vaccinia virus as vectors (Blancou et al. 1986). A large scale area (up to 435 square km) field trial of recombinant rabies vaccines occured in Belgium (OECD 1990). The vaccine was made using a vaccinia virus expressing the glycoprotein antigen of rabies virus (Bloom 1989). The results of these field trials has been a steady decline in the incidence of rabies in wild and domestic animals in the area of the field trials in Belgium. Recently this vaccine has been introduced in Belgium, and part of France, in bait to protect foxes against rabies. There has also been a field trial involving rabies virus approved in the USA, but there have been long delays with local officials about the proposed test site. There are several orally administered rabies vaccines being developed.

There are plans by the International Commission of Epizootics to immunise hundreds of millions of cattle with recombinant vaccinia expressing rinderpest antigens. There is currently an epidemic in West Africa of the cattle disease, rinderpest. This is a huge experiment, and there is bound to be adventitious infection of humans with the recombinant vaccine. Humans should be pre-immunised against vaccinia virus before this trial begins (Bloom 1989).

There has been testing of attenuated strains of Salmonella typhimurium as vaccines in Australia. Salmonella is important in sheep and cattle and poultry industries and is also a human health risk. There was initially prerelease testing in sheep in pens by the CSIRO. The vaccine proved effective in these trials (Davidson 1990). Tested were made to detect the bacteria in the feedlot, and also during shipping (since the sheep will be exported). There is currently a test involving less than 100 sheep underway (OECD 1990), and the project is expanding.

A rare fungus, sterile red fungus, found in Australia appears to protect plant roots that it grows in, against disease. The fungus makes the plants grow up to 40% faster, and infects most plants. It has much potential (Wood 1990). This fungus is not a GMO, but should still be considered carefully prior to free release into new ecosystems.

There are some human recombinant DNA vaccines being used. There is a recombinant AIDS vaccine being tested on twenty patients in London's Hammersmith Hospital in 1990 (Cherfas 1990). This vaccine may also be effective as a form of immune therapy, and uses virus-like particles which are like viruses but contain no viral nucleic acid. They are good delivery vehicles, and the trial will also test their efficiency. There are many groups around the world working on similar goals. The vaccine for Hepatitis B has been approved for use in most countries. Given the importance of human disease, if a vaccine is efficient, it is used widely.

Biological Pesticides

There have been many species introduced into the environment as forms of pest or weed control. Many insect parasites have been introduced, and none have been shown to have become pests themselves. The history of such releases is at least 100 years old. The strategies may rely on long term survival of the new species, as is the case with the grasshopper parasite Nosena locustae. It is being used to combat crop damage by grasshoppers, which cause US$ 400 million worth of damage annually in the USA alone (Cramer 1989).

In Britain there have been several field trials of GMOs. One series has involved baculovirus insecticides, which are viruses that only infect and kill a few species of insect. They have no effect on other types of insect or other species, and do not pollute the environment. Naturally-occurring baculoviruses have been used during this century, and more than a dozen have been employed commercially. In a series of studies at Oxford, Bishop et al. (1988) have used many precautions in their approved field release experiments of genetically modified and genetically crippled viruses (they self destruct after a certain time period). The objective is to improve their speed of action, as they normally take several days to have an effect. This may be done by inserting toxin genes into the virus. The results of the experiments have been good. The viruses killed all the caterpillars attacking the cabbages. After killing them the virus self-destructed (both in the soil or in the caterpillars), ensuring no enduring and unpredictable side-effects. If normal viruses are used, the site can be successfully disinfected by formalin treatment. The gene for insect juvenile hormone esterase has been expressed in a baculovirus, causing feeding larvae to die (Hammock et al. 1990).

The bacteria Bacillus thuringiensis has been used for twenty years as a pest control, and is standard practice for the control of mosquitoes. The spores can be dispersed in a water mixture and distributed. There are over 400 registered formulations of B. thuringiensis in the USA. There are a number of insecticidal proteins produced, the two most important are the beta-exotoxin and the delta-endotoxin. The beta-exotoxin is effective against a wider range of insects, and has a greater soil persistence, but it may be mutagenic and it is not permitted to be used in the USA. In Europe it is used in the USSR only. The strain used to produce delta-endotoxin is usually B. thuringiensis var. kurstaki. The protoxins are laid down as crystals in the spores, and only after ingestion, when the protoxin is hydrolysed in the insect gut does the active protein form (Andrews et al. 1987). It damages the membranes of the cells lining the larval gut; the insects stop eating within minutes, and are paralysed, dying within two days (Barkay et al. 1989). There have been several insecticides released using B. thuringiensis insecticidal proteins. Many of the preparations contain live bacteria (Brown 1989). Mycogen has a product called M-One which contains a variety (var. San Diego) of the bacteria.

There is only one reported case of resistance to this protein, and it was in a limited situation. If resistance does appear then other strains can be substituted. However, B.thuringiensis does not persist well in the environment, and the spores are sensitive to UV light, so the endotoxin gene has been expressed in other bacteria, which are being field tested. This would reduce the number of applications needed. The B. thuringiensis insecticidal protein genes have been cloned into a variety of bacteria. One test case was the application by Monsanto to field-test a soil bacteria (Pseudomonas fluorescans) which has been engineered to produce this endotoxin. One species of root-colonising Pseudomonas with the toxin gene can be used to treat seeds to make them resistant to cutworm. The toxin gene has been put into an endophytic bacterium, Clavibacter xyli subspecies cynodontis, that colonizes the xylem of plants to make the plants resistant to Lepidoptera. A field trial is being conducted (Lindow et al. 1989). It has also been introduced into blue-green algae which might serve as food for mosquito larvae in ponds and streams. This is in addition to the insertions directly into the genome of higher plants.

One solution to the release of live genetically engineered bacteria is to use dead bacteria. In September 1987 the U.S. company Mycogen received U.S. patents for the invention of a process that kills bacteria while preserving their cell wall as a gelatin-like capsule which remains intact until the insect pests eat them, only then releasing the contents such as a pesticide. Almost any product could be put in the capsules. Several genetically engineered toxins are being field tested by Mycogen and Japanese companies. This invention allowed Mycogen to be the first U.S. company to receive permission from the EPA to field test genetically engineered bacterial pesticides, in 1985 (Gaertner & Kim 1988). These are alternatives to chemical pesticides which are very damaging to the environment. The live bacteria began field tests in 1987. The MCap biopesticide delivery system may remain active for 6-7 days, which is 2-5 times longer than other biopesticides which last 1-3 days. There are large scale field tests of this pesticide, called "Mycogen Vegetable Product" (MVP) underway during 1990. Several manufacturer's are already producing MVP (Watts 1989b), and it should be commercially available very soon. Live bacteria will be more useful on fast growing plants such as lettuce, as bacteria will grow with the plant avoiding the need for reapplication. The capsules do have advantages in that they are better for transporting, and a higher concentration of insecticidal protein can be made in laboratory bacteria, compared to bacteria which have to grow in the open environment and thus spend much energy on survival with less on producing the pesticide.

The World's first commercial pesticide based on a live genetically engineered organism went on sale in Australia in March 1989. It is called "NoGall", and it protects stone fruits, nuts and roses from Crown Gall disease, which causes worldwide annual losses of at least US$ 150 million (Wright 1989). The "pesticide" consists of a harmless strain of the disease causing bacteria, that will live on the same leaves as, and produce an antibiotic which kills, the disease-causing strain. The gene for this antibiotic is on a plasmid, which has been engineered so it should not transfer to disease-causing bacteria to make them resistant to the antibiotic. They had an eighteen month trial prior to the commercial release. There are still some opponents to its release in Australia, and there has been a call for a review of the release guidelines. The genetic change adds nothing new to the bacteria, so it might be accepted for release in other countries. If its only ecological relationship is to the disease-causing bacteria the potential negative consequences are minimised.

Transgenic Plant Field Trials

Initially trials of genetically modified plants can occur in closed greenhouses, however many plants respond differently to conditions in a greenhouse and in the field. It is difficult in the greenhouse to simulate natural diurnal and stress conditions, and not all the plants can be grown there. Only when growing under natural conditions is it possible to determine if the genetic trait is expressed, effective, and does not have detrimental effects. Agricultural plants have been the economically most important group of exotic introduced organisms in the past. There have been a few disruptions of natural ecosystems by introduced crops. One example is the spreading of bamboo from cultivation to cover large areas of the mountain sides of Caribbean Islands.

There have been many field trials of plants containing the Bacillus thuringiensis insecticidal protein. They have shown excellent insect control, suffering no damage next to parent plants that have been totally defoliated (Delannay et al. 1989). Early market opportunities for caterpillar resistance are leafy vegetable crops, cotton and corn. Crop targets for beetle resistance are potato and cotton.

Transgenic tomatoes carrying the tobacco mosaic virus coat protein gene have been shown to be highly resistant to virus infection, with no yield loss (Gasser & Fraley 1989). Transgenic potatoes with multiple viral resistance have been tested (Lawson et al. 1990). Many crops have been tested with resistance to viruses, another example is cantaloupe and squash with resistance to cucumber mosaic virus, papaya ringspot virus, watermelon mosaic virus and zucchini yellow mosaic virus. When we consider that the first transgenic plant was made in 1983, and there were field tests with genetically modified tomatoes and tobacco well underway in 1987, we realise how rapid progress has been. Tobacco is very easy to modify, that is why it is often use, rather than for the interests of tobacco companies. The progress is exponentially increasing.

Genetically-modified crops were tested in Europe more than anywhere else during 1989. The countries that have recorded trials of transgenic plants include Australia, Belgium, Canada, Finland, France, Ireland, Israel, Italy, Netherlands, New Zealand, Spain, Sweden, UK and USA. Plant Genetic Systems alone conducted 30 field tests in seven European countries. Their programme in 1988 involved 13 field trials, more than the total number in the USA that year. The reason for their success includes good cooperation with government agencies and independent research institutes. They have also tested crops in the USA. The Belgium-based company involves a research staff of around 100, but conducts field trials only in joint ventures (Whitmore 1989). There have been a number of field trials involving plants resistant to the Basta herbicide. There have been field trials involving genetically modified alfalfa (Medicago sativa), Brassica napus, Lycopersicum aesculentum, Oilseed rape, poplar (Populus), potato (Solanum teberosum), sugarbeet and tobacco (Nicotiana tabacum). There has been a variety of other herbicide tolerant genes tested in different plants. There have also been trials with inserted Bacillus thuringiensis insecticidal protein genes, in potato, tobacco, and tomato.

There are already many types of agriculturally important plants that have been grown with genetic modifications in field trials. The range of species is growing, and is limited principally by the ability to regenerate entire plants from genetically transformed cells. The list in mid 1990 included alfalfa, apple, Arabidopsis, asparagus, bananas, cabbage, cantaloupe, carrot, cauliflower, celery, corn, cotton, cucumber, Douglas fir, flax, horseradish, lettuce, lotus, Medicago varia, Morning Glory, Orchard grass, peas, pears, pepinos, petunias, pinetrees, poplar, potato, rape, rice, rye, soybean, squash, sugarbeet, sunflower, tobacco, tomato, trefoil, Vigna aconitifolia, walnut, white clover, and I'm sure I have probably missed some out!

Release of Animals

There has been some success in the production of faster growing fish. The work has been largely confined to fish in secure containment facilities. In mid-1989, a release system for genetically-engineered carp was approved by the U.S. Department of Agriculture (USDA) in the USA (Ezzell 1989). The carp contains trout growth hormone genes, and in the laboratory they grow significantly larger than normal carp. They were to be tested in a pond at a University in Alabama, with barriers to prevent escape to the open water. However, at the time of writing there are still delays. The USDA formal environmental assessment has been released publicly (USDA 1990), but critics of the release still oppose the outdoor pond experiments, arguing that carp is a "weedy" species and a potential risk to other fish if it escapes.

There have been other ways used to genetically alter fish. One technique is to use heat shock to induce triploid salmon (triploid means three copies of each chromosome, instead of the normal two copies). The result is that the salmon do not spawn, but continue to grow. There are plans to introduce them into Lake Michigan. They actually have no genetic affect on the existing salmon, as they cannot breed and pass on the abnormality. Triploid carp are being made for control of waterweeds (OTA 1988b). The advantage of these fish is that they are sterile and cannot breed, so that if there is any adverse environmental effect it may be controlled by the number of fish introduced to the ecosystem. These genetic techniques will supplement the other methods used for breeding fish.

One success has been the breeding of salmon in Norway. During the last thirteen years the salmon involved over three generations have increased size by nearly 40%. Aquaculture in 1985 produced 13% of the world's fish, and it is increasing rapidly, which may take some pressure off the falling stocks of wild fish.

A new method of controlling the blowfly is being tried in Australia. The blowfly costs sheep farmers about A$ 200 million annually. CSIRO scientists are releasing 700 million genetically manipulated blowflies during the 1990 and 1991 summers on an isolated island for field trials. A sex-linked gene translocation will cause sterility in the male line, and another mutation causes blindness in females (Ewing 1990). This type of approach is also being used in California in attempts to control Mediterranean fruit fly. Until now, insecticides have been used, but the new approach is to introduce up to 400 million sterile flies every week.

There have only been limited trials of transgenic farm animals so far, as there have not been any useful traits consistently expressed. Small closed experimental farm trials have been underway for several years in Australia and the U.K. Larger trials are expected soon, and will challenge regulators. While the animals themselves can be kept inside fences, their excrement will enter the soil, raising the improbable but not impossible prospect of gene transfer.

Persistence of GMOs

The only examples of intentional release of GMOs into the environment have been recent and are few in number. Since most GMOs only differ from the parent strain in one or a few genes, they will often behave in a similar way. It is relevant therefore to examine existing cases of the release of organisms into the environment.

Many GMOs will probably be less fit than the parent organism. Part of the reason is that they have extra DNA to carry, which will slow the reproductive rate - the excess baggage hypothesis (Lenski & Hguyen 1988). However, this is not necessarily so, and even if it was it may take many generations to pass before the introduced organism disappears due to decreased fitness if the turnover rate of populations is slow. In the first field test of an engineered soil bacterium, Pseudomonas fluorescans, with a chromosomally inserted lacZY gene marker, the population sizes of the introduced parent and transgenic organisms were identical for 30 weeks. The Pseudomonas bacteria was applied to wheat plants, and the furthermost migration of the bacteria was 18cm. No transfer was observed to drainage areas (Drahos et al. 1988). There have even been cases where competitiveness has been enhanced with the possession of foreign DNA (Tiedje et al. 1989). An example where the recombinant strain has shown enhanced survival in the environment, was the survival of Escherichia coli with a transposon ColE1::Tn5, in a farm environment (Marshall et al. 1988). A variety of responses are also seen in plants. Some support for the idea of the intrinsic weakness of artificially bred plant varieties has come from experience with modern crops, which are often incapable of survival without human intervention. However, some crops such as potatoes, can become a weed in following crops.

Natural selection will act on all organisms, including GMOs. Genetic novelty provides the raw material on which natural selection acts (Regal 1988). Selection after the release of the GMO will tend to increase the fitness of the GMO, by reducing costs associated with novel traits. If increases in fitness occur they will probably increase the population growth rate and biological competitiveness, as well as other ecological effects. These need to be considered in risk assessment. Myxoma virus released in the UK and Australia in the 1950's to control rabbit populations evolved to become less virulent. Likewise, with GMOs natural selection will act only in a way to increase the fitness of the GMO, taking no account of human intentions.

Released microorganisms are highly likely to enter freshwater or marine environments via agricultural run-off, or faecal contamination. There are various environmental factors that affect persistence, including moisture, pH, temperature, nutrient level, sunlight and other organisms (e.g. predators) (Beringer & Bale 1988). The case of Rhizobium inoculants has provided some data. It has been found that introduced strains may not compete successfully for nodule formation with indigenous populations of Rhizobium. When indigenous Rhizobium populations in soil are high it may be necessary to use a thousand-fold number of inoculants (Lindow et al. 1989). Very little is known about the biochemical determinants of competitiveness. If there is no indigenous population, then the inoculants can establish as a complete population.

In some cases the GMOs may be intended to persist at a particular level in the environment. If the GMO is going to enter a new ecological niche, it may be safe. Those organisms used for biocontrol may need to remain in the environment at a low level in the absence of a pest outbreak. However some GMOs may be required to die out after use, like microorganisms that breakdown particular toxic wastes for emergency use. There will be a minimum inoculant population required if persistence is desired. This is also a reason why the deliberate introduction of GMOs to the environment is different to the accidental laboratory introductions that have frequently occured during the course of recombinant DNA experiments in scientific research. It has been argued that given that bacteria can rapidly grow in the environment, even if we introduced a few accidentally natural selection would determine whether they survive or not (Davis 1987, 1989). However, there is a minimum inoculation population that is required in many instances, so we can not say it is always going to be the same as the case of accidental escape of a few organisms from a laboratory. The types of organism may have very different ecological niches, also.

GMOs should be designed for safety, in addition to their function. This may mean that they include some biological containment attributes to reduce their longevity in the ecological niche to which they are released or which alter their ability to transmit genetic material to other organisms encountered in the habitat. At the same time the GMO must persist long enough to perform its task for which several means have been suggested. Mutations can be introduced, such as those to abolish the ability to synthesise cyclic AMP and its receptor protein, as strains of enteric bacteria with these mutations have a 50% longer generation time. A gene such as hok which encodes a 52 amino acid polypeptide that mediates a lethal collapse of the transmembrane potential, that kills a certain proportion of each generation or which can be linked to a chemical inducer, has been tested (Curtiss 1988). A plasmid vector that causes the destruction of any recipient cell could also be made. There has been successful use of self-destructive baculoviruses as mentioned earlier (Bishop et al. 1988).

Detection and subsequent elimination of organisms may be feasible if they are large, such as large animals or plants, however, insects, microorganisms or viruses may be difficult to exterminate after introduction. The detection methods used all have limits, and it may not be possible to ensure that a microorganism is eliminated, as past experience has shown (Tiedje et al. 1989). The absence of an immediate negative effect does not ensure that the effect will not ever occur, it may take time.

Transfer of Genes

To some it may be comforting that interspecific gene exchange occurs naturally, but this can also be interpreted as an argument to anticipate the spread of engineered genes to members of the natural community. Transfer of engineered genes from the GMO to other organisms may occur through hybridisation in higher organisms, or through conjugation, transduction or transformation in microorganisms. If lateral transfer occurs, an engineered gene may persist in the natural environment after the GMO is no longer present. An important unknown question is how often lateral gene transfer occurs. It occurs at a higher rate in the presence of selection pressure in the environment (Levin 1988). Among microorganisms the scientific evidence suggests that it is neither so rare that we can ignore its occurrence, nor so common that we can assume that barriers crossed by modern biotechnology are comparable to those constantly crossed in nature. Current evolutionary thought does not consider that organisms are perfectly adapted to their environment, as there are important restraints for this to occur: namely, the presence of suitable genetic variants, which is what may be introduced in GMOs. Techniques are being developed to reduce the potential for gene transfer (OTA 1988b), but there is little known outside of experience with a few laboratory species.

There are four essentials of enduring lateral gene transfer, and each component is improbable (Levy & Miller 1989):

1) the origin of an unusual genotype (to the recipient)
2) its transmission to an extraspecific cell
3) a resulting advantage to the recipient cell
4) multiplication of the recombinant to numbers that render it safe from random loss.

This scheme has often occured over the long period of evolution. Genetic engineering does increase the chances of gene transfer occuring.

If we are generating novel genetic combinations, experimental observation of organism behaviour is important. One possibility is that bacterial viruses could acquire a capacity to infect higher organisms, which may upset the extremely intricate ecological balance. However, some bacteria commonly pick up human genes naturally. The genetic barriers thought to exist between species are in fact often broken (Berry 1986), especially by viruses. The argument against genetic manipulation when we cross species barriers is weak, however, as it has been found that these barriers are broken in nature. Nature has set barriers to horizontal gene transfer in eucaryotes, but trans-species gene exchange by a process called conjugation is common among procaryotes. Another piece of evidence is that plant breeders have used comparatively crude techniques for decades to transfer genes en masse from wild species into crop plants, without adverse consequences. Traditional breeding may move the desired traits together with hundreds of other genes into a new variety. There have not been any cases of detrimental gene transfer from released organisms in the past, so there is unlikely to be problems with single, precisely engineered gene alterations, unless it is an inherent property of the gene vector to move to other organisms under field conditions.

If the trait that is being transferred is already in the environment then we will have less concern regarding its transfer, especially if the particular engineered gene is present in the system that the trial is conducted in.

Antibiotic Resistance

Some opponents of free-release of GMOs claim that natural gene transfer between Gram positive and Gram negative bacteria was never observed before genetic engineering. However, there has been a very dramatic transfer of genes, apparently naturally, in recent times. The Gram positive Streptomyces produce aminoglycoside antibodies and resistance genes, but these have been transferred to a wide range of Gram negative organisms. This type of transfer has also been observed experimentally (Hodgson 1989a, Levy & Marshall 1988, Levy & Miller 1989).

There are other antibiotic resistance determinants which have been found to be widely dispersed among different bacteria, especially the transfer of genes in transposons. There have been several specific bacterial ecosystems well studied, and these include human and animal gastrointestinal tracts. Observable gene transfer of plasmids occurs at a low frequency in the absence of selective pressure, but at a greater rate with selective pressure (Levy & Marshall 1988). The actual rate of gene transfer may be constant, but the stability may be increased with selective pressure. Resistance has been observed to be transferred between different species in these ecosystems. In animals, gene transfer in certain types of bacteria occurs more readily at 28C than at 37C, in faeces, so the gene transfer occured in the external environment. New bacteria can then be subsequently acquired by animals by ingestion or inhalation. Transfer of resistance traits has also been observed between human and animal (chicken, calf and pig) ecosystems.

Gene transfer has not been well studied in soils, most studies involved the inoculation of sterile soils with bacteria and observation of the transfer of marker genes. Gene transfer has been observed, as it has in natural soils containing an indigenous competing flora. Conjugation was observed at levels 10-57 fold lower than in vitro matings. It can be very difficult to measure the behaviour of organisms in controlled systems. For example, only about 10% of soil bacteria can even be cultured in the laboratory (OTA 1988b). Gene transfer has also been observed in aquatic environments, including lake water, and in sewage. Gene transfer appears to occur in a wide range of environments, involving many kinds of bacterial hosts. If there is a high concentration of recombinant modified bacteria applied to a crop there may be an increased rate of natural gene exchange with pathogens which could lead to harmful consequences, so this should be especially monitored, but so far no problems have been reported, though it does mean carefully monitored field tests are obligatory.

To avoid the selection of new genes, it has been suggested that genes should not be linked to antibiotic resistance genes as these are steadily being selected for by the extensive use of antibiotics in man's natural environments (Levy & Marshall 1988). Kanamycin resistance markers are commonly used in the laboratory selection of transgenic plants, but, kanamycin, the antibiotic, is only used in the laboratory. Even if kanamycin resistance was transferred from plants to bacteria, it would not have any significant effect as long as kanamycin is not used for animal health treatment. The chances of the gene in the plant being transferred to bacteria is very low, and has yet to be demonstrated. However, even if the transfer from plant to bacteria were to occur at a frequency of 1 in a million (the frequency in bacteria of spontaneous mutation to kanamycin resistance) there would not be any significant effect. In soil samples, about one bacteria in 100,000 are already resistant to kanamycin (IFBC 1989).

Eucaryote/Procaryote Transfer

There has been little investigation of DNA transfer between procaryotes (bacteria, fungi, and viruses) and eucaryotes, but it was recently found that DNA can be transferred between bacteria and yeast, a eucaryote (Heinemann & Spragne 1989). It has been known that gene transmission occurs between the bacterium Agrobacterium and certain plant species, in a process mechanistically similar to bacterial conjugation. The experiments used tested both broad-host range plasmids (that can promote conjugation between Gram negative and Gram positive bacteria), and a limited range plasmid, the F plasmid of Escherichia coli. The results suggest that the plasmids could be transferred between bacteria and animal cells, and plant cells, even when the bacteria do not contain specialised tumour-inducing plasmids. This lateral gene transfer has been suggested to be an explanation for discrepancies in phylogenetic evolutionary trees. So it follows that GMOs may not be significantly different from the organisms that might arise by normal genetic exchange, they just increase the rate of genetic reorganisation. It would take a large number of transferred genes to change species characteristics dramatically. However, the possibility that bacteria can transfer genes to animal cells, needs to be investigated, with some strict precautions (Stachel & Zambryski 1989).

Crop plants vary greatly in their potential for hybridisation. At one extreme are crops which are maintained in cultivation entirely through vegetative propagation; at the other is alfalfa, an obligate outbreeder. We still do not know the origins of many crops, which originated by hybridisation. It is undesirable for crops to transfer some genes to their wild relatives, because the wild relatives could become competitive weeds.

Dispersal of Gene Vectors

Several systems have been devised for tracing the fate of genes in genetically modified bacteria in the environment (Barkey et al. 1989). The stability of different gene constructs will vary, even within the same bacterial species (Winstanley et al. 1989). Different systems used to trace microorganisms include luminescence-based techniques, with a plasmid containing the enzyme luciferase, so that luminometry can be used to trace the spread of organisms (Meiklejohn et al. 1989). Other systems are based on immunological detection of organisms (Wipat et al. 1989). These antibodies can be bound to magnetic polystyrene beads to concentrate bacteria. A recent trial monitorred the persistence of genetically engineered microorganisms using in vitro amplification of target DNA by the DNA polymerase chain reaction. The inserted DNA could be detected over a longer period than the detection of live genetically engineered microorganisms by plate samples. This technique has the advantage that it focuses on the gene itself, so even if the gene was transferred to other organisms it would be detected (Chaudhry et al. 1990).

The first spray releases of genetically engineered ice-nucleation deficient Pseudomonas syringae bacteria were monitored. The EPA Office of Research and Development designed a sampling procedure to determine the drift of the bacteria during aerosol application, and to determine the movement of GMOs. Less than 0.001% of the total viable cells released entered the aerosol cloud at the plant level, the rest were directly deposited onto the plants and soil. Of the bacteria that entered the aerosol spray, 8% drifted out of the plot into buffer zones. Between 20-35m away was the maximum drift. On subsequent days, depending on the wind movement, some resuspension from plant and soil surfaces occurred. Low numbers of bacteria were spread (Lindow & Panopoulos 1988). The best way to measure spread in terms of cost efficiency was gravity plates (open petri plates at chosen sites). Factors such as the number of bacteria applied and the wind, and equipment features such as nozzle size, have effects which are being measured to determine the minimum spreading procedure. A computer model is being developed by the EPA with the results (Seidler 1988).

The range of organisms affected by a virus may be altered in less direct ways. A laboratory study of a pathogenic plant virus showed that by altering a single gene, the range of insects that could carry it was changed. This would enable the virus to come into contact with previously unaffected plant species (HMG 1989b).

There needs to be experiments to determine the distance that pollen spreads (on insect vectors this can be over large distances), and how it hybridises. Plants often show little barriers to hybridisation, so there is always a potential for exotic pollination (Young 1989). Hybridisation has often been observed in sorghum and wild relatives, but there is still no evidence that traits such as resistance to insect pests have spread from traditional crops to wild relatives (HMG 1989b).

There has been little research conducted on cross-fertilisation of GMOs There has been a study performed at DSIR Crop Research, New Zealand. Transgenic potatoes resistant to chlorsulfuron were grown with a border of wild type potato plants to measure pollen dispersal. The frequency of transgenic seedlings among the progeny of wild-type potatoes growing within the trial was about 1%, but only 5 in 10,000 of the progeny from wild-type potatoes planted up to 4.5m from the trial, were transgenic. There were no transgenic progeny recovered from wild-type potatoes growing 4.5 to 10m from the trial (Tynan et al. 1990). This type of study is very useful for risk assessment, though in many countries it would not have been allowed for fear of much greater spreading of pollen. However, since it has been performed, it should allay the fear of long distance pollen dispersal in potatoes, providing that the transgenic potatoes are surrounded by a suitable buffer zone of wild-type plants.

Most of the traits bred into domesticated crops by traditional breeding, such as rapid seed germination, would be detrimental to weeds. Many of the new traits will be beneficial in weeds also, such as pest resistance. Even if there is a very limited opportunity for cross-breeding, with large numbers, it is virtually inevitable that gene transfer will occur. Crops often grow in the same region as wild relatives (Ellstrand 1988).

Intrapopulation studies are constrained by the bounds of the study site, so do not serve as adequate models for interpopulation movement. The isolation distances that plant breeders use to protect their pure stock vary widely between species. The distance depends on the breeding system of the crop. For example, the isolation distances of some crops are; lettuce 10m, common bean 45m, clover 90m, cotton 400m, sunflower and watermelon 800m (Frankel & Galun 1977). There is much research currently underway to attempt to quantify this transfer.

Examples of Gene Transfer During Field Trials

There have been several examples of gene transfer occurring between different organisms. An experiment was performed to measure the fate of genetically engineered bacteria in activated sludge microcosms. Pseudomonas species carrying genes for the degradation of substituted benzoates could survive and degrade the pollutant. There was not any adverse effect noticed on the microbial populations, as this species buffered the system from pollutants. Transfer of genes in chromosomal insertions was not observed, but transfer of genes on plasmids did occur. This did not have an adverse effect because it increased the total ability of the ecosystem to cope with the presence of toxic pollutants (Dwyer et al. 1988).

The Rural Advancement Fund International issued a communique in 1989 stating that a gene had been transferred during a field trial (Krimsky 1989). The test involved a firm called Crop Genetics International, which carried out a small field trial, involving the injection of microorganisms containing the Bacillus thuringiensis insecticidal protein gene into corn plants. The company disclosed to the EPA that the gene was found in flea beetles during the tests. The explanation would be that an insect feeding on the corn plant picked up the gene, and became a vector for transferring the gene to nontarget species. There needs to be careful monitoring of gene transfer.

There have been several trials of recombinant vaccinia viruses in cattle. The transfer between inoculated animals and others was monitored. One trial was conducted at the Wallaceville Research Centre, New Zealand, with governmental approval. The vaccine was against Sindbis virus, and was shown to be infective and immunogenic for calves. It was not transmitted to uninoculated pen contact calves, even by intradermal inoculation (Wedman 1988). A contrasting field trial was that of a vaccinia-rabies recombinant vaccine, which was tested in Argentina without knowledge of the authorities. The trial was stopped when the authorities discovered it. At this time, from the samples taken, the virus was shown to have passed from vaccinated animals to all the contact animals, and at least to one human being involved in the handling and milking of the animals. Other animals are being tested to determine the extent of spreading of the virus (Torre 1988).

The release of animals into the environment can have equal risks as with plants, as the genes may in rare instances be transferred to other species, and may be present in the environment in faeces. The examination has to include the type of genes transferred. For instance extra growth hormone genes would not be a serious risk even if introduced to other animals. The first field test involving genetically engineered fish is in a contained pond, with a series of barriers to prevent their escape to open water (Ezzell 1989). It assumes that no smaller organisms will transfer genes to the open water, though growth hormone genes are probably harmless...

Reducing Risks of Transfer

There are several types of barriers to reduce the transfer and stability of genetic information introduced into an ecosystem. Environmental barriers include the avoidance of contact, which depends on the concentration of the components of the gene transfer system in the ecosystem. In many systems, the concentration is adequate for both transduction and conjugation. Conjugal transfer is efficient between 15-37C, and has been observed at 6C. The observed rates of plasmid transfer from soil bacteria to Escherichia coli may be 1000 times faster at 28C than at 37C. Temperature and concentrations affect the gene transfer rate (Miller 1988).

The particulate matter may have a significant effect. Conjugal transfer of many plasmids is more efficient on solid surfaces than in liquids devoid of particulate matter. There are differences observed between the types of surfaces (Miller 1988). There are different ways to stabilise recombinant plasmids. They can be joined with a DNA segment that produces a partition function, surplus DNA can be removed from the plasmid, the use of transposable elements can be avoided, as well as other factors (Imanaka 1986). More research is needed to improve stability, which will reduce the risks of transfer. Organisms have entry barriers to prevent exogenous genetic material entering the cell, and some plasmids have a limited host range. However, these barriers will not halt genetic exchange. There are also barriers to the stable integration of exogenous genes in the recipient cells, such as restriction endonucleases, and incompatibility of the system. Homologous recombination could probably occur. There are further barriers to gene expression if the gene is incorporated, though the gene transfer may still have opened up further routes for subsequent gene transfer. Many of these factors need further research.

Methods are being developed to manipulate intrinsic factors so that introduced organisms have a lower probability of transferring genetic material. The genes required for transfer, such as conjugation, can be removed from the plasmid used for the gene insertion. Disabled plasmids that are not capable of detectable gene transfer can also be made. It is also possible to use self-destructive GMOs, as previously described.

ES cell lines make the creation of transgenic animals easier, and give much more control over the exact genetic change transferred to the animals. If we are concerned about the targeting of gene changes then they have definite advantages. A consequence of this is that concerns about the safety of releasing transgenic animals into the environment, because of some unforseen low probability genetic event such as viral release, would be lessened by using animals that had only single copy insertions, the type of control which would currently require using ES cell lines. It is possible to determine the nature of genetic insertions in animals in the laboratory prior to field release, however, this is not always done and is impractical. In light of the weight given by many to fears of unsafe release, single copy insertions are desirable. The genes can be manipulated in their natural chromosomal environments, whereas the use of conventional methods for introducing DNA sequences into the germ line allows little control over the chromosomal site of integration and the number of integrated copies.

Potential Ecological Effects

Although it may not be considered by many protest groups, the most likely potential ecological effect of the release of GMOs will be to protect the environment from many harmful chemical pollutants. Overall these new genetic technologies promise much to aid world agricultural techniques. If plants were made to use fertiliser more efficiently it would mean less fertiliser would run off into rivers causing pollution, and if they were made disease resistant then less problems would arise from the poisoning of the environment by herbicides. They are cheaper and should help to solve the pollution problems caused by the current fertilisers, herbicides and insecticides.

In some cases where the ecosystem is changing due to climatic conditions, GMOs are being developed to maintain a species presence in the area, especially with the modification of trees, which may be required by many dependent insect and bird species. Rather than worrying about the inherent nature of any inter-species interchange of the genes, the major concern is safety of the GMOs and their genes on a case-by-case level to the ecosystem. Understanding ecological interactions is crucial to the planned introduction of GMOs. Most GMOs will pose minimal ecological risk. In addition to any ecological effects, we must be careful not to squander valuable genetic resources, such as important insecticidal proteins, which insects could develop resistance to, rendering some important existing uses ineffective.

Most planned introductions are likely to be agricultural, so the negative consequences probably would involve an agricultural problem. We could use a model from the affects of traditionally bred plants and domestic animals. New varieties have been introduced in the past, just as the GMOs will be. If successful the crop will be grown over a large area. Many of the traditionally bred genetic variants are chosen because they have disease or pest resistance, so they have similarities to the GMOs. If the GMO release had an impact on the natural community, the consequence would probably be a transitory disturbance of the community structure. There are, however, some examples in nature where the acquisition of a single gene can cause ecological problems, such as antibiotic resistance genes that have been acquired by bacteria, in the gastrointestinal tracts of animals.

From past mistakes we should learn the need to be cautious when applying new technology and introducing new organisms. About 10% of the eExotic species introduced into Britain have become established in the wild, and about 1% have become pests (Williamson & Brown 1986). Biological control has risks, and a dramatic illustration of this in Australia was the introduction of cane toads in Queensland to curb the sugar-cane beetle. Now, both the toad and the beetle are problems (Campbell 1989). New Zealand has several examples. About 80% of New Zealand weeds are introduced species, as are over 60% of its insect pests. However, while this is the total, the results of planned introductions of biological control agents is very good. Since 1874, 225 biocontrol agents have been released on 70 target species and there is no proof that any have been harmful. What this shows is that with careful evaluation and prerelease testing, new species can be safely introduced to New Zealand ecosystems. Nevertheless, we must be careful because isolated islands are more susceptible than continental areas to new species introductions (Hatchwell 1989). The only way to remove a gene that has integrated into the genome of many organisms, causing a detrimental effect, may be to attempt to destroy the organisms.

The distinction between a crop plant and a weed is very narrow. Of the 18 worst weeds in the world, only one is not used somewhere for an agricultural purpose (Williamson 1988). There are weed oats, barleys, rices, maizes, carrots etc. A similar situation exists with domestic versus wild animals that are pests. The lesson is that we must consider releases of new or GMOs carefully, taking into consideration the factors presented in Table 9-2.

The alternative which is probably more acceptable is to put into the environment only genes that we can tolerate spreading to other species. For instance if the gene for resistance to one herbicide is spread to weeds, then another could be used. One could even imagine the possible industrial sabotage of companies by the introduction of weeds that are resistant to their competitor's herbicides. There are several methods by which weeds could arise from the introduction of GMOs. One is through hybridisation with local weeds, which was mentioned previously, and which could be the most serious. There is also the low possibility that the GMO could evolve into a weed itself (Keeler 1989). This sort of risk is probably worth the large economic savings and increased crop production that is possible using herbicide resistant plants.

Bacteria can grow very rapidly, so there could be more rapid spreading of new genes among bacteria, such as in the many species found in contact with ground water. However, with alterations such as the removal of ice-nucleation proteins, serious reasons for concern have not been found so far from the recent results of preliminary field trials.

New pests could arise, as for instance in the escape of a salt-tolerant rice cultivar from cultivated fields to estuaries (Tiedje et al. 1989). It is possible for both native or exotic species with new traits to become pests. Sometimes the change in environment means that plants may become major weed problems. For example kudzu in America, where the climate is different, and it has a small competitive advantage, has over several generations spread to become a weed. Release of a genetically modified cultivar in regions which that cultivar has many wild relatives could see hybridisation between the cultivar and wild relatives. In Africa, sorghum cereal crops have hybridised with weedy relatives to produce a serious pest known as "shattercane" which can mimic the crop (Hauptli et al. 1985).

Enhancement of the effects of existing pests could occur if they hybridised with GMOs. Weeds could acquire herbicide resistance, or insect resistance, or other advantageous traits. Nontarget species might be harmed, such as insects other than the insect pests. The genes in farm animals might be transmitted via faeces. There has been a case with a recently applied antiparasitic drug, ivermectin, which is currently applied widely to cattle, horses, sheep and pigs in many countries. It has been found to have an insecticidal effect on the larvae of a fly which normally helps break down the dung. Most of the applied drug is quickly passed in the faeces, and the high concentration of the drug kills the insect. As a consequence, dung was not decomposed quickly enough, and it killed the pasture where it accumulated (Wall & Strong 1987).

One frequently expressed concern is the potential for GMOs to displace resident species in the receiving community, particularly microbial species performing key functional roles such as nitrogen fixation or lignin decomposition. Because redundancy of function appears to be common in microbial communities, in many cases there would be little concern over microbial species replacement caused by an introduced GMO. This effect would have the result of altering the diversity of species in the environment. However, an example of this type of displacement was the introduction of the highly competitive nitrogen-fixing bacterium, Bradyrhizobium serogroup 123, into fields, which made it more difficult to introduce the more efficient but less competitive rhizobia (Moawad et al. 1984). In this case the result was harmful. Bacteria, such as Rhizobium are being manipulated to enhance nitrogen fixation in the soil. The OTA has had a study conducted to investigate whether there could be consequences for the nitrogen cycle. Their conclusion was that the chances of adverse consequences were very remote, and that normal crop rotation could produce greater changes than microbial inoculations to the patterns of nitrogen distribution (OTA 1988b).

There are many other related scenarios. The worst possible ecological impact would be to disrupt a fundamental ecosystem process, such as the cycling of a mineral or nutrient, but that seems very unlikely. One scenario that is talked of is if organisms containing the cellulase genes (which break down cellulose, a major component of wood) were released the gene could transfer to other organisms and break down trees. However, these genes are already common in the environment, as one part of the carbon cycle, but living trees are not decomposed (HMG 1989b).

There are several models for ecological relationships, but all have limits (Tiedje et al. 1989). The disruption could be on the biotic community, or on ecosystem processes. The organisms need not totally replace another organism to disrupt the ecosystem: they can coexist. The overall record of little hazards stemming from the release of products of traditional agricultural breeding does not mean it is safe to proceed with release of GMOs. Some insight can be gained from the introduction of nonnative species. Although negative effects have not been seen from the experience with GMOs accidentally released from laboratories (Davies 1987), it does not mean that this is a valid model for release of GMOs. The laboratory itself may not be a reliable environment for modelling the larger environment, as shown with past experience.

Researchers must proceed with caution in tackling some problems, such as pollution. Incomplete degradation of hazardous chemicals can lead to the production of even more toxic by-products. The microbial degradation of trichloroethylene and tetrachloroethylene produces the more toxic vinyl chloride (Tiedje et al. 1989).

Resistance to Introduced Genes

The possibility of organisms developing resistance to GMOs is very real, and can be expected. The probability of it occuring will vary, as will the significance. We should plan strategies based on the assumption that resistance will develop, at some stage, in some trial or commercial application, and have a followup strategy.

A study in Southern Italy has shown how there have been some mutations induced by the vaccine to enable Hepatitis B virus to infect children that had been vaccinated. A total of 44 people, including infants of carrier mothers, became HBsAg positive, that is positive to hepititis B surface antigen, despite passive and active immunisation against it. In one infant this led to serious disease. The virus can mutate so that it loses the determinant that the vaccine binds to, so the remainder of the virus can infect the patient (Carman et al. 1990).

There is genetic variation in the herbicide resistance that has been found to occur in nature. Multiple herbicide resistance has arisen in the wheat weed Alopecurus myosuroides after selection due to treatment with a single compound. The mechanism of resistance is not understood which should make us very cautious in the introduction of large amounts of herbicide-tolerant plants (Oxtoby & Hughes 1990). In the laboratory it is possible to isolate mutant varieties of corn or tobacco or other plants resistant to sulfonylureas and imidazolinones, often the result of a single dominant mutation. Careful management is required, and broad farming strategies should be examined.

Organisms that have new phenotypes can provide selective pressure on other organisms. The evolution of resistance to the Bacillus thuringiensis insecticidal protein following the introduction of these crops has been modelled. The model suggested that the presence of the insecticidal protein will be a powerful selection pressure, which could result in acquisition of resistance (Gould 1988). There is also one situation where insect pests have evolved resistance to the insecticidal protein, when seeds stored in a silo were dusted with the spores (McGaughey 1985). This has been observed several times (OTA 1988b).

Resistance to the endotoxin has never been observed in nature, despite its continued use for many years. The spores are sensitive to UV light and degrade after several days, releasing the protoxin into the environment where it is broken down. However, if it is continually present, as it would be in transgenic plants, insects have an increased chance of becoming resistant. This has been achieved in laboratory selection experiments using Indian meal moth, where the lethal dose can be increased by 250 times compared to parent strains. Lower levels of resistance have been obtained in almond moth and tobacco budworm. The mechanism of resistance has been found to be due to alteration in toxin-membrane binding (Van Rie 1990). If this did occur it would reduce the efficacy of new resistant crops, and the efficacy of current uses of the toxin, and it could also change the role that the insecticidal protein plays in the natural ecosystem.

Modelling studies have shown that the time needed to develop resistance to the toxin could be significantly extended if some unprotected crops were planted in the mixture, as this increase in genetic variation is an important factor. This is an example of efforts directed at using the toxin genes in a way so as to reduce the rate that pests evolve resistance. This would limit the selection pressure on insects for the development of resistance. The probability of pests becoming resistant could be more substantially reduced if several genes were simultaneously used in the same plant. Combinations of insecticidal protein and protease inhibitors could be used. In a similar way, several methods of viral protection could be used to avoid viruses overcoming plant resistance. The combinations could be changed each year to ensure the pests were not able to overcome such methods, and the techniques of genetic engineering will allow this rapid breeding. During the last few years, 60% less insecticides have been used in US cotton fields in attempts to manage the selection of pests resistant to chemicals. Resistance to one variety of toxin can be countered by application of a different variety of B. thuringiensis with a variant toxin. The conclusion is that we need to employ good ecological management skills in addition to the new technology.

Optimal Procedure

In conclusion we could suggest that there are several major sequential steps that should be followed in generation of GMOs that are to be released into the environment. These include:

1) Choice of the useful gene or trait, and a suitable target organism
2) Well designed genetic alteration and expression
3) Laboratory and/or greenhouse studies
4) Small-scale field tests with extensive monitoring for gene transfer and any ecological effects, over a variety of climates and habitats
5) Step by step addition of new laboratory tested (steps 1-4) characters in each trial
6) Commercial scale release, with monitoring

Please send comments to Email < >.

To Shaping Genes contents
To Shaping Genes chapter 9
To Eubios Ethics Institute home page