Phylogenomics AND biogeography of Castanea (chestnut) and Hamamelis (witch-hazel) – Choosing between RAD-seq and Hyb-Seq approaches

Abstract

Hyb-Seq and RAD-seq are well-established high throughput sequencing technologies that have been increasingly used for plant phylogenomic studies. Each method has its own pros and cons. The choice between them is a practical issue for plant systematists studying the evolutionary histories of biodiversity of relatively recent origins. However, few studies have compared the congruence and conflict between results from the two methods within the same group of organisms in plants. In this study, we employed RAD-seq and Hyb-Seq of Angiosperms353 genes in phylogenomic and biogeographic studies of Hamamelis (the witch-hazels) and Castanea (chestnuts), two classic examples exhibiting the well-known eastern Asian (EA) -eastern North American (ENA) disjunct distribution, and compared them side by side. Our results showed congruences in phylogenetic inference and divergence time dating between the two data sets obtained through our customized procedures of library preparation and sequence trimming, although they differed in the number of loci and informative sites, the amount of missing data, and sampling within species. We provide recommendations regarding the selection of the two methods for phylogenomic study at generic level based on fund availability and sampling scale. If funds and time are not constrained, we recommend Hyb-Seq. If funds and time are somewhat limited and sampling is large, we recommend RAD-seq. However, we found greater conflict among gene trees from the RAD-seq data due to the short sequences per locus. Therefore, species tree building and network detecting with the RAD-seq data with short RAD-seq loci (e.g., <150 bp) should avoid using analytical methods relying on gene trees of individual locus, but using site-based methods such as SVDQuartets and D-statistic method. Our phylogenetic analyses of RAD-seq and Hyb-Seq data resulted in well-resolved species relationships. Analyses of the data using the D-statistic test and PhyloNet revealed ancient introgressions in both genera. Biogeographic analyses including fossil data using total evidence-based dated tree and DEC model applying specific inter-area dispersal probabilities revealed a complicated history for each genus, indicating multiple interareal dispersals and local extinctions within and outside areas of the taxa's modern ranges in both the Paleogene and Neogene. The study demonstrates the importance of including fossil taxa for a more accurate reconstruction of biogeographic histories of taxa to understand the EA and ENA floristic disjunction. Our results support a widespread ancestral range in EA-western North America (WNA) followed by early diversification in EA and expansion to North America (NA) and Europe for Castanea and a more widespread ancestral range in EA-ENA-WNA for Hamamelis. The origins of the modern EA-ENA disjunction in both genera were suggested to be the result of vicariance from widespread ancestors in Eurasia-ENA of the mid-Miocene and in EA-NA of the late Oligocene, respectively.