A strange tale of collaboration between plants, pathogens and insects .

CUCUMBER mosaic virus is not restricted to its eponymous host. It can also ravage tomatoes—stunting them and causing them to produce contorted tendril-like leaves. Given this devastation, it is surprising susceptible plants continue to exist; natural selection should have produced resistance years ago.

A paper in this week’s PLoS Pathogens, however, explains the apparent contradiction.

The team that wrote it, led by John Carr of Cambridge University, found that the virus actually helps its host to reproduce. It does so using an unsuspecting accomplice: the bumble-bee.

Like many discoveries, this one was accidental.

Dr Carr had ordered some equipment to analyse the volatile chemicals emitted by infected tomato plants. While he was waiting for it to arrive, a colleague offered to lend him some bees, as these pollinators are known for their sense of smell. His team placed both healthy and infected plants in a greenhouse, covered them so that the bees could smell but not see them, and released the insects.

The bees could indeed tell the difference—but to Dr Carr’s surprise they did not favour the healthy plants. Rather, they strongly preferred the infected ones. For some viral strains, they were four times more likely to visit an infected plant than a healthy one. Subsequent experiments, carried out after the analytical apparatus had arrived, showed that this difference in behaviour was, indeed, a response to differences in the chemicals that infected and uninfected plants give off.

The question was how the bees’ preference played into the plants’ evolution.

Tomatoes are able to fertilise themselves, without need of a pollinator, yet even this self-pollination can be assisted by a bee visiting a flower. Vibrations caused by the insect’s buzzing help release pollen from the flower’s anthers so that it can fall onto the stigma. Could this process, Dr Carr wondered, make up for the evolutionary disadvantage the virus otherwise inflicts?

To investigate, he and his colleagues grew both healthy and infected tomatoes in a greenhouse with no bees, and a similar mixture in a greenhouse into which bees were released one at a time, and their movements tracked.

They then analysed the seeds of the fruits of each plant as a proxy for evolutionary fitness. All of the seeds they looked at seemed viable.

What differed was the number of seeds in each fruit. In the bee-free greenhouse, infected plants had only about 50 seeds per fruit, whereas healthy ones averaged 70.

When bees were present, though, something strange happened.

The seed count of fruits from healthy flowers visited by bees increased to 85—but that of the flowers of infected plants jumped much higher, to 115.

Further investigation showed a possible reason. Not only did bees visit the infected plants more, but they also spent longer buzzing around infected flowers. The extra pollen thus displaced probably explains how the sickly plants outproduced their healthy confrères.

Using these experimental data, the researchers modelled how the system might have evolved. They found that the pollinator’s preference for the smell of infected plants was big enough to stop the emergence of viral-resistance in tomatoes. That explains what is in it for the tomatoes and also for the virus.

What the bees get out of it, though, remains mysterious.