Resolution of the ordinal phylogeny of mosses using targeted exons from organellar and nuclear genomes

Abstract

Mosses are a highly diverse lineage of land plants, whose diversification, spanning at least 400 million years, remains phylogenetically ambiguous due to the lack of fossils, massive early extinctions, late radiations, limited morphological variation, and conflicting signal among previously used markers. Here, we present phylogenetic reconstructions based on complete organellar exomes and a comparable set of nuclear genes for this major lineage of land plants. Our analysis of 142 species representing 29 of the 30 moss orders reveals that relative average rates of non-synonymous substitutions in nuclear versus plastid genes are much higher in mosses than in seed plants, consistent with the emerging concept of evolutionary dynamism in mosses. Our results highlight the evolutionary significance of taxa with reduced morphologies, shed light on the relative tempo and mechanisms underlying major cladogenic events, and suggest hypotheses for the relationships and delineation of moss orders.

Fig. 1

The ordinal relationships of moss tree of life (liverwort outgroup pruned for clarity). a Phylogenetic relationships from 105 concatenated nuclear single-loci amino acid sequences based on RAxML analyses (Supplementary Fig. 20); all branches maximally supported (i.e., 100% bootstrap frequencies) unless otherwise marked. Supports from inferences of the plastid (pt), mitochondrial (mt) and nuclear (nu) DNA, and amino acid (AA) sequences by maximum likelihood (ML), and Bayesian inferences (BI) were marked on the main nodes, a square with filled color indicates a strong support on the node (ML-BS ≥ 95; BI-PP ≥ 0.99); a square with a cross indicates a conflict with strong support; an empty square indicates conflict lacking strong support; b Coalescence based inferences of these nuclear single-loci amino acid sequences; ASTRAL tree with local posterior probabilities (Supplementary Fig. 24), branch lengths in coalescent units (2 × N generations) and are directly proportional to the amount of discordance; PhyParts Pie Charts of DNA (left) and AA (right) gene trees. The central sporophyte outlines characterize the major dehiscence types and their phylogenetic distribution; major arthrodontous peristome architectures are illustrated by sections of 1/8th of the amphithecium (see ref. ). Asterisk indicates exemplars for which transcriptome data complemented data recovered via targeted enrichment (see method section for details). Image credit for Sphagnum: Dr. Adam Wilson (University at Buffalo). All other images in Fig. 1 were taken by the senior author (B. Goffinet)

Fig. 2

Total gene tree depth in synonymous (dS) and non-synonymous (dN) substitutions per site for protein-coding genes in three genomic compartments across Bryophyta. Units are synonymous (dS) or non-synonymous (dN) substitutions per site. Rates were calculated using individual gene trees in PAML with the liverwort outgroups removed. Rates for individual genes can be accessed in the online version of the figure (https://plot.ly/~mossmatters/15/)