MrBayes_max_cred_treeMaximum credibility clade tree from MrBayes analysis. This tree topology was fixed to remain constant in all subsequent BEAST dating analyses. It contains 138 species.Viburnum_alignment_3_partitionsThe concatenated 10 gene sequence alignment including all matK, ndhF, rbcL, petB, psbA-trnH, rpl32-trnL, trnC-ycf6, trnK, and trnS-trnG sequences. There are 138 species and 9772 bp.BEAST_crown_1_posterior_treesPosterior trees from the BEAST analysis in which the MrBayes tree topology was constrained to remain constant and the crown age of Viburnum was fixed to 1 (uniform prior 0.99–1.01). Trees from the two runs were combined in LogCombiner with a Burnin of 1000000.BEAST_pollen_fossil_posterior_treesPosterior trees from the BEAST analysis in which the MrBayes tree topology was constrained to remain constant and the divergence times were calibrated with fossil pollen. Trees from the two runs were combined in LogCombiner with a Burnin of 1000000.BEAST_leaf_fossils_posterior_treesPosterior trees from the BEAST analysis in which the MrBayes tree topology was constrained to remain constant and the divergence times were calibrated with leaf fossils. Trees from the two runs were combined in LogCombiner with a Burnin of 1000000.BEAST_crown_1_max_credMaximum clade credibility tree from the BEAST analysis in which the MrBayes tree topology was constrained to remain constant and the crown age of Viburnum was fixed to 1 (uniform prior 0.99–1.01).BEAST_pollen_fossil_max_credMaximum clade credibility tree from the BEAST analysis in which the MrBayes tree topology was constrained to remain constant and the divergence times were calibrated with fossil pollen.BEAST_leaf_fossils_max_credMaximum clade credibility tree from the BEAST analysis in which the MrBayes tree topology was constrained to remain constant and the divergence times were calibrated with leaf fossils.

We used a near-complete phylogeny for the angiosperm clade Viburnum to assess lineage diversification rates, and to examine possible morphological and ecological factors driving radiations. Maximum-likelihood and Bayesian approaches identified shifts in diversification rate and possible links to character evolution. We inferred the ancestral environment for Viburnum and changes in diversification dynamics associated with subsequent biome shifts. Viburnum probably diversified in tropical forests of Southeast Asia in the Eocene, with three subsequent radiations in temperate clades during the Miocene. Four traits (purple fruits, extrafloral nectaries, bud scales and toothed leaves) were statistically associated with higher rates of diversification. However, we argue that these traits are unlikely to be driving diversification directly. Instead, two radiations were associated with the occupation of mountainous regions and a third with repeated shifts between colder and warmer temperate forests. Early-branching depauperate lineages imply that the rare lowland tropical species are ‘dying embers’ of once more diverse lineages; net diversification rates in Viburnum likely decreased in these tropical environments after the Oligocene. We suggest that ‘taxon pulse’ dynamics might characterize other temperate plant lineages.