Nested whole-genome duplications coincide with diversification and high morphological disparity in Brassicaceae

Abstract

Angiosperms have become the dominant terrestrial plant group by diversifying for ~145 million years into a broad range of environments. During the course of evolution, numerous morphological innovations arose, often preceded by whole genome duplications (WGD). The mustard family (Brassicaceae), a successful angiosperm clade with ~4000 species, has been diversifying into many evolutionary lineages for more than 30 million years. Here we develop a species inventory, analyze morphological variation, and present a maternal, plastome-based genus-level phylogeny. We show that increased morphological disparity, despite an apparent absence of clade-specific morphological innovations, is found in tribes with WGDs or diversification rate shifts. Both are important processes in Brassicaceae, resulting in an overall high net diversification rate. Character states show frequent and independent gain and loss, and form varying combinations. Therefore, Brassicaceae pave the way to concepts of phylogenetic genome-wide association studies to analyze the evolution of morphological form and function.

Fig. 1. Brassicaceae maternal timeline.

A simplified version of the BEAST analysis of coding regions from 60 plastid genes (Supplementary Fig. 2) is shown. Two hundred and fifty Brassicaceae species as well as representatives of 15 additional Brassicales families and 47 species (indicated here as outgroups) from other Rosidae families were included. Within Brassicaceae, tribes included with multiple species are represented by gray triangles and the respective crown ages are given. Taxa indicated in red refer to the three categories ‘remote genera’, ‘newly assigned genera’ and ‘unclear and remaining unassigned’; taxa indicated in blue are in need of a new genus name, because they create a polyphyly problem (Supplementary Note 2). Divergence time estimates are based on combined 28 BEAST analyses, resulting in 21,038 analyzed states. Ninety-five percent HPD intervals are represented by blue node bars. Assignment to the three lineages are given, and mesopolyploidization events^, as well as significant rate shifts are indicated with yellow stars and green triangles, respectively. Source data are provided as a Source Data file.

Fig. 2. Morphological characters and their disparity across tribes and lineages.

a Pairwise differences in character state frequency for all 111 character states. b Heatmap of tribal disparity calculated directly from tribal level morphomatrix. Tribes are sorted by phylogeny (following Fig. 1), and characters are sorted by disparity, with the highest mean disparity values on the left. Assignment to the three lineages are given, and mesopolyploidization events^, as well as significant rate shifts are indicated with yellow stars and green triangles, respectively. c Boxplot of tribal mean disparity (direct) across the three lineages. No significant differences were detected between lineages. In the boxplot, center line represents median; upper and lower quartiles are indicated by box limits; whiskers represent 1.5× interquartile range and points are outliers. d DAPC analysis of tribal level disparity with a priori grouping of lineages. The 37 characters in six categories (A–F) are named in the grey box on the right. Characters with significant phylogenetic signal (Moran’s I) are highlighted in red below the pairwise differences and above the heatmap (see also Supplementary Fig. 7). Source data are provided as a Source Data file.

Fig. 3. Disparity and diversification.

a Boxplot of tribal disparity for tribes with and without mesopolyploidizations or rate shifts. Mesopolyploidizations (WGDs) and rate shifts are associated with higher disparity. Significance was tested using phylogenetic ANOVA and P-values are indicated above the boxplots. In the boxplot, center line represents median; upper and lower quartiles are indicated by box limits; whiskers represent 1.5× interquartile range and points are outliers. b BAMM best shift configuration of genus-level plastome phylogeny. The phylogeny was pruned to contain only one sample per (monophyletic) genus. Two significant rate shifts were detected at the onset of lineage diversification ca. 25 mya as well as before the split of tribes Brassiceae+Thelypodieae+Sisymbrieae from Isatideae, and are indicated with red circles; branch colors indicate net diversification rate. Tribe names, number of genera (sampled/total), mesopolyploidizations^,, species level rate shifts and lineage assignment are given on the right. Source data are provided as a Source Data file.