Phylogenetic analysis using Lévy processes: finding jumps in the evolution of continuous traits

Area of Science:

• Evolutionary biology
• Phylogenetics
• Statistical modeling

Background:

• Gaussian processes, like Brownian motion, are standard for continuous trait evolution in phylogenetics.
• These models assume multivariate Gaussian distributions, which may not fit evolutionary scenarios like punctuated equilibrium or adaptive radiation.
• Non-normally distributed trait evolution requires alternative modeling approaches.

Purpose of the Study:

• To introduce a computational method for modeling continuous trait evolution using Lévy processes.
• To assess the distinct patterns generated by Brownian motion versus Lévy processes with jumps.
• To analyze primate body mass and endocranial volume using these models.

Main Methods:

• Developed a method to compute posterior probabilities for trait evolution modeled as a Lévy process.
• Utilized data simulations and model testing to compare Brownian motion and Lévy processes.
• Applied the Lévy process model to empirical data of primate body mass and endocranial volume.

Main Results:

• Single-rate Brownian motion and Lévy processes with jumps produce distinguishable patterns in comparative data.
• Analysis of primate traits (body mass, endocranial volume) rejected single-rate Brownian motion in favor of a Lévy process with jumps.
• Strong evidence against single-rate Brownian motion was found in the lineage leading to the great apes' most recent common ancestor.

Conclusions:

• Lévy processes offer a more flexible framework for modeling continuous trait evolution than standard Brownian motion.
• The proposed method and findings highlight the importance of considering non-normal evolutionary models.
• Empirical data from primates support the utility of Lévy processes with jumps in phylogenetic analyses.