Buzz of the chase

A former US detective and an evolutionary biologist have got a handful of bees to help the police with inquiries into serial crimes. Geoff Watts joins the investigation.

There is a faint apian hum in the laboratory where Steven Le Comber is peering into one of several shallow oblong wooden crates spread across most of its work surfaces. Each crate is about 1.5m long and covered in a sheet of Perspex. They enclose a handful of bees that are, unwittingly, assisting the police with their inquiries.

"This is actually a foraging arena," Le Comber explains. "It has artificial flowers with a sucrose solution for the bees to feed on. If you look closely, you can see that the bees are individually marked with numbers, so we can record which flowers they choose."

If this leads you to think that Le Comber, an evolutionary biologist at Queen Mary, University of London, must be a bee expert of some kind, think again. And Kim Rossmo, his principal collaborator on this and on previous work featuring bats, works in Texas State University's department of criminal justice.

The intention of this unlikely academic pairing is to take a forensic technique designed for catching criminals, apply it to animals and then feed back the experimental findings to fine-tune its use in humans.

The technique is called "geographic profiling". It is a second cousin to the better known "psychological profiling". But while this uses whatever is known about a series of linked crimes to define the sort of person who might have committed them, geographic profiling uses crime-site locations to narrow the search for the criminal's base - the "where" as opposed to the "who". In the bee experiments, the "crime site" is a flower.

Rossmo devised the technique during his previous career as a police officer, and Le Comber first heard of it a few years ago through an article in New Scientist . The theory behind geographic profiling is that anyone planning where to commit a criminal act is subject to two competing pressures. One is the "distance-decay" effect, Le Comber says. "This just means that for all sorts of reasons - time, money, effort and so on - criminals tend to commit crimes closer to home rather than farther away.

Seventy per cent of arson is committed within two miles of the arsonist's home. The second pressure is the buffer zone. Criminals also tend not to operate in the area immediately around their home, not least for fear of recognition."

Some years ago, Rossmo expressed these opposing tendencies as a mathematical equation and wrote some software incorporating it. Given a set of crime scene locations, geographic profiling can then prioritise the search for the criminal's base. Clearly, it cannot pinpoint the location, but it can suggest which areas are more likely candidates than others. Although suitable only for serial crime, it is now used by police forces around the world.

At the time he read about geographic profiling, Le Comber had been studying the mating habits of the three-spined stickleback and the vicious competition between males. He had noticed that sticklebacks foraging for material to build nests seemed to follow a pattern: on the one hand, they did not want to roam too far and leave the nest unguarded; on the other, they dared not denude the area around the nest, so drawing attention to it.

Although Rossmo had originally considered his technique relevant only to violent crime, he had already become aware that its use was extending to serial theft such as credit card fraud. So when he was contacted by Le Comber, he was intrigued at the prospect of a further application in such a novel direction but not entirely surprised. "People say that humans are nothing more than sophisticated animals," he reflects.

The two academics met in London and agreed to collaborate, though not on sticklebacks. At a conference in Poland, Le Comber had discovered that two Aberdeen University biologists, Barry Nicholls and Paul Racey, had some radio-tracking data on the foraging patterns of bats that would be suitable for a geographic profiling analysis. "It worked surprisingly well," Le Comber says. And because they already knew where the bats roosted (the equivalent, in profiling terms, of a criminal's home base) they were able to test Rossmo's model and fine-tune it.

Their current work was prompted by one of Le Comber's departmental colleagues, Nigel Raine, who works on foraging in bees. The density, number and location of the artificial flowers in his foraging arenas can be altered at will. "For instance, we could have twice as many artificial flowers," says Le Comber, "or cluster them in one part of the arena, or have flowers of different reward value". For "flower" read "potential crime site", and you can see how the set-up offers ways of performing experiments that would be impossible with human criminals.

Some critics will question whether modelling human behaviour on an animal system can teach us anything about ourselves. But you could argue equally that how humans behave has no bearing on animals. The evidence of the bats and the bees has already disproved this; there is reason to believe that the new knowledge works in both directions.

Le Comber is also working with a colleague in Africa who wants to use the location of animal snares to catch poachers. Rossmo talks of the military interest in geographic profiling to combat insurgency. Both are contemplating the technique as a way of pinpointing animal reservoirs from which humans become infected by malaria-carrying mosquitoes.

Talking to Le Comber, it becomes clear that he and Rossmo differ not only in their backgrounds but in their principal aims. "I'm interested in the model on purely theoretical grounds," Le Comber says. "I want to poke the equation and see what happens. Of course, if the work produces something useful - excellent. I'll be thrilled."

For Rossmo, the former detective, it is the possibility of improving the technique as a practical tool in police work that is the main driver.

Either way, so long as the bats, bees and who knows what else serve their respective purposes, the collaboration should remain a profitable one.