Introduction
To control some key insect pests, corn has been genetically modified to produce insecticidal proteins from the bacterium Bacillus thuringiensis (Bt). Unlike sprayed insecticides, the Bt toxins produced by such transgenic corn can kill pests inside corn stalks and ears. Further, compared with broad spectrum insecticides, Bt toxins cause less harm to most non-target organisms, including beneficial insects, wildlife, and people.

Bt corn grows on millions of hectares in the U.S. and elsewhere, but its usefulness will be cut short if pests evolve resistance to Bt toxins. So far, field-evolved resistance to Bt crops has not been reported. Although only one pest (diamondback moth) has evolved resistance to Bt sprays in the field, more than a dozen have evolved resistance to Bt toxins in the laboratory and in greenhouses.

To slow pest resistance to Bt crops, the U.S. Environmental Protection Agency (EPA) has mandated the "high-dose/refuge strategy" requiring farmers to grow non-Bt refuges near Bt crops. The non-Bt refuges are designed to enable survival of susceptible pests. Ideally, rare resistant individuals emerging from Bt corn mate almost exclusively with relatively abundant susceptible individuals emerging from refuges. If resistance is inherited as a recessive trait, the hybrid offspring resulting from matings between resistant and susceptible adults will be killed by the Bt crop. Models predict that the strategy will delay resistance substantially if these assumptions hold, but pollen-mediated gene flow from Bt crops to refuge plants could disrupt this.

Gene flow in plants occurs when pollen from one plant fertilizes another plant, carrying with it the genes from the first plant. We focused on corn in this study, but all genetically engineered crops produce pollen-containing transgenes. If pollen from transgenic plants moves by wind or if insect pollinators fertilize related non-transgenic plants, the transgenes may be integrated into the genome of the offspring, altering the traits of the offspring. Previous concerns about gene flow from transgenic crops have emphasized movement of transgenes to wild relatives of crops, landraces, and organic plantings while effects on pest resistance had been overlooked.

Gene flow to refuges from Bt corn can cause Bt toxin production in seeds (i.e., kernels) but not in vegetative tissues of refuge plants. Such refuge contamination, causing toxin production in corn kernels, could affect caterpillars of pests such as corn earworm (Helicoverpa zea) that eat corn kernels. In addition, during their second or later generations, 20 to 50% of populations of European corn borer, (Ostrinia nubilalis), the main target of Bt corn, also eat kernels.

Corn is primarily wind-pollinated, with approximately 97% outcrossing between plants and fertilization occurring at up to 200 m. To test the hypothesis that gene flow from Bt corn causes Bt toxin production in non-Bt corn refuges, we sampled kernels from ears of non-Bt corn along transects near Bt corn. Tests of non-Bt corn sampled from the field showed that the concentration of Bt toxin Cry1Ab in kernels and the percentage of kernels with Cry1Ab decreased with distance from Bt corn. These results imply that pollen-mediated gene flow from Bt corn caused Bt toxin production in kernels of non-Bt corn refuge plants.

Methods
Six pairs of near-isogenic Bt (4 Bt11 insertion event hybrids, and 2 Mon810 event hybrids) and non-Bt corn hybrids were planted in adjacent plots in the field, with each non-Bt hybrid planted in 36 rows downwind of 8 rows of its Bt counterpart (0.965 m row-spacing). Each commercially available Bt hybrid was produced by crossing a Bt line with a conventional line and backcrossing the resulting Cry1Ab-producing plants with the conventional line for several generations. Each backcrossed Bt line was crossed with a second conventional line to produce hybrid seed for planting that was hemizygous for the cry1Ab gene. Plants grown from this hybrid seed should carry the cry1Ab gene in half of their eggs and pollen. Thus, Bt plants fertilized with Bt pollen are expected to have the cry1Ab gene in 75% of their seeds and pollen (50% hemizygous and 25% homozygous). Ears were harvested from two rows of each Bt subplot and rows 1–4, 8, 16, 24, and 32 of each non-Bt subplot (0.96 to 31 m from the Bt subplot). From each sample, 30 randomly selected kernels were set aside and the remaining kernels ground into a powder.

We used enzyme-linked immunosorbent assay (ELISA) to test for Cry1Ab. To estimate the mean concentration of Cry1Ab in ears of each subplot row, we tested 2 g of each ground kernel sample with ELISA. To estimate the percentage of kernels containing Cry1Ab, 30 seeds from each sample were germinated and leaf punches from each seedling tested for Cry1Ab with ELISA by Mid-West Seed Services (Brookings, South Dakota, USA).

Results
Gene flow from Bt corn to non-Bt corn occurred at up to 31 m, the greatest distance we examined. ELISA tests of non-Bt corn showed that the concentration of Bt toxin Cry1Ab and the percentage of kernels with Cry1Ab decreased with distance from Bt corn (Figs. 1 & 2).

Fig. 1. Mean (± S.E.) concentration of Bt toxin Cry1Ab in kernels of Bt and non-Bt corn ears as a function of distance between Bt and non-Bt plots. Row distance = 0.97 m. Rows -1 and -3 were Bt corn approximately 2 and 4 m from the non-Bt corn, whereas, rows 1 to 32 were non-Bt corn approximately 1 to 32 m from Bt corn.

The mean concentration of Cry1Ab in kernels of Bt hybrids was 260 ± 19 ng/g (Fig. 1). The concentration of Bt toxin in non-Bt corn was highest within 2 m of Bt corn, decreasing rapidly with distance. The maximum Bt toxin concentration in kernels of non-Bt corn was 140 ng/g in refuge plants at about 1 m distance from Bt plants. This maximum toxin concentration in non-Bt corn was 45% of the mean concentration in adjacent Bt plants. At 31 m from Bt corn, the mean concentration of Cry1Ab in kernels of non-Bt corn was 4.4 ± 1.2 ng/g, which is 1.7% of the concentration in the kernels of adjacent Bt plants.

Fig. 2. Mean (± S.E.) percentage of Cry1Ab corn kernels in Bt and non-Bt corn ears as a function of distance between Bt and non-Bt plots. Rows labeled as in Fig. 1.

In Bt corn, the mean percentage of kernels with Cry1Ab was 74.5%, which is consistent with the 75% expected from hemizygous parents (Fig. 2). The maximum percentage of kernels of non-Bt corn with Cry1Ab was 60%, which occurred at 1 m from Bt corn. In proportion to the mean of 75% Bt kernels for the ears of adjacent Bt corn, the maximum of 60% Bt kernels in non-Bt corn was 0.8 (60% divided by 75%).

Discussion
The movement of Bt genes into non-Bt refuge plants has important implications for resistance management. If some target pests are killed by Bt toxins in refuge plants near Bt corn, refuges will produce fewer susceptible insects than expected. If so, the value of refuges for delaying resistance will be diminished. Further, a mosaic of kernels with and without Bt toxin in various proportions could cause some pests to eat a mixture of Bt kernels and non-Bt kernels in the same ear. This in turn might expose pests to a wide range of doses, rather than the uniform high dose envisioned in the ideal application of the high-dose/refuge strategy. Exposure to intermediate doses might favor survival of hybrid pests produced by matings between resistant and susceptible parents, which would undermine the strategy. Even in uniform plantings of Bt corn, toxin levels vary among leaves, silks, shanks, and kernels of a single plant, as well as among plants and through time. This variation calls into question the assumption that pests always receive a high toxin dose when eating Bt plants. Nonetheless, the additional variation in toxin concentration introduced by gene flow from Bt corn to nearby non-Bt corn could reduce the effectiveness of the refuge strategy.

The extent of Bt gene flow into refuges depends on many factors, including refuge size, shape, distance from the Bt crop, wind speed and direction, and similarity in maturation times between Bt and non-Bt hybrids. Thus, various physical, ecological, and molecular methods of gene containment could reduce contamination of refuge plants by gene flow from Bt crops. In our experiment, non-Bt corn was planted downwind from Bt corn. Planting non-Bt corn upwind of Bt corn would reduce gene flow into refuges, but increased gene flow in the opposite direction could produce Bt corn with intermediate toxin concentrations. This type of gene flow appears to be a less serious concern because vegetative tissues of Bt corn as well as some portions of all the kernels such as the seed coat would produce Bt toxin regardless of such gene flow.

Although movement of Bt genes into refuges has received little attention previously, the potential for movement of corn transgenes has been recognized in other contexts. In particular, the U.S. EPA mandates that experimental Bt corn hybrids must be planted more than 200 m from other corn to limit gene flow. Paradoxically, EPA guidelines for Bt corn resistance management suggest that non-Bt corn refuges in Bt corn fields can be as narrow as 4 m and that they must be planted within 800 m of Bt corn, preferably 400 m, to increase mating between insects from Bt corn and refuges. Revised guidelines should consider the effects of gene flow between Bt and non-Bt plants as well as gene flow between resistant and susceptible insects. In particular, the minimum width of non-Bt corn refuges should be increased to at least 30 m.

References
1. Chilcutt CF & Tabashnik BE. (2004) Contamination of refuges by Bacillus thuringiensis toxin genes from transgenic corn. Proceedings of the National Academy of Sciences USA 101: 7526-7529.

2. Gould F. (1998) Sustainability of transgenic insecticidal cultivars: integrating pest genetics and ecology. Annual Review of Entomology 43: 701-726.

3. Shelton, AM, Nyrop JP, Seaman A & Foster RE. (1982) Distribution of European corn borer (Lepidoptera: Pyralidae) egg masses and larvae on sweet corn in New York. Environmental Entomology 15: 501-506.

4. Tabashnik BE. (1994). Evolution of resistance to Bacillus thuringiensis. Annual Review of Entomology 39, 47-79.

Charles F. Chilcutt
Texas A&M Univ. Ag. Research & Extension Center
Department of Entomology
c-chilcutt@tamu.edu

Bruce E. Tabashnik
Department of Entomology
University of Arizona
brucet@ag.arizona.edu