There are a number of publications which address this issue. Maclean and Laight (2001) and Dunham (1999) have produced very useful reviews which discuss many of the points raised in this paper.
In our view the most important areas of risks which need to be considered in the use of transgenics are:
1. human health
3. animal welfare
4. poor communities
In each of these categories there exists a multiplicity of pathways by which effects could, in principle, be brought about. Rational and responsible assessment of risk requires that the following properties are all considered:
1. source of the DNA of the target gene;
2. source of the non target DNA segments of the construct used;
3. site(s) of incorporation of the transgene within the recipient genome;
4. product of the transgene;
5. interaction of the transgenic product with other molecules in host and consumer;
6. possible molecular changes in transgene product during processing;
7. pleiotropic effects of transgene;
8. tissue specificity of transgenic expression; and
9. numbers of transgenic organisms capable of interacting with natural systems).
The risks to health will depend upon all of the factors listed above. In practical terms the most important of these are likely to be the source of the DNA and the nature of the product.
The great majority (98 percent) of dietary DNA is degraded by digestive enzymes relatively quickly (Royal Society, 2001) but use of viruses (disarmed or otherwise) as vectors, must increase the risk factor significantly as these are organisms which are adapted to integrating into host genomes and some represent risk factors for cancer induction. The work of Zhixong Li et al. (2002) who induced leukaemia by using retroviral vectors in making transgenics for a commonly used marker gene in mice and a recent report of leukaemia induction in a child undergoing gene therapy for x-SCID using a retrovirus (Hawkes, 2002) show that this is not a trivial risk. Arguments about risks and benefits attached to this form of gene therapy are current (Kaiser, 2003).
At the other extreme the use of autotransgenics must be seen as posing a risk which is orders of magnitude lower than that for allotransgenics and probably negligible. The major risk from the production of the transgene will lie in the use of novel proteins or other molecules produced by the transgenic organisms. Either in the native form or, following modifications in the human body, such molecules could be inimical to human health (e.g. through allergies). It would seem sensible to avoid the use of such substances except where strictly necessary and under rigorous control.
Other potential risks may lie in incorporation of transgenic DNA into the genomes of resident gut microflora (though this is likely to be very improbable) or a change in the pathogen spectrum of the transgenic fish leading to it hosting a new pathogen which happens to be also a human pathogen.
Maclean and Laight (2000) assessed risks to consumers as very low.
The extent of aquatic diversity is both extremely large and relatively poorly understood (Beardmore, Mair and Lewis, 1997). This means that the task of estimating the risks to aquatic biodiversity at all of its levels from the use of GMOs or indeed, any genetically distinctive strain used in aquaculture is monumentally large. Aquaculture has a further problem in that the (almost always unintended) escapes of genetically distinct farmed fish are unpredictable and often large in numbers. Stenquist (1996) in discussing transgenics in open ocean aquaculture, quotes some relevant figures. Thus, 15 percent escapes for Atlantic salmon, escapes of 150 000 salmon and 50 000 trout in Chile and catch statistics for Atlantic salmon off Norway in which 15?20 percent of the fish caught were of farmed origin. In Scotland an escape of 100 000 Atlantic salmon was reported recently. It is clear that escapes of these magnitudes pose considerable problems and it is not surprising that in some parts of Norway fish of farmed origin represent a majority of the animals fished (Saegrov et al., 1997)
The major focus of attention in the literature lies, understandably, upon the effects of escapes upon natural populations of the same species, but we must always bear in mind possible impacts across an assemblage or ecosystem as a whole. The first general point to make is that there is, in principle, no difference between the biodiversity risks from escapes of GMOs and from fish genetically improved in some other way, e.g. by selective breeding or (in some respects) from exotic species.
The second general principle is that such genetically improved forms including GMOs, are developed for a specific set of environmental circumstances in which they enjoy an advantage conferred by human decisions. In nature, however, such genetically distinct forms may legitimately be regarded as mutant forms of the wild type. A considerable body of genetical knowledge tells us that the probability of survival of mutant forms is extremely low because they are disadvantaged in viability and/or fertility under natural conditions. Thus, for example, in the genetically distinct farmed Atlantic salmon in Norway the males are very much less successful than wild males in securing mates (Jonssen, 1997).
However, it must be conceded that in species like salmon where the farmed populations outnumber the wild populations by orders of magnitude, the effects of escapes of any genetically distinct genotype upon natural populations may be both deleterious and of significant size simply as a result of swamping
An interesting model of the effects on a medaka (Oryzias latipes) population of transgenic release has been produced by Muir and Howard (2001) using estimates of juvenile and adult viability, age at sexual maturity, female fecundity, male fertility and mating advantage. They were able to demonstrate that the transgene would spread in natural populations, despite low juvenile viability, if transgenes have sufficient high positive effects on other fitness components. It has been argued that this might lead to extinction but the selective pressure for recombinant genomes with higher viability would be expected to be immense.
Maclean and Laight (2000) simulated the changes in frequency of a transgene expected with different scenarios embracing a range of selective values including heterozyote advantage. They note that repeated small introductions [of the transgene] can have an effect on ... frequency ... since the frequency of advantageous alleles rises much more rapidly than if a single large introduction is considered.
A major problem in assessing risk to natural populations is that of scale. Even if farmed fish are at a selective disadvantage in natural conditions, the ratio of wild:farmed numbers may in some areas, be relatively small. In these situations significant modification of the native population and its role in the ecosystem is inevitable.
Whilst not providing a completely satisfactory answer, there is little doubt that making farmed fish sterile would go a long way towards reducing the pressure upon such threatened ecosystems. A number of research efforts to develop systems for sterile fish production are being made. The techniques include triploidisation, antisense transgenics, ribozymes and gene targeting (Maclean, 2002; Uzbekova et al., 2001; Maclean, pers. com.).
Provided that the best containment measures (physical and biological) are adopted, in our opinion, in general risks to biodiversity by GMOs per se are probably extremely small, but in specific cases, the risks and consequences may be large. As a general rule and adopting a precautionary approvah (OECD, 1995), it is, however, clear that each individual case needs careful study and appraisal and the best possible containment measures before approval for uptake into commercial production is given.
The direct or indirect effects of transgenesis upon the welfare of fish GMOs in aquaculture are very poorly understood. In part, no doubt, this is because notions of cruel or unnatural treatment in mammalian species translate, for a variety of reasons, imperfectly to fish. Nevertheless, as life forms with highly developed nervous systems and with a range of behavioural phenotypes which flow from this, fish qualify for welfare consideration.
There are a few studies which bear on this. Thus, for example, Devlin et al. (1995b) reported changes in colouration, cranial deformities and opercular overgrowth and lower jaw deformation in coho salmon transgenic for AFP and GH. After one year of development anatomical changes due to growth of cartilage in the cranial and opercular regions were more severe and reduced viability was evident.
The larger body of data on species farmed terrestrially shows dysfunctional development leading to acromegaly, lameness and infertility in some GH transgenics in pigs and sheep. However, in pigs dietary modification influencing nutritional levels of zinc proved successful in avoiding such abnormalities (Pursel and Solomon, 1993; Pursel, 1998).
We have been unable to find systematic data on the incidence, in fish GMOs, of effects such as those described by Devlin et al. (1995b) and this is probably because animal welfare is not sufficiently widely recognised as an issue in relation to the use of GMOs. This is well illustrated in the otherwise comprehensive and balanced review by Sin (1997) in which the section on ethical issues contains no reference to animal welfare. Nevertheless, if GMOs are to be used in aquaculture (and there are weighty arguments for so doing), concerns on this issue will need to be properly satisfied. The Royal Society report (2001) devotes a significant amount of space to this issue.
This term rather than poor countries is used because all poor countries contain rich people and rich communities. The possible economic disadvantages of use of transgenics centre on two issues:
If transgenic fish become widely grown because they are much more efficient, and if special broodstock are required to produce fry for on-growing to adults, which, cannot be used as broodstock, a dependency is created. This dependency may be benign or oppressive, depending on the arrangements made for seed supply.
This is a very difficult issue indeed. Since genes may now be patented and therefore, enjoy commercial value, the opportunities for dispute about equitable treatment of stakeholders in cases where ownership of genes and strains is contested, are legion.
A recently published report (Commission on Intellectual Property Rights, 2002) states that developing countries are frequently disadvantaged in the use of, and access to, IPR because of increasingly protective attitudes taken by owners of IPR. However, the report also indicates that developing countries are very heterogeneous in respect of their ability to use and develop IPR.