Written material from Professor Mark Williamson,
Professor Emeritus of Biology, University of York
There is some dispute about whether the study
of invasions is relevant to the assessment of risks from the release
of genetically engineered organisms. So, first, what are these
organisms and why should there be risks?
Molecular genetics now allows a gene to be taken
from one organism and inserted into some totally unrelated one.
Bacterial genes can be put into plants, arthropod genes into viruses.
There are, of course, limits on what can be done but, as the subject
is moving fast, I will not dwell on them here. This transfer of
genes is commonly called genetic engineering. The current fashion
is to refer to genetically modified organisms, or GMOs, rather
than genetically engineered ones, and that is followed in official
documents. For scientists, there is no reason to prefer an ambiguous
and obscure term to one that is reasonably precise (Williamson,
1992). Genetic modification is a term that can be applied to all
genetic programmes, and has no obvious association with restriction
enzymes and the other tools of molecular geneticists. Genetic
engineering is less ambiguous, and gives the flavour of experimental
manipulation, so I will use it here. It is also the term used
in US Congress OTA (1993).
Genetic engineering can be used to make organisms
with new properties that may be commercially useful. In crop plants,
herbicide- and pest-resistant varieties are being developed in
many species. It is possible to change the nature of the crop
product, changing the composition of the oil in oil seeds, manipulating
enzymes so that tomatoes do not go squashy, and many other features.
Pharmaceuticals could be made in plants or produced in milk. Fish
can be made to grow faster (US Congress OTA, 1993; Krattiger and
In principle, any commercially desirable trait
could be added or enhanced. It is not surprising that much research
has been funded, and that many commercial releases are near. On
Krattiger's (1994) count there have been 2,053 field trials of
transgenic plants world-wide up to mid-1994, and that, even allowing
for differences in the definition of a trial, is fairly certainly
too low (Anon, 1994b). Although almost all OECD countries have
some form of regulation, others outside the OECD such as China
and Israel apparently do not. Regulation, such as the European
Union's directive 90/220/EEC, is likely to keep only a few, rather
obviously undesirable, products from the market. It is reasonable
to assume that there will soon be many different genetically engineered
organisms marketed in large numbers world-wide.
Will there be ecological and environmental change
from genetically engineered organisms? Russo and Cove (1995) give
a good overview of all the benefits and hazards from these techniques.
Invasions show that damage can happen when an organism finds itself
in a new environment. For a novel genetically engineered organism
all environments are new. A familiar case where a change to a
new environment, accompanied by a small genetic change, has had
quite unforeseen terrible effects is AIDS. Maybe some day a genetically
engineered organism will produce a major, but quite different,
Human AIDS is caused by two viruses, HIV1 and
HIV2. These are closely related to a group of viruses found in
other primates, the Simian Immunodeficiency Viruses or SIVs (Morrison
and Desrosiers, 1994). These are all single-stranded, encapsulated
RNA viruses, retroviruses. Being RNA viruses, they are far less
stable genetically than DNA organisms. Various strains in one
virus have 80 to 100 per cent identity, closely related retroviruses
attacking other species have about 80 to 90 per cent identity,
more widely related ones 55 to 60 per cent. It would seem that
both HIV2 and SIVmac (which infects captive macaques,
Macaca) are derived from SIVsmm (which infects
the sooty mangaby, Cercocebus torquatus). Similarly, HIV1
is closely related to SIVcpz which is found in chimpanzee
Pan troglodytes. SIVs in wild monkeys and apes are, as
far as is known, non-pathogenic. Rhesus monkeys Macaca mulatta
with SIVmac develop an AIDS-like disease. Pigtail macaque
M. nemestrina with the same virus are killed in a week
or so. Small genetic changes and a new environment can produce
very drastic effects.
As I said at the end of section 5.3.1, major
invasions may come out of the blue at any time. Will genetically
engineered organisms add to these problems?
Some proposals, such as the engineering of non-specific
biological control viruses, are evidently bad practice (Williamson,
1991), but the unnecessary risk comes from the nature of the virus,
not the genetic engineering. It is often asserted that for most
commercial genetic engineering, the invasion model is not appropriate.
For instance, with crop plants, the argument is that the plant
is familiar, the new variety will have to undergo extensive performance
trials, and the genetic novelty is more precise and better understood
than the genetic novelty produced by traditional breeding programmes.
Hence the release of genetically engineered plants is different
from other invasions. It is also sometimes stated that many changes
are needed to change an organism into a weed or a pathogen (National
Academy of Sciences, 1987).
The unsatisfactory points in that argument are
covered in earlier sections of this book. Although the crop plant
is familiar, and the genes inserted are well known, the combination
is novel. There are no universal characters that distinguish weeds
and pathogens from their harmless relatives (section 3.3.2), and
the genetic differences between invasive species and those that
fail to invade may often be small (section 6.2). In fact, many
plants have become weeds merely by being taken to new regions.
It is not surprising that ecologists think that aspects of the
invasion model are relevant to the risk assessment of genetically
engineered organisms (Tiedje et al., 1989; Altmann, 1993;
Shorrocks 1993; US Congress OTA, 1993; Seidler and Levin, 1994).
It is an appropriate model.
Even in those countries where there is effective
regulation of small-scale trials, the study of invasions suggests
that the probability of detecting undesirable products at an early
stage is not large. Many pest invaders have not been recognised
as such for many years, often decades, Impatiens glandulifera
(section 1.3.3) and the muntjac deer (section 5.1) for example.
On the other hand others, such as zebra mussel (section 5.4) were
recognised as problems almost immediately, but spread so fast
that it was difficult to limit the damage. As those genetically
engineered organisms that become problems will usually be commercial
products, they will mostly be widespread quickly, and difficult
to control whether the problem arises soon or not.
Some problems may well be delayed. Texas cytoplasm
is a possible example of how this could happen; it is a genetic
modification of corn, Zea mays. In corn, male sterility
is a most useful agronomic trait, because it allows controlled
breeding without the work of removing the male inflorescences,
the tassels. Texas Cytoplasm varieties are male sterile because
of a change in a mitochondrial gene (Levings, 1990). Remarkably,
the same molecular mechanism that made the plants male-sterile
also made the plants susceptible to a fungal pathogen, Bipolaris
maydis race T. About two decades after the gene, T-urf13,
came into commercial use, the fungal disease devastated the corn
containing that gene, which was by that time 85 per cent of the
US corn hectarage, and made the innovation useless. The molecular
details were known, the pathogen was known, but the interaction
was not predicted and the consequences did not appear until the
new genotype was in full commercial use. It seems optimistic to
suppose that similar failures will not happen in future, however
the regulatory system is designed. Without regulation they might
even become common.
Texas cytoplasm was an agronomic problem. Will
there be ecological and conservation problems? There are two classes
of possibility. One is the spread of the genetically engineered
organism itself, the other is the spread of the engineered gene
in wild relatives of that organism (Raybould and Gray, 1993).
As with all invasions, and bearing the tens rule in mind (section
2.3), it is reasonable to say that neither will happen frequently.
Pests arise in around 1 per cent of organisms introduced at random.
If regulators can control excessive commercial enthusiasm, the
frequency could be much less (Williamson, 1988); that is taking
an optimistic view of the effectiveness of regulators and regulations.
But whatever the proportion, the number of proposed products is
so large, that some ecological damage seems likely, though it
may not be apparent for some decades.
Perhaps the most remarkable general feature
of invasions is how unpredictable they are. One possible gain
from the release of genetically engineered organisms may be a
better understanding of why most genetic variation seems to have
no relation to invasion success, but nevertheless some genes are