These three modification types (i.e., the inward, reduction, and outward types) were generally found in both sexes in the years 2002, 2003, and 2004, but their proportion among the total number of the modified individuals varied between sexes and among years (Fig. (Fig.2S).2S). In both sexes, but especially in females, the majority of the modified individuals were of the outward type. In males, the percentage of the outward type increased stepwise from 2002 to 2004. In males but not in females, the reduction type also occupied a significant proportion of the modified population, and its proportion was nearly constant over three years. The inward type was observed mostly in 2002 in both sexes, and in 2003 and 2004, its proportion was small or null.
Temperature simulation experiment
We here tested whether temperature dynamics recorded in Fukaura in summer could induce modifications in a laboratory. We obtained 32 successfully eclosed individuals from Cycle A (see Methods), which produced three modified individuals in total (one outward type, one inward type, and one reduction type). They were all just slightly modified and not comparable with the field-caught modified Fukaura individuals in their degrees of modifications (not shown). From Cycle B (see Methods), we obtained 30 adults, but none of them were modified. These results indicate that temperature fluctuations during the pupal period in summer would not be an immediate cause for the outbreak of the modified forms in Fukaura.
Artificial modifications: cold-shock experiment
The highest incidence of the modified forms in summer might at first suggest that heat shock in Fukaura induced modifications. We checked the past temperature records thoroughly but the highest temperature of Fukaura was not higher than those of southern parts in the mainland Japan. Indeed, Z. maha is mainly distributed in tropical or temperate area in Asia, and thus it should be resistant to "high" temperature in Fukaura. Furthermore, "high-temperature" treatment indeed produced no modification (not shown). Thus, despite the fact that the modified individuals were found during summer in Fukaura, we think that this is simply because of the increase of the entire population in summer.
On the other hand, the color-pattern modifications were indeed quite similar, if not identical, to the cold-shock-induced modifications previously reported in lycaenid butterflies [13,14]. At this point, we reasoned that northern expansion was made in response to warmer climate during summer, and Z. maha subsequently experienced cold stress in winter that had never been experienced before. We thus tested whether experimental application of cold shock could reproduce these color-pattern modifications. For this purpose, we first used pupae that were reared from eggs collected from normal females of Z. maha caught in Hiratsuka, a locality south of Fukaura (see Fig. Fig.1A1A inset). The Hiratsuka population would have never been exposed to the environmental conditions experienced by the Fukaura population; nonetheless, the Hiratsuka population is expected to possess similar phenotypic plasticity. This possibility can be examined by a cold-shock treatment, as if the Hiratsuka population migrated to the north and encountered the colder environment.