Introducing genes for herbicide resistance into a crop permits it to be sprayed with weedkiller that really does then kill nothing but weeds. But that works only until the weeds themselves develop resistance to the poison. One way this can happen is through crossbreeding with the crop originally protected—a risk if weed and crop are closely related.

That is the case for rice, where weedy, natural varieties are a perennial problem because of the competition they bring to the cultivars farmers actually want to raise. But, as he describes in Transgenic Research, Lu Baorong, an ecologist at Fudan University in Shanghai, thinks he has found a solution. By adding a second transgene to the crop, he can sabotage any weed that crossbreeds with it. Dr Lu’s transgene encodes a genetic “silencer” that shuts down the expression of a natural gene called SH4. In wild grasses SH4 promotes a phenomenon called “seed shattering” that releases seeds from the stalk when they are ripe.

Domestication selects against seed shattering because farmers want the seeds to stay attached to a plant as it is harvested. The best cereal crops are those which do not release their seeds until they are deliberately threshed. That means adding an SH4 silencer to them will, if anything, make them better crops rather than worse ones. Indeed, experiment shows that the silencer has no effect on the productivity of an otherwise genetically un-engineered cultivar, as measured by such things as the number of seed grains per plant, the weight of those grains, the percentage of them that germinate, and the survival rates of the resulting seedlings.

If a silencer-enhanced version does crossbreed with a weedy interloper, though, the offspring will end up carrying the silencer, too. And that, Dr Lu hoped, would damage them by stopping their seeds breaking off naturally, and thus preventing those seeds from spreading.

To test his idea, Dr Lu and his colleagues crossbred a weedy rice strain with a cultivar into which the silencer transgene had been introduced. They then allowed the crossbred offspring to breed with one another, creating second-generation hybrids of a sort that might emerge in the wild. They found that the expression of SH4 in those hybrids dropped sharply, to as low as 10%. That is a level similar to the one found in cultivars. This reduction in SH4 expression was accompanied by a reduction in the hybrids’ seed-shattering index, a measure of the strength of a plant strain’s stalks and the number of its seeds in the soil. In a rice-field, the consequence would be that the weedy grains get harvested along with those of the cultivar, removing them from circulation and thus suppressing the weedy population the following season.

In the long term, that might make herbicides obsolete. In the shorter term, however, Dr Lu hopes to make them more effective, by creating a cultivar in which silencers of SH4 and, perhaps, other seed-shattering genes are in close chromosomal proximity to the herbicide -resistance gene. That will mean any transfer of herbicide resistance automatically brings seed-shattering problems with it, stopping the spread of herbicide resistance within the weedy population.

Moreover, what works with rice might reasonably be expected to work, too, with other cereals, such as wheat and sorghum, which also have close relatives that behave as weeds. Dr Lu’s subtle approach of, in effect, domesticating weeds in order to destroy them, could therefore have a big influence on future crop yields.