[Paper Review] Phylogenomic Analyses of Large-scale Nuclear Genes Provide New Insights into the Evolutionary Relationships within the Rosids

Lei Zhao, Xia Li+6 more|arXiv (Cornell University)|Jun 30, 2016

Plant and Fungal Species Descriptions118 references9 citations

TL;DR

This study reconstructs the deep phylogeny of the rosids clade using large-scale nuclear gene data from 63 species across 14 orders, employing coalescence and concatenation methods. It reveals that coalescence methods outperform concatenation under high gene tree heterogeneity and missing data, particularly resolving the contentious placement of Zygophyllales.

ABSTRACT

The Rosids is one of the largest groups of flowering plants, with 140 families and ~70,000 species. Previous phylogenetic studies of the rosids have primarily utilized organelle genes that likely differ in evolutionary histories from nuclear genes. To better understand the evolutionary history of rosids, it is necessary to investigate their phylogenetic relationships using nuclear genes. Here, we employed large-scale phylogenomic datasets composed of nuclear genes, including 891 clusters of putative orthologous genes. Combined with comprehensive taxon sampling covering 63 species representing 14 out of the 17 orders, we reconstructed the rosids phylogeny with coalescence and concatenation methods, yielding similar tree topologies from all datasets. However, these topologies did not agree on the placement of Zygophyllales. Through comprehensive analyses, we found that missing data and gene tree heterogeneity were potential factors that may mislead concatenation methods, in particular, large amounts of missing data under high gene tree heterogeneity. Our results provided new insights into the deep phylogenetic relationships of the rosids, and demonstrated that coalescence methods may effectively resolve the phylogenetic relationships of the rosids with missing data under high gene tree heterogeneity.

Motivation & Objective

• To clarify deep evolutionary relationships within the large flowering plant clade Rosids, which includes ~70,000 species across 140 families.
• To overcome limitations of prior studies relying on organelle genes by using large-scale nuclear gene datasets.
• To evaluate the performance of coalescence versus concatenation methods in resolving rosid phylogeny under conditions of missing data and gene tree heterogeneity.
• To resolve the long-standing uncertainty in the phylogenetic placement of the order Zygophyllales.

Proposed method

• Constructed a phylogenomic dataset of 891 clusters of putative orthologous nuclear genes from 63 species representing 14 of 17 rosid orders.
• Applied both coalescence-based and concatenation-based species tree inference methods to the same dataset to compare results.
• Conducted comprehensive sensitivity analyses to assess the impact of missing data and gene tree heterogeneity on tree topology accuracy.
• Used statistical tests and model comparisons to evaluate the robustness of inferred relationships under varying data conditions.

Experimental results

Research questions

• RQ1 What is the most accurate species tree topology for the Rosids when inferred from large-scale nuclear gene data?
• RQ2 How do coalescence and concatenation methods compare in resolving rosid relationships under high levels of missing data and gene tree heterogeneity?
• RQ3 Why does the placement of Zygophyllales remain unresolved or inconsistent across analyses?
• RQ4 Can coalescence methods effectively mitigate systematic errors caused by missing data and gene tree conflict in species tree estimation?

Key findings

• Coalescence and concatenation methods produced highly congruent species tree topologies across all datasets, except for the placement of Zygophyllales.
• The placement of Zygophyllales was unstable and likely influenced by high gene tree heterogeneity and missing data, particularly in concatenation analyses.
• Missing data and gene tree heterogeneity were identified as key confounding factors that misled concatenation methods, especially when data were sparse.
• Coalescence methods demonstrated greater robustness and reliability in resolving species relationships under conditions of high missing data and gene tree conflict.