[To contribute to this conference, send your message to
For further information on the Electronic Forum on Biotechnology in Food and
Agriculture see Forum website.
NB - participants are assumed to be speaking on their own behalf, unless they state otherwise.]
Sent: 25 June 2002 09:13
Subject: 73: Re: Population genetic mathematical models // Human intervention
Professor Muir again, this is in response to Dr. Wayne Knibb (message 72, June 24) and also to address the Background Document to this conference which states:
"Gene flow may also be facilitated unknowingly by human intervention. For example, for GM crops, this may occur through aid or relief agencies accidentally providing GM seeds in programmes to replenish a ravaged country or region's seed stocks or through farmers using transgenic material, intended as food aid, as seeding stock. In some other situations, GM crop material may be illegally introduced by farmers to non-GM populations because they see an economic advantage in using them."
Risk assessment method based on population genetics/mathematical models, as detailed in my message 69 (June 22), can be applied regardless of how the initial exposure occurs, i.e. whether by escape of the plant/animal, pollen drift, or intentional release by humans. The assumption is that the ultimate fate of the transgene can be predicted based on fitness components estimated in a secure setting. Wayne's message (72) challenges that assumption because (he claims) the models cannot incorporate genotype x environment interaction (GxE) or changes in the fitness components themselves and because earlier models were inadequate.
First, GxE can be incorporated by estimating the fitness components in a range of environments the GM organism may escape to. These are then fed into the model. If the predictions of fate remain the same regardless of which environment they were estimated in, the predictions are robust to that assumption. If not, the GM organism must be restricted from those environments which are shown to be a risk. Secondly, regarding changes in the fitness components themselves, the robustness of the predictions can also be examined by comparing fate of the transgene using a range of fitness components. If the same outcome is predicted over a wide range, the prediction is then robust to that assumption.
The only way to really resolve this argument is to verify the model in a natural setting. However, this is a catch 22 situation because we cannot release GM organisms before they are deemed safe. To address this issue, we used the model to determine if it could predict spread of the Africanized honey bee in the Americas. Based on estimates provided to me, the model predicted almost perfectly not only that the Africanized honey bee would spread but also the rate of spread. Secondly, a paper by Rejmanek and Richardson (1996) showed that using just 3 of my six fitness components they were able to perfectly predict spread of invasive plants. Using all 6 would allow an even more precise prediction.
Finally, if one accepts Wayne's conclusion that we cannot predict fate of a transgene in natural settings from population genetics, the conclusion would be that there is no science based method that can be used to predict spread (or lack of). Using the precautionary principle, the only logical conclusion based on that argument is that we should never release a GM organism because the risk can never be predetermined. I personally feel that is a very pessimistic view that would doom the industry. To be sure, no model is going to be 100% accurate, and there is no such thing as no risk, but this is extreme.
Reference: Rejmanek, M. and D.M. Richardson, 1996. What attributes make some plant species more invasive. Ecology 77:1655-1661.
William M. Muir, Ph.D.
1151 Lilly Hall
W. Lafayette, IN 47906
bmuir (at) purdue.edu
Sent: 25 June 2002 10:42
Subject: 74: Models - Human intervention - Developing countries
This is Jane Morris from South Africa.
In the earlier phases of this conference (message 6, June 3), I pleaded for more research to enable scientists and regulators in developing countries to make informed decisions on GM release. I come back to this now.
Professor Muir's model [see message 73, June 25...Moderator] sounds great,
but is it really going to be a help in a developing country situation? A
model is only as good as the data available to populate it. In developing
countries, the environmental information is often not well documented.
Moreover, the spread of the GMO will often happen through human intervention
(e.g. "brown bag" seed passed from hand to hand amongst small farmers) which
cannot be easily documented or modelled. In many cases, it may be reasonable
to assume that even if such "uncontrolled" spread occurs it will not pose
any greater hazard than the spread of non-GMO seed - that is obviously part
of the total risk assessment. But if we are going to resort to modelling,
let's at least make sure
(a) that we understand the receiving environment adequately, and
(b) that we know how we are going to use the model to assist us in decision making (is an aid organization going to model the effects of distributing maize to the starving people in Africa? - seems somewhat unrealistic!)
E Jane Morris PhD
African Centre for Gene Technologies
P O Box 75011
Tel: +27 12 841 2642
Fax: +27 12 841 3105
Cellular: +27 82 566 2210
e-mail: jmorris (at) csir.co.za