TflEE vol. 5, no. 12, December 1990

bJt it lacks realism, while in the field the animals are exposed to all sorts of experiences that limit the con- clusions we can reach.

References 1 Slater, P.J.B. (1989) Rho/. Ecol. Evol. 1, 13-46 2 Payne, R.B. (1981) Anim. Behav. 29, 6 38-697 3 Marler, P. and Peters, S. (1987) Rhology 76,89-100 4 Pepperberg, I.M. (1985) Auk 102, 8 54-864

5 Petrinovich, L. and Baptista, L.M. (1987) Anim. Behav. 35,961-974 6 West, M.J. and King, A. (1988) Nature 334,244-246 7 McGregor, P.K. (1989) &ho/. Ecol. Evol. 1,124-127 8 Grant, B.R. (1984) Behaviour 89, 90-116 9 Millington, S.J. and Price, T.D. (1985) Auk 102.342-346 10 Nicolei, J. (1959) J. OmiTho/. 100, 39-46 11 Slater, P.J.B., Eales, L.A. and Clayton, N.S. (1988) Adv. Study Behav. 18, l-34

12 Immelmann, K. (1969) in Bird Vocalizations (Hinde, R.A., ed.), pp. 61-74, Cambridge University Press 13 Immelmann, K. (1962) Zoo/. Jahrb. Syst. 90, 1-196 14 Eales, L.A. (1985) Anim. Behav. 33, 1293-1300 15 Clayton, N.S. (1987) Anim. Behav. 35, 714-721 16 BBhner, J. (1990) Anim. Behav. 39, 369-374 17 Williams, H. (1990) Anim. Behati. 39, 745-757 18 Zann, R.A. Anim. Behav. (in press)

Releasing Genetically Engineered Plants: Present Proposals and

Possible Hazards Worldwide, the majority of current j;troposals to release genetically modified organisms concern plants; most are for small changes in com- mon crop plants and present little spparent hazard. However, there is much confusion about how plants become pests, and implausible risk factors are appearing in the literature 2nd before committees. It is danger- ous to estimate the risks from first principles because of the immaturity of plant population biology and the lack of empirical data. Regulatory agencies should concentrate on ob- taining realistic assessment of haz- ards, and not attempt to balance !,otional benefits and disbenefits.

In April 1988, at the time of the REGEM meeting’ in Cardiff, UK, TREE published (jointly with Trends in Bio- technology) a special issue on the lllanned release of genetically engin- eered organisms. REGEM was con- cerned only with microbes, but it is nevertheless remarkable how little was said about plants at that time. Now, only two years later, plant pro- posals are the commonest, and look likely to remain so for a while.

This article comments on the state of play regarding the release of what are now generally called genetically modified plants, and the present un- satisfactory means of assessing the hazards from these introductions. Two background events are relevant: Ihe European Community, at the cZouncil of Environment Ministers on .?2 March 1990, adopted a directive2 cln the release of genetically engin- eered organisms; and the Organis- *ation for Economic Co-operation and Ievelopment (OECD) is encouraging snide discussion of a paper on good .Jevelopmental practice3, which deals Nith safety assessment of small-scale

Mark Williamson, James Perrins and Alastair Fitter are at the Dept of Biology, University of York, York YOI 5DD, UK.

Mark Williamson, James Perrins and Alastair Fitter

field trials, particularly of genetically engineered plants. The European di- rective lists points to be considered, as do numerous other documents (e.g. Refs 4-7). Unlike any of these, which are very general, our intention is to concentrate on the hazards of proposals as they exist at present.

The proposals coming before committees The number of proposals coming

forward to national regulatory com- mittees is now increasing quite rapidly. Both OECD and GBFB (the Gesellschaft fiir Biotechnologische Forschnung mbH in Braunschweig, FRG) have started databases to track releases worldwide.

The GBF database is set up under the acronym BIKE (Biotechnologie Informations-Knotten fiir Europa), and Table 1 shows the counts pub- lished in March 1990. Some releases contain more than one construct, so the totals differ. Slightly over 60% of the recorded releases have been of plants. Only 25% are of microbes, increasing to 38% if viruses are added. The commonest releases among plants have been those with herbicide-resistance genes, consti- tuting about a third of all releases (apart from those that merely carry markers). This has caused contro- versy in Americas,lo, but not as far as we know in Europe, though there has been a conference on the topic”. The next commonest category, at just over a quarter of non-marker re- leases, contains those conveying re- sistance to insects, and includes both plants and microbes; 38 of these 46 involve the delta endotoxin from

Bacillus thuringiensis, almost equally divided between plants (20) and bac- teria (18).

Lists of possible environmental ap- plications of genetically modified or- ganisms (GMOs) have been compiled by bodies such as the Office of Tech- nology Assessment (OTA) of the Con- gress of the United States12. As well as plants resistant to herbicides, in- sects and diseases, it is expected that there will be plants with increased tolerance to adverse factors such as drought and heavy metals, and with the ability to fix nitrogen. A trickle of proposals in these fields has started and can be expected to produce more ecologically interesting problems than those of herbicide resistance. Also of ecological interest are pro- posals, not mentioned by OTA, to alter the constitution of the plant to make a more useful commercial product. Non-squashy tomatoes (Lycopersicon esculentum) and changed proteins in seeds of oilseed rape (Brassica napus) are two examples.

Plants as pests A major question with all these

proposals for plants is whether the genetic changes are more likely to make the plants into pests. With the European directive in place, plant breeders may have to think more about the possible ecological conse- quences of their work, about the effects of the changes they induce on factors such as seed survival or pol- len dissemination13, or more gener- ally about the pleiotropic effects14 of their genetic changes.

417 D 1990. Elsev~er Science Publishers Ltd (UK1 0169.5347/90'$02 00

TREE vol. 5, no. 12, December 1990

Table 1. Releases of genetically modified organisms (GMO+

USA Europe Elsewhere Total

(a) Recorded releases of GMOs

Viruses Microbes Plants Animals

Total

(b) Types of GM0 released

Markers Herbicide resistance Insect resistance Vaccines Virus resistance Frost protection Bacterial resistance Other

Total

41 60

109 66 36 211

22 25 35

7 10

1 0 17 12

124 83

17 8

41 0

35 18 10 5 2

3 4

28

10 15

6

4

39

27 53

129 2

67 58 46 18 13 9 2

33

246

“Data from Ref. 8.

It has been strongly stated that it is quite difficult to turn a plant into a weed. This mistaken view is based on Baker’s list of weed characters15, and was put forward for instance by the Kelman Report’“, which was rightly criticized17 for its ignorance of many sectors of ecology. That the view is basically flawed is shown by the fact that even one of its supporters18 quotes well-known examples of plants, particularly crops, becoming weeds without any known genetic change at all. Harlan’s list of weedy cropslg cannot be quoted too often: it includes eight cereals, as well as carrots, beets, radishes, lettuce, peppers, potatoes, tomatoes, sun- flowers, safflowers, hemp, water melon, Citrus, Manihot, Psidium, Car- ica, Punica, Mangifera, Passiflora, Prunus and Guizotis; and, even so, it is (as Harlan emphasized) a very par- tial list. Weed carrots are sometimes called devil’s plaguezO, while the case of pros0 millet (Panicurn miliaceum) in Canada2’, also quoted by Keeler18, has emerged since Harlan’s list. With so many weeds related to crops, a reasonable guess is that the prob- ability of changing a randomly chosen crop into a weed by changes in distribution or cultivation practice, or by natural selection, at some time and in some place, is perhaps IO-’ to 10m2, not IO-lo as suggested by Keeler18 - a difference of eight or nine orders of magnitudez2.

As is well known to most weed re- searchers, it is not just the character- istics of the plant but also the charac- teristics of the habitat that make a plant a weed. Weeds in regularly mown grass must be low-growing plants, and are mostly perennials; weeds in arable fields are mostly annuals, sometimes resembling the crop in interesting ways, and can only be perennials if they can tolerate ploughing; and weeds in woodlands

can have many sorts of life form. Our own studies show not only that there is considerable disagreement about which species should be called weeds, but that even in restricted sets, such as annual weeds in Britain, discriminant analysis using a large and statistically optimal suite of characters is only of the order of 85% efficient. Should any regulatory com- mittee be satisfied with a prediction that a plant is or isn’t a weed on the basis of its characters, when the pre- diction has a one in seven chance of being wrong?

Hazards from the release of genetically engineered plants

From the proposals coming for- ward, and with the knowledge that plants can become pests in a wide variety of different ways, what prob- lems might arise in the next few years? Three sets of hazards come to mind. The first is from the plant itself. For instance, altering the proteins in the seed might well alter the ecology of seeds, particularly their suscep- tibility to enemies and so their sur- vival in some habitats. This is most obviously a danger in crops that are already minor weeds, such as oilseed rape. Volunteer potatoes (Solanum tuberosum) - potatoes persisting into the next crop - are a frequent if not very serious problem in Britain. One of the cell-fusion experiments per- mitted in Britain produced hybrid Solanum plants that have small and irregular tubers23. If such plants were to be put out on a large scale, the volunteer problem might become more serious, since these mini-tubers will be difficult to remove even if they are less persistent than ordinary ones.

The second set of problems comes from the spread of the engineered genes into related species. Again, oilseed rape is an obvious candidate.

It almost certainly arose from a hy- brid between diploid cabbage (Bras- sica oleracea) and diploid turnip (B. raps), and, as there are tetraploid and otherchromosomeformsof both cabbage and turnip, the potential for gene flow between different bras- sicas is clearly worth considering. Herbicide-resistance genes in crops can be argued to be a good thing, but such genes in weeds are obviously undesirable. The more exciting gen- etic constructs that may come for- ward in the next few years, such as nitrogen fixation in cereals, are even more likely to produce ecological novelties that could be undesirable.

The third set of effects that need to be watched are the changes of hus- bandry and habitat that may come with the use of genetically engin- eered plants. Although the manufac- turer may like to say that a farmer shouldn’t spray herbicide except in the place where he means to and at the doses that are recommended, in practice some farmers disseminate herbicide all over the place, by spray drift, or just by carelessness. It is not reasonable for herbicide manufac- turers to claim that they have no part in the environmental problems that could arise. Deliberately cultivating herbicide-resistant plants can only increase these problemslo.

Estimating the risk The practice in laboratory-based

genetic engineering is to calculate risks on the Brenner scale24, which goes in factors of 10m3. These risk factors might bethought to be related to measured risks; in fact, they are conventional values that allow an ap- propriate degree of containment to be agreed without much argument. Similar factors are starting to appear in the literature about genetically modified plants18.

The history of other environmental hazards gives us little confidence in this a priori approach to risk assess- ment. If one believed the figures put out by nuclear engineers, Three Mile Island and Chernobyl could never have happened. Of course they shouldn’t have happened, but the cor- rect risk assessment has to allow for the vagaries of human mismanage- ment. It is well known to insurance companies that risks can only be esti- mated on the basis of proper experi- ence. For genetically engineered plants, that means much more exper- imentation than has been attempted so far. Most of the undesirable effects will be very small, but keeping a proper record of these may allow an estimate of unpleasant large effects25.

What is needed for the moment is

418

TIYEE vol. 5, no. 72, December 7990

an awareness by those developing committees’ main task is to ensure novel plants that, the more novel that releases are environmentally the plant, the greater the need for safe. Trying to consider as well irformation about the ecological whether they are justifiable in terms behaviour of the construct. Reassur- of balancing benefits with disbenefits ance about safety must come from is not a job that they are well de- empirical evidence, from properly signed to do, because such a balance designed and monitored field trials, is a political rather than a scientific n 3t weakly from first principles. one.

Bdancing the benefits References An argument in favour of the re-

le!ase of GMOs has been that they can produce environmental benefits. The fact that companies have chosen to ci>ncentrate on constructs such as herbicide-resistant plants is a cause of dissatisfaction9r10, though even here there could be a marginal ben- efit if some herbicides were replaced by others that were environmentally less damaging. Some ecologists certainly hope that GMOs that will a,ttack pollution, novel pesticides that will replace chemical pesticides, metal-recovery processes that won’t involve large-scale mining, and simi- lar proposals, will emerge’*. Inten- s ve farming, the loss of hedgerows, nitrate pollution of water supplies and the loss of diversity in meadows are all regarded by many as dis- benefits of agriculture; some GMOs ctIuld produce new problems.

In the present political climate, en- vironmental benefit is a factor that c!Dmpanies will want to pay regard to along with economic benefit. Is this a reasonable concern for regulatory c Jmmittees? We rather doubt it. Such

1 Sussman, M., Collins, C.H., Skinner, F.A. and Stewart-Tull, D.E. (1988) The Release of Genetically-engineered Micro- organisms, Academic Press 2 Council of the European Communities (1990) Off. J. Eur. Communities 33 (L117), 15-27 3 Organisation for Economic Co- operation and Development (1990) Good Developmental Practices for Small Scale Field Research with Genetically Modified Plants and Micro-organisms: A Discussion Document, OECD, Paris 4 Organisation for Economic Co- operation and Development (1986) Recombinant DNA Safety Considerations, OECD, Paris 5 Commissie ad hoc Recombinant DNA Werkzaamheden (1990) Rechtlijnen voor het Werken met Genetisch Gemidificeerde Organismen, Ministrie van Volkshuisvesting, The Hague 6 Health and Safety Executive (1990) The intentional Introduction of Genetically Manipulated Organisms to the Environment (ACGM/HSE/Note 3 revised), HSE, London 7 Tiedje, J.M. et a/. (1989) Ecology70, 298-315 8 Dijvell, A. (1990) Gentec Update 1, l-5 9 Shulman, S. (1990) Nature 344,371 10 Goldburg, IX, Rissler, J., Shand, H.

and Hassebrook, C. (1990) Biotechnology’s Bitter Harvest, National Wildlife Federation, Washington DC 11 Caseley, J.C., Cussans, G.W., Kemp, MS., Moss. J.R. and Atkin. R.K.. eds Herbicide desistance in Weeds and Crops: 1 lth Long Ashton International Symposium, Butterworth (in press) 12 US Congress, Office of Technology Assessment (1988) New Developments in Biotechnology - Field-Testing Engineered Organisms: Genetic and Ecological Issues, US Government Printing Office 13 Ellstrand, N.C. (1988) Trends Ecol. Evol. 3, 30-32 14 Williamson, M. (1988) Trends Ecol. Evol. 3,32-35 15 Baker, H.G. (1974) Annu. Rev. Ecol. Syst. 5, l-24 16 Committee on the Introduction of Genetically Engineered Organisms into the Environment (1987) introduction of Recombinant DNA-Engineered Organisms into the Environment: Key Issues, National Academy Press 17 Flanagan, P.W. (1989) Bull. Ecol. Sot. Am. 70,14-19 18 Keeler, K.H. (1989) Bioflechnology 7, 1134-1139 19 Harlan, J.R. (1965) Euphytica 14, 173-176 20 Hayden, W.J. (1986) Science 231, 103 21 Cavers, P.B. and Brough, M.A. (1985) in Studies on P/ant Demography (White, J., ed.), pp. 143-155, Academic Press 22 Fitter, A., Perrins, J. and Williamson, M. (1996) B;o/lechno/ogy 8,473 23 Pehu, E. et a/. (1989) Theor. Appl. Genet. 78,696-704 24 Health and Safety Executive (1988) Guidelines for the Categorisation of Genetic Manipulation Experiments (ACGM/HSE/Note 7), HSE, London 25 Hsii, K. (1987) Nature 328, 22

-. Sex and Flagellation The centriole is one of the cell’s more enigmatic structures. It lives a Jekyll and Hyde existence, changing from tile basal body, which seeds the pro- duction of cilia and flagellae, into the csntriole, in which guise it is of uncer- tain function. Recent work has indi- c.sted the possibility of DNA tightly packed into the structure’s core. This finding sheds light on theories of the evolutionary origins of the centriole and of its possible involvement in the evolution of sex. Recent experimen- tal work has been testing this latter possibility. _.

Why do organisms reproduce sex- ually? What is the evolutionary origin of the eukaryotic flagellum? At first

L,jurence Hurst is in the Animal Behaviour Re- search Group, Dept of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UK. Alan Grafen is at the Dept of Plant Sciences, University of Oxford, South Parks Road, Oxford OX1 3RA, UK.

sight these questions have little in common. They are, however, linked by one structure, the centriole. The centriole has a schizophrenic exist- ence. Assuming the guise of the basal body, it seeds the production of cilia and flagellae, the locomotory whip- like tails of protists and sperm. When not performing such a role, the fam- iliar ninefold (‘9+0’) circular array of triplet microtubules is surrounded by an aggregation of dense material that acts as a microtubule organizing centre (MTOC), polymerizing tubulin for cell division’. Recent studies have been clarifying the biology of this most enigmatic of structures, and they illuminate theories of its evolution and its possible involve- ment in the evolution of sex.

Immediately after cell division a

Laurence Hurst and Alan Grafen

cell has two centrioles, which be- come four prior to cell division. These move in two pairs to opposite sides of the nucleus and ultimately, when sur- rounded by the MTOC aggregate, form between them the mitotic spindle. Each daughter cell receives one pair of centrioles. Artificial ab- lation of the centriole indicates that it is not necessary for microtubule for- mation and cell division2. Further- more, those organisms (including most land plants) that have no flagellae or cilia in their life cycles lack centrioles’, but they undergo mitosis as efficiently as any organism with centrioles. The association of the cen- triole with a MTOC might just be a mechanism to ensure that both daughter cells receive their due of centrioles2, although there is evidence

419 :( 1990 Elsev~er Scence Puhltshers Ltd (UK1 0169~5347:90/$02 00