The National Academy of Sciences Press announced a (pre-publication) downloadable book entitled “Genetically Engineered Organisms, Wildlife, and Habitat: A Workshop Summary” . Being the result of a workshop, the content is both brief and choppy; however, it provides an interesting summation of the state of the science of GEOs (genetically engineered organisms), particularly with respect to their effect “in the wild”. GEOs outside of the lab may be more prevalent than one realizes. I was most surprised to read the following statistics in the introduction:
Less that two decades ago, genetically engineered organisms (GEOs) were the subject of much scientific study, but not part of everyday life. By 2006 — eleven years after the first commercial introduction of corn plants engineered to produce their own insecticide….more than 123 million acres of land in the United States were planted with genetically engineered crops. Today, 89 percent of all soybeans, 83 percent of cotton, and 61 percent of corn grown in the United States are the products of genetic engineering.
The next paragraph summarizes the overarching concern with GEOs released from the lab:
A key question related to GE crops has been their potential and actual effects on the environment, and numerous studies have been conducted to assess the risks and examine the outcomes of transgenic crops. Those studies have generally informed and strengthened the regulatory oversight of GEOs, but questions still linger in the scientific community about whether GE crops have been evaluated in a broad, long-term ecological context that might expose more subtle effects over time. Those questions also apply to the next generation of GEOs that are in development or poised for field study. Given the diversity of taxa involved and novel traits contemplated, ecologists wonder how the environmental effects of the new GEOs might be manifested, if at all, and how such effects can be detected.
These questions are particularly pressing with respect to GE micro-organisms, which by their very nature cannot be realistically confined to a finite geographical area or single ecological niche once released:
Microbes add complexity to the discussion of the effects of GEOs on wildlife and their habitats, asserted Michael Allen (University of California, Riverside). Although fewer than a dozen microbial traits have been approved for release, the diversity of GE microbes in development is vast and includes those developed for plant protection, improved nutrition, metal absorption, and other functions. Rather than list all the properties described in the hundreds of papers written about them, Allen suggested that he focus his presentation on the challenges and approaches to the study of microbes.
First, the dispersal of microbes cannot be contained, even with the kinds of facilities described earlier in the workshop for fish. Referring to the movement of microbes, Allen observed, “if it can happen, it will.” Microorganisms can travel long distances through events such as fire, hurricanes, and human or animal movement. However, even small distances can matter, especially at the interface of developed and wildland habitat. Ecologists have long observed that problems are often associated with the introduction of any exotic species into the natural environment, transgenic or not. For example, he said the presence of exotic grasses is believed to contribute to a more frequent fire cycle in California.
Of course, ultimately what makes GEOs so much more potentially dangerous than, say, a toxic chemical spill, is that organisms reproduce and perpetuate: GEOs can potentially multiply their effects on native species and environments, both temporally and spatially, by virtue of reproduction and propagation. Additionally, reproduction brings in the potential for interactions at a genetic level, and the issue of “gene flow” is a serious concern:
Because current environmental concerns and regulations prevent gene flow from being studied in the field, the group felt that simulations and alternatives using GEO proxies, such as natural mutants, are needed. The effect of factors, such as flowering, phenology, and pollen viability on gene flow rate and distance, need to be better understood. Eventually, if GE trees are to be released, there is likely to be some gene flow, because containment is never complete. Therefore, the group would like to see agreement on an acceptable threshold of gene flow, which might be established by research with proxies, at least in terms of risk quantification.
One participant observed that there are different opinions about whether it is possible to draw generalizations about the risk of categories or types of genes. He suggested that some research on gene-by-environment-by-organism (G x E x O) effects shows that hybrids that result from a cross of a transgenic and nontransgenic species have improved fitness characteristics, suggesting that containment must be maintained. However, he added, much can be learned by setting up pre-flowering systems, perhaps with some additional redundant systems built in.
Another participant noted that when the question was posed to a scientific advisory panel a few years ago about an acceptable level of gene flow, the response was that even a tiny level of gene flow just shifts the time frame of the effects: thus, the acceptable level, according to this panel, was zero. The Catch-22, summarized Harry, is that because we do not know what the fitness effects are, there can be no release—but without field study, the fitness effects will not be known. An acceptable level of gene flow perhaps could be on the order of a mutation rate, proposed a participant, because that kind of gene flow would be the same as a mutation occurring within the population.
In the Concluding Thoughts section, the following ideas are put forth as “takeaway messages” from the workshop:
proxy learning (using nontransgenic surrogates to study the effect a GEO would have on an environment)
large facilities and containment (creating a surrogate not of the GEO; but rather, of the environment
sensitive indicators (identify indicator species or other phenomena)
comparisons, baselines, contrasts
genetic background (probing genetic background to identify phenotype expression)
I would suggest that all of these are reliant on one of the other “takeaway messages” put forth: models. Whether they are explicitly invoked or not, the design of a surrogate facility, the choice of nontransgenic surrogate species, the choice of indicator species, etc., are all derived from one or more models.
Herein lies the central predictive difficulty with GEOs, for two reasons. First, a formal model, as a finite object, can model only a finite number of interactions; the model is thus effectively “closed” to all other interactions. It becomes a matter of art for the modeler to choose which interactions a model will include and which it will not. Thus, two things become apparent: 1) the modeler skews the behavior of the model based upon the selected and the ignored interactions, and 2) this selection process itself will be skewed by the richness (or sparseness) of the set of interactions from which the modeler may choose.
In this sense, the Catch-22 previously mentioned arises for the modeler in the form: We cannot know for sure which interactions to select as important to the model, and further, we cannot know for sure if our set of interactions is even a rich enough set to choose from, until we have the empirical data; but we cannot have the empirical data until our models assure us, to some degree, of the safety of releasing GEOs.
The second predictive difficulty with GEOs arises from the inherent unpredictability of the stability of any organism. That is, just as a model is closed to interactions it does not include, any temporal model in which the GEOs properties remains fixed over time is essentially modeling a purely non-mutating organism, which becomes an unrealistic very quickly as the timescale increases and as the population of the GEO increases. At best, mutation rates can be estimated statistically; but even then, the exact nature of those mutations cannot be predicted, and therefore, neither can the effects of those mutations.
At the same time, it must be acknowledged that we also cannot better predict what nontransgenic organisms will do. We do not know what mutations may arise in the wild, and the interactions that may result. We do not know what gene transfers will occur and their effects. We further do not know what, if any, extraterrestrial organisms are falling onto this planet from space and what effects they might have. And so on.
None of these limitations on our knowledge of the existing Nature can be used as justification for haphazard GEO release. At the same time, it must be acknowledged that this same lack of knowledge precludes any kind of guarantee that a world free of GEOs will be essentially safer or better in any way.
 Whitacre, P.T. (Rapp.). 2008. Genetically Engineered Organisms, Wildlife, and Habitat: A Workshop Summary. NAS. ISBN-10: 0-309-12085-3. PDF