Detecting Lineage-Specific Shifts in Diversification: A Proper Likelihood Approach

Area of Science:

• Macroevolutionary biology
• Phylogenetics
• Computational biology

Background:

• Phylogenetic branching patterns are assumed to reflect speciation and extinction rates.
• Current birth-death models often assume uniform diversification rates across lineages, which may be unrealistic.
• Models allowing rate shifts exist but rely on potentially flawed likelihood computations.

Purpose of the Study:

• To identify and correct inaccuracies in likelihood computations for phylogenetic models with varying diversification rates.
• To develop a new framework for accurate likelihood calculation in macroevolutionary models.
• To improve the inference of diversification histories from phylogenetic data.

Main Methods:

• Identified errors in existing likelihood computation methods for phylogenetic diversification models.
• Developed a novel mathematical framework for correct likelihood calculation.
• Validated the framework using simulations of single and multiple diversification rate shifts.

Main Results:

• The corrected likelihood framework yields more accurate parameter estimates compared to previous methods.
• The framework successfully handles time-dependent and diversity-dependent models with rate shifts.
• The approach clarifies the treatment of unobserved (extinct) lineage shifts in diversification analyses.

Conclusions:

• Existing methods for inferring diversification rates from phylogenies contain computational errors.
• The new framework provides a robust and accurate method for analyzing macroevolutionary dynamics.
• Accurate likelihood computation is crucial for understanding the history of life's diversification.