Abstract
Ehrlich and Raven proposed a model of coevolution where major host plant shifts of butterflies facilitate a burst of diversification driven by their arrival to a new adaptive zone. One prediction of this model is that reconstructions of historical diversification of butterflies should indicate an increase in diversification rate following major host shifts. Using reconstructed histories of 15 butterfly groups, I tested this prediction and found general agreement with Ehrlich and Raven's model. Butterfly lineages with an inferred major historical host shift showed evidence of diversification rate variation, with a significant acceleration following the host shift. Lineages without an inferred major host shift generally agreed with a constant-rate model of diversification. These results are consistent with the view that host plant associations have played a profound role in the evolutionary history of butterflies, and show that major shifts to chemically distinct plant groups leave a historical footprint that remains detectable today.