Predators' decisions to eat defended prey depend on the size of undefended prey.

Predators that have learned to associate warning coloration with toxicity often continue to include aposematic prey in their diet in order to gain the nutrients and energy that they contain. As body size is widely reported to correlate with energetic content, we predicted that prey size would affect predators' decisions to eat aposematic prey. We used a well-established system of wild-caught European starlings, , foraging on mealworms, , to test how the size of undefended (water-injected) and defended (quinine-injected) prey, on different coloured backgrounds, affected birds' decisions to eat defended prey. We found that birds ate fewer defended prey, and less quinine, when undefended prey were large compared with when they were small, but that the size of the defended prey had no effect on the numbers eaten. Consequently, we found no evidence that the mass of the defended prey or the overall mass of prey ingested affected the amount of toxin that a predator was willing to ingest, and instead the mass of undefended prey eaten was more important. This is a surprising finding, challenging the assumptions of state-dependent models of aposematism and mimicry, and highlighting the need to understand better the mechanisms of predator decision making. In addition, the birds did not learn to discriminate visually between defended and undefended prey based on size, but only on the basis of colour. This suggests that colour signals may be more salient to predators than size differences, allowing Batesian mimics to benefit from aposematic models even when they differ in size.