A relatively small group of scientists — including some Californians — have taken a hard and thoughtful look at the potential risks of transgenic crops. These varied scientists — including ecologists, soil biologists, agronomists, geneticists, entomologists, pathologists, horticulturists, botanists and molecular biologists — realize that traditional plant improvement and agriculture have, on occasion, created problems, and those problems can serve as models for anticipating the possible downsides of transgenic crops. A set of straightforward, scientifically based concerns has evolved. The most widely discussed concerns fall into two broad categories: (1) problems created directly by growing the crops themselves, and (2) problems created by unintended descendants of those crops.

Environmental biosafety is a relatively new and rapidly developing research area. An excellent source of information on this field is the National Research Council's (NRC 1989, 2000, 2002, 2004) series of peer-reviewed reports on the potential environmental impacts of agricultural biotechnology. The most up-to-date information can be found in peer-reviewed, disciplinary journals such as Environmental Biosafety Research, Ecological Applications and Molecular Ecology.

Direct impacts of crops themselves

Scientific consideration of the direct impacts of transgenic crops has focused almost exclusively on the evolution of pests that are resistant to new strategies for their control, and unwanted impacts on species in associated ecosystems. Another area of concern is the unwanted impacts on surrounding plant and animal communities from the use of transgenic herbicide-resistant plants (Firbank et al. 2003). Resistance to one or more herbicides is a general feature of most crops; also, resistance to herbicides can often be obtained through nontransgenic techniques (Duke et al. 1991). The impact of herbicide-resistant crops on surrounding community diversity depends largely on the type of herbicide, and where and how it is used.

Evolution of resistant pests

Insects, weeds and microbial pathogens often evolve resistance to controls used against them (Barrett 1983; Georghiou 1986; Green et al. 1990). When a pest evolves the ability to attack a crop, the results sometimes can be devastating. The 1970 corn leaf blight epidemic ravaged American cornfields, resulting in the loss of tens of millions of dollars to the industry (NRC 1972).

Resistance evolution is also expected to occur in pests targeted for control by or associated with transgenic crops. Although the evolution of resistance is a continuous process, the evolution of resistant pests has been considered a potential environmental hazard of transgenic crops because more environmentally damaging alternative treatments would then be needed for control. Furthermore, transgenic products at present have resulted in the use of a single, uniform control method over huge areas.

For example, most of the transgenic corn and cotton now grown in the United States, millions of acres, is engineered with a bacterial gene that allows them to manufacture their own pesticide to specifically target certain insect pests. Because the gene comes from the bacterial species known as Bacillus thuringiensis, these plants are commonly known as “Bt corn” and “Bt cotton.” Bt cotton is the most important transgenic crop in California (Taylor et al. 2004). Because the transgenic product does not kill all insect species, it is considered relatively environmentally benign. But the evolution of resistance to Bt crops is considered inevitable (NRC 2000). The U.S. Environmental Protection Agency has issued guidelines mandating that farmers plant “refuges” of non-Bt varieties in plantations of Bt varieties to prevent or delay the evolution of resistance. Despite the commercialization of Bt crops for almost a decade, no pests have yet evolved resistance to Bt crops in the field, suggesting that the refuge strategy has been effective (Tabashnik et al. 2003).

Effects on nontarget species

A crop engineered to interfere with the reproduction or viability of one or more pest species might also interfere with other nonpest species. For example, Bt corn was developed to control certain moth species that damage the crop. Reports of potentially toxic effects of Bt corn pollen and flower parts eaten by monarch butterfly larvae captured widespread attention (Losey et al. 1999). A flurry of subsequent research demonstrated that the effects of Bt pollen on monarch larvae are highly variable, depending on factors such as pollen density, the crop's Bt genotype and environmental factors (Sears et al. 2001). Current commercial Bt corn varieties are not considered hazardous to monarch larvae, but one variety no longer grown would have been. This example illustrates that risk assessment research can clarify whether a putative risk is, in fact, a problem.

But is this a new environmental problem? One might ask, “Isn't it better to deploy a pesticide through a plant that kills only a subset of insects than to spray one on a field that kills all insects willy-nilly?” The answer, of course, would be, “Yes, Bt in corn reduces environmental impacts relative to spraying broad-spectrum insecticides.” However, prior to the advent of Bt corn, many American corn farmers did not spray insecticides to control the pests controlled with Bt (NRC 2000). Those farmers simply took their chances without any control, possibly because the damage from the lepidopteran pests of corn varies so much from one year to the next. In the latter comparison, the addition of Bt, if it carries adverse nontarget effects, does pose a new problem.

Unintended crop descendants

All plants — including crops — are capable of some type of reproduction. The possibility of unintended reproduction by transgenic crops has raised questions about whether their descendants might cause problems. These problems have fallen into two broad categories: first, that the direct feral descendants of the crops may prove to be new weeds or invasives, and second, that unintended hybrids between transgenic crops and other plants could lead to certain problems.

Progeny of the transgenic crop could become a problem if the transgenic trait alters their ecological performance such that they evolve increased aggressiveness. Some crop plants — especially those with a long history of domestication (e.g., corn and soybeans) — pose little hazard because traits that make them useful to humans also reduce their ability to establish feral populations in either agroecosystems or nonagricultural habitats (NRC 1989). But other cultivated plants (e.g., certain forage grasses and turf grasses, ornamentals, rice, rye, alfalfa) often volunteer after cultivation, founding feral populations that create problems (Gressel 2005). In some cases, the tendency to found feral populations could increase as the result of acquiring new traits.

The factors that foster or limit invasiveness are not well understood (Sakai et al. 2001). Most of the current transgenic crop traits — insect, virus and herbicide resistance — are expected to confer a fitness advantage in certain environments. Empirical evolutionary-genetics studies have demonstrated that a new allele that confers a fitness advantage will usually spread rapidly through a population, but it will not necessarily result in the evolution of invasiveness (Bergelson 1994). The mere presence of a transgene that increases fitness cannot be taken as certainty that the invasiveness of a population has increased. Many crops are unlikely to become weedier by the addition of a single trait (Keeler 1989). In a few cases, however, the consequences might be obvious. The evolution of herbicide resistance in a weed population that was previously controlled by that chemical will force the consideration of new control options.

Scientifically based assessment

Genetically engineered crops are a heterogeneous group. It is no more reasonable to lump them all together to argue that, as a group, they pose an environmental danger than it is to lump them all together to argue that, as a group, they will feed the world and cure disease. It is fair to say that just like the products of traditional plant improvement, certain products of genetic engineering will create problems. To the extent that those products can be compared to traditionally improved plants, scientifically based hazards can be identified.