Crops of Uncertain Nature? Controversies and Knowledge Gaps Concerning Genetically Modified Crops An Inventory Plant Research International B.V., Wageningen r.12 Aug00

Commissioned by Greenpeace Netherlands, Amsterdam
Plant Research International B.V., Wageningen August 2000 Report 12

Address : Droevendaalsesteeg 1, Wageningen , The Netherlands
P.O. Box 16, 6700 AA Wagenmgen, The Netherlands
Tel. : + 31-317-477000
Fax : + 31-317-418094
E-mail: post@plant.wag-ur.nl
Internet : http://www.plant.wageningen-ur.nl

Cover illustration: Sanne N. de Visser

Commissioned by : Greenpeace Netherlands, Amsterdam
Project leader : A.J.C. de Visser*
Execution: A.J.C. de Visser, E.H. Nijhuis, J.D. van Elsas, T.A. Dueck

*Corresponding author:
Dr A.J.C. (Ries) de Visser
Plant Research International
P.O.Box 16, 6700 AA Wageningen, The Netherlands

visiting address: Bornsesteeg 65, Bldg 122, Wageningen
Tel. +31 317 475852
Fax. +31 317 423110
Email: a.i.c.devisser@plant.wag-ur.nl

Internet: http://www.plant.wageningen-ur.nl

This literature study aims to provide an inventory of gaps in current knowledge of environmental effects of genetically modified agricultural crops. Greenpeace Netherlands has commissioned the report but Plant Research International (Wageningen University and Research Centre) is the sole author and Greenpeace did not place any constraints on the substance of this literature review.

We thank the external reviewers of the report. These are the following scientists, in alphabetical order.

After initial contacts between Miriam van Gool MSc and Dr. Ries de Visser, this study was further defined and executed upon request by Greenpeace Netherlands, involving several meetings of Plant Research International and Greenpeace Netherlands. We like to express special thanks to Miriam van Gool for promoting constructive communication in `building bridges' between Greenpeace and Plant Research International. We would like to thank Prof. Dr. Arjen van Tunen and Dr. Anton Haverkort for continued support and Miriam van Gool MSc, Jan Bijlsma MSc and just van den Broek MSc for their comments, inspiration, trust and constructive and open communication. We thank Rina Kleinjan-Meijering for technical support. We are grateful to the external referees, and to Isabelle Meister of Greenpeace Switzerland and colleagues (Hendrik Terburg BSc; Coen Willemse MSc) for their valuable comments and suggestions at various stages of the preparation of this report. However, as authors of this report we stress that its final text and conclusions are our own sole responsibility.

Summary

This study has identified several areas of controversial, fragmentary or missing knowledge concerning the design, functioning and use of genetically modified (GM) crop plants, from a standpoint of the natural sciences. These areas concern the biological and agronomical sciences which are discussed here, while philosophical, ethical, social-economical and legal scientific areas are indicated to stress their relevance for the public debate on GM (food) crops. Controversies and knowledge gaps appear to be present at all levels of biological organization ranging from the levels of DNA and cellular metabolism to organism and ecosystem levels.

These findings demonstrate the fragmentary nature of current knowledge of genome structure and function and regulation of gene expression in general, and the limited understanding of several physiological, ecological, agronomical and toxicological aspects relevant to present-day and planned genetic modifications of crops. Whether and in which case this limited understanding can be judged as relevant to the public debate on GM and as sufficient for adequate risk assessment are questions requiring further interdisciplinary study.

1. Introduction

1.1 Background

In a recent extensive study by the Nuffield Council on Bioethics in the UK, Ryan et al. (1999) observe that the introduction of genetically modified (GM) crops into the environment and the food chain has become highly controversial in the United Kingdom, parts of Europe and in other parts of the world. These authors note a contrast with the near-indifferent attitude of consumers in the US and Canada with regard to GM crops, which is seen differently by others (e.g. Lacy et al. 1991). The ethical and social issues dealt with in the study by Ryan et al. (1999) are highly relevant to the mentioned controversy but beyond the aim of the present report, which mainly deals with unresolved life-science issues related to GM crops.

Through the use of genetic modification (GM), new organisms (GMOs) have been and are being created which would not likely have been formed otherwise through spontaneous or selection-driven evolution. GMO is defined as an organism in which the genetic material has been altered in a way that does not occur naturally by mating and/or natural recombination (involving several techniques which are listed in annexes of EU Directives 90/219, 220; see web site 14 in section 5.2).

This uniqueness of GMOs, even on an evolutionary time scale, might be based on scientific arguments dealing with the following characteristics of present-day GM (c.f. Gilissen & Nap 1999; Lewin 1990; Maessen 1997; Travers 1993):

Today, about 14 years after the first field introduction of a GM crop, concerns exist among parts of the nearly world-wide society about some of these very basic characteristics of GMOs, especially fundamental aspects like random genome changes, `breaching species barriers', and their potential environmental effects like decrease of biodiversity, and effects on health via food safety and antibiotic resistance (e.g. Van Gool 1999, 2000). It is mainly these new unprecedented aspects which appear to be fundamental to public concerns about the unpredictability of the outcome of interactions of GMOs with agro-ecosystems (e.g. Website 1). Regarding the possibility of the occurrence of unexpected and unintended side-effects of GM-crops, uncertainties and controversies exist among various groups in society, also among scientists (e.g. Lacy et al. 1991; Ewen & Pusztai 1999; Mae-Wan Ho 2000; Van Dommelen 1999, Van Erkelens 1999; De Lange 1999, Voormolen 1999). Such uncertainties and controversies also exist among regulators of EU member states and the European Commission (web site 15). Illustrative examples are the discussions on the `precautionary principle' (e.g. Pape 1999) and the concept of `substantial equivalence', e.g. Millstone et al. 1999; Trewavas 1999; Trewavas & Leaver 2000). Apart from these issues in the area of natural sciences, bases of concern have been identified in other scientific areas which are outside the scope of the present study, but nevertheless may be just as or more important in the public debate (e.g. ethical aspects; Ahuja 1997; Banner et al. 1998; Bruce & Bruce 1998; Ryan et al. 1999).

Greenpeace Netherlands has requested more information on the availability of knowledge relevant to potential decisions on field-introduction of genetically modified crops, in the form of an inventory of scientific knowledge gaps and controversies. Greenpeace is interested in biological-scientific arguments for raising fundamental questions that should be answered when considering field-introduction of genetically modified crops.

This study aims to identify gaps in and borders of the present knowledge of the functioning of transgenic crop plants at the various interacting levels of biological organisation, from molecular to agroecosystem levels, focussing on genetic modification, genome functioning, biochemistry, physiology, toxicology, ecology and agronomy. We do not attempt to answer the question whether the observed knowledge gaps and controversies are serious to the extent that responsible decisions about the field-introduction of (specific) GM crops can or cannot be made, because this requires a wider interdisciplinary study which is outside of the scope of this report.

1.2 Aim, starting-point and scope

This inventory aims to list unresolved issues regarding design, creation, testing, breeding, field-introduction and use of GM crops, by identifying knowledge gaps and controversies based on the current state of biological and agronomical sciences on these subjects.

While we recognize the need for interdisciplinary studies on (the acceptability of) the use of GM crops, we need to stress here, that our starting-point is strictly the natural sciences arena. Within this area, the scope of the study includes unresolved issues in a range of areas, from transgenic crop design and molecular biology to functional genomics, physiological and ecological interactions, and toxicological aspects. Where evident, we indicate the reasons for controversies and the relevance of unresolved questions for the current social debates on uses of GM crops. An extensive, detailed and methodological analysis of the causes of knowledge gaps and controversies requires additional, interdisciplinary research.

1.3 Subjects outside the scope of this report

The scope of the present study is an inventory of unresolved questions relevant to field introduction of GM crops, based on identification of knowledge gaps and controversies. Some subjects and approaches that we consider highly relevant to current debates on the use of GM crops, were necessarily out side the scope of the present study:

1.4 Outline

In the following, an inventory is given of knowledge gaps, controversies and fundamental questions on the topic of agro-ecological risks of (field introduction of) GM crops by reviewing the literature available in June 2000, following the structure shown below. According to Van Dommelen (1999) `all research depends on the questions raised'. This may be so for the research process, however, it does not necessarily apply to all research outcome, because of the important element of serendipity. There is in research practice more `trial and error' and `unexpected success in error' than many may want to admit.

From a natural science stand point, this study deals with biological and agronomic issues, and is structured according to the the different levels of biological organisation. Molecular, cell and ecosystemaspects are strongly related and intertwined (Fig. 1). Examples are: antibiotic resistance (`selectable marker' tool in GM) & human health (increased antibiotic resistance in pathogenic organisms); farmers' interest in better pest control & more rapid resistance development to Bt-toxins in pest insects due to continuous exposure to the transgenic crop Bt-toxins; e.g. Van Dommelen 1999, 33). Effects of GM may occur at all levels of biological organisation, each affecting the potential final use of a GM crop. Two phases of GM crop development can be distinguished, a lab phase and a field phase. The various interactive- steps in the process from gene construct design to GM crop product, or, paraphrasing, `from DNA to food' are:

As these five aspects are interrelated by nature, the choice for this structure should not be taken to imply a unidirectional or step-wise flow of information. Indeed, each of these aspects is affected by and has consequences for each of the other aspects (Fig. 1). For instance, a change in a biochemical pathway (biochemistry/physiology) by introducing a new enzyme may potentially affect crop yield and quality (agronomy), but may also lead to emergence of new allergenic proteins (toxicology) or affect (agro) biodiversity.

An advantage of this -largely disciplinary- systematics includes a distinction between two phases in GM crop development. These phases differ in their level of control of risks: the laboratory phase with nearly complete risk control, and afield phase with virtually no risk control (e.g. Goewie 2000).

1.5 Terminology

The terms genetic manipulation (e.g. Harlander 1991) and genetic engineering were the first to be poised in the scientific literature to indicate direct anthropogenic transfer of DNA between living organisms as early as the seventies. Since then other terms have been used for the same purpose, including gene technology, genetic transformation and genetic modification.

Since the present report is a literature overview, we have decided to reflect as much as possible intentions of authors by clinging to their terminology. We think that it is the only real option in a review such as this, covering various disciplinary fields, each with its own historically developed jargon. A shortlist of some essential terms not explained in the text is presented at the end of this report.

For more extensive and detailed information, the reader is referred to excellent recent overviews on various aspects of GM crops which include extensive glossary lists, descriptions of genetic engineering concepts and techniques, and terminology definitions (e.g. Bruce & Bruce 1998; Kahl 1995; Lotz et al. 2000; Ryan et al. 1999). Definitions of terms like GM and GMO are also available on the Web (sites 14 to 17, section 5.2); for instance, definitions within the Directive 90/219/EEC are listed by the Belgian Biosafety Server (website #14).

2. Methodology

2.1 General

In the view of Van Dommelen (1999: 37) the biosafety debate is in need of a more basic interpretative framework to be able to analyse the contested claims and to evaluate the persisting controversies.

Identifying knowledge gaps and controversies is to a large extent a scientific enterprise. Whether such unresolved questions are considered relevant to (widely accepted) problems, however, is determined by the `window of concern' of the actor involved (Van Dommelen 1999). Definition of a problem is subjective and individual. To make this explicit Van Dommelen (1999) has proposed the concept of `set of relevant questions' (SRQs) as a practical analysis tool for weighing opposing arguments and claims. While we see this approach as highly valuable for analysing controversies, we must stress here, that such an analysis is outside the scope of the present study.

Given the objectives of this study, it is clear that no one scientist will be sufficiently competent in all of the disciplinary fields. In part, this problem was solved by consulting internal and external specialists, within budgetary and time limitations. Where possible, unfounded claims and theoretical possibilities were distinguished from sound conclusions derived from empirical data and observations of practical situations. Furthermore, the rules of logic were applied to lines of reasoning, to identify improper conclusions. For example, frequently found conclusions like `no proof of risk' and `no evidence' are not equivalent to `proof of no risk', and `absence of evidence', respectively, in the absence of a statistically adequate number of reliable risk assessment studies. It is interesting to note that the order of the EU concerning novel foodstuffs Number 258/97 requires `a proof of no risk' to the consumer of genetically modified products (Hammes & Hertel 1997). Reproducibility and statistical significance are important scientific rules. Therefore, single, unique studies are not acceptable as conclusive evidence, but may merely indicate a gap in knowledge, a result requiring independent confirmation.

2.2 Technical

The present review is based on information from various sources. These included computerised literature databases (see list below; use of search profiles), intemet web sites (see section 5 `Sources' for a list of hyperlinks), newspapers and other periodicals (not) available via literature databases or the internet, personal communications, as with participants of national and international meetings and symposia (e.g. New Frontiers in Biotechnology, April 2000, Toulouse). The data for this review were collected mainly during the period of March-June 2000. The larger part of the resulting collection was incorporated into a relational database using MS-Access, in order to facilitate the categorising of references according to various criteria related to specific subjects or issues.

Literature databases used:
BA 1996-2000/03;
CAB 1996-2000/04;
AGRICOLA 1998-2000/03;
TROPAG & RURAL 1998-1999/12;
FSTA 19902000/06;
Current Contents Search 2000/01-2000/06.

2.3 Approach

Unresolved research questions have been formulated on the basis of data in the literature indicating knowledge gaps or controversies. As a basis for structuring the present report we have chosen the level of biological organisation of life, from the level of molecule and gene to agro-ecosystem (field) and food.

For analysing controversies, Van Dommelen (1999) has proposed the adoption of `sets of relevant questions' (SRQs), and has characterised controversies as competing SRQs and associated `windows of concern'. Such an analysis of controversies is outside the scope of this study, since identifying the boundaries of knowledge is our main objective. However, in a few cases we provide arguments for the relevance of the listed questions for the identification of potential hazards.

This study deals with unresolved questions which may be pertinent to various aspects of GM crops, from design and creation, to selection, screening, field-testing and market-introduction. Controversies or knowledge gaps have been identified preferably based on data in scientific journals and other peer-reviewed publications, in some cases based on non-peer-reviewed papers, and extended by expert opinions and commentaries. Therefore, these three types of publications have been separated in the list of references.

General scientific criteria for relevance of research questions cannot be given. The perspective of the person raising the question plays a major role (cf. `window of concern', Van Dommelen 1999). Relevance has been based on widely recognized or debated, current public concerns, notions and opinions, often presented in media other than scientific journals. General notions among the public regarding issues like health, food safety, environmental pollution, biodiversity, etc. are widely recognised to be associated with, for instance, the right of free choice, rejection of unsubstantiated claims and unclear argumentation, fear for the unknown, association of randomness with uncertainty and risk, suspicion of utopian predictions by scientists, and the consideration of `breaching species barriers' as unnatural.

In the section Results', we list a number of subjects which have raised questions that evidently have not been answered unequivocally (controversy) or have not been receiving (adequate) attention in the literature (gap), but are likely to be relevant to decisions regarding field introduction of GM crops. The criteria for these qualifications are described below.

Whether the occurrence of such knowledge gaps and controversies constitute risks is in some cases scientifically well established. In such cases, qualifications like `risk' or `benefit' are also assigned to an open question. Following an unresolved question, a conclusion is given regarding the current state of knowledge, followed by data reported on the subject (from the literature, consultation or other source). It should be noted that questions and issues concerning genetic modifications are often proposed to be evaluated on a case-by-case basis (e.g. Gilissen & Nap 1998), which may indicate in itself a lack of knowledge for developing general theories and models.

In order to clearly distinguish conclusions in the literature from our own conclusions, it should be stressed that text presented under `literature data' generally does not include our interpretations, conclusions or opinions, unless printed in italics.

2.4 Criteria for classification of (scientific) issues

The main criterion for in- or exclusion of references was the scientific quality of the information, as reflected by independent reproduction of data (i.e. result demonstrated at least twice) and a peer-reviewing process. However, the relatively large number of anonymously authored reports, generally not published in regular scientific journals, could not be ignored Such information is evident from the list of references, which has been split up into three parts: 1. Peer-reviewed scientific publications, 2. Reports and proceedings (may have been peer-reviewed), and 3. Non-peer-reviewed articles, opinions and commentaries. More specifically, in drawing conclusions from such literature data the following criteria were used for classification into the categories `Controversy' and `Gap in knowledge':

The term `risk' has been defined by most authors as `the product of hazard and chance of its occurrence (e.g. Van Dommelen 1999, 50; a hazard is an unwanted, unintended or undesirable (by whom?) effect that has been demonstrated (more than once) , with a chance being >0. `Benefit' may be defined as an event and its desirable effects (to whom?), which have been demonstrated to occur (chance >0).

3. Results: Account of present controversies and knowledge gaps

Controversies and knowledge gaps concerning GM crops have been noted world-wide, since the early years of field introductions, especially in the USA (e.g. Lacy et al. 1991) and for a wide range of issues. Most of these issues are not strictly of biological nature, and are therefore not within the scope of this review. These controversies on the use of GM crops are of ethical, social, economic, or legal nature and are merely noted here, in order to avoid the impression that controversies and knowledge gaps would be limited to biological issues.

Such other areas of controversies include: bioethics (Abuja 1997; Banner et al. 1998; Bruce & Bruce 1998; Ryan et al. 1999), GM-products trade negotiations (Zach 1999), general economic aspects (Bogdanovic 1993), regulatory policies regarding genetically engineered foods (McCullum 1997), intellectual property tights and the patenting of biotechnological innovations such as improved varieties (Anonymous/FAO 1995; Baumann et al. 1996; 1. Kroes in: Spiertz & Dons eds 2000), legal constraints on biotechnological research (Tucker et al. 1993), legal obligations of producers (Michael 1999), legal aspects of GM food (Schubert et al. 1990; Baumann et al. 1996; Law 1997), and risk assessments (e.g. Kasanmoentalib & Longino 1996).

Biotechnologists have particular concerns about controversies. Most of GM scientists (>90%) appear to see the present public concerns as a threat to their competitive edge in the field of GM (Rabino 1992).

The problem of proper communication between all the actors involved in the public debate on GM crops (Scholderer et al. 1999) has received attention for a long time in both Europe and North America: e.g. Lacy et al. (1991) mention that `Biotechnology [in the USA] is fraught with concern and controversy within both the scientific community and the broader public'; the first passage of the Genetic Engineering Bill through the Bundesrat in Germany showed that a great number of fundamental and detailed issues are controversial (Schubert et al. 1990). The latter authors concluded that sustained discussion is necessary so that satisfactory results are ultimately achieved. This last concern is indeed receiving more attention nowadays, e.g. with regard to respecting the public as `competent civilians' rather than as ignorant consumers (HJ Achterhuis in: Spiertz & Dons eds 2000; cf. Baumann et al. 1996), as neglected for instance by Penman (1994) and Reynolds (1997), and to avoiding making unbalanced utopic predictions (HJ Achterhuis in: Spiertz & Dons eds 2000).

Differentiating between opinions and experimentally established scientific data.

In the following, all controversies and knowledge gaps regarding genetic modification (GM) are presented in the form of questions, with answers listed in two parts:

3.1 Design of DNA constructs and GM crop varieties

The phase of design of a GM crop is considered to be crucial, not only in dealing with potential benefits and risks, but also in exploiting all of the relevant knowledge available on the functioning of plants, crops, ecosystems and food webs. In the design phase of a GM crop, theoretical and experimental analyses, risk and technology assessments and feasibility studies may be performed in order to deal with those aspects. Such assessments should cover all relevant levels of biological organisation, from the level of DNA to the level of (agro-) ecosystem. A literature review on such assessments is beyond the scope of the present study. The next sections of this study deal with the various levels of biological organisation, from DNA to GM food.

Important biological limitations to successful design of GM crops include the lack of knowledge of biochemistry and physiology (e.g. signalling pathways and gene regulation, metabolic regulation, pest and stress tolerance), ecology (e.g. biotic interactions), agronomy (e.g. pest management), and toxicology (food safety; allergies), and a lack of knowledge exchange between these areas.

Several authors stress the importance of the design phase of GM crops for increasing the chances to successfully achieve goals regarding genetic, physiological and agronomic performance and the ecological and toxicological interactions of the GM crop, e.g. Arndt & Rank (1997; genome structure and regulation of gene expression); Ebinuma et al. (1997; selection markers); Kapley et al. (1999; bioremediation); Hum-Musser et al. (1999); Lotz et al. (2000; discuss interaction of scenarios and designs); Nguyen et al. (1997; stress tolerance); Tucker (1993); Zhang et al. (1999; drought resistance). Other aspects requiring attention in this phase include social, economic, legal, and ethical aspects (see first part of section 3).

3.2 Unknown Mechanisms of genetic modification (GM) and breeding

General information regarding current practise of genetic modification may be found in several extensive reviews and other studies (e.g. De Maagd et al. 1999; Dunwell 2000; Ellstrand et al. 1999; FranckOberaspach & Keller 1997; Hilder & Boulter 1999; Maessen 1997; Ryan et al. 1999; Tucker 1993).

Current practise of GM mainly involves the following techniques of transformation:
Agrobacterium tumefaciens; Micro-particle bombardment (`biolistics'); Electroporation.

There is a gap in knowledge of scientific-fundamental differences between interspecific (or xeno-) transformation (as in genetic modification) and intraspecific transformation.

This question is relevant to public concerns about GM leading to `breaching of species barriers' of which subject very little (experimental) knowledge is available.

Interspecific transformation will be refered to as xenotransformation, in analogy to xenotransplantation. Xenotransformation is one of the prime features of genetic modification (GM), constituting a `breaching of species barriers', if not in principle (e.g. because of viral DNA transfer, and actions of transposons) then likely with regard to its current scale, occurrence and nature (e.g. involving genes of various origin, including hazardous antibiotic resistance genes; Van Dam & Schenkelaars 2000). It should be noted that the classical (bacterial) transformation is not in any way related to conventional or classical breeding of crop plants.

Unlike modern GM, classical transformation occurs naturally between closely related species of bacteria (Schlegel 1972). In the literature, little attention is paid to mechanistic differences of xenotransformation as opposed to transformation between related species. There is, however, a scientific basis for distinguishing xeno-transformation from the classical transformation.

The term transformation has been -and is being- used in microbiology to indicate the transfer of genetic information between closely related species, especially bacteria, resulting in surviving cells displaying the newly transferred characteristic in an inheritable way (Schlegel 1972). This still is the general rule, with only very few exceptions (e.g. the first demonstration of conjugal plasmid transfer from Escherichia coli to Bartonella henselae has been reported by Dehio & Meyer in 1997). This transformation contrasts with the anthropogenic transfer of DNA between organisms of widely different phylogenetic origin, as is common practise in GM (see also 3.5.4). In the light of recent findings on differences in structural and functional aspects of genomes between -for instance- micro-organisms and eukaryotes, it is necessary to make a distinction between transformation and xenotransformation. In analogy, the terms transplantation and xenotransplantation are used in the medical literature to indicate transfer of organs between individuals of the same and different species, respectively.

Characteristic for different species are their differences in functionalities, based on often minor but significant differences in genome structure (e.g. non-coding regions or `Junk' DNA, Fransz et al. 2000), DNA structure (e.g. Shimizu et al. 1997); enzyme structure & function and corresponding gene sequence and regulation.

Interspecific transformation has been referred to as `Breaching species barriers' (e.g. Ryan et al. 1999; Van Gool 1999). Some examples of unintended side-effects, potentially related to species-specific differences in regulatory characteristics are discussed here for the case of aluminium tolerance (see also paragraph 3.2.1).

Aluminum soils are unproductive for most crop varieties, while common in the world, especially in the tropics (e.g. Herrera-Estrella 1999). Biotechnologists have attempted to solve the problem of Al-toxicity in crops by overexpression of a bacterial gene coding for citrate synthase (CS) in tobacco plants. The choice of CS was based on the outcome of physiological research showing increased immobilisation of Al by organic acid exudates from roots of Al-tolerant plant species (de-la-Fuente-Maitinez & Herrera-Estrella 1999; Sasaki et al. 1996). This approach was successful in generating Al-tolerance in tobacco.
This is an example of adequate integration of physiological and biotechnological knowledge, and one important step forward. However, there are more steps needed towards a properly functioning crop. Overexpression is still a primitive way of gene regulation often leading to unintended side-effects (e.g. Winicov 1998; Delhaize et al. 1999). It remains to be seen whether the bactenal enzyme can be properly regulated in the plant, preferably only in the roots, and induced by aluminum (gap). Agronomic performance, ecological interactions and toxicological effects of the new crop were not evaluated.

In another case, a plant cDNA was selected on its ability to confer Al-tolerance in yeast, i.e. TaPSS1 encoding a phosphatidylserine synthase (PSS; [CDP-diacylglycerol-serine O-phosphatidyltransferasen from wheat, Triticum aestrivum (Delhaize et al. 1999). When the TaPP1 gene was overexpressed in Arabidopsis and tobacco (Nicotiana tabacum), the plants showed necrotic lesions on leaves, associated with accumulation of large amounts of phosphatidylserine (PS) at the expense of other phospholipids in the cells. These phenomena indicated poor regulation of PSS activity (Delhaize et al., 1999).
Whether this poor regulation was caused by species-specific differences in regulation of PSS activity, t: e. between wheat (donor) and Arabidopsis and tobacco (acceptors), was not clear.

Ezaki et al. (1999) isolated eleven aluminium (Al) stress-induced genes derived from plants (wheat, Arabidopsis and tobacco) and introduced these genes into Saccharomyces cerevisiae (yeast) to test if expression would confer Al-tolerance. Al sensitivity tests showed that expression of two (out of eleven) genes, either an Arabidopsis gene for blue copper binding protein (BCB), or a tobacco gene for the GDP dissociation inhibitor (NtGDI1), conferred Al-tolerance. Determinations of total content and localization of Al ions in these transformants suggested that the BCB gene product functions in restricting Al uptake, while expression of the NtGDI1 gene promotes release of Al ions after uptake (Ezaki et al. 1999).
It remains unclear what the reasons were for absence of Al tolerance m nine out of eleven transformants. Also, no study was conducted to test for normal growth and development i7 the two Al tolerant transformants.
Another question, raised by findings in organisms other than higher plants, concerns the impact of changes in GC content of the plant genome (e.g. Mouchiroud et al. 1997; Ricroch & Brown 1997).

The question is relevant to general concerns among the public about random changes to well-functioning organisms. This is a conspicuous issue of many unknowns and fundamental problems. There is hardly any report of systematic experimental studies comparing the functioning of crop plants originating from several different (xeno)transformation events; the few reports found are conflicting.

Genome structure and function are important for gene expression (regulation); yet, there are many fundamental unknowns, including the nature of the differences between bacteria, yeast, plants and animals in structure and functioning of the genome in relation to other differences like cellular compartmentation.

Indirect, non-specific effects of (random) DNA insertions are known. Any currently used technique of genetic transformation is accompanied by an insertion mutation, that may or may not occur in a coding region of the genome (De Jong et al. 1999; Kahl et al. 1995; Von Wettstein 1995), OR, if within a noncoding region, may occur in a DNA-folding-instruction region (c.f. Fransz et al. 2000).

Such effects are evident from `empty vector' transformants, i.e. transformants without a `gene of interest'. Empty vector effects, mechanistically related to insertion mutations, are well known, but seldomly described in the literature (Gap), highly variable effects depending on the transformation event, even without any involvement of a gene of interest. The origin of such `empty-vector' effects is generally not known.

Gilissen & Nap (1999) support the relevance of the above question, and show the absence of adequate and safe methods circumventing the problem. They confirm that current methods of GM of plants

result in insertions of DNA into the host genome at random and often multiple sites, and that, associated with this, position effects, copy number differences and multigene interactions make gene expression highly variable and the occurrence of desired phenotypes less predictable. These authors pose the question, whether the tools that are presently being developed to avoid such problems by insertion of DNA copies at predetermined sites in the plant genome, have risks associated with them (Gilissen & Nap 1999). These tools involve enzyme systems that are site-specific recombination systems and homing (rare-cutting) endonuclease systems. In an attempt to identify and evaluate possible biosafety issues related to applications of these systems, Gilissen & Nap (1999) conclude that, without technological improvements, endonuclease-mediated targeting is unlikely to become a popular method for agronomically relevant gene targetting in plants, because of various unpredictable changes in the host genome. Also they conclude that GM plants with high constitutive expression of a recombinase gene require special consideration when aimed at field and market introduction, and that the occurrence of homing endonuclease system phenotypes with effects on plant fitness and toxicological characteristics cannot be excluded They argue that plants expressing highly constitutively a transgenic endonuclease gene, deserve special consideration in biosafety assessment. Also, chromosome modifications from transgenic homing endonuclease activity at an introduced transgenic recognition site will not cause ecological and toxicological effects other than possible effects caused by the nature of the incoming DNA and the associated illegitimate recombination events (Gilissen & Nap 1999).

Maessen (1997) notes that the level of gene expression varies with the site of integration. Since the positioning of the integration is a random, unpredictable process, this means that gene expression is un

predictable for each transformation event. Also, instability of inheritance generally becomes visible within a few generations, but once it is stable it is supposed to remain so provided the environment does not change dramatically. Transposons (TEs) affect the functional stability of the genome (cf. Lonnig & Saedler 1997). However, the frequency of transposing in most crop species is not known (Maessen 1997).

Gaps in the knowledge of the control of gene expression in (transgenic) plants are also illustrated in the following studies:

Arndt & Rank (1997) review the well-documented use of complementary RNA sequences and argue that despite the simplicity of this approach, the technique usually results in only partial suppression of gene expression and, in some instances, even fails to regulate the gene of interest.

The authors expect that recent advances in the understanding of the global architecture of the nucleus, chromatin structure, and RNA metabolism provide useful and necessary information for designing novel approaches to improving antisense RNA and ribozyme regulation. Available data suggest that the position of genes within the nucleus is not random and that transcripts produced from these genes follow specific tracks in migrating to the cell cytoplasm.

Random insertion will have (unknown) consequences for the structure of DNA and thus, for DNAprotein interactions and genome functioning and regulation of gene expression (cf. Travers 1993). This author demonstrates the importance of role of dynamic interactions between proteins and nucleic acids, which is increasingly being recognized as an important regulator of gene expression, with differs between eubactenia and eukaryotic organisms like plants.

Confirming the notions of Travers (1993), Strahl & Allis (2000) observed distinct modifications of posttranslational histone (DNA binding protein), on one ore more tails of the molecule, and propose that these are acting to form a `histone code' that is read by other proteins to bring about differential regulation of gene expression.

An intriguing process that is playing a role in controlling (the timing of) gene expression is DNA (de)methylation.

DNA (de)methylation plays a role in gene regulation (e.g. Regev et al. 1998). Tatra et al. (2000) observed that lower levels of cytosine methylation are associated with low red/far-red ratios of light incident on ecotypes of Stellaria longipes. These data led them to suggest that DNA demethylation (=gene activation) is involved in the shade-avoidance response of these plants. An important question that remains is: what is the role of DNA (de)methylation in genome functioning and gene expression in transgenic crop plants? (Shimizu et al. 1997).

To our knowledge, little information is available on differences in methylation between taxons or species (however, cf. Shimizu et al. 1997). While data on higher plants are scarce, some interesting studies exist on animals. Monk (1995) reviewed data on changing patterns of DNA methylation and the regulation of gene expression in mouse embryonic development. Global demethylation of the DNA occurs from the eight-cell stage to the blastocyst stage in preimplantation embryos, and global de novo methylation begins at implantation. Monk raises the question whether methylation could be causal to gene inactivation, as well as being involved in its maintenance, and presents a picture of the inheritance of methylation imprints. On a more general level, an hypothesis of evolution by adaptive epigenetic/ genetic inheritance' is considered. This proposes modification of germ line DNA in response to a change in environment and mutation at the site of modification (e.g., of methylated cytosine to thymine). Epigenetic inheritance could function to shift patterns of gene expression to buffer the evolving system against changes in environment. If the altered patterns of gene activity and inactivity persist, the modifications may become 'fixed' as mutations; alternatively, previously silenced gene networks might be recruited into function, thus appearing as if they are 'acquired characteristics.' An extension of this hypothesis is 'foreign gene acquisition and sorting' (selection or silencing of gene function according to use). 'Kidnapping' and sorting of foreign genes in this way could explain the observation that increased complexity in evolution is associated with more 'junk' DNA.

Adaptive epigenetic/genetic inheritance challenges the 'central dogma' that information is unidirectional from the DNA to protein and the idea that Darwinian random mutation and selection are the sole mechanisms of evolution.

There is a clear controversy over potential differences between CB and GM. While no systematic research on (crop) plants has been found on this subject, only one study on yeast and one on wheat were found where a comparison is made between organisms with one particular new trait created by either CB or GM. The question is highly relevant to the debate on the use of GM crops, because it deals with the potential uniqueness of GM crops, which, when confinned, should have consequences for political decisions and regulations.

Regarding the relevance of the question: In the perception of the public, breaching species barriers is often seen as a source of potential risk to health and environment (e.g. Van Gool 1999, 2000; Franck-Oberaspach & Keller 1997). How important are such differences between CB and GM for assessing risks and benefits in ecophysiological functioning? Many biotechnologists seem to be convinced of a negative answer. However, there are no data available for an adequate evaluation of the question.

The question may be considered from various standpoints other than `natural-science' (cf. Lammerts van Bueren et al. 1998, 1999). For instance, socio-economically, several reports indicate considerable differences, e.g. in the required levels of capital investment (Nguyen et al. 1997; Vasil 1998), knowledge infrastructure (Burnell & Dowds 1996, Hum-Musser et al. 1999, Tucker 1993).

The issue of a (fundamental) difference between CB and GM is controversial within the area of natural science, since both the answers `no' and `yes' to the question are found in the literature. For instance: Some authors consider a comparison of CB and GM as incorrect or irrelevant, arguing that it is not the breeding technique by which a plant was produced that should matter, but the characteristics of the plant produced (e.g. Gilissen & Nap 1998).

Others see no difference between CB and GM: Trewavas (Univ. of Edinburgh) and Leaver (Univ. of Oxford) in Nature of January 6, 2000: ` ... all the food we eat has been continuously engineered by natural phenomena in ways that do not differ in any fundamental way from the current GM technology'. Ryan et al. (1999) present a similar opinion, although with some reservation: `Although techniques required to create GM crops are recent and relatively sophisticated, genetic modification is in most respects an extension of what has been happening for ten thousand years'. With regard to the dominating influence of the `genotype x environment'- interactions Brennan et al. (1999) see no difference between CB and GM in the level of disturbing the outcome of breeding efforts. The authors note that genotype x environment fluctuations in fruit quality exist, potentially reducing the quality in the genotypes developed by both classical and biotechnological means.

On the other hand, differences have been mentioned by other authors, with viewpoints outside (Lammerts van Bueren et al. 1998, 1999) and within the natural sciences, e.g. Franck-Oberaspach & Keller (1997), while an implicit `yes' to our question is evident in articles by other authors, e.g. Von Wettstein (1995), Honee (1999) and Daniell et al. (1998). Their opinions are presented in more detail because of the wealth and variety of arguments.

GM differs from CB according to Franck-Oberaspach & Keller (1997) discussing the consequences of classical and biotechnological resistance breeding for food toxicology and allergelu'city. The first food products derived from transgenic plants that are resistant to diseases, insects or viruses are now reaching the market and there is growing public concern about problems of allergenicity and toxicological changes in such transgenic food plants. The source of the transgene is of great importance for the application of immunological assays. Several 'self defence' substances made by plants are highly toxic for mammals, including humans. The source of the transgene is of no relevance in assessing the toxicological aspects of foods from transgenic plants. Food safety can also be severely influenced by invading pathogens and their metabolic products.

The authors conclude that a proper comparison may be a trade-off between 'nature's pesticides' produced by transgenic plants or by varieties from traditional breeding programmes, and synthetic pesticides and mycotoxins or other poisonous products of pests.

An implicit `YES' to the question of difference between GM and CB is also found in other papers. Darnell et al. (1998) see GM as an alternative to CB: `In the past, cotton fibre quality has been improved by classical plant breeding; however, this approach is limited by species incompatibility and available traits. An alternative approach is to introduce foreign genes to confer desired traits into cotton via genetic engineering. Protein-based polymers (PBPs) are available in nature as materials with extraordinary mechanical properties, such as spider webs composed of silk threads tougher than steel and elastin. Therefore, introducing this PBP into cotton fibre should increase fibre strength, water absorption, thermal characteristics and dye binding.'

The view that GM differs from CB, is also implicitly evident from a paper by Honee (1999) who notes that genetic engineering has provided strategies in an addition to other strategies like those based on CB; most of the GM strategies are based on the overproduction of one component of the plant's own defense response. These approaches to increase resistance to fungi have been successful under laboratory conditions. Incorporation of these strategies in resistance breeding programs of agriculturally important crops will depend on the results obtained from field experiments.

Burnell and Dowds (1996) argue that the methodologies of classical genetics and genetic engineering (GM) can both be used, in their specific case for the genetic improvement of entomopathogenic nematodes (EPNs) and their symbiont bacteria. However, they see a greater need for basic knowledge in the case of GM. They conclude that much basic research is needed for any progress to expect. The authors conclude that there is a lack of basic information on the genetics and biochemistry of the characteristics that might be altered by transgenic methods in EPNs, and their bacteria.

As a beneficial difference with CB, GM is often suggested to be faster in producing new improved crop varieties (Boerjan et al. 1999; Jung & Wyss 1999; Nguyen et al. 1997; Renard et al. 1997).
However, no experimental data have been found to support this c/aim.

For instance, a specific advantage of GM is seen in the case of tree-breeding. Boerjan et al. (1999) discuss the genetic improvement of trees which is a slow process in comparison to that of annual crops. Tree breeding though is important, given the ever increasing demand for wood and wood products. The authors intend to show that the classical genetic improvement of trees by breeding and selection can be assisted and accelerated by the application of molecular biology tools that have been developed over the last decade. They give two reasons. First, it is now possible to develop a set of diagnostic markers that predict the characteristics of new hybrids soon after they have germinated, thus long before the traits are displayed. Second, genetic engineering allows the modification or addition of a

given trait that would be difficult or impossible to obtain by conventional breeding. Case studies in both fields, with respect to disease resistance and wood quality, are presented.

A fair comparison of CB and GM may be difficult, because GM is a relatively young technique (e.g. in barley breeding: 6 yrs) as opposed to e.g. selection (since 5000 yrs) and the use of induced mutation (since 1927). VonWettstein (1995) reviews the history of breeding of barley which has been bred for food, feed and beverages by selecting for spontaneous mutations and random hybrids. Genetic transformation of barley has been a routine procedure since 1994 and permits the introduction of tailored genes for adding quality value to the grain. The author argues that it complements, but does not replace, existing efficient breeding methods, and that radiationand chemically induced mutations, as well as genes introduced by transformation, have to be fitted into the genome, which may take 50 years of breeding effort and testing for agronomic and industrial performance.

Only ONE example of an experimental study has been found where a comparison was made between new organisms of one species with identical acquired traits, while originating from either CB or GM: this concerned a yeast utilizing melibiose (Vincent et al. 1999), with similar success reported in both ways of breeding.
However, no long-term risk assessments were made, nor was anygenome structure analysis undertaken to investigate potential differences.

A study on crops coming close to a proper comparison, is that of Baga et al. (1999), on increasing levels of unbranched starch in wheat, by GM or CB (see next page).
However, the authors do not present a proper comparison, and are unclear about the actual quantitative aspects of the changes in unbranched starch levels brought about by GM (BE enzyme levels are no conclusive evidence; cf Fell 1997), while reporting large effects achieved by CB (>95% unbranched starch).

What evidence is there to support the term `tampering with nature' as a characteristic of GM more than of CB (van Gool 1999)? This question cannot be answered from our standpoint in the present study, which is strictly in natural-science. The discussion on `natural' vs `unnatural' is of a philosophical nature, and outside of the scope of our study (however, see for a discussion, Visser & Verhoog (1999); Dr. H. Verhoog has started an NWO funded project on this subject at the Louis Bolk Institute, Driebergen, Netherlands).

However, one aspect of this question that is within our `window of natural-science expertise' concerns the socalled `breaching of species barriers'. This topic is discussed in section 3.2.1.

Regarding the distinction between natural (CB) and unnatural (G4, Ryan et al. (1999) state: `The question of how to decide whether GM crops are `unnatural' to an unacceptable degree is more difficult to address'. These authors mention a theoretical (natural-scientific) argument for their conclusion `no difference': i.e. `because the same gene complement is achieved by CB and GM). This is surprising, since it is known that in practise `the same gene complement' may only be achieved at present at a certain degree, and has not yet been reported in the literature (knowledge gap). A clear definition of the term `same gene complement' will be required before any further discussion of this issue. This should be part of a methodological analysis identifying the various `windows of concern' and `sets of relevant questions' (cf. Van Dommelen 1999) of the participants in the debate.

Also, Ryan et al. (1999; 1.34) give the impression to deny the very existence of differences between CB and GM, by stating: 'In what way is a gene that is found in a fish and which might be very similar in structure and function to one found in a micro-organism, plant or animal, a `fish' gene? Some would say that it is no more than a defined stretch of DNA in a fish cell. But that does not seem to help. What lies behind such concerns?'
Comment. In order to throw some light on this issue: indeed, the literature provides some evidence for such concerns. First, the assumption `which might be very similar in structure and function' is purely theoretical. In practise, even a highly conserved gene like the one encoding the wide-spread respiratory enzyme cytochrome oxidase (ca. 104-108 amino acids; with a similar function) varies greatly in structure, e.g. by about 20 to .306 between yeast and fish (Vogel & Angermann 1971). Secondly, in the light of current knowledge about the complex mechanisms of regulation and control of metabolic pathways and modern control analysis of (allosteric) enzyme activities (Fell 1997), such structural differences are bound to have profound effects on the functioning of the enzyme in the transgenic organism (cf. Dressler & Potter 19921. Also, protein trafficking differs greatly between bacteria and eukaryotic organisms (signalling sequences, etc.). Fell (1997) argues that the lack of understanding of metabolic regulation has been revealed by poor results from attempts to increase the rates of selected metabolic pathways by genetic engineering techniques. Current biochemical theory predicts that it is relatively easy to decrease the activity of a metabolic pathway, as opposed to increase it (Fell 1997).

While widely recognized as a risk, few reports of a systematic experimental study are available; there is now increasing attention for bioinformatics.

The question is relevant to the debated issue of the `precision' of GM practise.

Genome databases and annotations are `Not sufficiently reliable; not operative in a well-organized manner', according to Clavene (2000) and Rouze (2000). Many errors have been detected in existing genome databases, in preliminary studies, especially in annotations.

These questions are relevant to the level of predictability of genetic modifications, which is generally claimed by biotechnologists to be higher than in classical breeding (e.g. Harlander 1991; De Jong et al. 1999).

The term `Junk DNA' reflects the obvious gap in knowledge of molecular biologists, illustrated by recent evidence for a function of a non-coding DNA stretch in chromosome folding, up to then identified as junk DNA (Fransz et al. 2000). The question is relevant to the debated issue of the `precision' of GM practise.

The following example Illustrates the poor know/edge of genome structure and function, and how to deal with these in genetic engineering.
The term `Junk DNA' has been poised to indicate stretches of DNA without a known function, that happen to constitute the major part, often more than 80%, of large genomes. Recently, this `junk DNA' was shown to have an -obviously unexpected- function, i.e. in the folding of a chromosome, by Fransz et al. (2000) in one of the leading scientific journals, Cell.

These authors started correlating the DNA sequence with the structure and function of the chromosome, which is required for elucidating the role of heterochromatin (condensed, non-expressing regions) in the regulation of gene expression in eukaryotic organisms. Their object of study was the short arm of chromosome 4 (4S) of Arabidopsis thaliana, a cosmopolitan plant species with a relatively simple structure of its chromosomes (Fransz et al. 2000) and a relatively small genome containing small amounts of `junk DNA'. According to Maessen (1997), its genome size is only 1.1% of the size of the wheat genome. Thus, it must be clear that we have only just started to uncover the role(s) of `junk DNA', with undoubtedly many surprises ahead of us, when exploring the larger genomes.

Genome sizes in higher plants vary by three orders of magnitude (a factor 1000). Large genomes, which contain more non-expressed heterochromatin or `Junk DNA', may be expected to allow a higher (relative) incidence of inactive transgenes, although evidence is lacking. Any insertion may result in a destabilized genome and lower viability (Maessen 1997). According to Ryan et al. (1999), such unexpected effects do not necessarily mean that the GM technology used is unsafe, as plants showing such side-effects are generally discarded in the subsequent breeding process (cf. Maessen 1997), but they do point to the need for vigilance in the regulatory procedure (Ryan et al. 1999). However, in practise, several examples are now available of unexpected side-effects of primary GM modifications, including physiological disorders (e.g. Delhaize et al. 1999; Gertz et al. 1999; Murray et al. 1999).
The very fact of naming the unknown junk; whether or not exemplifying the attitude of biotechnologists towards nature, may reflect an interesting aspect of the communication problems that some scientists have in the current debate on GM0s.

Kahl et al. (1995), in an article entitled `Junk DNA -not so junky after all', presented the various experimental functions of a particular species of 'Junk DNA', i.e. microsatellites, simple sequence repeats. Their exploitation in DNA profiling techniques has expanded the repertoire of useful sequences for plant biologists. Some old and new techniques involving microsatellites are presented, such as microsatellite primed polymerase chain reaction (PCR), anchored microsatellite primed PCR, random amplified (microsatellite) polymorphic DNA (RAPD) analysis, the generation of microsatellite fingerprints in RAPDs and the production of sequence tagged microsatellite sites.

Schmidt & Heslop-Harrison (1998) suggest that it is now possible to obtain a better understanding of the structure and functioning of plant genome. They show that plant species from wide taxonomic groupings have similar genes and ordering of genes along the chromosomes. However, the repetitive DNA, much of no known function and often constituting the majority of the genome, varies extensively from species to species in absolute amount, sequence and dispersion pattern. Despite this, it is known that families of repeated DNA motifs each have a characteristic genomic location within a genus, and that there are different constraints on the evolution of repetitive DNA and genes.

The authors conclude that there are now enough data about different types of repetitive DNA, from sequencing, Southern analysis and in situ hybridization, to build a model of the organization of a typical plant genome, and apply it to gene cloning, evolutionary studies and gene transfer.

The gap in present knowledge of the regulation of gene expression is evident from the large number of failures and unintended side-effects reported in the literature (see also 3.3.3). This subject is relevant to the debated issues of `well-defined' changes and the `precision' of GM practise.

At present, the answer must be that knowledge on the control of gene expression in higher plants has increased dramatically but nevertheless often appears to be inadequate for successful application. This means that at present it is generally not possible to control the expression of transgenes in an adaptive way in time and space, i.e. under the proper conditions and at the desired developmental stage and in the target plant part. In a review on mechanisms underlying stability of gene expression, Maessen (1997) noted that it is not possible to draw final conclusions about the stability of transgene expression because of limited and preliminary data, identifying a number of gaps in knowledge. While there are no indications that the introduced GM trait is less stable than the non-modified trait, insufficient data are available about the mechanisms underlying the high variability of transgene expression, about the randomness of transgene integration, about the mechanism of the (positive) effect of MAR (matrix-associated regions of DNA) elements on gene expression, and about the utility of MAR elements in other species than tobacco.

Other recent developments support the above notions. One of the major challenges in genetic modification is the development of adequate controls of gene expression. An example of one of the first steps taken in this direction is the use of a leaf senescence-associated promoter by Gan and Amasmo (for a relent report, see Jordi et al. 1-000): Other recent attempts include the use of inducible promoters (e.g. Salter et al. 1998; Goddijn & Van Dun 1999) and the use of a cascade of chimaeric genes (including toxin and anti-toxin genes) to correct for promoter leakage (Mouradov et al. 1998). However, the general picture is, that properly adaptive transgene regulation is not yet achievable.

3.3 Genetic and physiological performance of GM crops

The questions below are considered to be relevant to debated issues like `unintended effects on crop performance under specific environmental conditions', `the need for GM', `alternative approaches to GM', `stability of GM crop characteristics', `feasibility and risks of gene-pyramiding, and `predictability of (unintended) effects of GM on crop performance' and `effects on (agro)ecosystems and food webs'.

In most areas not enough is known about plant physiology and biochemistry to provide clear-cut designs for GM crops with improved performance. Metabolic control analysis theory predicts and has successfully predicted that in biotechnological applications, to ensure a significant increase in a metabolic flux without a marked change in metabolite concentrations, it is necessary to simultaneously activate many, even most of the pathway enzymes (Fell 1997). This view is supported by others. Effects of GM on secondary metabolism is an open and not researched aspect according to Richard D. Firn and Clive G. Jones, Secondary metabolism and the risks of GMOs, Nature 400, 13 - 14 (1999).

For instance, Zhang et al. (1999) mention current controversies on effective mechanisms of drought resistance. Plant water deficit is a component of several different stresses, including drought, salinity and low temperature, which severely limit plant growth and crop productivity. Genetic modification of plants to allow growth and yield under unfavourable conditions is an important component of the solution to problems of environmental stress. While disagreement and even confusion may characterize some of the discussions on what constitutes a significant and an effective mechanism of drought resistance in crop plants, osmotic adjustment (OA) is receiving increasing recognition as a major mechanism. The paper starts with the review of OA functions, genetic variation and inheritance, and theories and principles involved in commonly used protocols for quantifying OA. Emphasis is placed on a summary of current molecular strategies and advances in the improvement of plant stress resistance through manipulating OA. Zhang et al. include a genetic engineering approach and a QTL mapping approach. They suggest that future promising strategies for improving drought resistance lie in molecular technology that allows genes or QTLs controlling OA to be tagged and isolated, these genes to be expressed in transgenic plants, and efficiency of breeding via marker-assisted selection to be improved, combined with physiological research.

There is a insufficient information on the subject, while both risks and benefits of GM have been described concerning issues like pesticide use, stress tolerance, and productivity and quality.

This question is relevant to issues like `the need for GM' and `alternative approaches to GM'.

There are few studies which enable a proper comparison of CB with GM. Some of these mention both positive and negative results of GM, on insect resistance (e.g. Bowen et al. 1998) and aluminum tolerance breeding. The latter example is discussed below.

Aluminum tolerance has been bred into various crop plants by both classical breeding (CB) techniques and by a combination of CB and GM. Aluminum soils are unproductive for most crop varieties, while common in the world, especially in the tropics (e.g. Bouton et al. 1997; Herrera-Estrella 1999). Biotechnologists have attempted to solve the problem of Al-toxicity in various ways: one based on a bacterial gene coding for citrate synthase (de-laFuente-Martmez & Herrera-Estrella 1999), an other based on a plant gene, i.e. a cDNA (TaPSS1) encoding a phosphatidylserine synthase (PSS; [CDPdiacylglycerol-serine O-phosphatidyltransferase]) from wheat (Delhaize et al. 1999), and yet another dealing with testing eleven Al-stress induced genes derived from higher plants (wheat, Arabidopsis and tobacco) for conferring Al-tolerance to yeast; Ezaki et al. 1999). The first approach was successful (de-la-Fuente-Martinez & Herrera-Estrella 1999), the second was not, due to leaf necrosis (Delhaize et al. 1999), while in the last case (Ezaki et al. 1999) only two out of eleven types of transformants showed increased Al-tolerance. Ofcourse this may have been pure coincidence. Alternatively, the rate of success may have been largely determined by available knowledge of the physiology of aluminum tolerance.

Duncan et al. (1995) provide an example of successful breeding for multiple stress tolerance using biotechnological techniques without the use of GM, i.e. in vitro cell selection and somaclonal variation. A field selection protocol was developed for the three soil stresses and inter-stress evaluations were conducted in an effort to find multiple, stress-tolerant genotypes of Sorghum bicolor which is generally quite sensitive to salt and acid (high aluminium) soil stresses, but quite tolerant of drought stress. They report a variant frequency of 0.1 to 0.2% for stress tolerance and acceptable agronomic traits among the surviving somaclones. The authors conclude that the stress-tolerant regenerants had superior acid soil and drought stress tolerance to that of the donor parents, their yield capabilities under stress were superior to their parents, and their stress tolerance attributes were transferred in hybrid combinations.

Most authors propose a combined approach, using both GM and CB, e.g. Santos et a/. 1997) and:

Morpurgo et al. (1997) discuss the problems associated with plantain (Muses paradisiaca and banana (Musaceae) breeding and development of novel biotechnological techniques to overcome these

problems. Aspects considered include: limitations of conventional breeding methods for improving banana quality and yield; need for an increased variety of banana crops with good quality and yield, disease resistance, storage and shelf life characteristics; development of techniques aimed at facilitating banana and plantain breeding (etc., they give an extensive list of required types of research and facilities, etc); examples of traits that have been introduced by genetic engineering (herbicide, virus, insect and fungus resistance, genes for production of essential amino acids and for delaying fruit ripening); and the need to combine classical and novel approaches for continued development of improved crops.

There is a controversy regarding the level of predictability of characteristics of transgenic plants growing under various environmental conditions; some findings suggest risks (e.g. for farmers).

The physiological functioning of plants is found to be more complex than expected, as is evident from reported shortcomings of GM crops. The question is relevant to the claim of high precision of GM.

Several authors suggest that the performance of transgenic plants is highly predictable, e.g. Harlander (1991), and De Jong et al. (1999).

However, there are data indicating the opposite. For instance, some recent field crop and experimental laboratory observations indicate unexpected (and unintended) responses of GM soybean with glyphosate resistance based on a modified EPSPS (CP4 EPSPS; Gertz et al. 1999; discussed by Coghlan 1999b). These transgenic glyphosateresistant soybean plants show splitting stems and yield reduction (up to 40%) under growth conditions of nonexceptional, high soil temperatures (ca. 45 °C), and about 20% higher lignin levels at normal temperatures (25 °C). Gertz et al. (1999) hypothesize that the addition of glyphosate-resistant EPSPS in these varieties might have altered the product distribution in the shikimate pathway which leads to aromatic amino acids, lignin, some vitamins and other secondary metabolites.
The absence of a timely discovery of such an unintended side effect seems remarkable in the light of the existing extensive knowledge on lignin biosynthesis (for a review., e.g. Boudet et al. 19961 and effects of high lignin levels (Sasaki et al. 19961. Such knowledge and an interdisciplinary approach might have enabled a proper and timely prediction of such a side- effect of glyphosate resistance based on the modified EPSPS.

Other examples of unexpected side-effects (pleiotropy), described in other sections, are: necrotic lesions on leaves of plants overexpressing phosphatidylserine synthase (Delhaize et al. 1999; 3.3.1), phytotoxicity of glucose oxidase expression confering resistance to fungal infection (hurray et al. 1999).

There is a gap in present knowledge of the biochemistry and physiology of quantitative, multigenic traits of crops, like stress tolerance and food quality, increasing the risk of unexpected side effects of GM.

Predictability of characteristics of GM crops is seen as a problem by many among the public, and appears to be a key-issue undermining public trust in GM (e.g. Website 1; Van Gool 2000; Schalk 2000). This adds to the relevance of the question.

Most authors provide a positive answer to the question. Tucker (1993) confirms the above-mentioned gap, in a review on the application of recombinant DNA technology to quality and processing properties of tomatoes, discussing genetic manipulation of pectolytic activity in tomato fruit (action of pectinesterase, reduction of polygalacturonase activity) and the control of ethylene synthesis. The author concludes that application of genetic engineering to improvement of food crops has great potential, but is limited by a sparsity of biochemical knowledge regarding the traits to be manipulated.

Also, Hum-Musser et al. (1999) see gaps in present knowledge of the physiological mechanisms of thermotolerance, while noting potential prospects for applying GM in improving heat tolerance.

Winicov (1998) stresses the present lack of knowledge on the physiological mechanisms of salt and drought stress, hampering molecular approaches to improving salt tolerance in crop plants. Improvements to salt and drought tolerance in crop and ornamental plants have been elusive, partially because they are quantitative traits and part of the multi- (or poly-) genic responses detectable under salt/drought stress conditions.

However, Winicov (1998) also stresses the great potential of GM and proposes new strategies of transgenic manipulation more sophisticated than the rough type of overexpression accomplished till now. The author sees many practical limitations of overexpressmg all of the genes required for stress tolerance in a plant in a tissue specific manner that would maintain developmental control as needed. New approaches are being developed towards being able to manipulate expression of functionally related classes of genes by characterization of signalling pathways in salt/drought stress and characterization and cloning of transcription factors that regulate the expression of many genes that could contribute to salt/drought tolerance. Transcription factors that regulate functionally related genes could be particularly attractive targets for such investigations, since they may also function in regulating quantitative traits. Transgenic manipulation of such transcription factors should help us understand more about multigene regulation and its relationship to tolerance.

In a review on transgenic approaches to crop improvement, Dunwell (2000) observes a trend towards genetic modification of more complex agronomic traits such as growth rate and increased photosynthetic efficiency.

There is a risk of trans-genome instability, i.e. leading to loss of desirable traits, but this is usually restniced to the plant breeding phase where instable traits are lost or discarded in the selection process.

It seems that instability is generally not seen as a great problem?; Maessen 1997 (see 3.2.5); Metz et al. (1997). Welin et al. (1996) review ways for improving plant cold acclimation, examining the genetics of cold acclimation and frost tolerance (classical genetics, low temperature (LT)-induced gene expression, signal transduction), the function of LT-induced proteins (dehydrins, antifreeze proteins, cryoprotection), and genetic engineering of freezing tolerance.

Several other areas of genome functioning have been identified where knowledge is fragmentary or absent. Maessen (1997) mentions the following knowledge gaps in relation to possible factors influencing stability of genes and genomes:

There is a controversy over whether there are limitations to inserting large numbers of genes which are supposed to be required; also a general lack of knowledge exists, especially in the areas of biochemistry and physiology (regulation, control and interactions of metabolic pathways).

There are presently limitations to inserting large numbers of properly regulated genes via GM (a/though a novel technique 'multiple co-transformation' may expand the present possibilities to groups of about 20 genes by an unknown mechanism (Ryan et al. 1999)), an important limitation being the lack of knowledge of the nature and control of metabolic and signaling pathways and then encoding genes involved In polygenic plant traits. Current biochemical theory predicts that it is relatively easy to decrease the activity of a metabolic pathway, as opposed to increase it (Fell 1997).

See also Winicov 1998, there are practical limits to gene stacking. Alternative strategies involve genetic engineering of signalling pathways, thereby changing gene expression of families of functionally related genes in a coordinated way. Another example of such a new strategy involves modification of trehalose metabolism in higher plants (Goddijn & Van Dun 1999).

Santos et al. (1997) tested transgenes for insect resistance using Arabidopsis. One possible strategy to delay the selection of resistant insect populations is the pyramiding of multiple resistance genes into a single cultivar. However, the transformation of most major crops remains prohibitively expensive if a large number of transgene combinations are to be evaluated. Arabidopsis thaliana is a potentially good plant for such preliminary evaluations. The authors determined that four major agricultural pests, the four caterpillar species Spodoptera exigua, Helicoverpa zea, Pseudoplusia indudens, and Heliothis virescens grew as well when feeding on 'Landsberg Erecta' Arabidopsis as th

3.2.1	Does the interspecific (or xeno-) transformation in GM differ from the classical transformation that occurs naturally between closely related species?
3.2.2	Are there predictable, unintended effects of random insertion of DNA constructs on gene expression and genome functioning and stability?
3.2.3	Are there differences in mechanisms between classical, or conventional, breeding (CB) and genetic modification (GM)?
3.2.4	How reliable are present-day genome databases and annotations? Is there an adequate proofreading system operative?
3.2.5	What is known about DNA & chromosome structure and function?
3.2.6	Do we understand the mechanisms of regulation of (trans)gene expression in higher plants?

3.3.1	Are present GM techniques, in combination with conventional breeding (CB), more successful than CB alone in exploiting genetic variation?
3.3.2	Can physiological characteristics of transgenic plants be predicted for relevant environmental conditions?
3.3.3	Are pleitropy and polygenic characteristics a barrier towards designing and predicting traits of GM crops like pest and stress tolerance, productivity and food quality?
3.3.4	Are Genetic modifications in crop plants stable during several generations?
3.3.5	Are pyramidal (trans)gene systems feasible, such as proposed in future stress tolerance and pest resistance management?

3.4.1	Can GM-DNA be contained within experimental field plots?
3.4.2	Do GM crops pose a threat to maintaining biodiversity? (E.g. by outcrossing, running wild, weediness, gene spread, effects on non-target pests, etc.)
3.4.3	Are there (unintended) interactions between neighbouring GM- and GM-free agro-ecosystems?
3.4.4	To what extent are the effects on agro-ecosystems of introduction of GM crops into the field (un)predictable?
3.4.5	Are effects of field introduction of GM crops irrevocable and unprecedented?
3.4.6	Does introduction of GM crops in the field lead to `genetic pollution' as the outcome of gene flow, i.e. gene establishment?
3.4.7	Is there evidence showing that GM-insecticide production in planta (`Bt crops) differs in its effects on agro-ecosystems from crop applications of the same insecticide?
3.4.8	Will (pest) insects develop resistance against Bt-toxins faster in GM-Bt systems than in Bt-application systems?
3.4.9	Does the use of transgenic herbicide-resistant crops lead to unintended and unavoidable continuation of herbicide use and accumulation in the environment?
3.4.10	Can GM plants safely be used for the purpose of fighting pollution, as in phyto-remediation? (e.g. hydrocarbons, heavy metals)?
3.4.11	Are there unintended effects of `gene stacking' on (agro-)ecosystems?
3.4.12	Will sustainable agriculture be supported and promoted by GM?

3.5.1	Are theoretical models available predicting successfully the allergenicity of proteins in general and of proteins from GMOs?
3.5.2	Can the concept of `substantial equivalence' be used as a reliable and adequate guideline in toxicological studies of GM-crop products?
3.5.3	Are there specific (long-term) effects of new GM crops on human and animal health? Is adequate knowledge presently available concerning the toxicological and allergenic aspects of GM food?
3.5.4	Will horizontal gene transfer (HGT) possible from GM crops or microorganisms to (pathogenic) micro-organisms? [Relevant to: hazardous gene spread; antibiotic resistance of commensal and pathogenic bacteria; food safety]
3.5.5	Is it possible to separate GM and non-GM food chains?
3.5.6	In testing for toxicological side-effects, is it sufficient to test the isolated geneproduct rather than the whole plant part?

5.1	References
5.2	Web sites