Pollen for our assay was collected from N4640-Bt corn and an unrelated, untransformed hybrid, and was applied by gently tapping a spatula of pollen over milkweed (Asclepias curassavica ) leaves that had been lightly misted with water. Pollen density was set to visually match densities on milkweed leaves collected from corn fields. Petioles of individual leaves were placed in water-filled tubes that were taped into plastic boxes. Five three-day-old monarch larvae from our captive colony were placed on each leaf, and each treatment was replicated five times. Milkweed leaf consumption, monarch larval survival and final larval weight were recorded over four days.

Larval survival (56%) after four days of feeding on leaves dusted with Bt pollen was significantly lower than survival either on leaves dusted with untransformed pollen or on control leaves with no pollen (both 100%, P =0.008) (Fig. 1a). Because there was no mortality on leaves dusted with untransformed pollen, all of the mortality on leaves dusted with Bt pollen seems to be due to the effects of the Bt toxin.

Figure 1: Survival and leaf consumption of second- to third-instar monarch larvae on each of three milkweed leaf treatments: leaves with no pollen (light blue), leaves treated with untransformed corn pollen (green) and leaves dusted with pollen from Bt corn (dark blue).
figure 1

a, Mean ( ± s.e.m.) survival based on the proportion of larvae surviving in five replicates of each treatment. b, Mean ( ± s.e.m.) cumulative leaf consumption based on the total amount of leaf area consumed per larva in five replicates of each treatment. The amount of leaf area consumed per larva in each experimental unit was calculated for each time interval by dividing the amount of leaf area consumed in that interval by the number of larvae alive during the time interval. Cumulative consumption was calculated by summing the leaf area consumed per larva at each interval. Colours of lines correspond to those of the bars in a.

There was a significant effect of corn pollen on monarch feeding behaviour (P =0.0001) (Fig. 1b). The mean cumulative proportion of leaves consumed per larva was significantly lower on leaves dusted with Bt pollen (0.57 ± 0.14, P =0.001) and on leaves dusted with untransformed pollen (1.12 ± 0.09, P =0.007) compared with consumption on control leaves without pollen (1.61 ± 0.09). The reduced rates of larval feeding on pollen-dusted leaves might represent a gustatory response of this highly specific herbivore to the presence of a ‘non-host’ stimulus. However, such a putative feeding deterrence alone could not explain the nearly twofold decrease in consumption rate on leaves with Bt pollen compared with leaves with untransformed pollen (P =0.004).

The low consumption rates of larvae fed on leaves with Bt pollen led to slower growth rates: the average weight of larvae that survived to the end of the experiment on Bt -pollen leaves (0.16 ± 0.03 g) was less than half the average final weight of larvae that fed on leaves with no pollen (0.38 ± 0.02 g, P =0.0001).

These results have potentially profound implications for the conservation of monarch butterflies. Monarch larvae feed exclusively on milkweed leaves4; the common milkweed, A. syriaca, is the primary host plant of monarch butterflies in the northern United States and southern Canada5. Milkweed frequently occurs in and around the edges of corn fields, where it is fed on by monarch larvae6. Corn fields shed pollen for 8-10 days between late June and mid-August, which is during the time when monarch larvae are feeding7. Although the northern range of monarchs is vast, 50% of the summer monarch population is concentrated within the midwestern United States, a region referred to as the corn belt because of the intensity of field corn production8. The large land area covered by corn in this region suggests that a substantial portion of available milkweeds may be within range of corn pollen deposition.

With the amount of Bt corn planted in the United States projected to increase markedly over the next few years9, it is imperative that we gather the data necessary to evaluate the risks associated with this new agrotechnology and to compare these risks with those posed by pesticides and other pest-control tactics.