DID FOREST ISLANDS DRIVE THE DIVERSITY OF WARNINGLY COLORED BUTTERFLIES - BIOTIC DRIFT AND THE SHIFTING BALANCE

Species of the South American butterfly genus Heliconius have undergon e remarkably wide racial divergence in their patterns, and most of the resulting races are muellerian mimics. As warning coloration normally imposes stabilizing selection on the pattern, this divergence is much in need of explanation. Two models have been suggested. Brown, Sheppa rd and Turner proposed that the divergence results from 'mimetic captu re', the switching of patterns between adaptive peaks generated by cha nges in the overall composition of the local biota ('biotic drift') an d hence of the mimicry rings to which each species belongs; these chan ges have in turn been generated by long term patterns of species extin ction in island refuges as biota became progressively isolated and con tiguous during contraction and expansion of the rain forest during the Pleistocene. An alternative model, proposed by Mallet, is that truly novel colour patterns became established by mutation and random drift, then spreading to become predominant in local areas; subsequently the novel patterns spread over wide areas by the migration of dines. Unde r this application of Wright's shifting balance model, refuges are not necessary for divergence, and muellerian mimicry evolves after diverg ence rather than being the driving force for race formation. Although our respective models appear diametrically opposed, the hypotheses are difficult to distinguish and there are broad areas of agreement; in b oth models there is an initial stochastic event, followed by natural s election for mimicry, and both will operate either in parapatry or all opatry. The diversity of warning patterns is better explained by the s hifting balance model, but there are alternative selectionist explanat ions such as sexual selection.