Mimicry in motion and morphology: do information limitation, trade-offs or compensation relax selection for mimetic accuracy?

Many animals mimic dangerous or undesirable prey as a defence from predators. We would expect predators to reliably avoid animals that closely resemble dangerous prey, yet imperfect mimics are common across a wide taxonomic range. There have been many hypotheses suggested to explain imperfect mimicry, but comparative tests across multiple mimicry systems are needed to determine which are applicable, and which - if any - represent general principles governing imperfect mimicry. We tested four hypotheses on Australian ant mimics and found support for only one of them: the information limitation hypothesis. A predator with incomplete information will be unable to discriminate some poor mimics from their models. We further present a simple model to show that predators are likely to operate with incomplete information because they forage and make decisions while they are learning, so might never learn to properly discriminate poor mimics from their models. We found no evidence that one accurate mimetic trait can compensate for, or constrain, another, or that rapid movement reduces selection pressure for good mimicry. We argue that information limitation may be a general principle behind imperfect mimicry of complex traits, while interactions between components of mimicry are unlikely to provide a general explanation for imperfect mimicry.