Guide to Risk Assessment and Biosafety in Biotechnology, GRABB

Forces affecting the rate of spread and increase of hybrids between genetically modified crop plants and their related species remain qualitatively similar, irrespective of whether genetic modification was achieved using traditional methods, those of biotechnology or as a result of the natural evolutionary process. However, the precise magnitude of the forces and, consequently, the likely environmental impact of such hybrids, may depend stronend strongly on the nature of the gene or genes introduced into the native species. While many classes of transgenes are similar to those manipulated by conventional breeding techniques or evolution, biotechnology offers the potential to introduce genes into crops which are novel both from the point of view of function and origin. The qualitative similarity between transgenes and the products of conventional or evolutionary modification suggests that a historical view of the environmental impact of hybrids between traditionally produced crops or exotic species and their relatives would be of use in estimating the probable fate of hybrids containing transgenes in the environment. However, with certain classes of transgenes for which there are no existing analogues, there will need to be greater care in assessing the possible risks associated with release into the environment.

This chapter will give a brief overview of some of the consequences of hybrids produced naturally and between conventionally bredd crop plants and their related species. Some specific case studies involving the impact of such hybrids on the agricultural and natural environment will be presented. Finally, a brief review of the forces responsible for changing the frequency of individuals in populations will be given, together with examples of important interactions between these forces and other ecological parameters.

Keywords: biotechnology, risk assessment, hybridization, gene flow, genetic introgression, invasion.

The consequences of releasing transgenic crop plants into the environment may have an impact on the environment in a number of different ways, ranging from the effects of the transgenic organism itself on indigenous species, to the use of a product derived from the crop. The purpose of this chapter is to describe the forces that may act on a hybrid formed between the transgenic crop and its wild relatives. It. It is highly likely that such hybrids will be formed when GM plants are released into their centres of origin. It is extremely important therefore, to discuss the possible impact of such hybrids on that environment, particularly since centres of origin are usually synonymous with centres of diversity and may contain very valuable genetic resources.

Since it is a reasonable expectation that transgenes will be transmitted from generation to generation in a manner which is familiar to geneticists, that is, under the laws of Mendelian genetics for nuclear genes and under more or less uniparental (often maternal) inheritance for cytoplasmic genes, examples attesting to the fate of inter-specific hybrids between non-transgenic species will be considered appropriate biological models in this context.

The possible influence of the hybrid on existing plant communities will be discussed in several sections: a) a perspective on the origin and evolution of crop plants, b) plant invasion and inter-specificic hybridization, c) the effects of evolutionary forces on the fate of hybrids and d) the effects of structure of the indigenous recipient plant population on the rate of spread and increase of a hybrid.

A study of the evolutionary history of modern crop species shows that a large number are the products of inter-specific hybridization. Notable examples include wheat, which has changed from the no longer cultivated, diploid, Einkorn and Emmer, through several allotetraploid species (including some still in commercial cultivation, e.g. Triticum durum) to the allohexaploid bread-wheats including T. aestivum. The bread-wheats contain three different haploid genomes and probably arose as interspecific hybrids between early crops and indigenous weedy grasses followed by a doubling in ploidy to recover sexual fertility. The different hybridization events os occurred as the geographic range of cultivation radiated out from the centre of diversity in the Fertile Crescent, both during and after the Neolithic expansion. While wheat is perhaps an extreme example with respect to the number of wild species involved in hybridization, it is by no means an exception (see Simmonds, 1976, for examples).

The above examples are indicative of the extent to which interspecific hybridization occurred between early crop plants and their weedy relatives. It should be noted that these hybridization events were of great significance for crop improvement, but they represent an impact on the crop itself and not on the indigenous species. Regrettably, it is inevitable that evolutionary studies should have concentrated on the crop plant and tended to ignore the weed species. In many respects, this is due to a lack of suitable material from non-domesticated species on which to work. Nevertheless, it leaves the question unanswered as to the degree of impact of the crop on its inddigenous weedy relatives. However, it is almost inconceivable that such a reciprocal effect did not occur, altering the indigenous weeds to an extent that unfortunately cannot be estimated.

More recent studies, using modern techniques, have been applied to looking at the relationship between the three forms of beet, Beta vulgaris (Boudry et al., 1992). The three forms are inter-fertile and are:

- Beta vulgaris ssp. vulgaris sugar-beet, fodder beet & beetroot

- Beta vulgaris ssp. maritima sea beet or wild beet

- Beta vulgaris ssp. unknown weed beet or ruderal beet

Weed beet was first reported as a problem to agriculture in the USA during the 1920's, with the situation worsening rapidly in California in the 1950's. The problem in northern Europe became apparent in the 1970's and is a serious problem for some sugar-beet growers in continental Europe and, to a lesser extent, the UK. The major difference betwebetween weed- and sugar-beet is the requirement for plant vernalization to induce flowering in the crop. This character is under a simple one-locus, two-allele control with the allele for bolting dominant to that for non-bolting. The maritime populations of sea-beet in Europe show great variability for vernalization requirement. However, in the regions of sugar-beet seed multiplication in France, ruderal beet is homozygous for the bolting allele. Recurrent hybridization between seed multiplication lines and ruderal beet has resulted in the transfer of the bolting characteristic to individuals within the sugar-beet varieties, with the resulting "weed-beet" problems when the crop is grown in northern Europe. An analysis of cytoplasmic markers in the sugar-beet varieties and weed-beet suggests that weed beet is the result of a cross between ruderal beet as the male parent and the crop as the female parent. Studies of ruderal populations around the peripheries of the main areas of sugar-beet seed production in France show that both nuclear genes and, less frequently, cytoplasmic genes from the crop have introgressed into wild or weed populations. The environmental impact of this exchange of genes is not, however, clear.

Interspecific hybridization is one of the major mechanisms for generating new taxa in the plant kingdom (Stebbins, 1959, Stace, 1975, Grant, 1981). There are many cases in which hybridization between an indigenous and alien plant species has resulted in a new, sexually fertile taxon. One well-documented example of this is the result of a hybridization event between Senecio vulgaris, common groundsel, and S. squalidus, the Oxford ragwort, (see Abbott, 1992, for a review). S. vulgaris is a native herb of the British Isles. The common form, S. vulgaris var. vulgaris, produces a capitulum containing only disc florets, which tend to be be highly self-fertilizing and to produce relatively non-dispersed seed. A more recent form, S. vulgaris var. hibernicus, differs from the common form in that ray florets are produced in the capitulum. Seed obtained from ray florets tends to be more outcrossed and to disperse further than that of disc florets. The difference in capitulum type is due to variation at a single locus (Abbott et al., 1992).

The first report of wild radiate groundsel was in 1832 from Oxford, England, following the escape from the Oxford Botanic Garden in 1794 of the radiate species S. squalidus, a native of Sicily. Since this time, S. squalidus has spread widely throughout the British Isles, often following railway tracks, whose limestone ballast provides a well-drained medium somewhat reminiscent of its native Sicily. At the present time, S. squalidus is often found growing alongside S. vulgaris in open, disturbed habitats. Highly self-sterile hybrids (S. x baxterii) are formed at low frequency with S. vulgaris as the female parent but backcrosses to S. vulgaris are fertile. The hybrid radiate groundsel, S. vulgaris var. hibernicus, is postulated to be a stabilized introgressant, resulting from repeated backcrossing of S. x baxteri with S. vulgaris var. vulgaris, on the basis of three lines of evidence: (1) the strong similarity of the pattern of spread of S. squalidus and S. vulgaris var. hibernicus throughout the British Isles over the past 160 years; (2) tetraploid plants, produced by backcrossing S. x baxteri to S. vulgaris var. vulgaris, bear a close resemblance to S. vulgaris var. hibernicus; and (3) an allozyme marker, absent from non-radiate groundsel populations but present at high frequency in S. squalidus, is found at intermediate frequencies in radiate groundsel.

Another species appears to have resulted from the original sself-sterile hybrid, S. x baxteri, namely S. cambrensis. In this species, the original sterility barrier has been overcome by a doubling in ploidy from the hybrid (2n=30) to S. cambrensis (2n=60). Evidence from isozymes and cpDNA (Ashton & Abbott, 1992, Harris & Ingram, 1992) has confirmed the allopolyploid nature of S. cambrensis and shown it to have originated in at least two separate localities.

The example above is one of several indicating that self-sterility of a hybrid is not a sufficient barrier, per se, to invasion by genes derived from an alien into an indigenous species. Indeed, the production of fertile allopolyploids and stabilized introgressants are well documented, and not uncommon, phenomena whereby genes from exotic species have been combined with those of the native flora.

There is a tendeendency for workers attempting to quantify the impact of a hybrid between a genetically modified crop and an indigenous species to concentrate on measuring the 'fitness' of that hybrid. Fitness is only one of a number of evolutionary forces that may affect the frequency of genes in a population. When a hybrid first appears as the result of an 'escape' from a transgenic crop, it is likely to be at a relatively low frequency in the indigenous population and the major determinant of its fate will be random genetic drift. In addition, the early rounds of sexual reproduction, after such an escape, will be mainly backcrosses between the hybrid and the native species, making estimation of the fitness of the hybrid even less relevant in terms of predicting the fate of the crop genes that it carries. Another force, of particular relevance to these hybrids, is the recurrent migration of genes from crops into indigenous species. This force will have a magnitude which will depend on the probability of gene transfer (=hybrid formation) and the scale on which the crop is grown relative to the size of the indigenous recipient species.

Consider, for example, the case of a transgenic crop which can form hybrids with an indigenous ruderal or weedy species. As the scale on which the crop is grown increases, so does the magnitude of gene flow from the crop to the weed. As soon as the magnitude of this gene flow increases to beyond that of gene exchange of the weed with other local populations of the same species, the frequency of genes of crop origin will increase in frequency in the weed population, irrespective of whether the hybrids show increased relative fitness. Indeed, the magnitude of this directional gene flow may be sufficient to overcome slight fitness disadvantages of the hybrid. This type of phenomenon is clearly dependent on the scale on which the crop is grown and, as such, spread of genes of crop origin may not be detected in the small scale field trials which would normally precede the full scale commeercialization of a transgenic crop plant.

It is, therefore, perhaps more pertinent when measuring fitness of hybrids between transgenic crops and their relatives to be reassured about the lack of risk only when the hybrid can be shown to be markedly less fit than the weedy species or when the introduction of crop genes into indigenous weeds can be expected to pose no environmental threat.

In addition to the evolutionary forces acting on the genotypes of hybrids determining their rate of increase, another factor that needs to be taken into account is the way in which the recipient population is distributed in space and time. Simplistic estimates can be made of rates of increase of genotypes, given estimates of the magnitude of the evolutionary forces acting (e.g. Kareiva et al., 1991). However, these estimates generally fail to take account of any spatial orl or temporal heterogeneity of the recipient populations. This heterogeneity may be due to different magnitudes and/or directions of evolutionary forces acting in time or space or may simply be due to partial barriers to gene flow in patchily distributed populations.

Considering a simple neutral gene model, it is easy to show that the probability of a particular gene reaching fixation in a population depends only on the frequency of that gene and is independent of the structure of that population (Maruyama, 1970,1971). However, both the rate of spread of a gene and its rate of increase are strongly dependent on the structure of that population, being determined by the connectivity of the patches (number of connections and magnitude of gene flow among them).

In particular, in a highly subdivided population, it is relatively easy for a gene to become locally abundant (even in the face of a weak selective disadvantage), but it will rarely spread far. In contrast, the distance spread by a gene in a a highly connected population may be large but it rarely reaches a high frequency in any locality.

Since many agricultural weeds and ruderal species are very patchily distributed, their spatial structure may well be the major determinant of the rates of increase and spread of genes introgressed from transgenic crops. Although population genetic theory, outlined above, allows certain approximate statements to be made concerning the fate of such genes, there has been insufficient work on more 'ecologically realistic' population structures to allow reasonable inferences to be made concerning rates of spread of genes. In particular, if a patchily distributed species suffers repeated local extinction-recolonization events (a metapopulation, Levins, 1971), while the connectivity of the patches will be relatively high (and as a consequence the wide spread of genes outlined above), the prediction of classical theory that a gene will not become locally abundant no longer necessarily holds. The probability of loocal abundance will be determined primarily by the number and origin of founders of a recolonized patch. That is, the local gene frequencies will be determined principally by founder effects.

1) Given the qualitative similarity of hybrids produced by the natural evolutionary process, conventional modification and biotechnology, more emphasis should be placed on the relevance of existing information concerning the fate of naturally or conventionally produced hybrids in the environment.

2) In the absence of existing information, it may often be preferable to design appropriate large-scale experiments using conventionally produced 'biological models' rather than to carry out smaller, more costly, experiments with the products of biotechnology.

3) The presence of selective advantage in a hybrid is not necessary for its spread in indigenous species. It is i is important to recognise the existence of other, scale-dependent, evolutionary forces such as recurrent migration.

4) The rate of spread and rate of increase of a gene in a population may depend as much on the spatial and/or temporal structure of the recipient population as on the biological attributes of the hybrid itself. There is a need for an improved theoretical basis of description of the fate of genes in more ecologically realistic situations such as those described by metapopulation models.

Abbott, R.J. (1992) Plant Invasions, Interspecific Hybridization and the Evolution of New Plant Taxa. Trends in Ecology and Evolution, 7, 401-405.

Abbott, R.J., Ashton, P.A. and Forbes, D.G. (1992) Introgressive origin of the radiate groundsel, Senecio vulgaris L. var. hibernicus Syme: Aat-3 evidence. Heredity, 68, 425-435.

Ashton, P.A. and Abbott,bbott, R.J. (1992) Multiple origins and genetic diversity in the newly arisen allopolyploid species, Senecio cambrensis Rosser (Compositae). Heredity, 68, 25-32.

Boudry, P., Saumitou-Laprade, P., Vernet, Ph. and Van Dijk, H. (1992) Les betteraves mauvaises herbes: origine et évolution. In: IXème Colloque Internationale sur la Biologie des Mauvaises Herbes. Editions INRA.

Grant, V. (1981) Plant Speciation (2nd. edn.). Columbia University Press.

Harris, S.A. and Ingram, R. (1992) Molecular systematics of the genus Senecio L. I. Hybridization in a British polyploid complex. Heredity, 69, 1-10.

Levins, R. (1971) Evolution in Changing Environments (2nd edn) Princeton University Press.

Kareiva, P., Manasse, R. and Morris, W. (1991) Using models to integrate data from field trials and estimate risks of gene escape and gene spread. In: MacKenzie and Henry (eds.), Biological Monitoring of Genetically Engi Engineered Plants and Microbes, ARS.

Maruyama, T. (1970) On the fixation probability of mutant genes in a subdivided population. Genetcal Research, 15, 221-226.

Maruyama, T. (1971) An invariant property of a structured population. Genetical Research, 18, 81-84.

Simmonds, N.W. (1976) Evolution of Crop Plants. Longman, London.

Stace, C.A. (1975) Hybridization and the Flora of the British Isles. Academic Press.

Stebbins, G.L. (1959) The role of hybridization in evolution. Proceedings of the American Philosophical Society, 103, 231-251.