Food stress causes differential survival of socially parasitic caterpillars of Maculinea rebeli integrated in colonies of host and non-host Myrmica ant species

Final instar larvae of Maculinea rebeli Hirschke (Lepidoptera, Lycaenidae) are social parasites of Myrmica Latreille (Hymenoptera, Formicidae) nests. In the populations of the southern French Alps and Spanish Pyrenees, >95% adult M. rebeli emerge from colonies of Myrmica schencki Emery, despite >60% caterpillars being adopted by other Myrmica species (non‐hosts). However, in laboratory culture caterpillars can be reared successfully by many of the non‐host species. This contradiction, which has led some to question the existence of host specificity, has been explained by the lack of stress, particularly food stress, in laboratory cultures compared to wild conditions. Here, we report the results of an experiment that tested the survival of M. rebeli caterpillars that had been growing well, after being socially integrated into a series of host and non‐host cultures, and were then subjected to a 4‐week period of stress induced by a ‘starvation diet’ estimated to be less than the minimum for ant survival. Significantly more M. rebeli survived in M. schencki cultures than with any of the other Myrmica species (all died in most non‐host cultures). Under a starvation diet, caterpillars are killed and eaten along with dead workers – this never happens under an ample diet – rather than simply starving to death. It was noted that the proportion of young M. rebeli caterpillars that survived initial integration into an ant colony (including some M. schencki colonies) was a good predictor of subsequent survival under starvation conditions. We concluded that two key phases of host specificity exist in the life of this social parasite: initial integration, in which the caterpillar simply has to be accepted into a host society, followed by full integration, when a relatively high hierarchical status within the host society becomes essential for a caterpillar's survival during periods when the host colony is stressed, e.g., by food shortage. This experimental regime provides a useful bioassay for testing host specificity in other populations of Maculinea.