Global Phylogeny of the Brassicaceae Provides Important Insights into Gene Discordance

Abstract

The mustard family (Brassicaceae) is a scientifically and economically important family, containing the model plant Arabidopsis thaliana and numerous crop species that feed billions worldwide. Despite its relevance, most published family phylogenies are incompletely sampled, generally contain massive polytomies, and/or show incongruent topologies between datasets. Here, we present the most complete Brassicaceae genus-level family phylogenies to date (Brassicaceae Tree of Life, or BrassiToL) based on nuclear (>1,000 genes, almost all 349 genera and 53 tribes) and plastome (60 genes, 79% of the genera, all tribes) data. We found cytonuclear discordance between nuclear and plastome-derived phylogenies, which is likely a result of rampant hybridisation among closely and more distantly related species, and highlight rogue taxa. To evaluate the impact of this rampant hybridisation on the nuclear phylogeny reconstruction, we performed four different sampling routines that increasingly removed variable data and likely paralogs. Our resulting cleaned subset of 297 nuclear genes revealed high support for the tribes, while support for the main lineages remained relatively low. Calibration based on the 20 most clock-like nuclear genes suggests a late Eocene to late Oligocene icehouse origin of the family. Finally, we propose five new or re-established tribes, including the recognition of Arabidopsideae, a monotypic tribe to accommodate Arabidopsis. With a worldwide community of thousands of researchers working on this family, our new, densely sampled family phylogeny will be an indispensable tool to further highlight Brassicaceae as an excellent model family for studies on biodiversity and plant biology.

Synteny identifies reliable orthologs for phylogenomics and comparative genomics of the Brassicaceae

“…The selection of genes may nonetheless influence the topology. We found that while the Brassicaceae species tree based on the complete Brassicaceae set as used in the original publication supports Arabideae as sister to clades A, B and C (Nikolov et al 2019), the tree based on the syntenic subset we analyzed here followed the same topology we detected for other gene sets without paralogs, with clades A and C as sister to Arabideae and clade B. Interestingly, a recent genus-level phylogeny of Brassicaceae also found different topologies at deep nodes depending on filtering level, with the species tree inferred from the combined Angiosperm353 and Brassicaceae gene sets supporting topology B and stricter filtering resulting in a species tree more similar to our topology A (Hendriks et al 2022). Altogether, we conclude that the use of synteny for selecting genes for phylogenomics does not necessarily perform better than more established methods when it comes to the topology of the resulting species tree with regards to tree resolution.…”

“…Additionally, we ran ASTRAL for all gene sets against the Brassicaceae phylogeny following the prevailing species tree hypothesis (also shown in Fig. 1) derived from three recent comprehensive phylogenomics studies (Huang et al 2016; Nikolov et al 2019; Hendriks et al 2022) and compared the normalized quartet scores. The scores roughly followed the same trend as mean bootstrap support, with datasets comprised mainly of orthologs having higher values than those comprised of or including paralogs (with the exception of the two smallest gene sets, for which no reliable estimate could be obtained).…”

“…3b, c). Interestingly, the syntenic ortholog subset of Angiosperm353 and Brassicaceae sets as well as the OrthoFinder subset of the latter also followed topology A, although the use of the entire Brassicaceae gene set in the respective study resulted in topology B (Nikolov et al 2019), as did the combination of both sets (Hendriks et al 2022). Additionally, we ran ASTRAL for all gene sets against the Brassicaceae phylogeny following the prevailing species tree hypothesis (also shown in Fig.…”

“…However, the family’s phylogeny has so far proven difficult to resolve despite considerable efforts. The branching order of the main lineages differs between plastid based phylogenies (Walden, German, et al 2020; Hendriks et al 2022) and phylogenies based on nuclear genes (Nikolov et al 2019; Huang et al 2016; Kiefer et al 2019; Hendriks et al 2022), and support values for deeper nodes can be low. All Brassicaceae share the At-α WGD (Bowers et al 2003; Schranz & Mitchell-Olds 2006) that occurred after divergence from the sister family Cleomaceae ∼ 40 million years ago (Edger et al 2015), and many Brassicaceae tribes have undergone additional meso- and neopolyploidizations (Hohmann et al 2015; Mandáková, Li, et al 2017).…”

Modular transcriptional responses to environmental changes

“…Accordingly, the emergence of standardized practices for generating and analysing RNAseq data (Conesa et al ., 2016) offers support towards meaningful conclusions on transcriptional plasticity under environmental changes. Here, we address transcriptional responses of Biscutella laevigata , a widespread species belonging to an early diverging Brassicaceae genus (Couvreur et al ., 2010; Hendriks et al ., 2022) to different environments. Being a textbook example of autopolyploidy linked to ice ages (Manton, 1937; Parisod & Besnard, 2007), diploids of B. laevigata occur across major ecological gradients (e.g.…”