Inaccurate mimicry and predator ecology.

Batesian mimicry, the resemblance of palatable organisms to an aversive model, is a well documented phenomenon in biology. Resemblance to the model, however, frequently seems to be inaccurate for unknown reasons (Edmunds, 2000). Recently, Johnstone (2002) modelled a scenario where kin selection can lead to the evolution of inaccurate mimics. The model assumes that if mimics are relatively frequent compared to models or if attacking the model incurs only moderate costs, the predator's incentive to attack will be high. As a consequence, the evolutionarily stable attack rate of a predator increases with an increasing local similarity of mimics to their model. Johnstone argued that under conditions where predation risk of the mimic increases with local similarity to the model, kin selection can favour the evolution of inaccurate mimics. Here, we would like to suggest an addition to the model that may improve its resemblance to nature. In accordance with Lima (2002), we want to emphasize that expected outcomes of predator–prey interactions can radically change when predators are allowed to respond strategically to prey profitability. We suggest that a realistic estimate of predation risk would have to include not only attack rate but also predator density. Predator density is likely to be influenced by the net benefit of predation and will thus differ between patches with accurate and inaccurate mimics. In Johnstone's model, predators were assumed to optimize attack rate in individual patches, but patch choice or other predator responses to the available net benefit were not included. Generally, predators should prefer patches providing the largest net benefit. In the case of accurate and inaccurate mimics, the net benefit for predators to forage in patches with relatively inaccurate mimics will be higher because the risk of attacking models is reduced there, compared to patches with more accurate mimics. Thus, predators able to choose between patches that differ in local similarity between model and mimic should forage preferentially in areas with lower local similarity. Unspecialized predators may even refrain completely from attacking in patches with accurate mimics if increasing local similarity to the model leads to a lower net benefit in comparisons to the profitability of available alternative prey species. Under such conditions , an increased local frequency of predation resulting from an increased local net benefit to predators in patches with less complete mimicry thus at least has the potential to override the impact of kin selection proposed by Johnstone (2002). …