A genome triplication associated with early diversification of the core eudicots

Abstract

Background: Although it is agreed that a major polyploidy event, gamma, occurred within the eudicots, the phylogenetic placement of the event remains unclear.

Results: To determine when this polyploidization occurred relative to speciation events in angiosperm history, we employed a phylogenomic approach to investigate the timing of gene set duplications located on syntenic gamma blocks. We populated 769 putative gene families with large sets of homologs obtained from public transcriptomes of basal angiosperms, magnoliids, asterids, and more than 91.8 gigabases of new next-generation transcriptome sequences of non-grass monocots and basal eudicots. The overwhelming majority (95%) of well-resolved gamma duplications was placed before the separation of rosids and asterids and after the split of monocots and eudicots, providing strong evidence that the gamma polyploidy event occurred early in eudicot evolution. Further, the majority of gene duplications was placed after the divergence of the Ranunculales and core eudicots, indicating that the gamma appears to be restricted to core eudicots. Molecular dating estimates indicate that the duplication events were intensely concentrated around 117 million years ago.

Conclusions: The rapid radiation of core eudicot lineages that gave rise to nearly 75% of angiosperm species appears to have occurred coincidentally or shortly following the gamma triplication event. Reconciliation of gene trees with a species phylogeny can elucidate the timing of major events in genome evolution, even when genome sequences are only available for a subset of species represented in the gene trees. Comprehensive transcriptome datasets are valuable complements to genome sequences for high-resolution phylogenomic analysis.

Figure 1

Schematic phylogenetic tree of flowering plants. BR1 to BR4 denote potential time points when the γ event may have occurred. BR1, monocots + eudicots duplication; BR2, eudicot-wide duplication; BR3, core eudicot-wide duplication; BR4, rosid-wide duplication.

Figure 2

Exemplar maximum likelihood phylogeny of Ortho 1202. RAxML topology of an orthogroup (Ortho 1202) indicating that the γ paralogs of Vitis were duplicated before the split of rosids and asterids and after the early radiation of eudicots. The scored bootstrap (BS) value for this duplication is over 80%, because nodes #1 and #2 (and/or #3) have BS > 80%. Legend: green star = core eudicot duplication; colored circles = recent independent duplications; numbers = bootstrap support values.

Figure 3

Exemplar maximum likelihood phylogeny of Ortho 1083. RAxML topology of an orthogroup (Ortho 1083) indicates that the γ paralogs of Vitis were duplicated before the split of rosids and asterids, and after the early radiation of eudicots. The scored bootstrap (BS) value for this duplication is over 50%, because nodes #1 has BS < 80%. Legend: green star = core eudicot duplication; colored circles = recent independent duplications; numbers = bootstrap support values.

Figure 4

Age distribution of γ duplications. (a) The inferred duplication times for 161 γ duplication nodes that support core eudicot-wide duplication (BS ≥50%) were analyzed by EMMIX to determine whether these duplications occurred randomly over time or within some small timeframe. Each component is written as 'color/mean molecular timing/proportion' where 'color' is the component (curve) color and 'proportion' is the percentage of duplication nodes assigned to the identified component. There is one statistically significant component: green/117 (mya)/1. (b) Distribution of inferred γ duplication times from 66 orthogroups that support a eudicot-wide duplication with BS ≥50%. There is one statistically significant component: blue/133 (mya)/1. (c) Distribution of inferred γ duplication times from combination of (a) and (b) shows one significant component: purple/121 (mya)/1. (d-f) Corresponding distributions of inferred duplication times from orthogroups with BS ≥80%. One significant component in (d), green/116 (mya)/1; one in (e), blue/135 (mya)/1; and one in (f), purple/120 (mya)/1.

Figure 5

K_s distributions of paralogs in Vitis from syntenic block analysis. Methods for sequence alignment and estimation of K_s were as reported (Cui et al. 2006), but were here limited to paralogous gene pairs retained on syntenic blocks in the Vitis genome. Colored lines superimposed on K_s distribution represent significant duplication components identified by likelihood mixture model as in Figure 4 (Materials and methods). a, K_s distribution of 168 Vitis pairs supporting core eudicot-wide duplication in phylogenetic analysis. One statistically significant component: green/1.03/1. b, K_s distribution of 70 Vitis pairs showing all eudicot-wide duplications on phylogenies. One significant component: blue/1.31/1. c, K_s distribution of combination of Vitis pairs supporting core eudicot- (a) and eudicot-wide duplications (b) on phylogenies. Three significant components: black/0.3/0.01, green/1.02/0.70, blue/1.40/0.29. d, K_s distribution of 2191 paralogous pairs were identified from syntenic block analysis. Four significant components: black/0.12/0.02, green/1.09/0.74, blue/1.85/0.22, yellow/2.7/0.02.