Simulating trees with millions of species

Area of Science:

• Phylogenetics and Evolutionary Biology
• Computational Biology

Background:

• The birth-death (BD) model is fundamental for understanding speciation and extinction dynamics.
• Phylogenetic tree simulations are crucial for evaluating reconstruction tools, bootstrapping, and outlier detection.
• Existing simulation tools struggle with the scalability required for large, modern phylogenies.

Purpose of the Study:

• To develop efficient software for simulating phylogenetic trees under time-dependent BD models.
• To provide a tool that scales effectively to large phylogenies with thousands or millions of tips.

Main Methods:

• Simulation of reconstructed phylogenies in backward time.
• Conditioning on the number of sampled species and optionally the time since the most recent common ancestor (MRCA).
• Implementation of time-dependent birth-death models.

Main Results:

• Developed highly efficient software for BD model simulations.
• Achieved 1000-10,000 times speed improvement over existing tools for large trees.
• Software is integrated into the R package 'castor'.

Conclusions:

• The new software overcomes scalability limitations of previous phylogenetic simulation tools.
• Provides a faster and more efficient method for analyzing speciation/extinction dynamics.
• Facilitates robust performance assessment and outlier detection in phylogenetics.