Abstract

Reconstructing the biogeographical history and timing of the diversification of temperate forests is essential for understanding their history and resolving uncertainties about how flowering plants emerged from their deep tropical origins to dominate in todays freezing terrestrial environments. The angiosperm order Fagales, comprising iconic components of temperate forests worldwide with an extensive fossil record, are an excellent plant system in which to apply a fossil-aware paradigm, such as the fossilized birth-death (FBD) process, for investigating the macroevolution of temperate forest biomes. Here, we improve upon previous efforts to resolve phylogeny and incorporate fossils in Fagales using low-copy nuclear loci and an expanded morphological matrix to reevaluate the Fagales fossil record and: (1) infer the phylogenetic relationships and the time of origin of the clade using the FBD model as implemented in RevBayes, (2) provide a framework for evaluating the climatic and biogeographic history of Fagales, and (3) investigate how the inclusion of fossils via the FBD method influences ancestral reconstruction and diversification estimation. The phylogenetic relationships we recovered are conventional except for the position of Nothofagaceae, while our inferred ages support older timelines than previously proposed, with a mid-Cretaceous date for the most recent common ancestor (MRCA) of the order. Biogeographical analysis shows an origin of Fagales consistent with an ancestral circumboreal temperate distribution corroborated by ancestral niche reconstructions. While distributions today largely reflect the general conservatism of temperate forests, we identified two episodes of high diversification, one at the mid-Cretaceous origin of the clade and the other continuing from the Miocene to the present. Removing fossil taxa from the tree reveals a different story, shifting the origin of extant families from North America to East Asia, reflecting refugial distributions in this biodiversity "museum" and implying a general bias towards low extinction areas in biogeographic reconstruction. Likewise, without fossil data, diversification estimates were higher and unable to detect an early diversification burst. Based on our analyses, we close with recommendations regarding the interpretation of estimates of diversification and ancestral state reconstruction using phylogenetic trees with only extant species as tips.