Agricultural biotechnology seems to attract a lot of attention these days. Unfortunately, most of it is the wrong kind.

The genetic engineering of plants will almost certainly be needed to feed a human population that will stabilize at 10 billion or more sometime during the next century. Furthermore, it’s the first arena where the techniques of genetic manipulation, developed during the 1970s, are being applied on a large scale. Therefore, the attention is justified, since we need to be mindful of what we as a species are doing to the other species on the planet when we modify the genes of our food crops.

The problem is that the debate on this topic seems to have been hijacked by extremists on both sides: environmentalists crying disaster and corporate spokespersons telling us soothingly that there’s nothing to worry about-that there’s nothing sinister or even unusual about genetic engineering, which, they argue, simply extends methods farmers have used since the beginning of agriculture to improve crops and livestock.

Both extremes are wrong. Certainly global disaster isn’t just around the corner. On the other hand, there is something novel and extraordinary about our ability to insert the genes of one species among the genes of another. And, as with the introduction of any remarkable new technology, unforeseen consequences are inevitable.

One possible consequence is addressed in detail by Charles C. Mann in this issue’s cover story, “Biotech Goes Wild.” Mann, one of the nation’s very best science and technology journalists, reveals that crops genetically engineered to produce or resist pesticides could interbreed with wild plants, giving rise to “superweeds” that could be very hard to eradicate and might choke food crops. Many biologists have dismissed the threat of such phyto-promiscuity, but some scattered data are unsettling. And the fact is that science can’t yet tell us whether to be alarmed.

Why are we still in ignorance on this important issue? One reason is that the biologists who engineer genes into plants aren’t trained to think about the whole organism and its habitat; they’re molecular folk, lacking an ecological perspective.

Another reason is the patchwork of U.S. regulatory agencies that govern agricultural biotech: the Food and Drug Administration (FDA), the Environmental Protection Agency (EPA), and the Department of Agriculture (USDA). Mann writes that none of these agencies is responsible for superweeds, which could grow between the bureaucratic cracks. The FDA doesn’t look at plants engineered to express pesticides, because pesticides are exempt from the agency’s reach. The EPA is required to treat such foods as pesticides and simply establish human tolerances for each compound. USDA tries to make sure the crop grows as the producer says it will.

There’s only one thing omitted from this patchwork: the public interest. To stitch it back in two things are needed. One is additional research, from biotech companies and the federal government on the ecological consequences of genetic engineering. The other is a revision of the regulatory framework to make at least one of the agencies responsible for testing the environmental consequences of agribiotech. These changes would acknowledge the growing impact of this remarkable new technology and help us steer a course between wrongheaded extremes.