be based on scientific information. The information may come from the existing scientific literature. Depending on the relevance and completeness of the existing literature, agencies may require companies to generate original data to address environmental and food safety questions. USDA and EPA do not appear to be comparably inclined to require original data to support decision-making, and therefore might not always review products with comparable rigor.
At least two published studies have analyzed the use of scientific data by USDA in making regulatory decisions about transgenic crops (Wrubel et al. 1992; Purrington and Bergelson 1995). Both studies conclude that the agency relies heavily on existing scientific literature, rather than requiring that applicants and petitioners develop new experimental data directly relevant to risks that may be posed by individual transgenic plants. Purrington and Bergelson (1995) argue that there are “serious shortcomings in the content of the petitions ” approved by USDA. Another analysis (Mellon and Rissler 1995) concludes that field trials conducted under USDA's oversight produce little information of value to risk assessment when it is time to commercialize transgenic crops.
USDA's approval in 1994 of a petition to deregulate transgenic squash that contained viral coat protein genes illustrates well the agency 's reliance on existing information as the basis of agency determinations. Commercialization of the squash was controversial because some believed that it would probably transfer its acquired virus-resistance genes via pollination to wild squash, which is an agricultural weed in some parts of the southern United States. An analysis commissioned by USDA strongly recommended that new data be gathered for assessment of the risks that may be posed by commercialization of the squash (Wilson 1993), but USDA largely disregarded the recommendation. The agency deregulated the squash, relying almost entirely on existing information to find that commercialization of the squash would have no significant environmental impact (section 3.1.4). As the committee recommended in chapter 3, when published data are insufficient, USDA should require original data to support agency decision-making concerning transgenic crops.
In contrast, EPA generally requires that developers of transgenic pest-protected plants provide more scientific evidence, often including new data, before it makes regulatory decisions. The squash with viral coat proteins cannot be examined for comparison, because EPA was not required to review it before it was commercialized. However, the difference between the agencies' reviews can be illustrated by examining their use of data in their decisions concerning commercialization of Bt cotton discussed above: USDA deregulated the cotton on the basis of existing information about gene flow to wild cotton, and EPA placed geographic restrictions on the planting of Bt cotton until additional data could be provided.