TREE vol. 3, n0.4; T/5 TECH ~01.6, no.4, April I988

References I Colwell, R.K. (1987) inRiskAn&sis Approaches for Environmental Releases qf Genetically Engineered Organisms (Fiksel, J. and Covello, V., eds), Springer-Verlag (in press) 1 Brill, W.J. (1985) Science 227,381384 3 Colwell, R.K., Norse, E.A., Pimentel, D., Sharples, F.E. and Simberloff, D. (1985) Science 229,111-112 4 Levin, D.A. and Kerster, H.W. (1974) Evol. Biol. 7, 139-220 5 Hamrick, J.L. (1982) Am. j. But. 69, 1685-1693 6 Levin, D.A. (lY84) in Perspectives on

Plant Population Ecology (Dirzo, R. and Sarukhan, J., eds), pp.242-260, Sinauer 7 Ellstrand, N.C. and Marshall, D.L. (1985) Am. Nat. 126,60(X16 8 Karataglis, S.S. (1980) PlantSyst. Evol. 134.23-31 9 Levin, D.A. (1983) in Pupulution Biology: Retrospect and Prospect (King, C.E. and Dawson, P.S., eds), pp. 171-183, Columbia University Press IO Hamrick, J.L. and Schnabel, A. (1985) in Population Genetics in Forestry (Gregorius, H.R., ed.), pp. 50-70, Springer-Verlag I I Friedman, S.T. and Adams, W.T.

(1985) Theor. Appl. Genet. 69, 6099613

I2 Kemick, M.D. (1961) FAO Agricultural Studies 55, 181-547

I3 Schmitt, J. (1980) Evolution 34, 934-943

I4 Handel, S.N. (1983) inPollination Biology (Real, L., ed.). pp. 163-211, Academic Press

I5 Levin, D.A. (1981) Ann. MO. Rot. Gard. 68,233-253

I6 Willson, M.F. (1984) in Perspectives on PlantPofiulation Ecology (Dirzo, R. and Sarukhan, J.. eds). pp. 261-276, Sinauer

POTENTIAL EFFECTS OF RECOMBINANT DNA ORGANISMS ON ECOSYSTEMS AND THEIR

COMPONENTS

Will recombinant DNA organisms cause damage in ecosystems, both natural and man-made? The answer depends on how effective the regulat- ory systems that we are setting up will prove to be. There have been lots of problems in the past from man- made introductions of agricultural and other organisms, intentional or accidental. There is general agreement l that with recombinant DNA organisms it is the product - not the method - of producing the organ- isms that matters. As many of the products are intended to be new and unusual, new and unusual environ- mental problems can be expected.

Although recombinant DNA organ- isms have, happily, proved safer in the laboratory than was at one time predicted, the general experience with other environmental problems is different. For instance, in the case of pesticides, nuclear fall-out, acid rain and overfishing, there has not only been much controversy, but many experts predicted far less damage than in fact transpired. The prediction of risk is known to insurance com- panies and other such bodies to be best derived from experience’; pre- dictions from first principles can be many orders of magnitude out, either way. ‘The release of geneti- cally-engineered organisms into the environment and the growing of new crops may have unforeseen adverse effects”, is a sensible statement from a neutral body. My purpose in this review is to enlarge on the ‘may’ in that quotation.

Mark Williamson IS at the Dept of Bdogy. University of York. York YOI 5DD. UK.

MARK WILLIAMSON

There are many examples of organisms which, by new genetic comtrwctian or by find- ing -cores Fin @Sew habitats, have ~M&ied Iarge popu- lationr~ Some have changed yw&- ~~~~~~~

some scam& ax aill as (3r result. In the lit of these natkmal and unin- te&onal experkments, pro- tocols for the examinati of proposals for gelwstic release can be &k%f. Th8 risk Of damag;e will be small, but the damage that co&# be caused is kge. &cause of ,$he variety and s&&l&y of geological in- teract%Bns, some ous

Ibepdas ofcontrolneed

to be considered at the time of r&a%&

Perceptions It is quite frequently said that there

have been no problems from the de- velopment of organisms for cultiva- tion. That is not so, but the statement is usually based on the speaker’s own experience. So I start with three reasons why most developments for cultivation are perceived to be harm- less:

??The first is that most cultivated organisms are indeed harmless. But a small proportion have become or have produced severe pests that are frequently difficult to control. ?? Pests as a set are generally not of agricultural origin, so it is easy to overlook those that are. An organism

becomes a pest because of some special interaction between its properties and human activities, and, roughly, pests are a random sample of all organisms. Most organisms are not used in cultivation, and so most pests are not derived from it.

??The third reason is that many agri- culturists are only interested in dam- age to their own crops, and overlook damage elsewhere. In fact, pests de- rived from agricultural developments are common. Consider, for instance, domesticated mammals. Twenty- three species of mammals have been domesticated to the extent that some individuals in domestication are clearly different from their wild-type ancestors. For these, I can only find four for which there are no records of feral populations (populations estab- lished in the wild outside the original range of the species); these are the Guinea pig (&via porcellus), the llama and alpaca (Lama glama and L. paces - two names although they are part of the same species), the mithan @OS frontalis) and the yak @OS grunniens)“. Of the 19 with feral populations, only Bali cattle (Bos ban- teng) and the reindeer (Rangifer tarandus), have not formed pest populations, and in the case of the reindeer that is merely because the severe ecological damage it has done has not been in a place of interest to agriculturists”. In other words, about three-quarters of domesticated ani- mal species have given rise to pest populations somewhere, normally well away from the area in which domestication took place or the wild ancestor lived.

8~ / 988 Elsewer Publramnr Cambr,dge 0 169.534718830: 00

TREE vol. 3. 120.4; TIBTECH ~01.6. no.4, April I988

The distinction between a crop plant and a weed is very narrow. ‘Those who are not acquainted with the distribution across the world of our worst weeds often seem confused that a plant which is an important weed in one area may be a valuable crop in another place’6. Of the 18 worst ‘World’s Worst Weeds’, only one (Eichornia crassipes) is not used somewhere for an agricultural pur- pose. The range of situations is shown in ‘worst weed 13’, the species group wild oats, which is part of the genus Arena. The main cultivated oat (A. sativa), is essentially only a crop, though it may persist for a short time after cultivation. Avena str&osa, the black oat, is both a fodder crop on poor soils, and a minor weed. Avena fatua, the commonest and most troublesome wild oat, has apparently be,en cultivated on occasion in the past’. It is very probably derived from A. sativa. Auena sativa and A. fatua are hexaploid, A. strigosa is diploid. The history and taxonomy of Avena are complicated7, but there seem to have been frequent switches from crop to weed and vice versa.

Starting from the crop end, the picture is the same. ‘There are weed barleys and weed wheats. There are weed ryes, weed rices, weed sorghums, weed sugarcanes, weed maize and weed oats . . . there are weed carrots, weed beets, weed radishes, weed lettuce, weed pep- pers, weed potatoes, weed tomatoes, weed sunflowers, weed safflower, weed hemp, weed watermelon and many, many more. Weed races are not confined to annuals nor to small herbs. Citrus spp., Manihot, Psid- ium, Carica, Punica, Mangifera, Passijlora and Prunus, have weed races’n. Unlike feral animals, weed races often occur close to the point of the development of the crop, but equally occur in distant places.

Weeds and feral animals are par- ticularly conspicuous in agricultural land, but they do also get into natural communities, and occasionally cause much damage.

Types of effect in ecosystems ‘Jrbitrarily, I will consider recorded

effects of ecological damage in the general environment under five head- ings. These headings, perhaps, indi- cate those types of organism which, if they are developed by recombinant DNA methods, would need particular sautiny.

The first category is generalist spe- cies (see Box 1). These are much more dangerous than specialistsY. There are plenty of examples of this

from HawaiilO, such as ants and mon- goose. Rats and goats are other well- known dangerous generalists. It might be said that such cases are irrelevant to discussions of recom- binant DNA organisms, but the pro- posal to put genera1 insect toxins into genera1 soil bacterial’, for instance, comes in this category.

The second category is where an organism, by being moved to a new site, is set free from its enemies (predators, parasitoids, etc.), or, to put it another way, when an organism introduced sits at the top of the food web9. Some of the worst examples come from fish introduced for food. In Lake Gatun in the Panama Canal Zone a predatory fish species which escaped from a fish farm in the river exterminated half of the lake’s diur- nally foraging species of fish12. Even more serious things have been hap- pening in the African lakes, where native fisheries have been destroyed, and the well-known, important and much-studied endemic fish forms of the lakes endangered13. Similarly, the introduction of predatory molluscs for biological control has exterminated irreplaceable sets of endemic snails14. In these last two instances, species have been introduced deliberately (either new species for commercial fishing, or a biological control agent) and yet at the same time carelessly, in that the consequences were not thought of. To put it another way, there was inadequate, indeed no effective, regulation of the introduc- tions. This lack of sensible regulation

has led to the loss of many species, and these species take millions of years to develop.

Perhaps one of the commonest causes of trouble, and certainly a cause that will be difficult for regu- lators to deal with, is where species have, on being introduced to only a slightly different environment, changed from being innocuous, or nearly so, to being major pests. Plant pests in semi-natural vegetation, such as rhododendron15 and many of the ‘weeds of agricultural land discussed above, are simple examples. Others have had a more direct effect on man. Legionella pneumophila is encouraged by certain plumbing practice@ and becomes noticeably lethal when dis- tributed in a new way. Myxomatosis was a minor disease of a South Ameri- can species; its devastating effect in a new habitat, the common rabbit Oryctolagus cuniculus, is well known. It seems that AIDS has come in the same way, the HIV virus having jumped from some African monkey to man.

Of more immediate concern are those cases where genetic change has led to an ecological change (see also Box 2). The best documented exam- ple, at the molecular level, is influenza17; accumulated mutations lead to the avoidance of the immune response, and hence to successive epidemics. There are plenty of cases of ecological change following chro- mosomal changes. The combination of genotypes of two species leads to a new species with an ecology different from either parent. Proposals for similar changes are quite likely in genetic manipulations falling within regulations or guidelines. A well- known example is the cord-grass Spartina. The new species, which is both a useful one and a weed, is S. anglica. As a species it is only about 100 years old, and originated in the Solent, England. It is the fertile tetra- ploid formed from the sterile diploid S. x townsendii, which itself arose earlier in the 19th century in the same area by hybridization between the native S. mtitima and the introduced American S. alterniflora. Spartina anglica has the remarkable ability to grow on otherwise barren tidal mud- flats. It has therefore been planted in many parts of the world to prevent erosion and to reclaim land. In other places it is a pest, narrowing channels and blocking harbours in many parts of western Europe. The important feature - an ecological range not matched by either parent - is a com- mon feature of plant allopolyploids, and allopolyploidy is one of the

TREE vol. 3. no.4; TIBTECH vol.6, no.4, April 1988

Box 2. Genes and ecology Part of the power of genetic engineering is that it enables precise changes to be made in the genome - changes specified exactly by the nucleotides. Does this mean that precise changes can be made to the ecological characters of a species? From the environmental viewpoint, is it enough to say that either the gene will be expressed or it won’t? Unfortunately, the answer to both questions is a firm ‘no’.

If there were a simple connection between genes and ecology, much of the very extensive work in ecological genetics and in evolutionary ecology would be far simpler. The reason why the relationship is complicated can be looked at from either end.

From the genetical end, it is well known - indeed elementary - that genes have pleiotropic effects. Products of different genes interact to produce the phenotype: each gene affects many characters. Some of these characters will be of ecological import- ance. So, in general, any genetic change may affect a variety of characters of ecological importance. The primary gene product is. from the point of view of knowing what will happen in an environmental release. far from being the primary point of interest.

From the ecological end, characters like the intrinsic rate of natural increase, tolerance to physical factors and palatability are affected by many genes and by their interactions.

As an exercise, try this one: what could be the environmental consequences of inserting an amylase gene into Saccharomyces? And when you’ve answered that, say whether you think that the release of such a construct should be free of regulation, be permitted without experiments to determine the actual environmental effects.

commonest modes of speciation in plants.

Little is known as yet about the ecological effects of plasmids. but it is likely that they will, in some cases, be as surprising as those of chromosome sets. Plasmids of E. roli would appear to affect dispersal; plasmids in Khizo-

hium do affect host range. The final ecological effect in my

classification, and one that certainly needs to be watched for in recom- binant DNA and related proposals, consists of the side effects of the invader. Pinus strobus, the American eastern white pine, was grown in Britain as Weymouth pine from the early 18th century. It is a most useful tree, and was initially widely planted. However, neither this species nor most other five-needle pines can now be usefully raised in forestry plan- tations because of invasion by pine blister rust, Cvonartium ribicola. The life cycle of the rust requires both a pine and a species of Ribes, particular- ly blackcurrant, K. nigrunz. The rust is said to have been introduced from Asia via the two-needle European P. cembva, the Arolla or Swiss stone pine, around 1900 (Ref. 18). The presence of the rust on blackcurrant and the absence of plantations of tive- needle pines in Hritain are both easily overlooked.

Other examples of side effects come from insects introduced. with inadequate screening, for biological control into Hawaii”‘. ‘The import- ation of parasites to control various moths of economic importance has resulted in wholesale slaughter and near or complete extermination of countless species. It is now impos- sible to see the Hawaiian lepidoptera in the natural proliferation of species and individuals of ]-the late 19th century]““. Predation of native cater- pillars led to the rarity and perhaps

extinction of native predators, es- pecially Odynerus wasps. This decline of native arthropods, both herbivores and predators, may welt have been a factor in the decline and extinction of Hawaii’s insectivorous birds.

All these five categories of effects can perhaps be summarized by saying that the most important single thing for regulators to look at is the ques- tion of the specificity of the construct being released.

Probabilities and risks The intention in the formation of

new recombinant DNA organisms is to produce organisms that survive, at least for a short time and often per- manently, in the general environ- ment. and to produce organisms that are different from those that are there in the first place. Ecologists regard such introductions as invasions; a lot is known about the history of different invasions, but rather little is known about why some species invade and others fail. A recent SCOPE inter- national programme has surveyed the ecology of biological invasions2imz7. From these surveys, the oft-repeated assertion that the probability of an invasion succeeding is small, while the risk from a successful invasion is large. is evidently true. With recom- binant DNA techniques, we should attempt to design organisms where the probabitity of establishment, ex- cept exactly where they are wanted, is kept small. The ways of doing this are discussed below. All experience of safety regulations, insurance and such like, indicates that the probabil- ity cannot be reduced to zero. How small are the probabilities for natural invaders? GrovesZ”.2’3 considers that perhaps less than 10% of plants intro- duced to Australia have become naturalized and perhaps less than 2.5% have become weeds. William-

son and Brown::” suggested about 10% of invaders to Britain became established, and 10% of these, that is to say l%, became pests. From these two surveys, it looks as though the probability of invasive species becom- ing a nuisance of some sort would be in the range 0.1-l%, or possibly. a little higher. If genetic engineers can reduce this by an order of magnitude they wilt have done well, but the frequency will still be high enough to cause great economic and ecological loss.

To minimize the risk, there are six points that we should look for. The first, and probably the most import- ant, is specificity. It is well known in biological control that an exact matching of the control agent against the pest is often necessary, using sub-specific categories. Conversely. nowadays new biological control agents have to be screened against all manner of alternative targets, and in well-regulated cases are only accepted for release when their specificity has been demonstrated. The same types of rule are needed for recombinant DNA organisms.

The second point, related to the first but not the same, is predictabil- ity. There have been cases where very unexpected invasions have occurred, as for instance by the planktonic unicellular alga Biddulflhia sinensis. This was known from warm tropical salty seas from the Red Sea to Japan. At the beginning of this century it invaded Northern Euro- pean waters, and became a major species with a population peak in mid- winter in the cold neritic waters of the southern North Sea:“‘,:“. At the same time, it is an example of a typical invader in that although maintaining a permanent, quite large population. it is in no sense a pest. The African and Indian ring-necked parakeet (Psitta- cula cromeri) has established feral populations in south-east England::‘. Such cases of species behaving in an unpredicted way are not common amongst invaders, but it is a phenomenon to beware of with re- combinant DNA organisms, particu- larly when considering very novel genetic combinations.

The third point is that of reversibil- ity. The muskrat, Ondatra zibethico. an invader from North America, was exterminated in Britain with consider- able difficulty and much damage to other species’“, while in Ireland it was removed reasonably easily’:‘. Kut it has established itself beyond recall on the continent of Europe”“. If the release of a recombinant DNA organ- ism is not reversible, extensive tests

TREE vol. 3, no.4; TBTECH ~01.6, no.4, April 1988

on its ecological range and specificity are needed.

The next two points, namely, type of reproduction and dispersability, are found particularly in the weed litera- ture. High reproductive rates, an abil- it:’ for reproducing forms to persist and high dispersal rates all lead to difficulties of control. Nevertheless, there are many invaders which are not well-endowed with these charac- ters, merely sufficiently so.

Finally, the importance of the scale of use has been emphasized by the K&an Committee’. With recom- binant DNA organisms for commer- cial use, regulators will have to guess the eventual commercial success of products and devise tests to see the likely effect on the success of the invasions. An example of the import- ance of this can be seen with feral mammal populations. Almost all of these come from stocks where there has been marked genetic modifi- cation, which on common sense grounds one would think would re- duce the ability to invade. Common sense, here, is frequently wrong; nevertheless the success of these forms compared with unmodified yet domesticated forms is best ascribed tc the scale and geographical extent of’ the potential for establishing feral populations, i.e. their large domesti- cated populations.

Implications for the regulation of recombinant ONA organisms

Natural environments are complex, with many surprising things going on in them. Nevertheless, experience of past disruptions of ecosystems gives reasonably clear indications of at least some of the dangers that can be avoided by sensible design and sen- sible regulation of recombinant DNA organisms. Take just two of the pro- posals for release that have come fc’rward in Britain. For a virus to be used to control a particular lepidop- teran pest, the requirements are specificity to that pest, predictability in the sense of genetic stability and other characters, and, preferably, re- versibility by self-destruction outside the laboratory. The team under David Bishop (see p. 00, this issue) has acted with great responsibility to- wards all these aspects. Another proposal is for a hybrid between the domestic potato and another potato species from South America. The trials so far have ensured the hybrid will neither reproduce nor disperse. If future trials are called for, these aspects will need careful study, to ensure that hybrid potatoes become neither an agricultural weed, nor an

invader of non-agricultural habitats. 16 Cobourne, T.S. and Ashworth, T. (1986) These two examples show the need Luncetii, 583 for case-by-case examination in the I7 Palese, P. and Webster, R. (1987)iVature early stages, the possibility of general 329,486 rules at a later stage, and a need for 18 von Broembsen, S. in Ref. 27 (in press) close collaboration between ecol- I9 Howarth, F.G. (1983) PYOC. Hawaii. ogists, genetic engineers, and admin- Entomol. Sot. 24,239-244

istrators. 20 Zimmerman, E.C. (1958) Insects ofHawaii 7: Mwolqhii@teru, University of Hawaii Press 21 Groves, R.H. and Burden, J.T., eds (1986)

Acknowledgements Ecology of Biological Invasions: An Australian I thank J.D. Currey, A.H. Fitter, S&S. Hardy, R.

Law, J.H. Lawton and C.C.D. Williamson for f’ey$e&e, Au&&n Academy of Science and

comments on the manuscript, and also B.P. Ager, Cambridge University Press

M.J. Crawley, S.L. Pimm and M.B. Usher for much 22 Mooney, H.A. and Drake, J.A., eds (1986)

useful advice. Ecology of Biological Invasions of North America and Hawaii (Ecological Studies 581, Springer-Verlag

References 23 Komberg, H. and Williamson, M.H., eds I Committee on the Introduction of Genetically (1987) Quantitative Aspects ofthe Ecology of Engineered Organisms into the Environment Biological Invasions, Royal Society (1987) Introduction of Recombinant DNA- 24 Joenje, W., Bakker, K. and Vlijm, L., eds Engineered Organisms into tti Environment: (1987) PYOC. K. Ned. Akak. Wet. 90, l-80 Key Issues, National Academy Press 25 Macdonald, LA. W., Kruger, F.T. and 2 Hsii, K. (1987) Nature 328,22 Ferrar, A.A., eds (1986) The Ecology and 3 House of Lords Select Committee on the Management of Biological invasions in European Community (1987) Biotechnology in Southern Africa, Oxford University Press the Community, Her Majesty’s Stationery 26 Gray, A.J., Crawley, M.J. and Edwards, Office P . J . , eds (1987) Colonization, Succession and 4 Mason, I.L., ed. (1984) Evolution of Stability, Blackwell Domesticated Animals, Longman 27 Drake, J.A., di Castri, F., Groves, R.H., 5 Williamson, M. (1981) Island Populations, Kruger, F., Mooney, H.A., Rejmanek, M. and Oxford University Press Wiison, M.H., eds Biological Invasions: a 6 Hahn, L.G., Plucknett, D.L., Pancho, J.V. Global Perspective, Wiley (in press) and Herberger, J.P. (1977) The World’s Worst 28 Groves, R.H. (1986) in Resilience in Weeds, University Press of Hawaii Mediterranean-type Ecosystems (Dell, B., 7 Holden, J.H.W. (1976) in Evolution of Crop Hopkins, A.T.M. and Lamont, B.B., eds), Plants (Simmonds, N. W., ed.), pp. 86-90, pp. 129-145, Junk Longman 29 Groves, R.H. (1986) in Ecology ofBiological 8 Harlan, J.R. (1965) Eqbhyticu 14,173-176 Invasions: An Australian Perspective (Groves, 9 Pi, S.L. (1987) Trends Ecol. Euol. 2, R.H. and Burden, J.J., eds), pp. 137-149, 106-108 Australian Academy of Science and Cambridge IO Vitousek, P.M., Loope, L.L. and Stone, University Press C.P. (1987) Trends Ecol. Evol. 2,224-227 30 Williamson, M.H. and Brown, K. (1986) I I Marx, J.L. (1987) Science 237,1413-1417 Phil. Trans. R. Sot. LondonSer. B. 314, I2 Zaret, T.M. and Paine, R.T. (1973)Science 505-522 182,4494155 31 Williamson, M. (1987) Symp. BY. Ecol. Sot. 13 Barel, C.D.N., Dorit, R., Greenwood, 26,353-371 P.H., Fryer, G., Hughes, N. et al. (1985) 32 Sharrock, J.T.R. (1976) The Atlas of Nature 315,19-20 Breeding Birds in Britain and Ireland, British I4 Clarke, B., Murray, J. and Johnson, M.S., Trust for Ornithology and Irish Wildbird (1984) Pacific Sci. 38,97-104 Conservancy I5 Usher, M.B. (1986) Phil. Trans. R. Sot. 33 Moffat, C.B. (1938) Proc. R. Irish Acad. B London Ser. B 314,695710 44,61-128

Trends in Biotechnology - some forthcoming articles: ??Mass spectrometry in proteins and peptides (Application to protein

structure analysis?)

??Adhesive fermentation -engineering and cell physiological aspects

??Bubblefree aeration reactors for mammalian cells

??Continuous analyzers based on enzyme detectors

0 Genetic engineering of food proteins

??Assessing deactivation-disguised kinetics in biotechnological processes

??Seeds and plant genetic resource banks in the Third World

??Scale-up engineered parameters in animal cell culture

0 The nature of recombinant protein inclusion bodies and their formation

??CPG chromatography of proteins