Releasing genetically modified organisms: will any harm outweigh any advantage? *


  • *

    The text of the Eighth BES Lecture delivered on 21 December 1999 at the University of Leeds during the Winter and Annual General Meeting.


1.  The public debate about genetically modified organisms has concentrated largely on concerns about food safety and potential risks to the environment. In both cases there appears to be an assumption that existing crops and animals are safe. I discuss the experience we have to date from traditional methods and conclude that most concerns about environmental harm are more relevant to existing crops.

2.  The flow of genes among species, and even within different genera, is discussed with due attention being paid to the need for inherited genes to confer a selective advantage on hosts.

3.  A reason why so many people are critical of intensive agriculture and biotechnology is that virtually all changes in agricultural practice have an adverse impact on wildlife, particularly when such change leads to increased intensification. The problem of deciding how to manage agriculture to ensure that we maintain or enhance species diversity of wild plants and animals is discussed against the background that most of the UK environment is the result of human intervention.

4.  Nature and dense human populations cannot coexist without the former suffering. Our objective should be to develop and exploit our understanding of ecology to provide the information required to enable us to develop a far more enlightened future for agriculture and wildlife.


The answer to the question posed in the title to this review is no, of course not. Why are we so lacking in imagination that we can only see perceived problems and few benefits? Advantages will outweigh disadvantages because we have a technology that allows the precise modification of species and, as our understanding of genes and gene expression expands, will enable almost any imaginable change to be made to our crops and livestock. However, it is quite clear that even though potential benefits are limited largely by our imagination, the realization of them without continued concerns about environmental harm will depend upon a much better understanding of factors that affect the ability of released plants and animals to survive and spread in the environment. Our limited understanding of the factors driving environmental change, and a widespread lack of knowledge of the environmental risks associated with traditional plant breeding, bedevil the present genetically modified (GM) debate. We see possibilities for environmental harm because we are looking at new types of organism without making the appropriate comparisons. I know of no example of a GM crop for which environmental harm greater than for equivalent traditionally bred crops has been demonstrated.

This brief review provides comment from a microbial geneticist on some of the issues surrounding the current debate about the potential for genetically modified organisms (GMOs) to cause harm to the environment. It is not exhaustive, nor is it intended to ‘sell’ the advantages of GMOs. I at best paint part of a picture the final appearance of which is far from clear. In doing so, I hope to facilitate debate that will eventually lead to a clear view of how we should take the technology forward in this new millennium.

Now that gene cloning is a routine laboratory procedure and has reached the marketplace, we should recognize that it is a technology that is here to stay, regardless of presently very eloquently articulated attempts to stop its use in agriculture. I believe that, after an inevitable lag due to present concerns, there will be a dramatic expansion in the production and release of GM plants, microbes and animals. Before the expansion occurs there is an absolute need for the public to have a much better understanding of the risks involved in proceeding, and the risks associated with not doing so (Taverne 1999). For this to happen, politicians and the public need to be more aware of the reasons for existing declines in wildlife, and we need to have robust policies for ensuring that we achieve the environment we desire in the future. Whether or not these policies will require GM crops, or will merely be compatible with their use, is yet to be determined. It is always worth remembering that in a democracy people are quick to demand that environmentally sound policies are adopted, but are usually very reluctant to pay financially, or by changing their behaviour, to ensure that such policies are realized.

What is genetic modification?

At its simplest this is a series of techniques that allow us to move genes among different species and to modify genes so that their expression occurs when and how we desire. Genetic modification is possible because evolution appears to have found only one mechanism for conveying instructions for the production and existence of organisms, namely the use of nucleic acids to encode the instructions. Not only are nucleic acids ubiquitous, but also all organisms use essentially the same code and share a high proportion of the same genes. Having a single code means that moving a gene from one organism to another is simplified because the ability of the new host to understand the information coded by the gene is conceptually not greatly different from an American speaking to an English person. For distant transfers, such as from bacteria to plants, it is usually necessary to change the regulatory region in front of the gene to ensure that it functions (speaks); once it is expressed it will usually be understood. Quite how well a gene functions in a new host is almost as variable as the ease of communication among different speakers of the same language.

While our ability to move genes among species is now relatively well established, our ability to modify the expression of genes is not as well advanced. In most of the crops commercialized to date, the cloned genes are modified so that they are expressed in most cells under most cultural conditions. This is because the regulatory regions available for molecular geneticists to use have been limited. As our understanding of genomes and gene functioning advances, new regulatory sequences will become available that will allow much more tissue-specific or temporal gene expression. After all, we would not be as we are if our genes were not so well regulated that our cells knew when to start and stop division to form the right amount of tissue in the correct place and orientation.

Genes do not belong to an organism, they are at best part of a family of genes that may be very widespread. Genes coding for haemoglobin synthesis provide a good example. They are found in plants and animals, and are extremely similar. In general, moving genes between species, and even families, is not perceived by many molecular geneticists to be contravening any ‘natural laws’ or ‘divine plan’. Here we have a fundamental problem of comprehension and opinion that has not been properly debated, and must lie at the heart of concerns about GM crops and food derived from them. For many scientists the concept that genes have unique identities derived from the organisms from which they have been isolated makes little sense. But it is quite clear that for many people a component of their sense of being human is the uniqueness of their make-up and that of other organisms.

From the point of view of the fitness of organisms for particular environments, it is the combination of genes that have come together during natural selection that determines their ability to flourish. Thus, when we import birch trees from continental Europe we introduce combinations of genes that have evolved together for different environments. Crosses with native tress will necessarily change the genetic make-up of the progeny. How much such change will affect the fitness of the progeny for a particular environment cannot be predicted. One might assume that the greater the difference between the environments of the two parents the lower would be the chance of the progeny being well adapted for either environment, although it is also possible that such crosses could lead to progeny with hybrid vigour. It may not be until the next ice age that the true effect of mixing birch gene pools becomes relevant; or chance recombination might already have led to very weedy progeny that will become an environmental nuisance.

A common concern about the use of GM crops is that they will give rise to combinations of genes that could not arise naturally, and that such combinations might confer extra selective advantages, or disadvantages, on progeny derived from crosses between the crops and wild relatives. Our inability to predict with certainty what might happen when such plants are released into the environment strengthens the hand of those who wish to stop all uses of gene cloning in plant breeding. However, before we should agree that a new technology is necessarily going to cause problems, it is sensible to assess the risks posed by the technologies presently in use for breeding new varieties of crop plants.

It is commonly assumed that the genomes of existing species are unique and do not undergo change, but this is totally wrong. Chromosomes are subject to change due to normal mutations, damage during chromosome replication, and the recombination that occurs during natural crossing within species. If this was not the case there would be no evolution and we would not see the chromosomal changes that have arisen from the migration of parts of chromosomes to other chromosomes. From the very beginning of the domestication of plants and animals, our distant ancestors have kept alive rare variants that have undergone chromosomal changes, or have even inherited parts of chromosomes from related species. They have then bred such variants with each other (initially by chance and more recently purposefully) to produce combinations of genes and changes in metabolic pathways that would not have arisen naturally. In recent years plant breeders have used distant relatives of crop plants from parts of the world in which the parents of the crop were not present to introduce unknown numbers of genes, some of which may not have been present in our crops. We thus have a history of production of novel combinations of genes, and might assume that if novel combinations are likely to pose serious problems for the environment we would have seen them. While I would argue that for crop plants such adverse effects have yet to be demonstrated, the other form of introduction of novel gene arrangements that we have encouraged, namely the importation of non-indigenous species for our gardens, is well known to have led to problems of environmental harm (Williamson & Brown 1986). In making this comparison I do not suggest that Williamson & Brown's observation that about 1% of introduced species have become problems would be reflected in the release of GM organisms. Introduced species are very different in that they contain a full complement of at least 50 000 genes that have evolved together to produce organisms that are very well adapted for certain environments. There is every reason to believe that they will become established and compete with indigenous species if they find similar environments in the country into which they have been introduced. Domestic animals and crop plants, on the other hand, generally have low environmental fitness and have only been genetically modified by the introduction of a few genes, usually for characteristics that are very unlikely to affect environmental fitness.

The very strong selection for better variants of plants and animals that has been conducted since the time that animals and plants were first domesticated has weakened the domesticated species involved, such that most crop plants and domestic animals have very poor abilities to survive without human intervention. While from a species’ point of view this is unfortunate if human intervention ceases, from the point of view of the environmentally safe use of animals and plants by humans it is a great benefit. It is presumably the reason why crop breeding has not produced plants whose invasiveness of uncultivated environments in the UK has been remotely as serious as that from rhododendron species.

The very negative nature of the debate about GM crops has largely ignored the possibly negative effects that the spread of genes from traditionally bred crops to wild relatives might be having on wild populations. Whether a crop is genetically modified or not, genes will flow from it to wild relatives (Hails 1999). In the absence of evidence to the contrary we must assume that the spread of genes that affect environmental fitness is not a great problem. Two major UK crops have such relatives, oilseed rape Brassica napus and sugar beet Beta vulgaris. If we were concerned about the impact of gene flow from crops on the survival of their wild relatives would we seriously consider preventing the growth of traditionally bred crops to protect them?

Promiscuous gene transfer

A very hotly debated topic over the last few years has been the risks associated with the possible transfer of genes from released GMOs to other species and even families of organisms. The debate has been very polarized because objectors to the use of gene cloning in agriculture have chosen to assume ‘worst possible cases’ as the expectation, rather than possibilities that must be considered. There is ample evidence to show that genes can indeed be transferred widely among micro-organisms (Fry & Day 1990) and from bacteria to plants (Fraley, Rogers & Horsch 1986; Buchanan-Wollaston, Passiatore & Cannon 1987). Recently evidence has also been published to demonstrate the integration of viral DNA into the nuclei of animal cells (Doerfler et al. 1998). Very good evidence that intergeneric transfer of DNA has occurred during evolution comes from various phylogenetic studies (Doolittle 1999), and genome sequencing is showing the presence of regions of different base composition that could only have arisen through horizontal gene transfer (Andersson & Kurland 1998).

While sequence information is both interesting and informative, it can only show us the horizontal gene transfers that have occurred during the past few million years that have resulted in the stable inheritance of the DNA sequences concerned. For the movement of genes from GMOs to be a concern to us it is important to consider how likely it is that particular genes would be inherited after horizontal gene transfer.


Sufficient biological barriers to cross-pollination are in place to ensure that the possibilities of intergeneric gene transfer are extremely limited. Work on interspecific transfer confirms that while it is common, the frequencies at which vigorous hybrids are formed are very low (Raybould & Gray 1993). In part this is because speciation has evolved because such hybrids are genetically unstable and are often sterile or grow very poorly. Also it is because the hybrids very seldom have a selective advantage, so that their frequency in populations can rise. Spartina anglica is a rare example of a very vigorous hybrid that has perhaps surprisingly remained impossible to reproduce through crosses of its parent species. The inheritance of cloned genes from GM crops that have interfertile relatives can be expected to be at least as frequent as for other genes conferring selective advantage that have moved from the two crops since the time they were first domesticated.

Evidence is gradually accumulating for frequencies of the transfer of genes from traditionally bred crops to wild relatives, but as yet the information is sparse and suggests that such frequencies are usually low. When we consider that crop breeding has for years concentrated on disease resistance, seed yield, seed size and general fitness, it is surprising that the flow of these ‘fitness’ genes to related species has not led to enhanced weediness. It seems most unlikely that the absence of observed problems is due to gene transfer not occurring. It is more likely that the complexity of the genetic make-up of a successful species is such that even when potentially important genes for environmental fitness are inherited their ability to modify the host is very limited. Were this not the situation traditional crop breeding, which can involve the use of weedy distant relatives, or even pernicious weeds such as wild oats Avena fatua, would surely have produced a number of very weedy derivatives of crop plants.

Much more research on characteristics that affect ‘fitness’ and ‘weediness’ is needed before we are in the position to be able to predict with any certainty whether a new crop or introduced plant species has the potential to cause environmental harm. However, I would not argue that until such research is done no releases of GM crops should occur. Given that so many genes of unknown function are routinely introduced into commercial crops from weedy relatives, risks arising from their cultivation must be greater than for most types of GM crop. All new varieties of crops should be subject to screening for potentially adverse agronomic/environmental characteristics, with the intensity of screening being related to the nature of the donor species. For example, a new oat variety derived from a cross with wild oats should surely need a more stringent risk assessment than a GM oat in which a modification to starch synthesis has been engineered using oat genes? Existing variety trials for seed certification ensure that readily identified undesirable changes to varieties are recognized, but at present there is no systematic procedure to collect and interpret information derived from the very large-scale cultivation of such crops by farmers. There should be such a system. It should be based on the present system for medicines, in which any unexpected behaviour of patients is reported centrally so that rare events can be identified. Increased weediness, potential to spread and changes in wildlife are characteristics that would be of great interest. While it would be very helpful to gather such information for all crops, attributes that would most concern ecologists, such as those that lead to decreased populations of weeds or insects, are likely to be those that many farmers would deem as desirable traits. I return to this issue later.


The potential for gene transfer is much greater for bacteria than plants because most species appear to be able to use at least two of the three known methods for the promiscuous transfer of genes by microbes. These are: transfer via infecting viruses (transduction), the uptake of naked DNA (transformation), and DNA transfer through pairing of cells (conjugation). Gene transfer via conjugation is probably the only sexual activity in which all bacteria indulge. They do this because the genes used to bring about pairing and gene transfer are coded on extrachromosomal elements called plasmids. If a bacterium inherits a so-called conjugative plasmid, genes on the plasmid will ensure its distribution by making the cell produce appropriate proteins to initiate pair formation and plasmid transfer. Some types of conjugative plasmids carry genes that enable their host cells to pair with any other bacteria, so that they are transferred within a species or intergenerically. Other genes on plasmids code for functions that can greatly enhance the fitness of hosts in certain, often adverse, environments. Classical examples of such genes are those for antibiotic resistance, and the spread of such plasmids among different pathogens has become a problem for the successful treatment of certain diseases with antibiotics.

Some plasmids are rapidly and efficiently transferred among bacteria, such that usual observed frequencies of transfer range from 1% to 100%. Sometimes plasmids can combine with bacterial chromosomes, resulting in strains in which chromosomal genes are transferred at frequencies of up to 100%. Plasmid-mediated conjugation may also be an important way in which chromosomal genes are transported among different genera. This is because some plasmids can interact with chromosomes and ‘pick-up’ chromosomal genes that can then be moved among bacteria in matings mediated by the plasmids concerned (Chatterjee et al. 1985; Holloway 1986). This phenomenon allows chromosomal genes to be inherited without the need for homology and recombination, but it does not obviate the need for the genes concerned to confer a selective advantage to the host if they are to be maintained in the host organisms.

While plasmid transfer is clearly important and well documented, there is also very good evidence for the movement of genes among bacteria that have efficient mechanisms for allowing transformation to occur. Analysis of bacteria that have been subjected to very strong selective pressure due to the use of antibiotics to control the diseases they cause (e.g. gonorrhoea and pneumonia) has shown that resistance has spread through the horizontal transfer of regions of appropriate genes (Maynard Smith, Dowson & Spratt 1991). The extent to which transformation provides opportunities for extremely widespread gene transfers to bacteria is unknown; there are no obvious reasons why some species of soil bacteria should not take-up DNA from any organism. However, there is an enormous gap between DNA entering cells and being integrated into the genome such that the gene(s) is replicated and transferred to progeny. The primary obstacle is the need for recombination between the chromosome and the incoming DNA, and this requires a high degree of DNA sequence homology. The need for incoming DNA to be sufficiently similar to a region of chromosomal DNA effectively blocks most opportunities for the widespread dissemination of genes. If this were not the case, species that take up foreign DNA very efficiently would soon become so genetically heterogeneous that they would no longer be taxonomically identifiable, which is not what is observed.

Examples of concerns associated with the release of GM crops

The transfer of genes from crops to bacteria

Perhaps one of the most acrimonious and unnecessary debates about risks emanating from the use of GM crops has been that surrounding the inclusion of a bacterial antibiotic resistance gene in a herbicide-tolerant maize being marketed by Novartis. It is unnecessary because the gene concerned, which confers resistance to penicillins, is not expressed in maize and is not required. The debate is very acrimonious because the genetic construct that has been introduced into the plant contains part of a bacterial plasmid and a regulatory region to drive the expression of the antibiotic resistance gene in bacteria. The following characteristics of the genetically modified maize greatly facilitate the possibility of bacteria inheriting and expressing the plant gene. The resistance gene is for an antibiotic (penicillin) that is widely used and would provide very strong selection for the multiplication of bacteria that had inherited the gene. The gene can be expressed in bacteria without the need for modification of its regulatory region, unlike other plant genes. There is a chance that the DNA from the plant could be replicated as a plasmid and thus escape the need for homology to enable the sequence to be inherited within a chromosome. It almost appears that this GM crop has been designed to test if it is possible for genes from GM crops to be inherited by bacteria.

As presented above, it seems very likely that we have a potentially damaging GM crop, but is this indeed the case? For harm to arise there is a need for the resistance gene to be inherited by bacteria whose populations are controlled by penicillins. It is possible that such bacteria exist in soil, and that if such bacteria inherited the gene they could grow uncontrollably. However, despite the widespread occurrence of antibiotic production and resistance among soil bacteria there is no evidence that concentrations of antibiotics ever reach levels at which they control populations of sensitive bacteria (Huddleston et al. 1997).

Of real potential concern is the possibility that pathogenic bacteria might inherit the resistance gene, and it is important to have an indication of the likelihood that this will occur. The resistance gene would be made available to bacteria through the breakdown of maize plants in fields and from the release of DNA from maize-based animal feed in the same manner as all genes from crops are released into the environment. The DNA, together with all other DNA in the environment, would be taken up by species of bacteria that have efficient uptake mechanisms. Some of these would be soil bacteria and some would be bacteria associated with animals. DNA taken up by bacteria will only be inherited and replicated if it integrates into the chromosome or plasmids of the bacteria concerned, or if it circularizes and replicates as a plasmid. Recombination would be extremely unlikely if a similar region of DNA was not already present in the host. It would also be very unlikely that the incoming DNA would become a circular plasmid molecule because there is no mechanism that would facilitate such an event. However, both events are theoretically possible, even if at extremely low frequencies.

In the absence of selection, genes entering populations remain at about the frequency at which they enter. Any use of penicillin on animals harbouring resistant bacteria would provide very strong selection both for the bacteria that had inherited the resistance gene from maize and those bacteria that might have inherited the genes through some form of interbacterial gene transfer. Transfer of chromosomal genes between species of bacteria is very infrequent, as also would be the transfer of any plasmids formed from the plant-derived DNA. The reason for the latter is that the plasmid used for constructing the genetically modified maize is one that cannot induce conjugation and is not moved during conjugation induced by other plasmids.

From what has been stated, harm could arise as a result of at least two very unlikely events producing an antibiotic-resistant pathogen under conditions in which penicillin is being used to treat a disease. I believe that the expected frequency of such events occurring is extremely low, but is feasible if extremely large amounts of GM maize carrying the gene are cultivated and fed to animals. Various European regulatory authorities and pressure groups have advised that the maize should be banned on the basis that any increase in risk is unacceptable. This is indefensible. We should compare risks when deciding if something is unsafe, or likely to become a serious problem. In this case it is necessary to be aware of the existing frequency of penicillin-resistant strains of pathogenic bacteria isolated from sick humans. Data from the Central Reference Laboratory in Colindale, England, indicate that for most of the species tested about 1–10% are resistant (D. Spellar, personal communication). Given this extremely high frequency of known pathogens already carrying penicillin-resistance genes, I remain totally unconvinced that GM maize presents a measurable risk to humans. I also remain totally incredulous that Novartis should persevere with a crop that is so widely perceived to be a threat to human health, and whose construction is so obviously flawed from a public perception point of view.

Cross-pollination of organic crops

Recently the Soil Association in the UK announced that the standard for their approval of organic food would require that there was no contamination with pollen from GM crops. This standard requires no cross-pollination and has been interpreted further to include the requirement that no pollen grains from a GM crop are present on the food. Scientifically based reasons for zero tolerance have yet to be produced, but this has not prevented a major press and pressure group offensive to require that GM crops are separated from organic farms by at least 6 km.

There is no doubt that the organic food lobby offensive has caused a considerable ‘regulatory headache’ in the UK. If zero tolerance were to be supported by government it would lead to the exclusion of GM crops from the UK unless the recently announced European Union (EU) organic standard of less than 1% cross-pollination becomes mandatory. As with so much of the GM debate, the issues surrounding the organic standard have become confused and often have been severely misrepresented. It is not for me to judge what an organic food standard should be, and the organic lobby would not expect such standards to be required to stand up to scientific scrutiny. However, there is a need for a reasoned defence of the policy if a minority of farmers is to determine what the majority can grow.

Why would it be impossible to introduce GM crops if zero tolerance was approved? The reason is that pollen is transported over long distances by wind and insects. The 6-km distance is presumably derived from the sort of distance that would not be exceeded by most bees travelling from a hive to a flowering crop. On the other hand, pollen can be dispersed very long distances by wind, although its fertility and concentration decline rapidly with distance. It would thus seem that cross-pollination over long distances is not only possible but also likely, and indeed the Advisory Committee on Releases to the Environment (ACRE) was made aware of the possibility many years ago. Recent criticism of ACRE and the Government for approving releases of GM plants with segregation distances of only tens of metres have completely ignored the reality of what actually happens in the environment. ACRE's advice has been based on many decades of measurement of gene flow during plant breeding. Standards have been set for seed purity that require less than 0·5% or 0·1% cross-pollination, depending on the type of seed, and such standards are reliably achieved with separation distances in the order of 50 m for some crops. An analysis of the frequency of cross-pollination with distance from the pollen source shows that the frequency declines very rapidly, as would be expected from dilution of the pollen, and falls to a very low level (less than 0·1%) which is maintained for a great distance (Anonymous 1994). Therefore, zero cross-contamination is unlikely to be achievable without segregation distances greater than the 6 km that the Soil Association presently considers adequate.

Why is it that so much attention is being given to ‘evidence’ that gene transfer over long distances is likely? The reason is that the evidence often quoted is derived from people who monitor the physical dispersal pollen, or the distance that it can be carried by very hungry bees. Confusion has arisen over the presence of a substance and its biological activity and, as has so often happened recently, reality has not been allowed to interfere with a good story. Somehow the GM debate has to move from an overemphasis on what might happen to more realistic assessments of what is likely to happen.

Are cloned genes likely to behave differently from other genes?

The simple answer is that it depends on the gene, the site of insertion, and the selective pressures acting on the gene. Chemically speaking, a cloned gene is integrated within the genome of the host organism and is intrinsically no more able to move from that site, or be lost, than any other gene. There is therefore no scientific support for the often-articulated belief that genes from GM crops are more likely to spread in the environment than other genes. For genes derived from bacteria, the likelihood of them spreading in the environment is considerably lower than from the very widely distributed bacteria from which they are derived.

If the cloned gene confers a strong selective advantage, such as frost tolerance, it is possible that its inheritance by weedy relatives could present serious ecological problems, an obvious example being inheritance of such a gene by a frost-sensitive garden plant that in warmer climates is a pernicious weed. However, what would be observed would not be a reflection of an enhanced ability for the cloned gene to spread, but simply natural selection operating very efficiently on a characteristic that confers a strong selective advantage. The risk would, as always, be composed of the nature of the harm and the frequency with which the harm could arise.

Changes in gene expression due to the location of cloned genes in the genome and to the modification of cloned DNA so that gene expression has been lost, have been observed. Neither need be a problem as long as the possibilities are considered in risk assessments during commercialization, and lines showing such variability are not commercialized. In practice the impact of this occurring is a quality issue, because if the genes ceased to be expressed the host would revert to the characteristics of the parent from which it was derived. Exceptions would be crops in which the gene cloning has been used to stop the production of a harmful substance or characteristic. Before such crops could be considered for commercial use there would need to be very stringent testing and there should also be a requirement for careful monitoring during cultivation.

What is the relevance of the GM crop debate to ecologists?

We have serious problems with the application of GM technologies, not only because many people have serious emotional reservations, but also because our understanding of the environment and how it functions is so limited. We are seeing dramatic declines in many species of farmland bird and wild plant species, which we know to be associated with changes in farming practice over the last few decades but find very difficult to tie down to specific agricultural practices (Fuller et al. 1995). For example, autumn sowing of crops is certainly important in restricting the food available for some bird species, and the use of agrochemicals is also important (Fuller et al. 1995). However, even for such well-documented organisms as birds, direct cause and effect relationships are difficult to understand in detail. The position is very much worse with plants and invertebrates, as there have been far fewer long-term surveys that enable correlations of declines with specific agricultural practices (Donald 1998). As a result, when the release of herbicide-tolerant or insect-resistant crops is contemplated there is a real problem in deciding if such crops will further enhance declines in the species that are affected. Weed control is a direct elimination of biodiversity within fields, as well as a major reduction in food for wildlife. Likewise, if all crops are resistant to insects what will birds in the future eat? Controls of weeds and insect pests are an integral part of agricultural practice, whether it is intensive or organic. Intensive agriculture is a greater problem because such controls are more efficient, reflecting five decades of agricultural policy to enhance food production. Therefore, is there a logical reason why GM crops bred to reduce pest problems should be discriminated against? It is clearly unreasonable; but surely it is not unreasonable to strive to reduce the impact of pest control on wildlife, whether the methods involve GM crops or not?

In the UK we have the problem that many of the species of plants and animals we want to conserve are the products of previous agricultural practice, so that any change in farming leads to changes in wildlife. In the absence of human management, wildlife would be dominated by the plants and animals best adapted to living in forests. Our concerns about GM crops need to change from an overemphasis on the slight risk that crops and genes may move from farmers’ fields into our few remaining areas of uncultivated land, to a proper recognition of the dramatic impact that changes in agricultural policy can have on our landscapes. GM crops can be produced and managed in ways that are less environmentally harmful than present crops; they can also be bred and used in ways that will enhance existing declines in wildlife. The challenge for ecologists is to be able to provide clear advice on the changes in farming practice that are needed to produce the environment we want. The challenge for politicians is to determine what people really want from our environment and how much they are willing to pay farmers to provide it for them. The challenge for all of us is to decide which plants and animals we want, and how many there should be of each species.

I believe that the only effective way to manage farms to produce more biodiversity will be through a system in which the yield of wildlife has a commercial value to the farmer. Only then will the selection of seed, management practices and the desire to enhance wildlife be driven by those with the greatest ability to ensure that the changes we desire are achieved. I have no doubts that many, perhaps most, farmers would wish to be able to enhance the environmental quality of their farms, but are heavily constrained from doing so by the need to generate adequate incomes under the constraints of the Common Agricultural Policy. We should no more expect farmers to produce more flowers in their fields without remuneration than we should expect city dwellers to provide flowering window boxes without financial incentives.

A window of opportunity has opened up for ecologists to produce the research needed for us to understand better how agriculture can produce the biodiversity we want, and to facilitate risk assessments for the introduction of new crops, whether they be GM or not. Resolution of the first problem will certainly not be solved by an idealistically driven move to widespread organic farming; birds starve whether fields are ploughed in the autumn by intensive or organic farmers. Neither will it be resolved by the blind use of technology. Statements of intent to produce more biodiversity and greater numbers of individual species are meaningless unless linked to realistic means to achieve the goals. Once we have a clear picture of the agriculture we want in the future it will be much easier to judge GM or traditionally bred crops for their environmental benefits and disadvantages. It will also be much easier for breeders to develop crops that are best suited for an agriculture that is being managed to satisfy environmental needs.