Phylotranscriptomics Illuminates the Placement of Whole Genome Duplications and Gene Retention in Ferns

Abstract

Ferns are the second largest clade of vascular plants with over 10,000 species, yet the generation of genomic resources for the group has lagged behind other major clades of plants. Transcriptomic data have proven to be a powerful tool to assess phylogenetic relationships, using thousands of markers that are largely conserved across the genome, and without the need to sequence entire genomes. We assembled the largest nuclear phylogenetic dataset for ferns to date, including 2884 single-copy nuclear loci from 247 transcriptomes (242 ferns, five outgroups), and investigated phylogenetic relationships across the fern tree, the placement of whole genome duplications (WGDs), and gene retention patterns following WGDs. We generated a well-supported phylogeny of ferns and identified several regions of the fern phylogeny that demonstrate high levels of gene tree-species tree conflict, which largely correspond to areas of the phylogeny that have been difficult to resolve. Using a combination of approaches, we identified 27 WGDs across the phylogeny, including 18 large-scale events (involving more than one sampled taxon) and nine small-scale events (involving only one sampled taxon). Most inferred WGDs occur within single lineages (e.g., orders, families) rather than on the backbone of the phylogeny, although two inferred events are shared by leptosporangiate ferns (excluding Osmundales) and Polypodiales (excluding Lindsaeineae and Saccolomatineae), clades which correspond to the majority of fern diversity. We further examined how retained duplicates following WGDs compared across independent events and found that functions of retained genes were largely convergent, with processes involved in binding, responses to stimuli, and certain organelles over-represented in paralogs while processes involved in transport, organelles derived from endosymbiotic events, and signaling were under-represented. To date, our study is the most comprehensive investigation of the nuclear fern phylogeny, though several avenues for future research remain unexplored.

Keywords: biased gene retention; fern; phylogenetics; polyploidy; transcriptome; whole genome duplication.

FIGURE 1

The diversity of ferns. (A) Equisetum hyemale (Equisetaceae, Equisetales), (B) Sceptridium dissectum forma dissectum (Ophioglossaceae, Ophioglossales), (C) Osmundastrum cinnamomeum (Osmundaceae, Osmundales), (D) Angiopteris evecta (Marattiaceae, Marattiales), (E) Lygodium microphyllum (Lygodiaceae, Schizaeales), (F) Azolla filiculoides (Salviniaceae, Salviniales), (G) Sphaeropteris cooperi (Cyatheaceae, Cyatheales), (H) Myriopteris wootonii (Pteridaceae, Polypodiales), (I) Asplenium hybrid (Aspleniaceae, Polypodiales), (J) Telmatoblechnum serrulatum (Blechnaceae, Polypodiales), (K) Dryopteris ludoviciana (Dryopteridaceae, Polypodiales), (L) Polypodium virginianum (Polypodiaceae, Polypodiales). All images by JP.

FIGURE 2

Species tree generated from 2884 single copy nuclear loci (SCO60 dataset) under the multi-species coalescent. Divergence times are based on a penalized likelihood method in TreePL. Inferred whole genome duplications (WGDs) are placed on the tree (note that the age of the WGDs are not depicted, rather events are placed on the midpoint of the branches). Events inferred from K_S and MAPS analyses are shown as red circles, and events inferred from K_S evidence only are shown as blue squares. The size of the symbol reflects whether the event is large-scale (larger symbol, including more than one sampled taxon) or small-scale (smaller symbol, including only one sampled taxon). Asterisks for LEPTO and POLY events indicate there are no current names for the clades corresponding to the taxa affected by these events: LEPTO is shared by leptosporangiate ferns excluding Osmundales and POLY is shared by Polypodiales excluding Lindsaeineae and Saccolomatineae.

FIGURE 3

Proportions of gene trees supporting the inferred species trees and alternative topologies for relationships of particular interest: (A) Horsetails (Equisetales) and other ferns, (B) Eusporangiate-leptosporangiate fern relationships, (C) Gleicheniales and Hymenophyllales, and (D) Pteridaceae, Dennstaedtiaceae, and the eupolypods. Topological support from select plastid (green, Pryer et al., 2004; Testo and Sundue, 2016) and nuclear (blue, Rothfels et al., 2015; Qi et al., 2018; Shen et al., 2018; One Thousand Plant Transcriptomes Initiative, 2019) studies are given beside possible topologies.

FIGURE 4

Inferred whole genome duplication (WGD) events from uncorrected (left, blue) and corrected (right, green) K_S analyses. Unambiguous placements of inferred WGDs are depicted as solid squares; ambiguous placements are depicted as dotted squares. Trichom = Trichomanoideae, Hymeno = Hymenophylloideae.

FIGURE 5

Gene retention in ferns following large-scale duplications is biased. Gene ontology (GO) Slim composition of retained duplicates (paralogs) from 17 of the 18 large-scale whole genome duplications (WGDs) identified in this study are shown, where categories that are over-represented in paralogs are in red (Chi-squared residuals > 2) and under-presented as blue (Chi-squared residuals < -2). Terms with non-significant residuals are gray. Events are sorted by ascending median Ks values. Note that SALV is not shown here; while MAPS identified a significant proportion of duplicated gene trees, uncorrected K_S plots were ambiguous, although a WGD was identified with syntenic analyses in Li F. W. et al. (2018).