A reappraisal of the phylogenetic placement of the Aquilegia whole-genome duplication.

The accurate placement of an ancient whole-genome duplication (WGD) in relation to the lineage divergence is important. Here, we re-investigated the Aquilegia coerulea WGD and found it is more likely lineage-specific rather than shared by all eudicots.

Whole-genome duplications (WGDs) are frequent and common in plants, contributing to the evolutionary novelty and adaptation to extreme environments [1, 2]. Aköz and Nordborg reported a common WGD in the ancestral eudicot: while this ancestral tetraploid is preserved in Aquilegia (a basal eudicot), the hexaploid in the ancestral core eudicot was formed by hybridization of this tetraploid and another diploid with a subsequent WGD [3]. Nevertheless, this is contradictory to the studies of Nelumbo nucifera [4, 5]. Nelumbo, another basal eudicot, having a much closer relationship with core eudicots [6], shows a slower synteny loss and substitution rate than Aquilegia and core eudicots when aligned to outgroup species including Nymphaea colorata [7], rice, and Brachypodium distachyon (Figs. 1b and 2e) [4], and thus, Nelumbo should preserve more of its traces. However, the Ks peak corresponding to the “shared WGD” is absent in Nelumbo, and only a lineage-specific WGD after the Nelumbo-Macadamia split was found [4] (Fig. 2a), which raises doubt about their hypothesis of the common tetraploid origin.

Fig. 1

Genome structural analyses of Aquilegia and related species. a Scenario of rapid intraspecific synteny loss (diploidization) causing homoplasy of interspecific syntenies. b, c Distribution of interspecific and intraspecific syntenic blocks according to syntenic block size. d Scenario of a chromosomal fusion in the common ancestor of Aquilegia and Vitis. e Scenario of two independent chromosomal fusions in Aquilegia and Vitis, respectively. f, g Number of syntenic anchors between the orange-green fused chromosomes (Vitis chr7 and Aquilegia chr5) and their corresponding homologous regions

Fig. 2

Phylogenomic analyses of syntenic duplicates from Aquilegia. a–d Histograms and violin plots of rate-corrected Ks of gene pairs from intraspecific and interspecific synteny in Aquilegia and Nelumbo (a, b) and in Aquilegia and Vitis (c, d). Median Ks values are marked by the black line in violin plots. **p value < 0.0001. e A species tree based on a concatenated alignment of 697 strict single-copy genes. f Example gene tree supports the hypothesis of Aquilegia WGD only shared by Aquilegia and Papaver. g Example gene tree supports the hypothesis of Aquilegia WGD being Aquilegia-specific. h Example gene tree supports the hypothesis of Aquilegia WGD occurring in the ancestral eudicot. Node values: bootstraps. Branch length: the number of amino acid substitutions per site. Colored dots: WGD or triplication. Arrow: the difference in placement of the Aquilegia WGD

The authors’ assertion that the Aquilegia WGD is shared by all eudicots is mainly based on clustering by gene order similarity or structural similarity within species genome and between species, which depends on synteny evolutionary rate. When the trait evolves with rate heterogeneity, the simple clustering by similarity can hardly reflect the true phylogeny. Therefore, the more rapid loss of intraspecific synteny in Aquilegia might cause homoplasy such that higher similarity in gene order of Aquilegia-Vitis than Aquilegia-Aquilegia can be observed (Fig. 1a). Indeed, Aquilegia-Aquilegia shows the most rapid decay of synteny with the smallest size and the fewest number of syntenic blocks, which is even smaller and fewer than older divergent pairs (Aquilegia-Ceratophyllum and Aquilegia-Nymphaea) (Fig. 1c) [7]. Rapid convergent loss of the alternative copies, particularly housekeeping genes, by selection against redundancy after WGDs was found in angiosperms [8], and this is also congruent with more rapid syntenic gene loss that occurred soon after a WGD in teleost fish [9]. Therefore, their placement of a WGD simply by gene order similarity is flawed because of homoplasy caused by the heterogeneous rates of synteny loss. Meanwhile, the authors claimed that a fusion of ancestral eudicot chromosomes “green” and “orange” was observed in both Aquilegia chr5 and Vitis chr7, which is derived from a fusion in their tetraploid ancestor (Fig. 1d). Yet, they did not exclude an alternative possibility of two independent green-orange fusions in Aquilegia and Vitis, respectively (Fig. 1e). If their assumption is true, the fused green-orange chromosomes are expected to share closer homology (more syntenic anchors) to each other than the other green or orange regions. On the contrary, we found that the green and orange regions in the fused Vitis chr7 and Aquilegia chr6 share more anchors with non-fused green or orange regions (Fig. 1f, g). Therefore, the scenario of two independent chromosomal fusions is more likely.

Here, we showed that the Aquilegia WGD is more likely lineage-specific rather than common to all eudicots by two different phylogenomic approaches. (1) Ks values of syntenic paralogs and orthologs were measured to infer the sequential order of WGD and species divergence. Ks of each lineage were corrected using the relative rate (Aquilegia to Nelumbo to Vitis = 1:0.750:0.970) based on Ks branch length ratios of 1425 single-copy ortholog groups [10]. For comparison, Ks > 3 were removed to prevent substitutional saturation [11]. The Ks distances of syntenic orthologs for both Aquilegia-Nelumbo and Aquilegia-Vitis are significantly longer than intraspecific paralogs in Aquilegia (Mann-Whitney U test, both p value < 0.0001), suggesting this WGD exclusively occurred in Aquilegia (Fig. 2a–d). (2) To circumscribe Aquilegia syntenic duplicates in relation to the species divergence, 697 gene trees of ortholog groups from key taxa containing the Aquilegia syntenic paralogs were reconciled with the species tree using Notung2.9 (www.cs.cmu.edu/~durand/Notung) [12-14]. The species tree was constructed from 176 strict single-copy genes by OrthoFinder (www.stevekellylab.com/software/orthofinder) and IQ-TREE (www.iqtree.org) (Fig. 2e). When we applied bootstrap thresholds of 90, 70, and 50, we found the hypothesis that Aquilegia syntenic duplications occurred after the divergence of Aquilegia and the ancestor of Nelumbo and core eudicots is supported by 53, 117, and 194 subtrees, respectively (red and yellow dots in Fig 2e–g), whereas hypothesis that duplication occurred in the ancestral eudicot is supported by 34, 81, and 131 subtrees, respectively (gray dot in Fig. 2e, h). Due to the fact that Aquilegia WGD and Aquilegia-Papaver split are closely spaced in time, it is more difficult to resolve their sequential order. However, more subtrees support Aquilegia WGD being independent (red dot in Fig. 2e, g) than being shared by Aquilegia and Papaver (yellow dot in Fig. 2e, f), which is in line with a 2:2 syntenic relationship between Aquilegia and Papaver [15].