As expected, the resistant caterpillars survived after munching on cotton plants producing only that toxin. The surprise came when Carrière's team put them on pyramided Bt cotton containing Cry2Ab in addition to Cry1Ac.

If the assumption of redundant killing is correct, caterpillars resistant to the first toxin should survive on one-toxin plants, but not on two-toxin plants, because the second toxin should kill them, Carrière explained.

"But on the two-toxin plants, the caterpillars selected for resistance to one toxin survived significantly better than caterpillars from a susceptible strain."

These findings show that the crucial assumption of redundant killing does not apply in this case and may also explain the reports indicating some field populations of cotton bollworm rapidly evolved resistance to both toxins.

Moreover, the team's analysis of published data from eight species of pests reveals that some degree of cross-resistance between Cry1 and Cry2 toxins occurred in 19 of 21 experiments. Contradicting the concept of redundant killing, cross-resistance means that selection with one toxin increases resistance to the other toxin.

According to the study's authors, even low levels of cross-resistance can reduce redundant killing and undermine the pyramid strategy. Carrière explained that this is especially problematic with cotton bollworm and some other pests that are not highly susceptible to Bt toxins to begin with.

The team found violations of other assumptions required for optimal success of the pyramid strategy. In particular, inheritance of resistance to plants producing only Bt toxin Cry1Ac was dominant, which is expected to reduce the ability of refuges to delay resistance.

Refuges consist of standard plants that do not make Bt toxins and thus allow survival of susceptible pests. Under ideal conditions, inheritance of resistance is not dominant and the susceptible pests emerging from refuges greatly outnumber the resistant pests. If so, the matings between two resistant pests needed to produce resistant offspring are unlikely. But if inheritance of resistance is dominant, as seen with cotton bollworm, matings between a resistant moth and a susceptible moth can produce resistant offspring, which hastens resistance.

According to Tabashnik, overly optimistic assumptions have led the EPA to greatly reduce requirements for planting refuges to slow evolution of pest resistance to two-toxin Bt crops.

The new results should come as a wakeup call to consider larger refuges to push resistance further into the future, Carrière pointed out. "Our simulations tell us that with 10 percent of acreage set aside for refuges, resistance evolves quite fast, but if you put 30 or 40 percent aside, you can substantially delay it."

"Our main message is to be more cautious, especially with a pest like the cotton bollworm," Carrière said. "We need more empirical data to refine our simulation models, optimize our strategies and really know how much refuge area is required. Meanwhile, let's not assume that the pyramid strategy is a silver bullet."

More from Western Farm Press

Fire ant balls can wreak havoc

Roadkill a recipe of the bizarre and unexpected

Farmland real estate crash a growing concern

Biggest wine hoax in history reveals trade secrets