Effects of missing data on species tree estimation under the coalescent

Area of Science:

• Genomics
• Phylogenetics
• Computational Biology

Background:

• Genomic sequencing advances highlight the need to reconcile gene and species histories.
• Multilocus datasets often suffer from incomplete data coverage, impacting phylogenetic accuracy.

Purpose of the Study:

• To investigate the impact of missing data on species tree reconstruction accuracy.
• To evaluate the influence of sampling strategies and missing data patterns on phylogenetic inference.

Main Methods:

• Simulated 10- and 100-locus datasets under the coalescent model for shallow and deep speciation.
• Estimated species trees using maximum likelihood (STEM) and Bayesian (*)BEAST frameworks.
• Assessed accuracy using symmetric difference and SPR distance metrics.

Main Results:

• Species tree estimates demonstrated remarkable resilience to missing data across various scenarios.
• For limited loci, multiple individuals per species significantly enhanced accuracy with missing data.
• Larger multilocus datasets (25-100 loci) showed negligible impact from missing data, even at 50%.

Conclusions:

• Missing data has a limited effect on species tree reconstruction accuracy.
• Sampling strategies, particularly multiple individuals per species, are crucial for smaller datasets with missing data.
• Robustness of species tree methods to missing data is confirmed, especially for large multilocus datasets.