he team gave predators the same options at each site. They tied one of each replica—ringed, striped, and plain brown—to a nail. The nail went into something solid, such as a log, so predators wouldn't make off with the replicas.

The team expanded their operation to Arizona, where the Sonoran mountain kingsnake mimics the western coral snake. And Harcombe made a whole new batch of snakes—this time with some friends he enlisted as last-minute help.

After Pfennig and Harcombe collected the replicas and brought them back to Chapel Hill, the attacks were "scored" by Karin Pfennig, who had no knowledge of where each replica had been placed. "That way we wouldn't bias our results with any preconceived notions of what we expected to get," David Pfennig says.

The idea was to look at the total number of attacks, then determine what proportion of those was on the kingsnake replicas. "If this is truly mimicry, the proportion of attacks on our kingsnake replicas should be lower in areas where there are coral snakes than in areas where there are no coral snakes," Pfennig says.

But before they dived into the numbers, the team developed a null hypothesis, a kind of simplest plausible explanation. In this case, it meant defining what the results would look like if attacks were totally random. "If predators were just attacking these things willy-nilly, at random, and were not paying any attention to pattern, then we would expect that one third of the total number of attacks would have been on the ringed kingsnake replicas," Pfennig says. "So values above that one third indicate that predators are actually preferring to attack the ringed replicas. Values below that one third are indicating that predators are avoiding the ringed ones."

And how did the values add up?

In areas where coral snakes live, fewer than one third of the kingsnakes were attacked. A lot fewer, in fact—only about 13 percent of the total attacks in Arizona were on kingsnakes.

But outside the coral snake's range, "the kingsnakes got nailed by predators," as Harcombe puts it. In Arizona, the proportion of kingsnakes attacked was about 50 percent; in North Carolina, the proportion was 65 percent—much higher figures than the team would expect to see if predators were simply attacking the models randomly.

n Arizona predators behaved differently across a remarkably local scale. The western coral snake doesn't live in the higher elevations of Southeast Arizona's Chiricahua Mountains. But the Sonoran kingsnake, which mimics the western coral snake, lives in the Chiricahuas' higher and lower elevations. So the team sampled sites going up the mountain. "These sites are no farther apart than about three miles," Pfennig says. "And yet we're seeing significant differences in predator behavior, and it's even more fine grained than what we were seeing in the Carolinas."

Other researchers have shown that, from birth, predators avoid coral snake patterns without prior exposure to the snakes. "People have often pointed out that it isn't effective to have to learn this avoidance behavior," Pfennig says, "because if a predator makes a mistake, it's dead. There's no trial and error. So there's some evidence that this is probably genetically encoded. What's remarkable to me is that predators born just maybe eighty kilometers south of us may be born with different genes than predators born here."

And that keeps mimicry interesting. Consider the different and seemingly incompatible needs of the coral snake, the scarlet kingsnake, and their predators. The kingsnake needs to resemble the coral snake for protection. But the coral snake needs his color scheme to remain an effective warning—which is tricky if a bunch of harmless snakes look just like him. "I'm anthropomorphizing here," Pfennig says, "but if predators start to learn ‘Hey, most of these ringed things aren't venomous—I ate one, and I didn't die,' then they might attack both kingsnakes and coral snakes."

The predator simply needs to know whether he's looking at lunch or the last snake he'll ever see. "Predators will be under selection to develop greater eyesight, greater cognitive abilities, or whatever it takes to tell the difference," Pfennig says.

And since the predator exerts all the selection on the system, mimicry is effective even if the coral snake and the scarlet kingsnake never actually encounter each other. That's why biologists were excited by Bates' idea in 1862 and remain so today.

"Mimicry adds an interesting dimension to evolution because an individual's predators, prey, parasites, and hosts can evolve alongside the individual," Pfennig says. "So it becomes a three-way game in which each player must constantly out-evolve the others. These survival strategies create the ever-improving traits that we see throughout nature."

David Pfenning was a winner of the 2000 Hettleman Prize for Artistic and Scholarly Achievement.