EPA's approval of a natural refuge option for Bollgard II and WideStrike cotton last spring caught many growers by surprise. Some couldn't believe EPA would approve such a strategy. Others didn't think the agency had that much sense.

EPA's mandate that farmers plant a percentage of their acres in non-Bt cotton to delay Bt resistance in target pests has been a sore spot almost from the time Monsanto introduced the original Bollgard varieties back in 1996.

Cotton producers said the 5-percent unsprayed or 20-percent sprayed options, as they came to be known, were difficult to manage and cost them money because of the toll untreated tobacco budworms and bollworms took on yields in the acres not planted in Bt varieties.

In its defense, EPA was under pressure to require more refuge acres because of concerns insects that could survive on Bt cotton would also be tolerant to applications of foliar-applied Bt products, a mainstay of insect control for organic farmers.

Early on, the Union of Concerned Scientists accused EPA of failing to institute adequate measures to protect the Bt technology, which involves inserting a gene from Bacillus thuringiensis, a naturally occurring soil bacteria that is toxic to caterpillar insects, into a plant.

“First, the agency approved the sale of Bt crops without tested resistance management plans,” the UCS said. “Then, it failed to move expeditiously when control failures revealed a significant flaw in the Bt cotton resistance management plan — the lack of a high dose.”

Recognizing the need to delay resistance to the Bt toxin — and protect its considerable investment in the technology — Monsanto began developing a new, “second generation” Bt product almost at the same time it launched the original Bollgard.

“It was broadly believed that if you put two genes in the plant that somehow worked differently on the insect that that would reduce the probability of resistant insects surviving both genes and significantly decrease the time to full-blown resistance,” said Walt Mullins, Monsanto technology development manager.

By 2003, Monsanto had developed and launched a second generation Bt cotton, which it marketed as Bollgard II. The latter contained the Cry1Ac Bt gene in its Bollgard cotton and the Cry2Ab gene. Later, Dow AgroSciences introduced its WideStrike Bt cotton with two Bt genes.

Farmers had been growing increasingly restless over having to devote a percentage of their acres to non-Bt cotton so susceptible insects could mate with others of their species that might be developing resistance to the Bt toxin.

Groups such as the Mississippi-based Delta Council were asking why farmers had to plant 20 percent of their acres in lower-yielding varieties and place them in locations that might be better suited for Bt varieties.

Environmentalists and university entomologists argued the refuge acreage requirements should be set even higher. (The Union of Concerned Scientists recommended a 40-percent to 50-percent non-sprayed, non-Bt cotton refuge in Bt cotton fields.)

Monsanto believed its two-gene approach with Bollgard II could mean EPA would reduce rather than raise the refuge requirements.

University researchers had been using computer models to try to predict when bollworms or tobacco budworms could develop resistance to the Bt toxin in genetically engineered cotton.

Most of the models indicated that to prevent resistance from occurring in less than a theoretical 30-year “shelf life” for the Cry1Ac gene, the refuge requirements had to be set at a minimum of 5 percent.

“When you look at the refuge requirements for the two-gene system, the refuge requirement falls to 0.1 percent with the computer models,” said Mullins. “What that means is we should be able to manage resistance with less refuge and fewer refuge moths contributing to the total system.”

But where would the susceptible bollworm and tobacco budworm moths come from if they weren't growing on non-Bt cotton in close proximity to the Bt acres? That was the question EPA and environmentalists were sure to ask when Monsanto proposed giving growers a natural refuge option.

Driven by questions about the need for higher percentages of unsprayed refuges, researchers at North Carolina State University had begun conducting studies to try to determine the prevalence of alternate hosts for tobacco budworms and bollworms.

Scientists had long felt that the bollworm, also known as the corn earworm, a major pest of corn, had numerous alternate hosts in the Carolinas and in other parts of the Cotton Belt.

According to NC State researchers, a stable carbon isotope analysis of Helicoverpa zea or bollworm wings indicated that C4 plants, such as corn, are the major larval hosts in the early to mid-season in Louisiana and Texas; that is, 90 to 100 percent of tested moths developed as larvae on a C4 host.

“Later in the season when cotton is an attractive host, less than 50 percent of bollworm moths were produced from C3 hosts — cotton, soybean and other non-grass hosts,” the scientists said in a paper presented at the Beltwide Cotton Conferences in 2006.

The stable carbon isotope analysis could not easily differentiate between C3 host crops, but the scientists believed they could develop estimates of the contributions of different plants using data from unsprayed field tests. Those indicated that late in the season, 19.17 percent of the adults were produced from corn; 36.09 percent on cotton; 5.64 percent on peanuts and 39.10 percent on soybeans.

“Since nearly 100 percent of the early-season moths were produced from corn, non-Bt corn appears to be the most useful early-season crop host refuge,” the scientists said. “Soybean serves as a major late-season non-Bt host and may play a more pivotal role than a 20 percent sprayed or 5 percent unsprayed non-Bt cotton refuge in delaying resistance evolution to Bt in eastern North Carolina.

“This includes only production on the major crop hosts and ignores production in the numerous H. zea minor crop hosts, wild hosts and immigration of moths from other areas.”

In Heliothis virescens or tobacco budworm, the range of alternate hosts plants has been considered to be much narrower than that of the bollworm.

In the Mid-South, weedy plant species serve as hosts for the first generation of budworms, while subsequent generations are more likely to be associated with cotton. The same situation occurs with first generation budworms in the Southeast.

But later generations of the insect can be found on a number of weedy hosts and on tobacco, cotton, peanut and, to a lesser extent, on soybean in North Carolina, according to North Carolina State researchers.

To try to better assess the situation, Monsanto expanded the alternate host studies from North Carolina to five states across the Southeast and Mid-South in 2004 and 2005. In 2004, tobacco budworm populations were so low they could not capture enough moths to make the study worthwhile, said Mullins.

In 2006, the study was expanded to seven states, including Tennessee and southeast Texas.

Using the stable carbon isotope analysis and field evaluations, researchers found that along the East Coast states, 80 percent to 100 percent of the tobacco budworm moths are derived from sources other than cotton.

“In North Carolina and Georgia, we found significant numbers of tobacco budworm moths, but they were largely non-cotton moths all season long,” said Mullins. “We anticipated a certain degree of that based on tobacco as a host. But peanuts are also a fairly good host for tobacco budworms as are soybeans to an extent and other, weedy hosts for the area.

“But we were amazed at how much the cotton was getting drowned out by everything else that was going on. So we felt that data package was pretty solid based on our discussions with other people.

In the Mid-South states of Arkansas, Louisiana, Mississippi and Tennessee, the profile was different because of the larger fields and absence of larger blocks of weedy hosts that are often found in the Southeast.

“But, in the very worst case scenario we ran into, we saw no less than 10 percent of the moths that were non-cotton moths all season long in any location,” said Mullins. “That means if you have 100 moths at least 10 of those moths were non-cotton derived moths. And more typically in Mississippi and northeast Arkansas, it would be closer to 40 to 50 percent.”

Based on that 10 percent level and the computer modeling that has been done, Monsanto representatives believe the effective natural refuge that would be required based on the strength of a two-gene system would be 0.1 percent.

“So what we said in our proposal to EPA was that when we look at the amount of refuge necessary that, even in the worst case scenario, given the lower end of the statistical significance of the number suggests that we still have significantly more refuge than would be required to maintain efficacy for an extended period of time,” said Mullins.