Ecological risk assessment of release of GE-organims
The article below addresses important points about risk assessment, showing that the knowledge about the consequences of release of genetically engineered (GE) organisms is insufficient to make it possible to assess the safety of their release. Also, it shows that important arguments used to justify liberal regulation for release and safety assessment of GE-foods do not have proper scientific basis.
Professor Philip Regal has studied questions about the safety of GE-organisms since over 15 years and is recognized as and international authority in this field. He has recently signed our Declaration, demanding a moratorium on the release of GE-organisms and on their use as foods.
NOTE: This article is written for scientists. If you want to read more popularly written information about the safety issue, we recommend you to start with:
"Is Genetic Engineering a variety of breeding?" or
"Genetic Engineering Possesses Inherent Unpredictability"
Scientific principles for ecologically based risk assessment of transgenic organisms
P. J. REGALUniversity of Minnesota, St Paul, MN 55108, USA
Published in Molecular Ecology (1994) 3:5-13.
It is critical to base scientific risk assessment of genetically engineered organisms (GEOS) on appropriate scientific concepts. A variety of 'generic safety' models has now largely been recognized to have been based on outdated scientific thinking. One broad safety argument that is still used is that genetic engineering categorically is nothing but an extension of selective breeding. It is explained here, though, that the mechanisms and potentials of the two can be profoundly different. This does not mean that every GEO is ecologically dangerous; but some types of GEOs may be considerably more risky than what could be produced with selective breeding, especially when an ecologically competent host is supplemented with novel features that may increase its competitiveness. In addition, genetic 'side effects' raise food-safety issues; and the possibility that they may sometimes increase ecological competitiveness cannot be ruled out, though this would be quite rare. Field plots have a proper use: to gather particular data that could be used in analysing the risks of commercial releases. But it is not scientific to call a small, confined, field population, isolated from potential competitors, a 'test' or 'release' and then conclude that because 'nothing happened' the GEO will be safe when commercialized, or indeed that all GEOs will be safe.
Keywords: adaptive potential, domestication, genetically engineered organisms, pleiotropy, selective breeding
Received 17 April 1993; revision accepted 10 September 1993
Twilight of the generic safety arguments
A variety of arguments was used from about 1978 to 1986 to claim that all genetically engineered organisms (GEOS) should be safe. These have been criticized systematically by Regal (1985, 1986, 1988) and Colwell (1989) and analysed at numerous technical workshops and as a result are seldom now used. Yet the older arguments should be mentioned because they may still come to mind for many people who are unfamiliar with the literature and workshops, and one should be careful not to use them in risk assessment. They included the following assertions.
1 Genetic engineering is not different from ordinary sexual reproduction, therefore it presents no new risks.
2 Genetic engineering will always impose such a great added metabolic burden that transgenic organisms will always be ecologically incompetent.
3 Genetic engineering can create nothing truly new because millions of years of evolution have tested every possible combination of genes. And whatever does not exist today has been proved to be maladaptive.
4 Genetic engineering can only make an organism less perfect than evolution has made it.
5 Nature keeps all populations in balance, it will reject transgenic novelties or keep them in balance as it did the original hosts.
One reason that these arguments for generic safety have had any appeal has been that most biologists have been unfamiliar with conceptual progress in the fields of ecology and evolutionary biology in the last part of the 20th century.
The 19th and early 20th-century views did include ideas of the perfection of adaptations and a simplistic balance of nature. Although superficially secular, the older thinking was in fact squarely in the old deistic, natural theology tradition of nature as a watch-like integrated and perfect system that could scarcely be improved upon.
Modern ecology and evolution, however, leaves much more room for the idea that if a wild-type or near-wildtype organism is engineered with a novel adaptive combination of traits, then even if it is imperfect, if the ecological benefits of the new trait(s) outweigh its ecological costs, then the new creature could persist in loosely integrated nature and sometimes disrupt the ad-hoc organization of communities and ecosystems as we now understand them (Regal 1985,1988,1989). In any event, even A. M. Campbell, a long-time advocate of deregulation (and still commendably a foe of regulatory overkill) eventually agreed that there can be no generic arguments for the safety of genetically engineered organisms. (See also Colwell 1989; National Academy of Sciences 1987; National Research Council 1989; Office of Technology Assessment 1988; Tiedje et al. 1989.) ... there can be no generic tests for the safety of engineered organisms as a group .... The various generic arguments that have been raised so far for either the safety or the danger of engineered organisms can be dismissed as either unproved or illogical ... each engineered organism is a separate case. (Campbell 1991, pp. 38-40)
The claim that genetic engineering is basically the same as traditional domestication
The advocates for relaxed regulation seem to have come down to an argument with intuitive appeal but one that is scientifically unspecific. This argument goes as follows: The 'new biotechnology' (genetic engineering) with recombinant DNA (RDNA) and related techniques merely moves genes between chromosomes. Humans have been moving genes around for thousands of years in the traditional breeding of com, wheat, cows, yoghurt, beers, etc., and this has never created a dangerous creature. Hightech RDNA gene-splicing now allows us to move genes around much more precisely, and so this should be even safer than breeding by traditional methods where the results have been more unpredictable. For all its simplicity and appeal is this argument true?
It is, after all, reductionist rhetoric (breeding and RDNA both reduce to, are 'nothing but', moving genes around) and rhetoric is not the same thing as a professional, scientific comparison between the genetic mechanisms of traditional breeding and those of genetic engineering. Careless reductionism can be treacherous, for reductionism has too often been misused to make appealing arguments that superficially appear to be legitimate, but that have obscured rather than enlightened (Koestler & Smythies 1969; Regal 1990a, pp. 82-87) One could similarly (mis)use reductionist philosophy to argue that basically automobiles or bicycles or carts are 'nothing but' wheeled transportation, and that automobiles are merely faster and more efficient than bicycles or carts. Ergo, automobiles should present no special practical problems. However, automobiles, unlike genomes, are familiar to all; so anyone can see that the superficial truth in the reductionism is seriously misleading. Indeed, if one does not understand the actual mechanisms of how bicycles, carts and automobiles operate, then one could be very embarrassed to find out the hard way that automobiles can run out of gas even though bicycles do not, or that automobiles may explode when they collide even though carts do not, or that some automobiles pollute the atmosphere.
So despite (or perhaps because of) the aggressive campaign to convince that genetic engineering is basically the same as traditional breeding only more efficient (see also Davis 1987; and Sharples 1987 for reply), one is right to ask if this 'basically the same' litany is true. We shall see that it is not. And in that case can the new practices present any new hazards to human health or the environment?
Three important differences between genetic engineering and traditional breeding
There are at least three independent scientific reasons why rDNA genetic engineering is not basically the same in its mechanisms and potentials as the traditional breeding that humans have been conducting for thousands of years. These differences do not mean that most genetic engineering projects are inherently dangerous; only that the technology has more potential to be misused than does traditional breeding.
The numerical majority of all types of genetic engineering projects involve the introduction of nonadaptive traits into ecologically incompetent hosts, and thus most projects are in ecologically low-risk categories.
It is those projects in which novel adaptive traits are added to an ecologically competent host that could create unusual, ecologically high-risk, supercompetitive organisms, perhaps with dangerously altered ecological functions. What does unusual mean? If one thinks of nature as a chess match, 'unusual' would be like giving one player a powerful new piece that other players have not had experience in dealing with, such as knights that can move four squares (whereas opponents have traditionally had only experience with knights that can move two squares). Breeders have long introduced insect and disease-resistance traits into crops, with no dramatic increase in the ecological competitiveness of the crops. But the biological principles would be different if on the one hand modestly effective resistance traits were moved by any technique into ecologically incompetent crops from their relatives, or if on the other hand nearly wild-type forage or forest crops were provided with potent new resistance mechanisms with which the pests that attack them have had no coevolutionary history. Obviously there is no absolute line between qualitative and quantitative novelty, and for regulatory purposes it is best to think of adaptive novelty somewhat flexibly.
A summary of the major differences between the mechanisms and biological potentials of traditional breeding and of modern genetic engineering will help to explain why RDNA techniques have the potential to create adaptively enhanced organisms while similar projects using traditional breeding techniques would fail.
1 True ecological novelty by phylogenetic leapfrogging
Traditional breeding has typically only recombined genetic traits within species and between related ones (or more rarely between related genera, with wide hybridization). Since related species and genera have diverged only recently, their gene pools and phenotypic traits are basically similar. Humans and chimpanzees (Homo and Pan), for example, only differ in 1.6% of their DNA after 7 million years of separation. Random mutation sometimes produces adaptive novelties that are unique to one species within a genus. The production of novel metabolic functions has been observed long ago even in laboratory microorganisms (e.g. Hall 1983). But genetics among related species has turned out to be typically quite conservative. [To avoid confusion, bear in mind that sequence differences between homologous strands of DNA, are not the same thing as functional differences. Because of DNA code redundancy and other reasons, there are abundant DNA sequence differences between spe cies that have no functional significance. Thus, RFLP studies or sequencing show variations in the DNA code even among individuals within a species, they reveal extensive 'silent substitutions', pseudogenes, and differentiation of nontranscribed sequences, that are highly useful as genetic markers. But such polymorphisms can give an exaggerated impression of the amount of biologically, adaptively, significant genetic variation, that is actually expressed phenotypically and is relevant to adaptation and artificial and natural selection.]
Traditional breeding has been limited to exchange of genetic traits between populations where the genetic differences are typically minor to begin with.
Genetic engineering, however, can move fully functional genetic traits between completely different sorts of organisms. It has already produced rodents and pigs with functional human genes, plants and fishes with functional insect and bacterial genes, etc. Quite obviously, phylogenetic leapfrogging is not in and of itself dangerous or even ecologically significant. But phylogenetic leapfrogging offers unique possibilities to create populations of organisms with novel combinations of adaptive traits.
Most transgenic novelties to date have not had new combinations of ecologically adaptive traits. For example, it was certainly novel to put firefly genes into tobacco, but it seems unlikely that this could make tobacco more competitive in nature. The tobacco host was highly modified from the wild-type to begin with, and the firefly genes offered no obvious adaptive advantage from an ecological perspective.
But there are particular cases where the novel traits could well confer competitive advantages. For example take an ecologically competent host such as a forage grass or forest tree and add to it greatly superior defences against insects or diseases, novel mechanisms for drought or freezing resistance, and/or for mineral uptake from poor soils. Thus, genetic engineering does have the potential to create types of organisms that can interact with particular ecosystems and biological communities in novel competitive or functional ways or at new levels of impact. These sorts of transgenic organisms should be allowed to become self-reproducing in nature only after the most meticulous reviews, if at all.
2 Genetic modification without traditional debilitating trade-offs
Wild-type genetic traits have often been traded away in the process of domesticating plants and animals for human needs. Corn, wheat, white rats and mice, for example, seem unable to establish self-sustaining populations in natural communities. What happened? Sometimes the phenotypic features that were selected simply made the domesticated forms physically or physiologically 'clumsy'. But let us focus here on genetic trade-offs, since the question before us now is whether rDNA and traditional breeding differ in any significant genetic ways, and in terms of their potential greatly to improve the competitiveness in nature of a manipulated form.
Breeders have selected for phenotypic traits over the last several thousand years, and it is not completely clear what the underlying changes in genomes have been. Yet it is clear from modern genetic studies that the changes produced by most selection since the early 1900s have involved substitutions of alleles. (Alleles are mere variants, that occur in populations, of given genes.) So what?
First, note that if 20th-century experience reflects what happened historically, then the traditional breeding of the last several hundred or several thousand years probably has not in fact involved simply 'moving genes around' and introducing truly new and functionally proven genes to chromosomes or genomes, as the litany has long implied. Alleles are not genes.
Secondly, if traditional breeding has usually been allelic substitution, then this should explain in large part why some of the more radically changed domesticated forms have become ecologically incompetent. Extensive allelic substitutions are virtually guaranteed to produce reductions in an organism's fitness for nature. Such tradeoffs between fitness for nature and fitness for human needs result from the actual losses of wild-type genetic material, and from imbalances in previously integrated genomes.
There are two reasons why loss and imbalance have tended to happen. The first and most obvious cause of genetic loss has been due to the elementary fact that only one allele can occupy a locus on a chromosome. So if a rare mutant that has human interest but that has low fitness is artificially selected at a given locus, then the wild-type alleles cannot also occupy that locus and are lost from the line. The second cause of genetic loss is due to the fact that in selecting strongly for any allele, one is also unintentionally selecting particular alleles of other genes in the same linkage group.
Thus, certain alleles throughout the genome will be retained and others lost 'randomly' due to 'genetic hitch-hiking'. It would demand unrealistically enormous numbers of individuals and of genetic combinations, and perhaps unattainably wise decisions, to avoid this last problem if one were to try to use traditional breeding, without high-tech genetic markers etc. to try to establish a new adaptive trait in an organism and make a form that would be even modestly superior in competitiveness in nature to a known form.
The allelic variants, Mendelian traits, that have been selected by humans in the traditional breeding of radically altered domesticated forms presumably were often rare minor mutations that breeders substituted for the wildtype alleles that nature had found most fit and made most common in natural populations. Due to this and due to genetic hitch-hiking, the artificially selected combinations of alleles have understandably been inferior to wild-types in terms of their direct contributions to fitness in nature and in terms of their indirect pleiotropic and epistatic interactions in the integrated genome.
Thus, pronounced agricultural gains have robbed the new creature of some of its fitness for the wild. And thus, breeders often must depend upon wild and near-wild relatives of crops to obtain traits to reinvigorate the crops with defences against pathogens, insects, resistance to drought, etc., and to balance selected lines. No one intentionally turned mice, rats, and rabbits into pale and competitively incompetent organisms. Breeders constructed types that would thrive in small cages, and in the process the original combinations of alleles that made up the fit wild-types were inadvertently but understandably disrupted and lost.
It is true that new genes proper will arise in populations from gene duplication and certain types of recombination events. When these have been selected for in agriculture, one has truly introduced new genes into a line. But compared to the introduction of a transgene that long was completely adaptive in its donor and was the product of a long evolutionary history, it seems unlikely that a new gene from a recent gene duplication and with only one or a few base changes has a similar potential to increase the fitness of an ecologically competent form.
In summary, the traditional breeders of thousands of years have perhaps had less experience with moving completely new genes into genomes than it might have seemed. Certainly they have had little experience with moving novel and superior adaptive traits into ecologically competent organisms. What is most important to understand is that genetic engineering can, in principle, create organisms with novel combinations of adaptive traits, and at the same time avoid extensive abuse to wild-type genomes that have been balanced by aeons of natural selection. Traditional breeding, for all practical purposes, would fail in attempts to make radical adaptive improvements because of the genetic losses that are difficult to avoid with radical allelic substitutions and the genetic hitch-hiking effects that can accompany them.
Access to the non-Mendelian portions of the genome
A naive 'believer' in Darwinism might think that one should be able to alter dogs through artificial selection until one has made something as different as a cat, or modify corn until one has made something as different as wheat. But no breeder has even come close.
In part this is because many phenotypic features are highly conserved and vary little, if at all, within and between species in a taxon. For example, the number of backbone elements does not vary in mammals or in some taxa of salamanders. But the number does vary in other taxa of salamanders and in fishes. Rigorous phenotypic conservation implies that significant parts of genetic programs may normally be 'hidden' beyond the reach of selection. This seems to be due to functional genetic monomorphism (and/or to a buffering of underlying genetic polymorphisms, which for our purposes has much the same implications).
Traditional breeding has been largely restricted to modifications of those phenotypic traits that vary in Mendelian fashion. Indeed, natural selection may work extremely slowly for such reasons, and due to the unbalancing effects of radical allelic substitutions, even though there are vast amounts of neutral and balanced mutations, recombination, and gene-flow in natural populations (Schmalhausen 1986).
Evidence suggests that many genetic instructions are functionally monomorphic. Breeders have not produced dogs with four teeth, two tongues, fifty backbone elements, and so on, because the basic architecture of dogs is 'locked up' beyond the reach of artificial selection. There is not Mendelian variation in the number of vertebral elements in dogs, as there is in some fishes and salamanders, and thus there has been no way to change these by conventional breeding, though the dog's backbone is obviously under the control of genetic instructions.
It is true that relatively recently, various mutagens and transposons have been used to 'open up' parts of genomes that have been 'beyond the reach' of artificial selection. But these have been largely haphazard changes and have not allowed a systematic remodelling of an organism's basic architecture or physiology.
Biotechnology, though, offers the realistic prospect of systematically reprogramming sectors of an organism's basic genetic command systems even when there are not Mendelian polymorphisms. This potential to systematically reprogram the 'hidden genome' could in principle be used to create highly unusual and effective ecological competitors or adaptively competent forms with altered ecological functions. [It should be added that the arguments are invalid that internal rearrangements of a genome are not really genetic engineering and/or should categorically be excluded from regulatory oversight. In principle, internal rearrangements of an organism's genetic programming can produce profoundly significant changes in its biology (e.g. Hall 1983; Raff & Kaufman 1983). Then too, any natural DNA sequence could be mimicked by systematically rearranging the bases from a length of DNA. (This is something that cannot be done with conventional breeding.) In principle, there is thus no fundamental difference between a transgenic GEO, and one with an identical DNA code added to it that was made by rearranging its own DNA sequences. In principle, either molecular technique could be used to make an ecologically competent organism more competitive or alter its ecological functions.]
Summary of basic differences
Biotechnology has exciting promise precisely because it has the potential to allow one to modify genomes according to new principles, including phylogenetic leapfrogging, the bypass of traditional genetic trade-offs, and rearrangement of normally inaccessible genetic programs. Yet it is true that genetic engineers can, when they select to, use the new techniques to make trivial genetic modifications that are very similar to those that have been made by traditional breeding, or even to those that simple sexual recombination produces. Similarly, one could surely use horses to pull an automobile and use this as an example of how automobiles are basically like carts. But it is completely misleading to claim that the simple examples represent the full potential either of genetic engineering or of automobiles.
Reasons why ecologically incapacitated organisms should not, in any event, become supercompetitive
Regardless of the above, many transgenic forms will be non-competitive because,
1 The parent organisms were highly modified forms such as extensively domesticated corn or E. coli K-12 to begin with.
2 There may be cases in which the genetic engineering process itself does demonstrably incapacitate the transgenic form ecologically. (Usually, however, breeders eliminate the debilitating effects that are often seen in the first few transgenic generations. Natural selection would tend to improve fitness as well.)
3 If the host is the sort of foreign wild species that simply cannot persist without human help under local conditions of inappropriate weather, soil, etc., biotechnology is unlikely to tum it into a locally ecologically vigorous organism. Thus, no plant is likely to become an irreversible ecological problem unless it can persist and reproduce in uncultivated ground (see also Gillett et al. 1986).
So let it be clear that if the host is to begin with so genetically debilitated by centuries of artificial selection that it is an ecological incompetent (e.g. corn), biotechnology is unlikely to turn it into an ecologically vigorous organism.
Yet, what is in a word? Scientists and regulators should not try to take comfort in the notion that a transgenic organism will be safe simply because the host has been 'domesticated'. The particular biological traits of the domesticated organism must be understood. Obviously, there can be artificial selection without ecological incapacitation. Horses, for example were domesticated for thousands of years. Yet they became feral when the Spaniards reintroduced them to North America, and wild mustangs have even become ecological problems in some parts of the United States. The world has seen too clearly how so called 'domesticated' bees in Brazil became a spreading menace when the genes of African bees were added to their populations (Page 1989). The word 'domestication' only means 'to bring into the house'. It is not a guarantee of safety.
Serious ecological concerns nevertheless remain
Not all hosts, including some domesticated ones, will necessarily be ecological incompetents
Although many types of GEOs will be ecologically incompetent, serious ecological concerns remain first because not all hosts that are and will be engineered, and not even all row-crop plants, have been as genetically manhandled as corn, wheat, and white rats.
Forage grasses, alfalfa, trees, wild rice, cranberries, sunflowers, numerous species used in landscaping, fishes, shellfish, insects, and so on are much closer to wild-types than are corn or wheat.
When novel adaptive traits are taken from a donor and added to such hosts, the transgenic forms will be much more difficult to screen for ecological safety than genetically engineered corn in Iowa or genetically engineered tropical bromeliads in Minnesota, which are quite unlikely to cause ecological problems.
This group of species, of unknown membership, seems to present the greatest challenge for risk assessment of plants or animals at the present state of the technology. Ecologically competent GEOs with new adaptive features have been compared to exotic or introduced species.
The record of natural species that were brought to one continent from another suggests that 90% have not caused significant problems. But the remaining 10% have caused significant problems, and that includes a substantial number of great disasters (Sharples 1982; Simberloff 1985; Regal 1993). [See Keeler & Turner (1991) for a review of the potential for problems of weediness in transgenic plants, also Sukopp & Sukopp (1993), Williamson (1993). Vitousek (1985) has done an excellent job of summarizing the significant ways in which exotic plants have sometimes altered soil chemistry and hydrology.]
Lateral transfer of genes to ecologically competent relatives
Even ecologically incapacitated crop plants or animal species, might in theory pass novel adaptive genes on to relatives and enhance the competitiveness of these (Regal 1986, 1990b; Doebley 1990; Ellstrand & Hoffman 1990; Wilson 1990; Keeler & Turner 1991).
Thus, even the most ecologically incapacitated crop plants may be of concern in the sense that they have wild relatives with which they might cross somewhere in the world, such as corn in the Mexican highlands or wheat in the Middle-East.
A less obvious concern is that if adaptive transgenes enter a related wild population and give the progeny of some individuals a large competitive advantage, the genotypes of these could, by 'genetic hitch-hiking', in some cases sweep the population, eliminate other genotypes, and reduce the amount of genetic variation. This could have practical implications, because plant breeders at some time in the future might have had important uses for the genes that were eliminated from these natural 'germ-plasm banks'. Study of this issue is needed.
Many crop plants will be genetically engineered not simply for food purposes, but to produce drugs and industrial chemicals in large amounts. In this case the challenge for regulators will be to anticipate any effects of chemical residues in the fields on wildlife or on soils or on water quality, in cases where the chemicals are not quickly degraded. Agencies should also be alert to the possibility that these chemicals could enter the food supply if pollen from a field of such transgenic plants blows or is carried by insects into fields where food crops are being grown. The United States currently has growing distance standards for commercial seed purity that take the potential for cross pollination into account. But higher standards of quality will probably be desired when the issue is food safety rather than commercial effectiveness.
How to think about uncertainty due to universal pleiotropy
Ecologically adaptive pleiotropic effects?
The interconnectedness of biochemical systems in organisms means that commonly there are unintended effects of genes on phenotypes. 'Pleiotropy' is the best known term related to genetic 'side-effects' (Raff & Kaufman 1991). One would not usually expect pleiotropic effects to be adaptive, however, and a brief hands-on inspection by a qualified ecologist should easily verify this for the majority of projects. Yet pleiotropy makes it difficult to forgo all direct examinations and screenings whatsoever, because in principle some small subset of genetic side-effects could increase adaptiveness.
Pleiotropy does raise special issues with regard to food safety, toxicity and allergies. Plants defend themselves from insects and mammalian herbivores and pathogens with bioactive chemicals. The concern with transgenic plants is that biochemical novelty could sometimes alter biosynthetic pathways and cause a plant to produce bioactive compounds at new high levels, or produce completely new toxic compounds.
Thus, even ecologically safe plants could in theory present food safety problems and require appropriate caution. Indeed, while the links are not completely understood between the novel L-tryptophan dimer that may have been produced as a 'side-product' by genetically engineered bacteria (and that charcoal filtration apparently failed to remove) and the eosinophilia mylagia and deaths that were caused in many people who ingested the dimer, the tragic case does thus far resemble the sort of scenario that one would be concerned about (House of Representatives 1991).
Is pleiotropy enhanced in transgenic organisms?
Critics will argue that pleiotropy has caused few if any problems with traditional breeding and should present no new problems for transgenic organisms. And, of course, sometimes a transgene will not even express at all, or only weakly, in a new host. Campbell (1990) has predicted that pleiotropy will not create major adaptive novelties in transgenic forms.
Yet in fact both evidence and theory do suggest that pleiotropic effects are and should commonly, though not always, be greater in transgenic than in normal forms, even if pleiotropy is unlikely to create major adaptive novelties. The evidence that pleiotropic expression may often be unusually great in transgenic organisms suggests a need for at least some empirical observations to assure food safety, and to a lesser extent to assure ecological safety even for transgenic forms that one would idealistically suppose to be safe judging only from the primary effects of the transgenes and/or from teleological descriptions of the purposes of a given project alone.
Ninio (1983, pp. 109-110) for example, summarized studies that showed that certain types of molecular cross reactions between an enzyme and tRNAs are stronger when the molecules come from more distantly related species. Apparently this is because 'The perfected defences which the activating enzyme has against undesirable tRNAs of its own cell become ineffective with foreign tRNAs which it charges because there is nothing to tell it to exclude it.'
In a review of transgenic mammals, Palmiter & Brinster (1986) reported that often, though not always, there were a variety of large and unexpected phenotypic 'side-effects'. The rather common surprises in genetic engineering have even attracted the attention of the popular press: 'Genetic Vegomatics Splice and Dice With Weird Results' (Wall Street Journal April 13,1992, pp. Al and A5, midwest edition).
The bizarre nontarget effects commonly and indisputably seen in transgenic organisms may in theory be caused by many factors, not simply pleiotropy (which is a somewhat vague concept in any event). Gene-splicing processes may also cause mutations, for example.
In any case the result and the concem is similar; new factors may be added to the host's biochemical milieu and cause quantitative or qualitative changes in the output of existing biochemical pathways. Systematic research would very much help to develop a body of precise knowledge about the side-effects so commonly produced by genetic engineering.
Such often dramatic side-effects have come as a surprise to molecular biologists and the public who have had the incorrect belief that the relationship between genotype and phenotype is simple, but the side-effects should not be surprising to evolutionary biologists and developmental geneficists. Charles Darwin (1896) in The Variation of Animals and Plants Under Domestication, long ago cited numerous cases of reversions or atavisms that resulted from hybrid crosses and he concluded that there is a great amount of hereditary variation that normally remains invisible. In modem terms, we would say that for more than a century, similar and other abundant facts have suggested that there is extensive buffering or control of potential generic expression, and that this buffering can become ineffective under unusual conditions (see also Regal 1977; Schmalhausen 1986).
In short, theory and evidence have suggested that the host's buffering or control systems will often be ineffective for those transgenes that can express well.
Judging potential hazards: field plots and commercialization
Small field populations of GEOs can provide valuable data to help make decisions about widespread commercial releases. But one cannot claim that since plants in small confined and ecologically irrelevant field plots, plots used largely to study commercial features, have not 'caused problems' or have not 'caused surprises' then it will be safe to truly release any transgenic forms commercially.
For example, 'no adverse consequences have resulted from work in more than fifteen years in laboratories and in over 500 field releases' (Casper & Landsmann 1992, p. xiii). The term 'releases' is completely misleading. These were largely not scientific tests of realistic ecological concems.
It is hard even to imagine a case where one might have concerns that ecological problems might arise from widespread release, and where one would expect to see 'problems' by simple inspection of field plots, especially if they contained no potential native competitors!
After all, ecological problems are only apt to occur within the context of biological and physical interactions that take place on natural soils and within a natural community of competitors. Yet this sort of nondata on nonreleases has been cited in policy circles as though 500 true releases have now informed scientists that there are no legitimate scientific concerns.
Moreover, time scale and geographic scale can be important in determining whether a population will gain an ecological foothold before spread begins. The history of rogue species tells us quite counter-intuitively that in many cases there were series of deliberate attempts to introduce species that repeatedly failed, even though in the end the same species became enormous problems. The reasons for this oddity have not always been clear, but it is not hard to propose probable causes from the details of the evidence.
1 Particular weather conditions in critical seasons, and favourable series of years, may be necessary for a plant or animal population to establish and spread to form local demes.
2 Time and a range of ecological exposures may be required for a founder population to reach a critical size and high enough levels of genetic diversity for natural selection to balance it genetically to the spectrum of local microhabitats.
3 Local ecological success, especially in the face of local competition, can be a matter of the balance of rates of a population's deaths and demic extinctions against its rates of reproduction and dispersal. A population may have to attain a critical mass or a microhabitat foothold to begin to reproduce reliably despite statistical fluctuations in mortality.
Yet even though casual observations of field plots may not resolve ecological concerns, the careful and organized study of transgenic plants in field plots could be absolutely essential to detect and to understand biological properties that can be used in making scientific judgements of potential risks and hazards when the time comes to make decisions about a given true release. There is valuable information that can be gathered from agricultural field plots about seed dormancy, vegetative reproduction, seed dispersal, fecundity, pollination biology, water and mineral requirements, vegetative characteristics, growth and germination responses to changes in weather, resistance to natural diseases and predators, and so on. This is information that skilled plant ecologists may need in order to make informed judgements about the potential persistence and ecology of the plant in a variety of potential habitats if its general commercial release were allowed.
Thus, the strategy for the scientific use of field plots should be to:
1 survey and list the possible ways in which the particular transgenic form might or might not cause problems in general release, based on knowledge of the ecology and distribution of the parent species and its relatives, and the engineered trait(s);
2 use this list to define and focus on those particular questions about the biology of the plant that can be answered from studies in field plots and in the laboratory. The required data will vary from project to project;
3 collect the required data from field plots. Also, make an effort to observe and note any unanticipated new features that may have emerged from pleiotropic interactions and such.
4 armed with the actual data, skilled plant ecologists can begin to make informed judgements. More data may be necessary or not. Actual tests, experiments, may be needed or not. Permission for commercial releases will have to be based on informed judgements by experts with the proper ecological backgrounds and armed with critical data from properly designed field plot studies when appropriate. This would be a systematic scientific approach to making evaluations.
Yet Wrubel et al. (1992) surveyed the documentation for officially approved field trials of transgenic organisms and found that little thought had tended to go into what data should be collected or not to help make decisions about the safety of potential commercial releases. Such unsystematic approaches are not science-based decision-making.
Brill Wj (1988) Why engineered organisms are safe. Issues in Science and Technology, 1988 (1), 40-50.
Campbell A (1990) Recombinant DNA: past lessons and current concerns. In: Introduction of Genetically Modified Organisms into the Environment (eds Mooney H, Bernardi G), pp. 9-13. Wiley, New York.
Campbell A (1991) Microbes: the laboratory and the field. In: The Genetic Revolution: Scientific Prospects and Public Perceptions (ed. Davis BD), pp. 28-44. Johns Hopkins University Press, Baltimore.
Casper R, Landsmann j (1992) Proceedings of the 2nd International Symposium on The Biosafety Results of Field Tests of Genetically Modified Plants and Microorganisms, 11-14 May 1992, Goslar, Germany. Biologische Bundesanstalt fur Landund Forstwirtschaft, Braunschweig, Germany.
Colwell RK (1989) Ecology and biotechnology: expectations and outliers. In: Risk Analysis Approaches for Environmental Releases of Genetically Engineered Organisms (eds Fiksel J, Covello VT), pp. 163-180. Springer-Verlag, New York.
Darwin C (1896) The Variation of Animals and Plants Under Domestication. D. Appleton, New York.
Davis BD (1977) The recombinant DNA scenarios: Andromeda strain, chimera, and golem. American Scientist, 65, 547-555.
Davis BD (1979) Evolution, epidemiology, and recombinant DNA. In: The Recombinant DNA Debate (eds Jackson DA, Stich S), pp. 137-154. Prentice-Hall, Englewood Cliffs, Nj.
Davis BD (1987) Bacterial domestication: underlying assumptions. Science, 235,1329-1335.
Davis BD (ed.) (1991) The Genetic Revolution: Scientific Prospects and Public Perceptions. Johns Hopkins, Baltimore.
Doebley J (1 990) Molecular evidence for gene flow among Zea species. Bioscience, 40, 443-448.
Ellstrand NC, Hoffman CA (1990) Hybridization as an avenue of escape for engineered genes. Bioscience, 40,438-442.
Gillett JW, Stern AM, Levin SA, Harwell MA, Alexander M,
Andow DA (1986) Potential impacts of environmental release of biotechnology products: assessment, regulation, and research needs. Environmental Managemnent, 10, 433-463.
Grobstein C (1986) Asilomar and the formation of public policy. In: The Gene-Splicing Wars: Reflections on The Recombinant DNA Controversy (eds Zilinskas R, Zimmerman BK), pp. 3-10. Macmillan, New York.
Hall BG (1983) Evolution of new metabolic functions in laboratory organisms. In: Evolution of Genes and Proteins (eds Nei M, Koehn RK), pp. 234-257. Sinauer, Sunderland MA.
Halvorson HO, Pramer D, Rogul M (1985) Engineered Organisms in the Environment: Scientific Issues. American Society for Micrbiology,
Washington, E)C.House of Representatives (1991) FDA's Regulation of the Dietary Supplement L-Tryptophan. Human Resources and Intergovernmental Relations Subcommittee of the Committee on Government Operations, United States House of Representatives, Washington, E)C.
Knoll AH (1991) End of the proterozoic eon. Scientific American, 265, 64-73.
Koestler A, Smythies JR (1969) Beyond Reductionism: New Perspectives in the Life Sciences. Beacon Press, Boston.
Krimsky S (1991) Biotechnics and Society: The Rise of Industrial Genetics. Praeger, New York.
Laycock G (1970) The Alien Animals: The Story of Imported Wildlife. Audubon/Ballantine Books, New York.
McCormick D (1992) Frankenfood ... or frank discussion? Biol Technology, 10, 829.
National Academy of Sciences (1987) Introduction of Recombinant DNA-engineered Organisms into the Environment: Key Issues. National Academy Press, Washington, DC.
National Research Council (1989) Field Testing Genetically Modified Organisms: Framework for Decision. National Academy Press, Washington, DC.
Ninio J (1983) Molecular Approaches to Evolution. Princeton University Press, Princeton, NJ.
Office of Technology Assessment (1988) New Developments in Biotechnology, Vol. 3. Field-Testing Engineered Organisms: Genetic and Ecological Issues. US Government Printing Office, Washington, DC.
Palmiter RD, Brinster RL (1986) Germ-line transformation of mice. Annual Review of Genetics, 20,465-499.
Page RE (1989) Neotropical African bees. Nature, 339,181-182.
Raff RA, Kaufman TC (1991) Embryos, Genes, and Evolution: The Developmental-Genetic Basis of Evolutionary Change. University of Indiana Press, Bloomington.
Regal PJ (1977) Evolutionary loss of useless features: is it molecular noise suppression? American Naturalist, 111, 123-133.
Regal PJ (1982) Pollination by wind and animals: ecology of geographic patterns. Annual Review of Ecology and Systematics, 13, 497-524.
Regal PJ (1985) The ecology of evolution: implications of the individualistic paradigm. In: Engineered Organisms in the Environment: Scientific Issues (eds Halvorson HO, Pramer D, Rogul M), pp. 11-19. American Society for Microbiology, Washington, DC.
Regal PJ (1986) Models of genetically engineered organisms and their ecological impact. In: Ecology of Biological Invasions of North America and Hawaii (eds Mooney HA, Drake JA), pp. 111-129. Springer-Verlag, New York.
Regal PJ (1987) Safe and effective biotechnology: mobilizing scientific expertise. In: Application of Biotechnology: Environmental and Policy Issues (ed. Fowle J.R., 111), pp. 145-164. Westview Press, Boulder, CO.
Regal PJ (1988) The adaptive potential of genetically engineered organisms in nature. Trends in Biotechnology, 6 (4), s36-s38. (Special combined issue of TIBTECH and TREE, April 1988.)
Regal PJ (1989) Biotechnology jitters: will they blow over? Biotechnology Education, 1, 51-55. Regal PJ (1990a) The Anatomy of Judgment. University of Minnesota Press, Minneapolis.
Regal PJ (1990b) Gene flow and adaptability in transgenic agricultural organisms: long term risks and overview. In: Risk Assessment in Agricultural Biotechnology: Proceedings of the International Conference (Davis, California, 1988) (eds Marois JJ, Bruening G), pp. 102-110. Publication No. 1928, Division of Agriculture and Natural Resources, University of California, Oakland.
Regal PJ (1993)The true meaning of 'exotic species' as a model for genetically engineered organisms. Experientia, 49, 225-234. Schmalhausen II (1986) Factors of Evolution. University of Chicago Press, Chicago.
Sharples FE (1982) Spread of Organisms with Novel Genotypes: Thoughts from an Ecological Perspective. ORNL/TM-8473, Oak Ridge National Laboratory Environmental Sciences Division Publication No. 2040. 1982. (Reprinted in Recombinant DNA Technology Bulletin, 6, 43-56.)
Sharples FE (1987) Regulation of products from biotechnology. Science 235:1329-1332.
Sharples FE (1991) Ecological aspects of hazard identification for environmental uses of genetically engineered organisms. In: Risk Assessment in Genetic Engineering: Environmental Release of Organisms (eds Levin MA, Strauss HS), pp. 18-31. McGrawHill, New York.
Simberloff D (1985) Predicting ecological effects of novel entities: evidence from higher organisms. In: Engineered Organisms in the Environment: Scientific Issues (eds Halvorson HO, Pramer D, Rogul M), pp. 152-161. American Society for Microbiology, Washington, DC.
Sukopp H, Sukopp U (1993) Ecological long-term effects of cultigens becoming feral and of naturalization on non-native species. Experientia, 49, 210-218.
Tiedje JM, Colwell RK, Grossman YL et at. (1989) The planned introduction of genetically engineered organisms: ecological considerations and recommendations. Ecology, 70, 297-315.
Vitousek PM (1985) Plant and animal invasions: can they alter ecosystem processes? In: Engineered Organisms in the Environment: Scientific Issues (eds Halvorson HO, Pramer D, Rogul M), pp. 169-175. American Society for Microbiology, Washington, DC.
Williamson M (1993) Invaders, weeds and the risk from genetically manipulated organisms. Experientia, 49, 219-224. Wilson HD (1990) Gene flow in squash species. Bioscience, 40,449455.
Wrubel RP, Krimsky S, Wetzler RE (1992) Field testing transgenic plants: an analysis of the US Department of Agriculture's environmental assessments. BioScience, 42, 280-289.
This paper is part of an ongoing analysis to determine the proper theoretical foundations for attempting to predict the adaptability or non-adaptability to natural environments of organisms with novel features and genetic programs. The author is a Professor in the College of Biological Sciences, University of Minnesota, and has long studied patterns and mechanisms of adaptation to natural environments in animals and plants. His studies of the principles of adaptation have included research in physiological ecology, functional morphology, behaviour, community ecology, tropical ecology, basic evolutionary theory, and the history of ideas in science and philosophy.
(Published at this website with the permssion of the author)
"Genetically Engineered Food - Safety Problems"