Phylogenomics, co-evolution of ecological niche and morphology, and historical biogeography of buckeyes, horsechestnuts, and their relatives (Hippocastaneae, Sapindaceae) and the value of RAD-Seq for deep evolutionary inferences back to the Late Cretaceous.

In this study, we used RAD-seq data to resolve the phylogeny of the tribe Hippocastaneae (Sapindaceae) and conducted comparative analyses to gain insights into the evolution and biogeography of the group that had fossils dating back to the late Cretaceous. Hippocastaneae, including the horsechestnuts and buckeyes, is a well-supported clade in Sapindaceae that comprises 12-14 species in Aesculus, two in Billia, and one in Handeliodendron. Most species in the tribe are distributed in Eurasia and North America and exhibit a classic pattern of intercontinental disjunction in the Northern Hemisphere, while Billia occurs from southern Mexico to northern South America. The earliest fossils of Aesculus date back to at least the earliest Paleocene of eastern Asia and western North America, where there are also putative occurrences from the latest Cretaceous. The group provides an excellent system for understanding floristic disjunction in the Northern Hemisphere extending to the Neotropics. However, a strongly supported and well resolved phylogeny is presently lacking for the tribe. Previous phylogenetic studies using several gene regions revealed five well-supported clades in Aesculus, largely corresponding to five recognized taxonomic sections, but relationships among these clades and among Aesculus, Billia, and Handeliodendron were not well supported. In this study, we used RAD-seq data from 68 samples representing all clades and species of Hippocastaneae except Billia, for which we used one of two species, to further resolve relationships within the tribe. Our phylogenomic analyses showed strong support for a sister relationship between Aesculus and Handeliodendron, in contrast to previous findings which supported Billia as sister to Aesculus. Within Aesculus, relationships among sections were strongly supported as (sect. Calothyrsus, (sect. Aesculus, (sect. Macrothyrsus, (sect. Parryana, sect. Pavia)))). We found that the traditionally recognized section Calothyrsus was monophyletic, with all eastern Asian species sister to the western North American species, A. californica. Analyses of divergence times combined with biogeographic analyses suggested a Late Cretaceous origin of Hippocastaneae, in eastern Asia, western North America, and Central America (including southern Mexico), followed by isolation of Billia in Central America, extinction of the tribe ancestor in western North America, and divergence of Aesculus from Handeliodendron in eastern Asia. A Late Cretaceous origin of the common ancestor of Aesculus in eastern Asia was followed by dispersals into western North America, Europe, and eastern North America during the Late Cretaceous and the Paleogene. Our results support Aesculus as a relic of the boreotropical flora and subsequent intercontinental spread of the genus through the Bering and North Atlantic land bridges. We performed character mapping analyses, which revealed that biogeographic isolation and niche divergence may have played important roles in driving morphological evolution and lineage divergence in Aesculus. Our study demonstrates the value of RAD-seq data for reconstructing phylogeny back to the Late Cretaceous.