The green genes don't get out much.

Could genes from a genetically modified crop escape and create a superweed that could take over the world, as some people fear? One preventive measure might be to confine any transplanted genes to the cell's photosynthetic structures, or chloroplasts.

Escape of such genes would be "extremely rare and scattered," predict Susan E. Scott and Mike J. Wilkinson of the University of Reading in England, who for 3 years have tracked chloroplast DNA in fields of unmodified oilseed rape. They report their results in the April Nature Biotechnology.

The work grows out of proposals that tinkering with the DNA in chloroplasts poses less risk of runaway genes than the more common strategy of modifying DNA residing in the cell's nucleus. Henry Daniell of the University of Central Florida in Orlando, a pioneer of gene insertion in tobacco chloroplasts, last year advocated that approach. In many plants, he argues, chloroplasts are inherited maternally and thus don't show up in hard-to control, wind-blown, insect-riding pollen.

Wilkinson and Scott evaluated chloroplast-gene escape routes in Brassica napus, known as oilseed rape or canola. Farmers grow this mustard for its edible oil. The crop species hybridizes readily with wild mustards, including one of its ancestors, Brassica rapa.

To track any wandering of chloroplast genes, the researchers checked 47 crop-weed hybrids found near commercial fields. All hybrids showed the weed, or maternal, chloroplast DNA. That convinced Wilkinson that pollen wafting from fields does not carry chloroplast DNA.

In practical terms, errant crop pollen doesn't have that many places to go, Wilkinson notes. He and Scott found that of more than 140 patches of wild B. rapa in farmland, only 2 grew near oilseed rape fields.

In such patches, just 0.4 to 1.5 percent of the plants have mixed parentage, he and Scott reported last year. All in all, there will probably be "no or negligible" escape of chloroplast genes through crop pollen, he predicts.

The researchers also considered the other escape avenue: the female flower parts. If crop seeds spill near wild plants, the resulting plants may be pollinated by weed species to create hybrids carrying the modified chloroplast. In another generation or two, the wayward genes could get into highly fertile wild plants.

However, when a crop plant gets loose, "it doesn't last very long," Wilkinson says. He and Scott monitored 18 patches harboring crop plants that had gone wild. Fifteen of the patches disappeared or failed to set seed during the 3-year study.

Hybridization is "inevitable but will occur only extremely rarely," Wilkinson says. "It all comes down to what the transgene actually is." A transplanted gene that gives a plant whopping advantages in the wild might spread even through a tiny keyhole escape avenue.

"You have to look at each plant on a plant-by-plant basis," agrees Dean Chamberlain of the University of North Carolina (UNC) in Greensboro. In a commentary in the same issue of NATURE BIOTECHNOLOGY, he and UNC's C. Neal Stewart Jr. note that chloroplast genes are so difficult to work with that, to date, only tobacco has been transformed in this way.

Just wait, responds Daniell. He expects several researchers soon to announce transfers of chloroplast genes.

The chloroplast strategy is still no panacea, warns Joseph E. Cummins of the University of Western Ontario in London, Ontario. Chloroplasts are passed on through pollen in many conifers and through both parental lines in alfalfa. Also, Cummins points out that chloroplast DNA can leak into mitochondria, a cell structure that does show up in oilseed rape pollen.

Wilkinson speculates that both fans and foes of genetically modified crops will quote the new paper as supporting evidence. "To us, it's just data," he sighs.