G enetically modified organisms (GMOs) are now a part of everyday life in the US, with ingredients from GM crops present in the majority of our processed foods (Hopkin 2001). GM crops are also a major feature of our landscape. In 2003, GM crops were grown on 42.8 million hectares within the US alone, an area larger than the entire state of California (James 2003). Genetic engi- neering promises society everything from crops with improved agronomic and nutritional qualities to frivoli- ties, such as colored lawns and fluorescing pet fish (Figure 1). The possibilities seem to be limited only by our imag- inations (Dunwell 1999). Thus far, however, commer- cially available GMOs have been almost exclusively lim- ited to crops of major economic importance (eg corn, soybean, cotton, and canola), and the commercially introduced traits have been primarily agronomic (eg insect or herbicide resistance; Figure 2). Different degrees of confinement are warranted for dif- ferent types of GM crops, depending primarily on the nature of the genetically altered traits and the breeding system of the crops and related species. For those GM varieties that have been deregulated by the US Department of Agriculture (ie approved for widespread commercial production), confinement is typically not expected. However, there are special cases in which there has been an intent to locally segregate or contain trans- genes even for deregulated varieties. For example, some commercially approved GM crops are restricted from being grown in particular states where there are concerns regarding hybridization with weedy relatives; thus, GM cotton can be grown in all states except Florida and Hawaii (EPA 2000). In another well known case, it was assumed that potential risk could be avoided by requiring that seeds remain segregated according to their allowed use; for example, StarLink corn was intended as animal feed but not as human food. The accumulated experiences regarding containment of crops with altered agronomic properties – both before and after their deregulation – pro- vide clear lessons about our ability to contain transgenes. A second body of evidence regarding containment comes from the last 4 or 5 years of experience with crops that are engineered to cheaply and efficiently produce pharmaceutical and industrial proteins (Giddings et al. 2000; see Table 1 for examples of pharmaceutical proteins currently in development). For these varieties there are no realistic expectations of deregulated production – their cultivation will, in all likelihood, forever be limited to “confined” field trials. Examples of the confinement measures for the cultivation of these crops include geo- graphic isolation, scouting for and destroying escaped plants that sprout in subsequent seasons (volunteer plants), and the dedication of equipment for use only on the regulated crop. Inexpensive production of drugs, vac- cines, and enzymes would provide benefits to society, but these crops may also represent new risks and they cer- tainly pose new challenges to our ability to contain trans- genes while growing plants outdoors. The issue of containing transgenes has become a flash- point in the current debate about biotechnology. If trans- 93 © The Ecological Society of America www.frontiersinecology.org REVIEWS REVIEWS REVIEWS Can crop transgenes be kept on a leash? Michelle Marvier 1 and Rene C Van Acker 2 Debates about the benefits and risks of genetically modified (GM) crops need to acknowledge two realities: (1) the movement of transgenes beyond their intended destinations is a virtual certainty; and (2) it is unlikely that transgenes can be retracted once they have escaped. Transgenes escape via the movement of pollen and seeds, and this movement is facilitated by the growing number of incidents involving human error. Re-examination of our risk management policies and our assumptions about containment is essential as genes coding for phar- maceutical and industrial proteins are being inserted into the second generation of GM food crops. Even the best designed risk management can be foiled by human error, a reality that is underestimated by most GM crop-risk analyses. Thus, our evaluation of risk should assume that whatever transgene is being examined has a good chance of escaping. Front Ecol Environ 2005; 3(2): 93–100 1 Biology Department and Environmental Studies Institute, Santa Clara University, CA (mmarvier@scu.edu); 2 Department of Plant Science, University of Manitoba, Winnipeg, MB, Canada. In a nutshell: • The movement of transgenes beyond their intended destina- tions is a virtual certainty • It is unlikely that transgenes can be retracted once they have escaped • Human error can foil even the best designed strategies for risk management • Evaluation of risk should assume that transgenes have a good chance of escaping • The second generation of GM plants includes traits which could put humans, as well as ecosystems, at risk following transgene escape