The theory of mimicry explains how a mimic species gains advantage by resembling a model species [1-3]. Selection for increased mimic-model similarity should then result in accurate mimicry, yet there are many surprising examples of poor mimicry in the natural world [4-8]. The existence of imperfect mimics remains a major unsolved conundrum. We propose and experimentally test a novel explanation of the phenomenon. We argue that predators perceive prey as having several traits, but that the traits differ in their importance for learning. When predators learn to discriminate prey, high-salience traits overshadow other traits, leaving them under little or no selection for similarity, and allow imperfect mimicry to succeed. We tested this idea experimentally, using blue tits as predators and artificial prey with three prominent traits: color, pattern, and shape. We found that otherwise imperfect color mimics were avoided about as much as perfect mimics, whereas pattern and shape mimics did not gain from their similarity to the model. All traits could separately be perceived and learned by the predators, but the color trait was learned at a higher rate, implying that it had higher salience. We conclude that difference in salience between components of prey appearance is of major importance in explaining imperfect mimicry.

Mimicry occurs when one species gains protection from predators by resembling an unprofitable model species. The degree of mimic-model similarity is variable in nature and is closely related to the number of traits that the mimic shares with its model. Here, we experimentally test the hypothesis that the relative salience of traits, as perceived by a predator, is an important determinant of the degree of mimic-model similarity required for successful mimicry. We manipulated the relative salience of the traits of a two-trait artificial model prey, and subsequently tested the survival of mimics of the different traits. The unrewarded model prey had two colour traits, black and blue, and the rewarded prey had two combinations of green, brown and grey shades. Blue tits were used as predators. We found that the birds perceived the black and blue traits similarly salient in one treatment, and mimic-model similarity in both traits was then required for high mimic success. In a second treatment, the blue trait was the most salient trait, and mimic-model similarity in this trait alone achieved high success. Our results thus support the idea that similar salience of model traits can explain the occurrence of multi-trait mimicry.

Batesian mimicry evolution involves an initial major mutation that produces a rough resemblance to the model, followed by smaller improving changes. To examine the learning psychology of this process, we applied established ideas about mimicry in Papilio polyxenes asterius of the model Battus philenor. We performed experiments with wild birds as predators and butterfly wings as semiartificial prey. Wings of hybrids of P. p. asterius and Papilio machaon were used to approximate the first mutant, with melanism as the hypothesized first mimetic trait. Based on previous results about learning psychology and imperfect mimicry, we predicted that: melanism should have high salience (i.e., being noticeable and prominent), meaning that predators readily discriminate a melanistic mutant from appearances similar to P. machaon; the difference between the first mutant and the model should have intermediate salience to allow further improvement of mimicry; and the final difference in appearance between P. p. asterius and B. philenor should have very low salience, causing improvement to level off. Our results supported both the traditional hypothesis and all our predictions about relative salience. We conclude that there is good agreement between long-held ideas about how Batesian mimicry evolves and recent insights from learning psychology about the role of salience in mimicry evolution.

The importance of receiver biases in shaping the evolution of many signalling systems is widely acknowledged. Here, we show that receiver bias can explain which traits evolve to become warning signals. For warning coloration, a generalization bias for a signalling trait can result from predators learning to discriminate unprofitable from profitable prey. However, because the colour patterns of prey are complex traits with multiple components, it is crucial to understand which of the many aspects of prey appearance evolve into signals. We provide experimental evidence that the more salient differences in prey traits give rise to greater generalization bias, corresponding to stronger selection towards trait exaggeration. Our results are based on experiments with domestic chickens as predators in a Skinner-box-like setting, and imply that the difference in appearance between profitable and unprofitable prey that is most rapidly learnt produces the greatest generalization bias. As a consequence, certain salient traits of unprofitable prey are selected towards exaggeration to even higher salience, driving the evolution of warning coloration. This general idea may also help to explain the evolution of many other striking signalling traits found in nature.