An increasing number of transgenic crops are deliberately released into centres of biodiversity, which are mostly found in developing countries. Since these centres are important for the future of the world’s major food crops, there is great concern about the potential ecological impact of these releases. Information and knowledge about the ecological impact are, however, still marginal. This article focuses on the hypothetical risks, with potato serving as an example.

Biodiversity is described as the total variability within and among species of living organisms and their habitats. Consequently a centre of biodiversity is defined as the geographic region in which the greatest variability of a given species occurs. Since the term "species" includes highly domesticated varieties, farmer-developed cultivars (i.e. adapted landraces), selected lines and non-cultivated wild relatives, it is often difficult to draw a line between crop species and wild relatives, and between the cultivated and the wild area. Moreover, in chickpea, common bean, maize, potato, rice and other crops a "wild-weed-crop complex" is observed, which is characterized by a continuous exchange of genetic material (gene flow) between wild and cultivated forms. In the Northern hemisphere the gene flow might be regarded as a comparably rare event because wild relatives of cultivated crops are exceptional. The centres of biodiversity, which are mostly found in developing countries, however, provide an excellent environment to promote this flux of genetic material.
Biodiversity is continuously evolving and has to be regarded as a highly dynamic process. It is also a resource of genetic material for improving agricultural production by providing superior crop varieties via conventional breeding and genetic engineering.

Transgenic crops
Two obvious and specific characteristics determine the difference between transgenic and non-transgenic crops. Firstly, genetic engineering provides the opportunity to incorporate genetic material without "natural" boundaries. DNA-sequences, encoding for a specific plant characteristic such as virus-resistance or flower pigmentation that would otherwise never, or only in extremely rare cases, belong to a certain species, can be introduced into the plant genome. In other words, genetic engineering adds new genes to the crop’s genome, whereas conventional breeding usually adds variants of a given gene (new alleles). Secondly, the introduced DNA-sequence finds itself in a disturbed genetic context. Although empirical evidence is missing, this may theoretically result in unexpected side-effects. Two indications for these unexpected effects are the observed subsequent silencing of an introduced gene in the transgenic progeny and the variation in expression of a transferred gene depending of the location in the genome, the so-called "position-effects".
Notwithstanding these two differences between transgenic and non-transgenic crops, it is the trait or the acquired property itself, and not the process or method of introduction, that interacts with the environment and determines largely the potential ecological impact. From that point of view, it is not justified to define a specific intrinsic risk for transgenic plants per se.

Hypothetical risks
Traditional farming or the release of transgenic crops may each affect the stability and diversity of an ecosystem. A generalized prediction on the behaviour of transgenic crops is not possible. Consequently, only a few studies address the potential impact of the release of (transgenic) crops, of which most are theoretical in nature. This is surprising, since an ecological impact is not restricted to recombinant DNA or transgenic plants. In particular the behaviour of released exotic species in the past, provided the respective difference between "exotic" and "transgenic" is considered, may serve as a useful empirical model for the release of transgenic crops.
Based on the discussion on theoretical risks, the following three general scenarios that describe how transgenic crops may influence the composition and stability of a natural ecosystem are identified. The three scenarios are based on known or most probable mechanisms. Speculative risks will not be discussed, since proving the non-existence of an unknown risk is a logical impossibility.

(1) The transgenic plant itself becomes a weed, because of its added characteristic. It will leave the area under cultivation and displace wild species. The crop "escapes". It should be noted, that the transgenic crop may also escape because of minor genetic changes which are not related to the transferred DNA. Transgenic plants of highly domesticated species will have a lower probability of escaping compared to low-input crops like pasture legumes, sorghum or cowpea, since the former usually cannot compete with other plants outside the farmers’ fields.
It is however difficult to define what is a weed. Typical properties for weeds are described in the "Baker’s list of characters". Impatiens glandulifera (touch-me-not), for example, has only two hits out of Baker’s list of twelve, but has become invasive. Probably no list is very consistent or predictable.

(2) The introduced DNA is sexually transmitted into the wild population and may confer weediness if expressed in the progeny of a cross between the respective crop and its compatible wild relative (a hybrid). This can be regarded as a probable event, especially in centres of biodiversity. The transmission itself is not restricted to recombinant DNA, but only the recombinant DNA usually has no allelic counterpart in the wild relative.
Any detrimental effects resulting from a hybridization between the transgenic crop and its wild relative, however, are likely to be of subordinate importance. A hybrid may invade the area only if the stressor (e.g. a virus) represents an important pressure outside the area under cultivation. Usually, however, stressors relevant in agricultural areas are irrelevant in wild populations. Assuming that the introduced DNA confers a higher fitness, then spreading of this dominant and single trait into the wild population is faster compared to "natural" recessive or multigenic traits. On the other hand, "natural" recessive and multigenic traits might not be subject to immediate selection and may stay "latent" in a population and thus may not be subject to immediate selection and remain for a longer time as part of the crop’s gene-pool.
"Natural" resistance traits are often poorly understood on a molecular level. Sound comparison of the transfer of genetic material involved with resistance from engineered transgenic or non-transgenic crops to its wild relatives would require more detailed knowledge of the basic processes mediating natural resistances. Crops and their hybrids may establish themselves outside the cultured area (a process which is called "naturalization"), but there is no example of a crop or an interfertile hybrid which has become invasive.

(3) The introduced DNA is transmitted asexually to species of other kingdoms, such as bacteria, viruses and animals. Although there is no experimental proof up to now that this can happen, this horizontal gene transfer does occur in nature. In fact, the horizontal gene transfer mediated by the bacterial plant pathogen Agrobacterium tumefaciens has become the most successful tool in genetic engineering of plants. This poorly understood dynamic gene flow between species might represent a driving force behind evolution. In this respect, the underestimated role of DNA and micro-organisms of unknown function and distribution deserves more attention.
Assuming that a transgenic crop or hybrid has become a weed, this may result in the loss of indigenous species because of competition. Notably, the number of species in an ecosystem does not sufficiently describe biodiversity. The genetic variability within a species, which allows the adaptation to a variable environment, is probably of equal importance. In addition, the establishment of a transgenic crop or the introduced DNA in the natural environment depends on the number of plants, i.e. the size of the founder population. Also the number of introduced genes or alleles (it is not necessary to discriminate here between "genes" and "alleles") may influence the potential weediness. In principle, the more novel characteristics introduced into the system, the higher the probability of finding a respective ecological niche within the system which perfectly fits into the introduced pattern of characteristics. This principle, however, does not exclude the possibility that only one novel gene could be sufficient to become invasive.
The whole discussion on the impact on biodiversity of transgenic crops concentrates on the potential weediness of the new crop (scenarios 1 and 2). Surprisingly, the important question as to whether or to what extent engineered or conventionally acquired resistances against bacteria and fungi do affect the associated soil-borne microflora and endophytic micro-organisms (scenario 3) is usually neglected.

Potato as a model
A basic principle of risk analysis in centres of biodiversity is the assumption that gene flow might occur. Impact analysis has to focus on the consequence, not on the probability, of such a gene flow. The proposed strategy, therefore, consists of two elements:

• To characterize the species under concern (i.e. factors determining sexual compatibility, geographic distribution, climatic requirements and important diseases);
• A case-by-case analysis of the impact of the trait to be introduced.

Eight cultivated species with numerous varieties and several hundred non-cultivated wild relatives of potato are known in its genus Solanum. Many of these species contain resistance genes against pests and pathogens. The main centres of biodiversity are located in Mexico, Peru, Bolivia and northwest Argentina. The genus comprises a polyploid series, i.e. species in which 2 (diploid) up to species with 6 (hexaploid) chromosome sets can be found.
Transfer of introduced DNA from cultivated transgenic potatoes into its wild relatives is possible provided pollinators (mainly bumble bees) and a compatible genetic constitution are available. Gene flow is reduced by cultivating transgenic potatoes in areas in which compatible wild relatives do not occur, pollinators like bumble bees or beetles are absent, and by the cultivation of male sterile varieties.
Another example is maize. The maize genus Zea has four species, all of which are sexually compatible and produce fertile hybrids. However, the formation of hybrids is not observed under natural conditions because of geographic distances, different flowering times and incompatible flower-morphology. Nevertheless, Zea mays ssp. mays and Zea mays ssp. mexicana (teosinte) are a matter of concern since gene flow is observed under experimental conditions, although hybrids do not have a superior fitness. Because of the fact that maize and teosinte do form fertile hybrids, the out-crossing of any DNA-sequence (including recombinant DNA) from Zea mays ssp. mays into teosinte might occur.
 

Releases of transgenic plants in developing countries (by species and introduced trait)  Species Introduced trait Total field  releases Herbicide resistance  Insect resistance Virus   resistance Product   quality  Others  Maize  46 29 16 1 3 5 Soya bean 27 25 - 1 1 - Cotton 24 16 15 - - - Tomato 19 - 2 1 16 - Potato 13 - 1 6 - 6 Subtotal 129 70 34 9 20 11 Others species 30 Total 159 Note: The transfer of more than one trait is possible.  Source: André de Kathen (1996), Gentechnik in Entwicklungsländern: Ein Überblick: Landwirtschaft. Berlin, Germany: Umweltbundesamt.

Risk assessment
Since for both maize and potato gene flow of introduced DNA into the wild cannot be excluded, the following questions are valid for a risk assessment:

• To what extent do specific pests and diseases represent a limiting or selective factor for the development of a wild population?
• To what extent will a respective engineered resistance enhance the biological fitness of the target species and potential hybrids?

Other crops
Although the above considerations appear to be sound and in principle applicable to other crops such as cowpea, cassava and banana, the analysis is still based on theoretical assumptions rather than on empirical data. In order to allow substantiated statements on potential ecological impacts resulting from the release of transgenic plants, more empirical information should be gathered. Risk assessment demands a multidisciplinary approach and should include transgenic crops as well as non-native species and new breeding lines. In this respect, it is surprising that the release of selected cold-tolerant or herbicide-resistant cultivars or exotic species like Trithordeum (a cross between wheat and oats) and Triticale (a cross between durum-wheat and rye) do not receive the same attention as the release of transgenic varieties.
André de Kathen

PostDoc-fellow, Department of Molecular Genetics, University of Hannover, Herrenhäuserstr. 2, 30419 Hannover, Germany. Fax (+49) 511 762 4088; E-mail kathen@mbox.lgm.uni-hannover.de

This article is largely based on André de Kathen (1996), Gentechnik in Entwicklungsländern: Ein Überblick: Landwirtschaft. Berlin, Germany: Umweltbundesamt.

Sources
M. Crawley (1990), "The Ecology of Genetically Engineered Organisms: Assessing the environmental risks." In: H.A. Mooney and G. Bernardi (eds.), Introduction of Genetically Modified Organisms into the Environment. pp. 133-150

N.C. Ellstrand (1994), Are There Unique Risks When Testing in Centres of Diversity. Proceedings of the 3rd International Symposium on the Biosafety Results of Field Tests of Genetically Modified Plants and Microorganisms. University of California. pp. 311-313.

K. Harding and P.S. Harris (1994), Risk Assessment of the Release of Genetically Modified Plants: A review. MAFF -Chief Scientists’ Group, Ministry of Agriculture, Fisheries and Food.

R.J. Frederick, I. Virgin and E. Lindarte (1995), Environmental Concerns with Transgenic Plants in Centres of Diversity: Potato as a model. Stockholm: BAC/SEI and IICA.