Shift happens! Shifting balance and the evolution of diversity in warning colour and mimicry

2. Phase I, random local processes (including genetic drift and idiosyncratic selection), and phase III, interdemic selection are the most controversial phases of the shifting balance. Phase II consists of ordinary natural selection to a new adaptive peak within populations, and is uncontroversial. 3. Heliconius have bold patterns of iridescent blue, black, yellow, white, and red. These are clearly warning patterns, and near-perfect Mullerian mimicry has evolved among species, suggesting tight control by natural selection. Field experiments have also demonstrated strong selection (often s > 0.1 on single colour pattern loci), and the population structure of Heliconius is typically not conducive to phase I. Yet the colour patterns are clearly somewhat independent and incompatible signals of unpalatability. 4. There is empirical evidence in Heliconius for both controversial phases. For phase I, occasional and local polymorphisms of colour pattern in a number of species go against the generally expected (and generally observed) monomorphism for Mullerian mimics. 5. For phase III, one of the few colour pattern clines mapped in detail has been observed to move rapidly over a period of 20 years. There are also a number of curious 'leapfrog' geographical disjunctions in colour pattern races. Disjunctions are expected if successful races have spread from the centre of the range (e.g. the Amazonian rayed races) via phase III, in competition with earlier races that are now distributed in scattered places along the periphery of the range. 6. Evidence from the genomes of Heliconius may in the near future aid in understanding colour pattern 'supergenes' and to help test for origin and spread via shifting balance.