The performance of coalescent-based species tree estimation methods under models of missing data

Area of Science:

• Phylogenetics
• Computational Biology
• Evolutionary Biology

Background:

• Species tree estimation is challenged by incomplete lineage sorting, gene duplication/loss, and horizontal gene transfer.
• Existing methods for species tree inference are statistically consistent under incomplete lineage sorting when gene trees are complete.

Purpose of the Study:

• To assess the statistical consistency of coalescent-based species tree methods with taxon deletion from genes.
• To evaluate the impact of missing data on four species tree estimation methods (ASTRAL-II, ASTRID, MP-EST, SVDquartets).

Main Methods:

• Statistical consistency analysis of coalescent-based methods under taxon deletion models.
• Evaluation of species tree methods using simulated datasets with varying incomplete lineage sorting, gene tree error, and missing data.

Main Results:

• Statistical consistency was established for certain coalescent-based methods under specific taxon deletion models.
• Performance of ASTRAL-II, ASTRID, MP-EST, and SVDquartets was evaluated across diverse simulated conditions.

Conclusions:

• Species tree estimation accuracy improved with an increased number of genes.
• Highly accurate species trees were frequently obtained despite substantial amounts of missing data, indicating method robustness.