Displayed Trees Do Not Determine Distinguishability Under the Network Multispecies Coalescent

Area of Science:

• Phylogenetics
• Computational Biology
• Evolutionary Genetics

Background:

• Estimating species relationships often involves inferring phylogenetic networks to account for hybridization events.
• Probabilistic models utilizing the multispecies coalescent are employed for inferring network topologies and parameters like branch lengths and hybridization.
• A key challenge is determining the identifiability of species networks, i.e., whether distinct networks can produce the same gene tree distribution.

Purpose of the Study:

• To investigate the identifiability of species networks in the presence of hybridization and incomplete lineage sorting.
• To develop a new representation of species network likelihood for improved inference.
• To determine how allele sampling strategies affect network identifiability.

Main Methods:

• Developed a novel representation of species network likelihood based on the multispecies coalescent model.
• Expressed the probability distribution of gene tree topologies as a linear combination of gene tree distributions given species trees.
• Analyzed the impact of sampling one versus multiple alleles per species on network identifiability.

Main Results:

• Demonstrated that distinct species networks, which yield identical gene tree distributions with single-allele sampling, can be distinguished when multiple individuals are sampled per species.
• Showed that species network identifiability depends not only on the displayed trees but also on allele sampling within species.
• Provided an example where two networks displaying identical trees are distinguishable even with single-lineage sampling.

Conclusions:

• Species network identifiability is enhanced by sampling multiple alleles per species, allowing for the differentiation of previously indistinguishable networks.
• The findings highlight the importance of considering allele sampling strategies in phylogenetic network inference.
• This work contributes to more robust methods for reconstructing evolutionary histories involving hybridization.