Genome-wide macroevolutionary signatures of key innovations in butterflies colonizing new host plants

Abstract

The mega-diversity of herbivorous insects is attributed to their co-evolutionary associations with plants. Despite abundant studies on insect-plant interactions, we do not know whether host-plant shifts have impacted both genomic adaptation and species diversification over geological times. We show that the antagonistic insect-plant interaction between swallowtail butterflies and the highly toxic birthworts began 55 million years ago in Beringia, followed by several major ancient host-plant shifts. This evolutionary framework provides a valuable opportunity for repeated tests of genomic signatures of macroevolutionary changes and estimation of diversification rates across their phylogeny. We find that host-plant shifts in butterflies are associated with both genome-wide adaptive molecular evolution (more genes under positive selection) and repeated bursts of speciation rates, contributing to an increase in global diversification through time. Our study links ecological changes, genome-wide adaptations and macroevolutionary consequences, lending support to the importance of ecological interactions as evolutionary drivers over long time periods.

Fig. 1. Evolution of host–plant association through time shows strong host–plant conservatism across swallowtail butterflies.

Phylogenetic relationships of swallowtail butterflies, with coloured branches mapping the evolution of host–plant association, as inferred by a maximum-likelihood model (Supplementary Figs. 4 and 6). Additional analyses with two other maximum-likelihood and Bayesian models inferred the same host–plant associations across the phylogeny (Supplementary Fig. 5). Lue. Luehdorfiini, Zerynth. Zerynthiini, T. Teinopalpini. Pictures of butterflies made by Fabien Condamine.

Fig. 2. Synchronous temporal and geographic origin for swallowtails and birthworts.

Bayesian molecular divergence times with exponential priors estimate an early Eocene origin (~55 Ma) for both swallowtails and Aristolochia (alternatively, analyses with a uniform prior estimated an origin around 67 Ma for swallowtails and 64 Ma for Aristolochia; Supplementary Figs. 3, 8 and 9). Biogeographical maximum-likelihood models infer an ancestral area of origin comprising West Nearctic, East Palaearctic and Central America for both swallowtails and birthworts (Supplementary Figs. 10 and 11). Paleoc Paleocene, Pl Pliocene, P Pleistocene, Ma million years ago. Pictures of the plant and butterfly made by Fabien Condamine, and the world map made by Rémi Allio.

Fig. 3. Host–plant shifts lead to repeated bursts in diversification rates and a sustained overall increase in diversification through time.

a Diversification tends to be higher for clades shifting to new host plants, as estimated by trait-dependent diversification models. Boxplots represent Bayesian estimates of net diversification rates for clades feeding on particular host plants (see also Supplementary Fig. 12). b A global increase in diversification is recovered with birth–death models estimating time-dependent diversification (see also Supplementary Figs. 14 and 15). Taking into account rate heterogeneity by estimating host–plant and clade-specific diversification indicates positive gains of net diversification after shifting to new host plants (see also Supplementary Fig. 13). K Cretaceous, Paleoc. Paleocene, Oligoc. Oligocene, Pl Pliocene, P Pleistocene, Ma million years ago. Pictures of butterflies made by Fabien Condamine.

Fig. 4. Host–plant shifts promote higher molecular adaptations.

a Genus-level phylogenomic tree displaying branches with and without host–plant shifts, on which genome-wide analyses of molecular evolution are performed. b Number of genes under positive selection (dN/dS > 1) for swallowtail lineages shifting to new host–plant families (n = 14, green) or not (n = 14, grey). c Number of genes under positive selection for swallowtail lineages undergoing climate shifts (n = 5, orange) or not (n = 23, grey). d Number of genes under positive selection for swallowtail lineages shifting to new host plants (n = 9, green), shifting both host–plant and climate (n = 5, blue) or not (n = 14, grey). The proportion of genes was estimated with Dataset 2 (1533 genes, see Supplementary Fig. 19 for the results with Dataset 1 and 520 genes). This demonstrates genome-wide signatures of adaptations in swallowtail lineages shifting to new host–plant families. Genes under positive selection did not contain over- or under-represented functional GO categories (Supplementary Data 2). Wilcoxon rank-sum test: n.s. = not significant (P > 0.05), *P ≤ 0.05, **P ≤ 0.01. Pictures and icons made by Fabien Condamine.