Colonizations cause host shifts, diversification of preferences and expansion of butterfly diet breadth

Dynamics of herbivorous insect diet breadth are important in generation of novel pests, biological control of weeds and as indicators of global change impacts. But what forces and events drive these dynamics? Here we present evidence for a novel scenario: that specialization increases in persistent populations, but that, at the species level, this trend is countered by effects of colonizations. Colonizations cause host shifts, which are followed by non-adaptive evolutionary expansions of diet breadth, adding transitory hosts during adaptation to the principal novel host. We base this thesis on long-term study of 15 independently-evolving populations of Ediths Checkerspot butterfly, eight of which used fewer host genera in recent censuses than in the 1980s, while none used more - a significant increase in specializaton. At the same time, two extintion/recolonization events were followed by temporary expansions of diet breadth. Behavioural experiments showed that these expansions were driven by within-population diversification of individual oviposition preferences. These results may explain an old puzzle: a significant negative association between population-level diet breadth and mtDNA diversity. Populations with fewer mtDNA haplotypes had broader diets, suggesting that diet breadth increases in younger, recently-colonized populations. A recent global meta-analysis of butterfly diets, using biogeographic data, explains latitudinal patterns of diet breadth by showing that poleward range expansions have caused reduced specialization. This implies broad applicability of our results, which provide a plausible mechanism for the latitudinal trends: colonizations at expanding range margins would increase population-level diet breadths, while population persistence in range interiors would facilitate increasing specialization.