In her seminal work, Silent Spring, Rachel Carson writes:‘‘If Darwin were alive today the insect world would delight and astound him with its impressive verification of his theories of survival of the fittest. Under the stress of intensive chemical spraying the weaker members of the insect populations are being weeded out.’’(1) Evolution of insecticide resistance in 400 species of insects not only confirms Darwin’s theories, it threatens agriculture and human health worldwide (www. pesticideresistance. com/; ref. 2). To reduce reliance on insecticide sprays, corn and cotton have been genetically engineered to produce insecticidal crystal (Cry) proteins derived from the bacterium Bacillus thuringiensis (Bt). Transgenic Bt corn and Bt cotton grew on 42 million ha during 2007, with a cumulative total of 200 million ha planted worldwide since their commercialization in 1996 (3). However, the history of insecticide resistance informs us that adaptation by insects could diminish the long-term efficacy of Bt crops and the associated economic, health, and environmental benefits (4–6). To date, field-evolved resistance to Bt crops has been documented in only 3 insect species (Fig. 1)(7–10). Along with other evidence, the report by Meihls et al.(11) in this issue of PNAS suggests that refuges of plants that do not produce Bt toxins may be useful for delaying insect resistance to Bt crops.

The refuge strategy, which is mandated in the United States and elsewhere, is based on the idea that most of the rare resistant pests surviving on Bt crops will mate with abundant susceptible pests from nearby refuges of host plants without Bt toxins (12, 13; www. epa. gov/EPAPEST/1998/January/Day-14/paper. pdf). If inheritance of resistance is recessive, the hybrid progeny from such matings will die on Bt crops, substantially slowing the evolution of resistance. This approach is sometimes called the ‘‘high-dose refuge strategy’’because it works best if the dose of toxin ingested by insects on Bt plants is high enough to kill all or nearly all of the aforementioned hybrid progeny (12, 13). In principle, if a high dose is not achieved, resistance can be delayed by increasing refuge abundance, which lowers the proportion of the population selected for resistance to compensate for survival of hybrid progeny on Bt plants (12, 13). The most direct way to test the highdose hypothesis is to let resistant and susceptible adults mate in the laboratory and