The trouble with aposematism, though, is that it requires giving up another, more common defensive color scheme: camouflage. If you're a poisonous critter, and you evolve bright coloration for the first time, predators don't yet know that you're poisonous - but you're really brightly colored and easy to see. How, then, does aposematism evolve from non-aposematic ancestors?
A new study on early release from Biology Letters suggests that it isn't easy. The authors, Noonan and Comeault, set out to determine whether brightly-colored poison dart frogs are more likely to be attacked when they evolve new color patterns [$-a]. It's possible that the frogs' predators avoid all brightly-colored prey regardless of pattern, in which case new frog patterns would be just as good for predator deterrence as the old ones. But it's also possible that predators only avoid patterns they've run across (and spat out) before - so that new, rare patterns would have all the disadvantages of giving up camouflage with none of the benefits of aposematism.
Photo by dbarronoss.
Photo by dbarronoss.
Noonan and Comeault performed an elegant behavioral experiment, setting out clay model frogs in an area where frogs of one color pattern predominate. One set of models matched the local color pattern, another was brightly colored but different from the local pattern, and a third was drab and camouflaged. Birds were much more likely to attack the "new" color pattern than either the "local" version or the drab one. This result is hard to understand at the first pass - if new color patterns are vulnerable to attack, how can aposematism evolve in the first place? The answer is, not by natural selection, but by genetic drift.
Genetic drift is a natural, mathematical consequence of finite populations: imagine a bag full of marbles, half of them black and half white. If you pull a sample of marbles from the bag, you expect them to be half black and half white on average (i.e., over many samples) - but any individual sample might have a very different frequency of white and black marbles, especially if it's small. If the probability of picking a white marble from the bag is 0.5 (because half the marbles are white), then the probability of picking a sample of four white marbles is 0.5 × 0.5 × 0.5 × 0.5 = 0.0625. That's a small probability, but not zero. Drift is a very real effect in the natural world, especially during the establishment of new local populations, when the population size is initially quite small.
The key to understanding Noonan and Comeault's result is that aposematism is frequency dependent - it favors not the old pattern as such, but whatever bright color pattern is most common in the frog population. Birds attacked the "local" color pattern at a low rate, which suggests that they're always re-learning which pattern to avoid. A new color pattern might be hard to establish within a population of frogs that look very different from it, but if a new pattern pops up in the course of establishing a new population, then - thanks to genetic drift - it may be common enough for predators to learn to avoid it.
B.P. Noonan, A.A. Comeault (2008). The role of predator selection on polymorphic aposematic poison frogs. Biology Letters DOI: 10.1098/rsbl.2008.0586