Fishy reputations: undersea watchers choose helpers that do good jobs.

Coral reef fish do consumer research in a way that keeps a service-providing species honest, says a new study.

When parasites build up on the skin of a reef fish, it can get relief by swimming over to a small, so-called cleaner fish that nibbles off the encrustations. Now, lab tests show that potential clients choose cleaners that they've seen working attentively, says Redouan Bshary of the University of Neuchatel in Switzerland.

Other lab evidence suggests that the cleaners are on their best behavior when there's an audience of potential clients, Bshary and Alexandra S. Grutter of the University of Queensland in St. Lucia, Australia report in the June 22 Nature. In the team's previous observations, fish in the wild were less likely to bite one client when potential clients lingered nearby.

"It's much more complicated than simple reciprocity," says behavioral ecologist Lee Alan Dugatkin of the University of Louisville in Kentucky, who has studied altruism. The cleaner-client system appears to be a case of indirect reciprocity, where a fish works attentively and holds back on taking a tasty nibble of the client's protective mucus coating, which would incite a jolt and perhaps a chase. By so doing, it wins an opportunity to clean a future client--the fish that watched.

The idea that animals show generous behavior and reap benefits from those that observe it "has been floating around," says Dugatkin. "This is the strongest evidence so far that it's really happening in nature."

Bshary and Grutter put a possible client, the brindled monocle bream Scolopsis bilineatus, in the middle compartment of an aquarium, where it could watch two cleaner fish, Labroides dimidiatus, in the end compartments. The researchers provided clients to each of the cleaners.

One cleaner, which got a client smeared with delectable prawns, worked diligently picking off the treats. The other cleaner, given a familiar, already clean client, largely ignored the visitor. In 28 tests, the observer fish in the center compartment tended to hover nearer the compartment of the harder-working cleaner.

In that experiment, the cleaners couldn't see the observer. To see what might happen when the cleaners had an audience, Bshary and Grutter used pairs of food plates as stand-ins for client fish.

The researchers stocked two plates with both ho-hum fish-food flakes and delectable bits of prawn. In one scenario, as long as the cleaner nibbled the flakes, the researchers left the second mixed plate as if it were a waiting client. If the cleaner fish gulped a prawn, though, the researchers snatched away the second plate.

That setup mimics real life, says Bshary, because a cleaner has to forgo nibbling on a client's mucus coating if it's going to avoid scaring away its next meal.

When researchers trained the cleaner fish with plate-snatching rules, the fish ate more of the uninspiring flakes when the second plate was nearby than when there was only one plate. When the two plates were available, the trained fish also ate more flakes than a fish that hadn't been taught the rules.

"What makes this unique is that this study brings together audience effects and image-scoring in one system," says Dugatkin.